title
stringlengths 1
827
⌀ | uuid
stringlengths 36
36
| pmc_id
stringlengths 5
8
| search_term
stringclasses 44
values | text
stringlengths 8
8.58M
|
|---|---|---|---|---|
Reduced risks of influenza-associated hospitalization and complications following vaccination among over 2 million older individuals: a nationwide study using target trial emulation framework | 0adb0ac8-533f-4f7b-bfe5-02dba12d8073 | 11907999 | Vaccination[mh] | People aged ≥ 65 years are at an increased risk of influenza infection and development of serious influenza-related complications (e.g., cardiovascular [CV], severe pulmonary, and kidney diseases, and death ) due to the immunosenescence, multiple chronic health problems, and frailty in this population . In Taiwan, an influenza vaccine is therefore government-funded and strongly recommended to adults aged 65 years or above for reducing influenza transmission as well as averting influenza-associated morbidities and mortalities . Despite these efforts, the uptake rate of the influenza vaccine among the older population in routine clinical practice settings remains suboptimal (e.g., around 50–60% ). Influenza vaccine effectiveness (VE) is commonly reported to inform personal decision-making on vaccination . However, there are several caveats regarding current evidence for older populations. First, despite substantial evidence supporting VE in the general older population worldwide from annual epidemiological reports of sentinel surveillance systems and the potential vaccine benefits on avoiding influenza-related complications [ – ], only the subsets (i.e., those with gout , disability , or breast cancer ) or limited numbers (i.e., thousands ) of general older populations in Taiwan were analyzed. Therefore, current VE evidence might not be generalizable to general older populations in Taiwan to support decision-making. Second, the concern of residual confounding by indication and healthy vaccinee bias might remain in previous observational studies, with case–control or cohort designs used, leading to imprecise VE and thereby undermining the study validity and confidence to support clinical decision-making . Recently, the target trial emulation framework has been extensively employed in the field of vaccine studies (e.g., COVID-19 vaccine ), and this approach comprises an explicit study design framework along with rigorous methodologies which could minimize the possibility of confounding effects and biases commonly seen in the observational studies of real-world data . Lastly, the impact of the uptake of the influenza vaccine on healthcare utilization (e.g., medical costs and length of hospital stay) in Taiwan is also unclear. Against this background, the present study sought to determine the effectiveness of a standard-dose influenza vaccine on influenza infection, a series of influenza-related complications (i.e., CV, respiratory, and kidney diseases and death), and healthcare utilization and costs in adults aged ≥ 65 years using a target trial emulation approach. Empirical evidence from this study in particular vaccine effectiveness beyond influenza infection (i.e., associated complications and economic consequences) is important for facilitating personal decision-making on vaccination, improving the vaccine uptake rate, and maximizing the value of an influenza vaccine in real-world practice.
Emulation of hypothetical trial using nationwide claims data This was a retrospective cohort study using a target trial emulation framework. Our study design for the target trial was adapted according to a published randomized, double-blind, clinical trial . The specifications of the target trial are detailed in Additional file : Table 1. Taiwan’s National Health Insurance Research Database (NHIRD) for 2017–2019 was utilized. The National Health Insurance program in Taiwan covers healthcare services (e.g., outpatient, emergency room [ER] visits, hospitalization, and medication prescriptions) for over 99% of Taiwan’s population. Health-related information is therefore longitudinally collected and recorded in the NHIRD. The health records in the NHIRD are individual-level and de-identified. Details of the NHIRD are available elsewhere . This study was approved by the Institutional Review Board of National Cheng Kung University (111–458-2). Study population All individuals in Taiwan aged ≥ 65 years have been eligible for a government-funded influenza vaccine since 2001 . The government-funded vaccine is available each year from October until it runs out or the end of September the following year. Annually, 40% to 60% of older adults received the vaccination; of those vaccinated, > 90% of the vaccinations were administered in October through December. The present study identified all subjects aged ≥ 65 years in the NHIRD for 2018 as the study cohort. Subjects who were first vaccinated in October, November, or December 2018 were placed into the vaccinated group . Those vaccinated in other months (i.e., January to September 2019) or not vaccinated were placed in the unvaccinated group. To emulate the target trial, the eligibility for vaccination was assessed for each older individual (Additional file : Table 1). A sequential trial approach (Additional file : Fig. S1) with a two-step propensity score (PS) matching was performed to enhance the comparability between the vaccinated and unvaccinated groups regarding baseline characteristics. In the first step, vaccinated subjects were matched with unvaccinated subjects based on age, gender, and city/county in a 1: n ratio. A vaccinated subject was matched with multiple unvaccinated subjects (as much as possible) to ensure similar accessibility to healthcare within matched individuals. In the second step, which was performed on the matched pairs in each month as a stratum (i.e., October, November, and December 2018), 1:1 PS-matched pairs of vaccinated and unvaccinated subjects were obtained using 8-to-1-digit greedy matching. Of note, the index date for a vaccinated subject was the date of receiving the influenza vaccine, whereas that for an unvaccinated subject was set to October 15, November 15, or December 15 based on their matched month stratum to minimize the immortal time bias. Details of the inclusion and exclusion criteria and matching procedure and the operational definitions of the baseline characteristics are given in Additional file : Fig. S1 and Table . Each study subject was followed from the index date until the occurrence of an influenza event, reception of the influenza vaccine (i.e., the second vaccination for the vaccinated group and the first vaccination for the unvaccinated group), death, or the end of influenza season (i.e., September 30, 2019), whichever came first (i.e., observational analog of the per-protocol scenario). Measurements of vaccination status and influenza outcomes Vaccination status was ascertained from one of the following records in the NHIRD: (1) reception of the influenza vaccine, (2) reimbursement for vaccine administration, or (3) outpatient visit for influenza vaccination, indicating that the influenza vaccination was reimbursed by the Taiwan Center for Disease Control. The main study outcomes were (1) influenza-associated hospitalization, which was defined based on influenza diagnosis codes (the International Classification of Diseases, Tenth Revision, Clinical Modification: J09-J11); (2) influenza-associated outpatient visits, which were defined based on influenza diagnosis codes (J09-J11) with antiviral drug use; and (3) influenza-associated ER visits, which were defined based on influenza diagnosis codes (J09-J11) with antiviral drug use. Because systematic inflammatory responses triggered by influenza events could increase the risks of severe pulmonary , CV , and kidney diseases , these clinical conditions or complications are likely to occur following influenza infection among the older population . The prevention of influenza infection episodes through vaccination is, therefore, crucial to avoid the occurrence of these complications . In this regard, we included influenza-associated complications that occurred during influenza-associated hospitalization as study outcomes of interest. These complications included pneumonia, acute respiratory distress syndrome (ARDS) with ventilator use, sepsis, acute myocardial infarction (AMI), stroke, acute kidney injury, and death, which was ascertained from the Cause of Death files in the NHIRD. Of note, given that the protective effect of a vaccine generally starts 14 days after vaccination , study events that occurred ≥ 14 days after the index date were measured. Details of the operational definitions of the study outcomes are available in Additional file : Table 3. Statistical analysis The standard mean difference (SMD) was utilized to assess the between-group comparability in patient baseline characteristics. An absolute value of SMD ≥ 0.1 was considered to indicate a statistically significant between-group imbalance. The event rate of a study outcome was estimated as the number of events divided by 1000 person-years. The healthcare costs associated with influenza-associated healthcare utilization, including hospital admissions (and corresponding length of stay, measured in days) and outpatient and ER visits, were estimated from the perspective of the healthcare sector. Costs were standardized into 2022 values using the medical component of the consumer price index in Taiwan and are presented in United States dollars (USD). Student’s t -test was used to evaluate the between-group difference in healthcare costs. The Cox proportional hazard model was employed to assess the risk of the study outcome (e.g., influenza infection) with vaccination status. Any unbalanced variables between vaccinated and unvaccinated groups were further treated as covariates and adjusted in the Cox model analysis. The results are presented as hazard ratios (HRs) and associated 95% confidence intervals (CIs). The VE was then calculated as (1 − HR)*100%. A series of sensitivity analyses were performed to test the robustness of the study findings in the primary analyses. First, to enhance the validity of the measurement of the study outcomes, influenza-associated hospitalization was determined according to the principal diagnosis codes for influenza, and influenza-associated outpatient and ER visits were defined based on the principal diagnosis codes for influenza with the prescription of an antiviral drug. Second, considering influenza occurs throughout the year in Taiwan, the analyses focused on the peak season (month) of influenza, where the end of the follow-up period was restricted to March, April, or May 2019, respectively, to assess whether the VE waned following influenza vaccination . Third, considering the potential of unmeasured confounders (i.e., health awareness), negative control analyses were conducted , where influenza events that occurred 7 or 13 days following vaccination, influenza events in the sampled cohort with 50,000 subjects obtained from a vaccine-mismatched season (i.e., influenza season of 2014 or 2015 ) using the same patient selection procedures (i.e., the selection of patient cohort followed the target trial emulation framework), and traffic-accident-related hospitalization were treated as study outcomes. No significant risks of these control events with vaccination status were expected (i.e., 95% CI of HR overlapping 1), ensuring the validity of our study materials and procedures. Lastly, given the possibility of the presence of immunosenescence in geriatric populations , the modification effect of patients’ baseline characteristics (e.g., age) on vaccination outcomes (i.e., VE) cannot be ruled out. Therefore, a series of interaction tests were carried out for various patient characteristics, including gender (i.e., female or male), age (i.e., ≥ or < 75 years), frailty (i.e., fit or frail), and influenza risk (i.e., presence or absence of high-risk disease conditions such as infectious diseases, blood disorders, and endocrinologic disorders, as specified by Taiwan’s Center for Disease Control), and joint subgroups (i.e., aged ≥ or < 75 years and fit or frail, and aged ≥ or < 75 years and with or without high risk of influenza infection). A two-tailed p -value of < 0.05 was considered to indicate a statistically significant difference. All analyses mentioned above were conducted using SAS software version 9.4.
This was a retrospective cohort study using a target trial emulation framework. Our study design for the target trial was adapted according to a published randomized, double-blind, clinical trial . The specifications of the target trial are detailed in Additional file : Table 1. Taiwan’s National Health Insurance Research Database (NHIRD) for 2017–2019 was utilized. The National Health Insurance program in Taiwan covers healthcare services (e.g., outpatient, emergency room [ER] visits, hospitalization, and medication prescriptions) for over 99% of Taiwan’s population. Health-related information is therefore longitudinally collected and recorded in the NHIRD. The health records in the NHIRD are individual-level and de-identified. Details of the NHIRD are available elsewhere . This study was approved by the Institutional Review Board of National Cheng Kung University (111–458-2).
All individuals in Taiwan aged ≥ 65 years have been eligible for a government-funded influenza vaccine since 2001 . The government-funded vaccine is available each year from October until it runs out or the end of September the following year. Annually, 40% to 60% of older adults received the vaccination; of those vaccinated, > 90% of the vaccinations were administered in October through December. The present study identified all subjects aged ≥ 65 years in the NHIRD for 2018 as the study cohort. Subjects who were first vaccinated in October, November, or December 2018 were placed into the vaccinated group . Those vaccinated in other months (i.e., January to September 2019) or not vaccinated were placed in the unvaccinated group. To emulate the target trial, the eligibility for vaccination was assessed for each older individual (Additional file : Table 1). A sequential trial approach (Additional file : Fig. S1) with a two-step propensity score (PS) matching was performed to enhance the comparability between the vaccinated and unvaccinated groups regarding baseline characteristics. In the first step, vaccinated subjects were matched with unvaccinated subjects based on age, gender, and city/county in a 1: n ratio. A vaccinated subject was matched with multiple unvaccinated subjects (as much as possible) to ensure similar accessibility to healthcare within matched individuals. In the second step, which was performed on the matched pairs in each month as a stratum (i.e., October, November, and December 2018), 1:1 PS-matched pairs of vaccinated and unvaccinated subjects were obtained using 8-to-1-digit greedy matching. Of note, the index date for a vaccinated subject was the date of receiving the influenza vaccine, whereas that for an unvaccinated subject was set to October 15, November 15, or December 15 based on their matched month stratum to minimize the immortal time bias. Details of the inclusion and exclusion criteria and matching procedure and the operational definitions of the baseline characteristics are given in Additional file : Fig. S1 and Table . Each study subject was followed from the index date until the occurrence of an influenza event, reception of the influenza vaccine (i.e., the second vaccination for the vaccinated group and the first vaccination for the unvaccinated group), death, or the end of influenza season (i.e., September 30, 2019), whichever came first (i.e., observational analog of the per-protocol scenario).
Vaccination status was ascertained from one of the following records in the NHIRD: (1) reception of the influenza vaccine, (2) reimbursement for vaccine administration, or (3) outpatient visit for influenza vaccination, indicating that the influenza vaccination was reimbursed by the Taiwan Center for Disease Control. The main study outcomes were (1) influenza-associated hospitalization, which was defined based on influenza diagnosis codes (the International Classification of Diseases, Tenth Revision, Clinical Modification: J09-J11); (2) influenza-associated outpatient visits, which were defined based on influenza diagnosis codes (J09-J11) with antiviral drug use; and (3) influenza-associated ER visits, which were defined based on influenza diagnosis codes (J09-J11) with antiviral drug use. Because systematic inflammatory responses triggered by influenza events could increase the risks of severe pulmonary , CV , and kidney diseases , these clinical conditions or complications are likely to occur following influenza infection among the older population . The prevention of influenza infection episodes through vaccination is, therefore, crucial to avoid the occurrence of these complications . In this regard, we included influenza-associated complications that occurred during influenza-associated hospitalization as study outcomes of interest. These complications included pneumonia, acute respiratory distress syndrome (ARDS) with ventilator use, sepsis, acute myocardial infarction (AMI), stroke, acute kidney injury, and death, which was ascertained from the Cause of Death files in the NHIRD. Of note, given that the protective effect of a vaccine generally starts 14 days after vaccination , study events that occurred ≥ 14 days after the index date were measured. Details of the operational definitions of the study outcomes are available in Additional file : Table 3.
The standard mean difference (SMD) was utilized to assess the between-group comparability in patient baseline characteristics. An absolute value of SMD ≥ 0.1 was considered to indicate a statistically significant between-group imbalance. The event rate of a study outcome was estimated as the number of events divided by 1000 person-years. The healthcare costs associated with influenza-associated healthcare utilization, including hospital admissions (and corresponding length of stay, measured in days) and outpatient and ER visits, were estimated from the perspective of the healthcare sector. Costs were standardized into 2022 values using the medical component of the consumer price index in Taiwan and are presented in United States dollars (USD). Student’s t -test was used to evaluate the between-group difference in healthcare costs. The Cox proportional hazard model was employed to assess the risk of the study outcome (e.g., influenza infection) with vaccination status. Any unbalanced variables between vaccinated and unvaccinated groups were further treated as covariates and adjusted in the Cox model analysis. The results are presented as hazard ratios (HRs) and associated 95% confidence intervals (CIs). The VE was then calculated as (1 − HR)*100%. A series of sensitivity analyses were performed to test the robustness of the study findings in the primary analyses. First, to enhance the validity of the measurement of the study outcomes, influenza-associated hospitalization was determined according to the principal diagnosis codes for influenza, and influenza-associated outpatient and ER visits were defined based on the principal diagnosis codes for influenza with the prescription of an antiviral drug. Second, considering influenza occurs throughout the year in Taiwan, the analyses focused on the peak season (month) of influenza, where the end of the follow-up period was restricted to March, April, or May 2019, respectively, to assess whether the VE waned following influenza vaccination . Third, considering the potential of unmeasured confounders (i.e., health awareness), negative control analyses were conducted , where influenza events that occurred 7 or 13 days following vaccination, influenza events in the sampled cohort with 50,000 subjects obtained from a vaccine-mismatched season (i.e., influenza season of 2014 or 2015 ) using the same patient selection procedures (i.e., the selection of patient cohort followed the target trial emulation framework), and traffic-accident-related hospitalization were treated as study outcomes. No significant risks of these control events with vaccination status were expected (i.e., 95% CI of HR overlapping 1), ensuring the validity of our study materials and procedures. Lastly, given the possibility of the presence of immunosenescence in geriatric populations , the modification effect of patients’ baseline characteristics (e.g., age) on vaccination outcomes (i.e., VE) cannot be ruled out. Therefore, a series of interaction tests were carried out for various patient characteristics, including gender (i.e., female or male), age (i.e., ≥ or < 75 years), frailty (i.e., fit or frail), and influenza risk (i.e., presence or absence of high-risk disease conditions such as infectious diseases, blood disorders, and endocrinologic disorders, as specified by Taiwan’s Center for Disease Control), and joint subgroups (i.e., aged ≥ or < 75 years and fit or frail, and aged ≥ or < 75 years and with or without high risk of influenza infection). A two-tailed p -value of < 0.05 was considered to indicate a statistically significant difference. All analyses mentioned above were conducted using SAS software version 9.4.
We identified 3,394,238 subjects aged ≥ 65 years in the NHIRD for 2018 and 2019. After the study eligibility and matching procedures were applied, 1,214,392 pairs of vaccinated and unvaccinated subjects were obtained for the analysis (Additional file : Fig. 2). The baseline characteristics of the study cohort before PS matching are provided in Additional file : Tables 4–6. Table shows satisfactory between-group comparability in the baseline characteristics (SMD less than 0.1), except for the history of government-funded health examination in the year prior to the index date. In general, the study population had a mean age of 74.3 years and was 46.1% male. In Fig. , the primary analyses show that the event rate of influenza-associated hospitalization was 3.12 and 3.67 per 1000 person-years for vaccinated and unvaccinated subjects, respectively, resulting in a VE of 14% on influenza-associated hospitalization (i.e., adjusted HR [aHR]: 0.86, 95% CI: 0.82–0.90). The event rates of influenza-associated outpatient/ER visits were 0.52/0.20 and 0.56/0.22 for vaccinated and unvaccinated subjects, respectively, but the VE values estimated from these events were insignificant (i.e., VE: 10%/10%, aHRs: 0.90/0.90, 95% CIs: 0.82–1.03/0.75–1.08). The results of sensitivity analyses that assessed influenza events based on principal diagnostic codes and restricted influenza seasons (i.e., October 2018 to the end of March, April, or May 2019) were consistent with the primary analysis findings, showing significant VE for influenza-associated hospitalization (i.e., VE/aHRs [95% CIs]: 13%/0.87 [0.83–0.93], 25%/0.75 [0.70–0.81], 23%/0.77 [0.72–0.82], and 21%/0.79 [0.74–0.84], respectively). No significant effect of influenza vaccination on negative control outcomes was observed (i.e., VE: − 1% [− 42%, 28%] for influenza events that occurred within 7 days following vaccination, 0% [− 29%, 23%] for influenza events within 13 days following vaccination, 11% [− 3%, 22%] for influenza events in the sampled cohort from a mismatched season, and aHR: 1.01 [0.97, 1.05] for traffic accident-related hospitalization). Details of the event rates are given in Additional file : Table 7. Figure shows significant interactions of VE with age and multimorbidity frailty. That is, the VE (95% CIs) for influenza-associated hospitalization was 23% (18–29%) and 11% (5–16%) for subjects aged < and ≥ 75 years, respectively ( p -value for interaction < 0.0001), and 27% (19–33%) and 11% (7–16%) for fit and frail subjects, respectively ( p -value for interaction = 0.001). The joint subgroup analyses also indicate significant interaction of vaccine status with multimorbidity frailty in subjects aged < 75 years; i.e., the VE for influenza-associated hospitalization was 35% (25–43%) and 19% (12–26%) for fit and frail subjects, respectively ( p -value for interaction = 0.011). Details of the event rates and associated aHRs and 95% CIs are given in Additional file : Table 8. Table shows the detailed event rates and associated aHRs (95% CIs) for clinical complications that occurred during influenza hospitalization. Compared with non-vaccination, reception of the influenza vaccine was associated with significantly reduced risks of influenza-associated death (aHR: 0.70 [95% CI: 0.56–0.88]), infectious/pulmonary diseases (i.e., 0.88 [0.81–0.96], 0.85 [0.74–0.98], and 0.34 [0.19–0.61]) for pneumonia, sepsis, and ARDS with ventilator use, respectively), CV diseases (i.e., 0.61 [0.39–0.95] and 0.53 [0.38–0.75] for AMI and stroke, respectively), and kidney diseases (i.e., 0.77 [0.61, 0.96] for acute kidney injury). Figure shows significantly lower influenza-associated hospitalization costs (per admission) (i.e., $1866 versus $2377, p -value < 0.0001) and marginally significantly higher influenza-associated outpatient ($18 versus $16, p -value = 0.0215) and ER ($132 versus $112, p -value = 0.0789) costs (per visit) in vaccinated subjects compared with those for unvaccinated subjects. The results of influenza-associated healthcare costs and length of hospital stay are detailed in Additional file : Table 9.
This study of over 2 million older individuals showed the beneficial effect of influenza vaccination on influenza-associated hospitalization and a wide range of related complications, including infectious/pulmonary, CV, and kidney diseases and death. These findings extend the current evidence, which mostly focuses on high-risk older populations, to the general older population in real-world settings. We adopted a target trial emulation framework and a series of sensitivity analyses, which not only enhanced the transparency of study design and procedures but also guaranteed high-quality study results. This strengthens confidence in the reported VE and associated clinical/economic benefits, encouraging individuals to receive an influenza vaccination. Comparison of vaccine effectiveness between previous studies and this study In the present study, for the older population in the influenza season of 2018–2019, the estimated VE values (95% CIs) for influenza-associated hospitalization, outpatient visits, and ER visits were 14% (10%, 18%), 10% (− 3%, 18%), and 10% (− 8%, 25%), respectively, which fall in the range of VE values reported in previous studies (i.e., 12% [− 31%, 40%] to 26% [20%, 31%] ). However, caution should be taken when comparing findings across studies due to differences in the influenza viruses circulating worldwide, which are affected by antigenic drift in local or regional geographic areas, operational definitions of influenza events (i.e., laboratory-confirmed influenza versus clinical diagnoses in the present study), and study procedures (i.e., previous cohort or test-negative case–control studies versus the present study using a target trial emulation design) across studies. In this study, a substantial effort was made (in terms of methodology) to minimize the confounding effects and biases that are commonly seen in studies . First, a series of sensitivity analyses that restricted influenza events to those confirmed by the principal diagnosis were carried out. The results of these analyses were consistent with the primary findings, strengthening confidence in VE in real-world settings and providing clinical insights to facilitate real-world decision-making. Specifically, in these sensitivity analyses, statistically significant protection by influenza vaccination was only shown for influenza-associated hospitalization (i.e., VE [95% CI]: 13% [7%, 17%]), but not for influenza-associated outpatient and ER visits (9% [− 3%, 20%] and 12% [− 12%, 31%], respectively). These findings imply that influenza vaccination may be effective in alleviating the severity of influenza infection (e.g., avert severe cases that require hospitalization), but it does not decrease infection episodes or mild cases (e.g., influenza-associated outpatient and ER visits) [ – ]. Second, we performed several sensitivity analyses using negative control outcomes. The non-significant results supported the success of implementing a target trial emulation design with two-step PS matching in eliminating the concern of unmeasured confounders. Lastly, considering the possibility of VE attenuation over time following vaccination in older populations , sensitivity analyses were conducted to restrict the study follow-up period to different lengths of the peak influenza months in winter (October 2018 to March 2019, October 2018 to April 2019, October 2018 to May 2019, October 2018 to September 2019). It was found that the VE for influenza-associated hospitalization decreased as the time interval increased (i.e., 25% [19%, 30%], 23% [18%, 28%], 21% [16%, 26%], and 14% [10%, 18%] for the considered influenza season lengths, respectively). These results suggest the importance of continuous influenza vaccine uptake over influenza seasons for enhancing immunogenicity against influenza infection in older populations. Also, the provision of high-dose or adjuvant influenza vaccines and the development of a new vaccine platform (e.g., mRNA) are suggested to achieve optimal protection against influenza infection among older individuals. Variation of VE for influenza hospitalization by age and multimorbidity frailty in older patients In this study, aging (which is typically associated with immunosenescence) and frailty were found to be strong effect modifiers. There was a large disparity in the VE values across the subgroups stratified by these two variables. That is, subjects who were younger (i.e., 65 ≤ age < 75 years) or fit in the multimorbidity frailty category had higher VE values compared with those of their counterparties (i.e., aged ≥ 75 years and frail), as supported by significant interaction results ( p for interaction < 0.05, Fig. ). These findings were confirmed by interaction tests in the joint subgroup analysis for subjects aged < 75 years and fit or frail. Therefore, these results indicate that the underlying health status (e.g., old age, frailty) may affect VE even in a season with a good match between the vaccine and circulating strains . Young and healthy subjects may have greater VE than those old and frail subjects. Nevertheless, we found that reception of the influenza vaccine was statistically significantly associated with a reduced risk of influenza-associated hospitalization, irrespective of age, frailty status, and with and without high risk of influenza infection; e.g., although low VE values (95% CIs) were obtained for subjects aged 75 years (i.e., 6.6% [1.0%, 11.9%]), frail individuals (11.8% [7.1%, 16.3%]), and high-risk older individuals (13.1% [8.1%, 17.8%]), all estimates were statistically significant. Such specific subgroups in older populations should be prioritized for receiving high-dose, or adjuvanted influenza vaccines whenever those are available in Taiwan. Moreover, the caregivers of these subgroups are recommended to be vaccinated to optimize protection from influenza infection for older individuals through the cocoon strategy . Also, the number needed to treat (NNT) results of influenza vaccination for influenza infection and associated complications can provide explicit insights for clinical decision-making. For example, 47 patients would need to be administered an influenza vaccine relative to non-vaccination for a mean follow-up of 0.8–0.9 years to avert one case having influenza-related hospitalizations (Additional file : Table 7). Also, per recommendations from the World Health Organization, a 75% influenza vaccination coverage rate shall be achieved to avoid the annual epidemics, irrespective of age, country, healthcare systems, and race/ethnicity . With consideration of the suboptimal uptake rate of the influenza vaccine in current practice (e.g., 50–60% ), promoting the additional benefit of influenza vaccination on the reduction of the CV, pulmonary, and kidney risks to society, adopting high-dose influenza vaccines to the annual vaccination program, and improving vaccine accessibility in rural areas may improve influenza vaccination coverage rates for the general older population, thereby diminishing the disease burden attributable to influenza . Vaccine effectiveness beyond influenza infection in older patients In addition to a reduced risk of influenza events, the protective effect of influenza vaccination on a wide spectrum of influenza-related complications (i.e., death and infectious/pulmonary, CV, and kidney diseases) was shown in this study. Previous studies only analyzed the additional benefits of the influenza vaccine among specific subgroups (e.g., patients with gout, CV diseases, or respiratory diseases) of older populations [ – ]. Empirical evidence derived from large-scale general older populations is limited. This study bridges this knowledge gap. It found that influenza vaccination reduces the risks of influenza-related events (i.e., influenza infection, pulmonary disease, and death) by 12–66%, CV diseases by 39–47%, and acute kidney injury by 23%. Mechanisms to support the reduced risks of complications associated with influenza vaccination may be derived from its prevention on influenza infection, which could increase systematic pro-inflammatory cytokines and directly act on vasculature and myocardium, resulting in plaque destabilization and MI or stroke development . Beyond these clinical benefits, the savings from influenza-associated hospitalization following influenza vaccination were also remarkable (i.e., approximately $3,000,000 in total, Additional file : Table 9). It is expected that this economic benefit will increase as the uptake rate of the influenza vaccine increases. The savings could be re-allocated to support the universal coverage of the influenza vaccine for the older population and to maintain the national health insurance program. Study limitations First, given the implementation of exclusion criteria according to the study target trial setting, our results may not be generalized to the patients excluded from this study. In particular, a certain number of individuals with dementia (~ 6% of the general older population) were excluded. These individuals are vulnerable to and usually have multiple comorbidities , and are thus at high risk for influenza infection and associated complications. Influenza events are likely to be under-recognized due to the decline in cognitive function in these patients, affecting the validity of the VE estimates presented in this study. Hence, to understand the VE in patients with dementia, a prospective pragmatic trial design that relaxes the strict trial patient inclusion criteria to accommodate disadvantaged patients in routine care settings could be adopted in the future. Second, despite a large amount (i.e., over two million) of older subjects included in the current study, a certain proportion of the older population was lost in the process of PS matching. Therefore, in future research, other matching methods (e.g., inverse probability of treatment weighting) can be utilized to strengthen the robustness and generalizability of the current study findings. Third, the present study did not include self-paid influenza vaccination because such data are unavailable. However, given the universal coverage of influenza vaccination for individuals aged ≥ 65 years in Taiwan, this concern might be negligible. Also, potential unmeasured confounders (e.g., health awareness) might exist, which were likely to increase the effect size and thereby lead to underestimated VE. To minimize such a concern, we measured several surrogate indicators (e.g., the receipt of health examination and cancer screenings within 1 year prior to the index date) and adjusted them in analysis (e.g., matching procedures). Fourth, the clinical complications that occurred during influenza hospitalizations (Table ) were likely affected by the timely receipt of antiviral prescriptions, treatment with antibiotics, and receipt of pneumococcal vaccination, which were not measured and adjusted in our analyses. Fifth, there is a lack of laboratory confirmation (e.g., polymerase chain reaction testing) to identify influenza cases. Lastly, the present study was conducted from the perspective of the healthcare sector and, therefore, did not consider non-medical benefits (e.g., the loss of productivity of family members due to having to care for influenza patients) of influenza vaccination. These additional outcomes following vaccination deserve to be included in a future analysis of the overall benefit of influenza vaccination.
In the present study, for the older population in the influenza season of 2018–2019, the estimated VE values (95% CIs) for influenza-associated hospitalization, outpatient visits, and ER visits were 14% (10%, 18%), 10% (− 3%, 18%), and 10% (− 8%, 25%), respectively, which fall in the range of VE values reported in previous studies (i.e., 12% [− 31%, 40%] to 26% [20%, 31%] ). However, caution should be taken when comparing findings across studies due to differences in the influenza viruses circulating worldwide, which are affected by antigenic drift in local or regional geographic areas, operational definitions of influenza events (i.e., laboratory-confirmed influenza versus clinical diagnoses in the present study), and study procedures (i.e., previous cohort or test-negative case–control studies versus the present study using a target trial emulation design) across studies. In this study, a substantial effort was made (in terms of methodology) to minimize the confounding effects and biases that are commonly seen in studies . First, a series of sensitivity analyses that restricted influenza events to those confirmed by the principal diagnosis were carried out. The results of these analyses were consistent with the primary findings, strengthening confidence in VE in real-world settings and providing clinical insights to facilitate real-world decision-making. Specifically, in these sensitivity analyses, statistically significant protection by influenza vaccination was only shown for influenza-associated hospitalization (i.e., VE [95% CI]: 13% [7%, 17%]), but not for influenza-associated outpatient and ER visits (9% [− 3%, 20%] and 12% [− 12%, 31%], respectively). These findings imply that influenza vaccination may be effective in alleviating the severity of influenza infection (e.g., avert severe cases that require hospitalization), but it does not decrease infection episodes or mild cases (e.g., influenza-associated outpatient and ER visits) [ – ]. Second, we performed several sensitivity analyses using negative control outcomes. The non-significant results supported the success of implementing a target trial emulation design with two-step PS matching in eliminating the concern of unmeasured confounders. Lastly, considering the possibility of VE attenuation over time following vaccination in older populations , sensitivity analyses were conducted to restrict the study follow-up period to different lengths of the peak influenza months in winter (October 2018 to March 2019, October 2018 to April 2019, October 2018 to May 2019, October 2018 to September 2019). It was found that the VE for influenza-associated hospitalization decreased as the time interval increased (i.e., 25% [19%, 30%], 23% [18%, 28%], 21% [16%, 26%], and 14% [10%, 18%] for the considered influenza season lengths, respectively). These results suggest the importance of continuous influenza vaccine uptake over influenza seasons for enhancing immunogenicity against influenza infection in older populations. Also, the provision of high-dose or adjuvant influenza vaccines and the development of a new vaccine platform (e.g., mRNA) are suggested to achieve optimal protection against influenza infection among older individuals.
In this study, aging (which is typically associated with immunosenescence) and frailty were found to be strong effect modifiers. There was a large disparity in the VE values across the subgroups stratified by these two variables. That is, subjects who were younger (i.e., 65 ≤ age < 75 years) or fit in the multimorbidity frailty category had higher VE values compared with those of their counterparties (i.e., aged ≥ 75 years and frail), as supported by significant interaction results ( p for interaction < 0.05, Fig. ). These findings were confirmed by interaction tests in the joint subgroup analysis for subjects aged < 75 years and fit or frail. Therefore, these results indicate that the underlying health status (e.g., old age, frailty) may affect VE even in a season with a good match between the vaccine and circulating strains . Young and healthy subjects may have greater VE than those old and frail subjects. Nevertheless, we found that reception of the influenza vaccine was statistically significantly associated with a reduced risk of influenza-associated hospitalization, irrespective of age, frailty status, and with and without high risk of influenza infection; e.g., although low VE values (95% CIs) were obtained for subjects aged 75 years (i.e., 6.6% [1.0%, 11.9%]), frail individuals (11.8% [7.1%, 16.3%]), and high-risk older individuals (13.1% [8.1%, 17.8%]), all estimates were statistically significant. Such specific subgroups in older populations should be prioritized for receiving high-dose, or adjuvanted influenza vaccines whenever those are available in Taiwan. Moreover, the caregivers of these subgroups are recommended to be vaccinated to optimize protection from influenza infection for older individuals through the cocoon strategy . Also, the number needed to treat (NNT) results of influenza vaccination for influenza infection and associated complications can provide explicit insights for clinical decision-making. For example, 47 patients would need to be administered an influenza vaccine relative to non-vaccination for a mean follow-up of 0.8–0.9 years to avert one case having influenza-related hospitalizations (Additional file : Table 7). Also, per recommendations from the World Health Organization, a 75% influenza vaccination coverage rate shall be achieved to avoid the annual epidemics, irrespective of age, country, healthcare systems, and race/ethnicity . With consideration of the suboptimal uptake rate of the influenza vaccine in current practice (e.g., 50–60% ), promoting the additional benefit of influenza vaccination on the reduction of the CV, pulmonary, and kidney risks to society, adopting high-dose influenza vaccines to the annual vaccination program, and improving vaccine accessibility in rural areas may improve influenza vaccination coverage rates for the general older population, thereby diminishing the disease burden attributable to influenza .
In addition to a reduced risk of influenza events, the protective effect of influenza vaccination on a wide spectrum of influenza-related complications (i.e., death and infectious/pulmonary, CV, and kidney diseases) was shown in this study. Previous studies only analyzed the additional benefits of the influenza vaccine among specific subgroups (e.g., patients with gout, CV diseases, or respiratory diseases) of older populations [ – ]. Empirical evidence derived from large-scale general older populations is limited. This study bridges this knowledge gap. It found that influenza vaccination reduces the risks of influenza-related events (i.e., influenza infection, pulmonary disease, and death) by 12–66%, CV diseases by 39–47%, and acute kidney injury by 23%. Mechanisms to support the reduced risks of complications associated with influenza vaccination may be derived from its prevention on influenza infection, which could increase systematic pro-inflammatory cytokines and directly act on vasculature and myocardium, resulting in plaque destabilization and MI or stroke development . Beyond these clinical benefits, the savings from influenza-associated hospitalization following influenza vaccination were also remarkable (i.e., approximately $3,000,000 in total, Additional file : Table 9). It is expected that this economic benefit will increase as the uptake rate of the influenza vaccine increases. The savings could be re-allocated to support the universal coverage of the influenza vaccine for the older population and to maintain the national health insurance program.
First, given the implementation of exclusion criteria according to the study target trial setting, our results may not be generalized to the patients excluded from this study. In particular, a certain number of individuals with dementia (~ 6% of the general older population) were excluded. These individuals are vulnerable to and usually have multiple comorbidities , and are thus at high risk for influenza infection and associated complications. Influenza events are likely to be under-recognized due to the decline in cognitive function in these patients, affecting the validity of the VE estimates presented in this study. Hence, to understand the VE in patients with dementia, a prospective pragmatic trial design that relaxes the strict trial patient inclusion criteria to accommodate disadvantaged patients in routine care settings could be adopted in the future. Second, despite a large amount (i.e., over two million) of older subjects included in the current study, a certain proportion of the older population was lost in the process of PS matching. Therefore, in future research, other matching methods (e.g., inverse probability of treatment weighting) can be utilized to strengthen the robustness and generalizability of the current study findings. Third, the present study did not include self-paid influenza vaccination because such data are unavailable. However, given the universal coverage of influenza vaccination for individuals aged ≥ 65 years in Taiwan, this concern might be negligible. Also, potential unmeasured confounders (e.g., health awareness) might exist, which were likely to increase the effect size and thereby lead to underestimated VE. To minimize such a concern, we measured several surrogate indicators (e.g., the receipt of health examination and cancer screenings within 1 year prior to the index date) and adjusted them in analysis (e.g., matching procedures). Fourth, the clinical complications that occurred during influenza hospitalizations (Table ) were likely affected by the timely receipt of antiviral prescriptions, treatment with antibiotics, and receipt of pneumococcal vaccination, which were not measured and adjusted in our analyses. Fifth, there is a lack of laboratory confirmation (e.g., polymerase chain reaction testing) to identify influenza cases. Lastly, the present study was conducted from the perspective of the healthcare sector and, therefore, did not consider non-medical benefits (e.g., the loss of productivity of family members due to having to care for influenza patients) of influenza vaccination. These additional outcomes following vaccination deserve to be included in a future analysis of the overall benefit of influenza vaccination.
This empirical study with a large-scale general older population adds supporting evidence regarding the effects of influenza vaccination on severe influenza events (i.e., those requiring hospitalization), influenza-related complications (i.e., infectious/pulmonary, CV, and kidney diseases and death), and potential health care savings. Beneficial effects were found irrespective of individual age, frailty status, and underlying high risk for influenza infection, thereby promoting a wide adoption of the influenza vaccine in this population. To avert severe infection episodes, undesirable complications, and associated economic consequences while maintaining immunogenicity against influenza, the uptake of annual influenza vaccination is recommended for older populations.
Additional file 1: Table 1. Target trial emulation framework. Table 2. Operational definitions of exclusion criteria, baseline characteristics, government-funded medical examinations, and exposure to cardiovascular/pulmonary medications. Table 3. Operational definitions of clinical outcomes. Table 4. Baseline characteristics of the study population identified from October 2018 before the two-step propensity score matching. Table 5. Baseline characteristics of the study population identified from November 2018 before the two-step propensity score matching. Table 6. Baseline characteristics of the study population identified from December 2018 before the two-step propensity score matching. Table 7. Influenza event rates and vaccine effectiveness in primary, sensitivity, and negative control outcome analyses. Table 8. Event rates and hazard ratios of vaccination versus non-vaccination for influenza hospitalization in subgroup and joint subgroup analyses. Table 9. Descriptive results of influenza-associated healthcare resource utilization stratified by status of vaccination. Fig. S1. Study scheme of sequential trial approach. Fig. S2. Flowchart of cohort selection.
|
Comparison of Nintendo Wii and PlayStation2 for Enhancing Laparoscopic Skills | 08dd13a7-126c-4fe4-b442-33ba9071855d | 3558901 | Gynaecology[mh] | The growing use of laparoscopy in gynecology has led to an increasing need for resident training. Virtual reality simulators and box trainers have both been used to provide a controlled setting in which residents can develop technical skills outside of the operating room. – Several studies have demonstrated a strong correlation between performance on a virtual reality simulator and performance in the operating room. , Virtual reality simulators display high-fidelity images and scenarios, but one major disadvantage is the paucity of tissue feedback or haptics. On the other hand, the low-fidelity, low-budget box trainers provide tactile feedback. Students trained on box trainers have also shown significant improvement in their intra- and extracorporeal suture-tying skills, precision, and overall surgical time, compared with the controls. Residents may have better skill retention after training with a box trainer. The debut of the Nintendo Wii in 2006 introduced the world to a novel gaming experience; 3-dimensional movement mimicry and tactile feedback. Wii controllers consist of a Wii Remote paired with a Wii Nunchuk. This duo functions on accelerator technology and triangulation, thus allowing for much more realistic movements than traditional controllers. Accelerators are a vector-based technology that can sense orientation and shock to provide feedback vibration. They allow the player to move freely and mimic true motion. Players grasp the controllers and move their arms, hands, and wrists. Many Wii games involve fine motor skills and precise coordinated maneuvers of the controllers, which more accurately mimic movements in laparoscopy. They necessitate timing, depth perception, quick reflexes, and precise movements. At the end of each game, the console scores the performance. Studies on the correlation between manual dexterity, speed, and acquisition of laparoscopic skills with video game experience are conflicting. – Several recent studies have focused on the Wii platform. Similar to traditional video games, Wii has been found to be associated with better laparoscopy skill. We hypothesized that the Nintendo Wii would be superior to traditional joystick/push button video games such as PlayStation2 (PS2) for laparoscopy training.
This was a single-center, stratified, randomized, controlled trial conducted at the Beth Israel Medical Center in New York City. Randomization was done using balanced randomization (1:1) in blocks of 10. Allocation concealment was maintained by central randomization until after the pretest was completed. A sample size of convenience was used. Eligible participants were Department of Obstetrics and Gynecology house staff and faculty as well as third- and fourth-year medical students. Recruitment was performed via E-mail and during grand rounds. Testing was performed during daylight hours whenever a participant had an uninterrupted block of time during the workday or immediately afterwards. No testing was done post call or in the middle of the night. The trial was approved by the Beth Israel Medical Center Institutional Review Board. The trial is registered: ClinicalTrials.gov Identifier: NCT01483677. A background questionnaire gathered demographics, experience in laparoscopy, and video games. A stratified enrollment by the number of laparoscopy cases performed in the past year was conducted. This categorization was used to take into account the range of laparoscopy experience from intern to chiefs, and within the attending pool. An intern may not have more cases than a medical student, and a chief may have had a similar number of cases as certain attendings in the past year. Each stratum was then randomized to either the Nintendo Wii or PlayStation2. All test participants completed a box trainer pretest after watching a short instructional video where an American Association of Gynecologic Laparoscopists fellowship trained Minimally Invasive Gynecology Specialist (Dr. Wang) demonstrated the box trainer task. After a 10-min break, during which they watched a short video tutorial on the assigned video game, they played their respective video games for 30 min. Upon completion of the game, the participants received a 10-min break before taking the post-test. The post-test was identical to the pretest ( ) . The first author conducted all proctoring. Box Trainer Pretest and Posttest The optical system comprised a camera, a light source, a monitor and a zero-degree 10-mm laparoscope secured at a fixed angle and distance directed away from the participant. The image was projected onto the monitor placed in line with the participant. The 2 tasks were intracorporeal suturing and bead transfer. Suturing Task Two tourniquets were spaced 0.5 cm apart and secured to a base. Mirrored targets were marked on the tourniquets at 1.5-cm intervals and 0.5 cm from the edge. The goal was to place 3 interrupted stitches using polyglactin suture and secure them with intracorporeal square knots. Accuracy was measured by the distance between the suture and the premarked targets. An error was defined as a gap >2 mm. Points were deducted for sutures placed more than 2 mm from the premarked targets. The quality of the knot was evaluated by attempting to insert a small suture scissor under the knot. If it passed, the knot was considered to be loose or unsatisfactory and thus was not included in the final score. Five minutes were allotted to the task. The score was produced by counting the number of secure knots made in 5 min and subtracting any errors with regards to accuracy and loose knots. Bead Transfer Task Two pegs were secured on the left and right sides of a board. Three beads were stacked on the left peg. The task was to grasp each bead with the right grasper, transfer it to the left grasper in midair and then stack it on the right peg. After all 3 beads had been transferred onto the right peg, they were then transferred back to the left peg in a similar fashion. Five minutes were allowed for the task. The time to completion was subtracted from a starting score of 300 (seconds). Each bead not on a peg at the end of the task incurred a 20-point deduction. Video Games PlayStation2 Trial (Time Crisis 2) The participants received a scripted video introduction to the PlayStation2 system (Sony Playstation 2 Model No: SCPH-30001 Serial Number: U0288336. Sony Computer Entertainment America LLC: 919 & 989 East Hillsdale Boulevard, Foster City CA, 94404, 650 655 8000), controllers, goals, and rules of the game. Time Crisis 2 was the game played on PlayStation2. The purpose of this game was to shoot as many enemies as possible in a set period of time. Participants played as many times as possible in a 30-min time span. Nintendo Wii Trial (Boomblox) The participants received a scripted video introduction to the Wii system (Nintendo Wii Model No. RVL-001, Serial Number: LU30472824 Nintendo of America Inc., P.O. Box 957, Redmond, WA 98073-0957 USA), controllers, goals, and rules of the game. Boomblox was the game played on the Wii. The goal was to remove as many blocks as possible in a steady and controlled fashion. Participants played as many times as possible in a 30-min time span. Data Analysis Final Wii and PS2 scores were compiled by averaging all scores from the 30 min of game play. Paired t tests were used to analyze mean and standard deviations of the pre- and post-test scores. This study was reported using the 2010 CONSORT guidelines.
The optical system comprised a camera, a light source, a monitor and a zero-degree 10-mm laparoscope secured at a fixed angle and distance directed away from the participant. The image was projected onto the monitor placed in line with the participant. The 2 tasks were intracorporeal suturing and bead transfer.
Two tourniquets were spaced 0.5 cm apart and secured to a base. Mirrored targets were marked on the tourniquets at 1.5-cm intervals and 0.5 cm from the edge. The goal was to place 3 interrupted stitches using polyglactin suture and secure them with intracorporeal square knots. Accuracy was measured by the distance between the suture and the premarked targets. An error was defined as a gap >2 mm. Points were deducted for sutures placed more than 2 mm from the premarked targets. The quality of the knot was evaluated by attempting to insert a small suture scissor under the knot. If it passed, the knot was considered to be loose or unsatisfactory and thus was not included in the final score. Five minutes were allotted to the task. The score was produced by counting the number of secure knots made in 5 min and subtracting any errors with regards to accuracy and loose knots.
Two pegs were secured on the left and right sides of a board. Three beads were stacked on the left peg. The task was to grasp each bead with the right grasper, transfer it to the left grasper in midair and then stack it on the right peg. After all 3 beads had been transferred onto the right peg, they were then transferred back to the left peg in a similar fashion. Five minutes were allowed for the task. The time to completion was subtracted from a starting score of 300 (seconds). Each bead not on a peg at the end of the task incurred a 20-point deduction.
The participants received a scripted video introduction to the PlayStation2 system (Sony Playstation 2 Model No: SCPH-30001 Serial Number: U0288336. Sony Computer Entertainment America LLC: 919 & 989 East Hillsdale Boulevard, Foster City CA, 94404, 650 655 8000), controllers, goals, and rules of the game. Time Crisis 2 was the game played on PlayStation2. The purpose of this game was to shoot as many enemies as possible in a set period of time. Participants played as many times as possible in a 30-min time span.
The participants received a scripted video introduction to the Wii system (Nintendo Wii Model No. RVL-001, Serial Number: LU30472824 Nintendo of America Inc., P.O. Box 957, Redmond, WA 98073-0957 USA), controllers, goals, and rules of the game. Boomblox was the game played on the Wii. The goal was to remove as many blocks as possible in a steady and controlled fashion. Participants played as many times as possible in a 30-min time span.
Final Wii and PS2 scores were compiled by averaging all scores from the 30 min of game play. Paired t tests were used to analyze mean and standard deviations of the pre- and post-test scores. This study was reported using the 2010 CONSORT guidelines.
A total of 42 (Wii: 22, PS2: 20) participants were enrolled. Demographic characteristics, video game experience, and laparoscopic experience were similar in the 2 groups ( ) . Twenty-three participants had fewer than 12 cases per year, and 19 had more. The Wii group consisted of 12 less-experienced and 10 more-experienced surgeons. The PS2 group consisted of 11 less-experienced and 9 more-experienced surgeons. Participants included third-year medical students, fourth-year medical students, Obstetrics & Gynecology attending physicians and residents PGY 1-4. All participants assigned to Wii showed significant improvement in bead-transfer scores (pretest 129±18 vs. post-test 191±9, P <.001). Of this intervention group, 86% performed better, and 14% performed worse on their post-test bead task. No statistically significant improvement in suturing scores (pretest 0.45±0.14 vs. post-test 0.59±0.17) ( ) was observed. In fact, 18% performed worse, 64% the same, and 18% better on their suturing task ( ) . PS2 participants also showed statistically significant improvement in bead-transfer scores (pretest 139±17 vs. post-test 178±14, P = .012). Eighty percent performed better, and 20% performed worse on their post-test bead task. Suturing scores were not significantly improved (pretest 0.45±0.18 vs. post-test 0.55±0.20) ( ) . Fifteen percent performed worse, 55% the same, and 30% better on their suturing task ( ) .
Laparoscopic bead transfer skills were significantly improved in both Wii and PS2 participants. Although the degree of improvement between the 2 cohorts was not statistically significant despite a large difference (Wii: 62±14, PS2: 40±14), this pilot study was not powered to detect a difference of this magnitude. However, the difference of 22 points or seconds on the bead transfer task is encouraging and may carry clinical importance. Several studies have focused on examining the impact of Wii gaming with laparoscopy performance. These investigations compared intervention with the Wii versus no intervention. Badurdeen et al. found that participants with previous Wii experience demonstrated higher baseline laparoscopic skills. Boyle et al. noted improved bead transfer skills after practicing on the Wii (Super Monkey Balls) for 3 h over the span of 7 d. A study by Schilkum et al. found Wii players to move more proficiently in laparoscopy than nonplayers did. To our knowledge, no study has compared the impact of Wii versus traditional video games on laparoscopic skill improvement. This study is unique in its attempt to identify any advantage derived from the tactile feedback and novel Wii mechanics that mimic laparoscopic movements better than traditional joystick/pushbutton systems. We noticed a decline in the bead transfer box trainer score after the intervention in Wii and PS2 participants (14% and 20%, respectively). This paradoxical outcome can be attributed to participant fatigue and distraction. The study was conducted in the call room lounge. Foot traffic and telephone calls in this area were unavoidable. We tried to minimize distraction by keeping the door closed and asking those who passed through to keep noise to a minimum. This study had both strengths and weaknesses. The strengths include clear, specific, and measurable outcomes, high participation rate, and no loss to follow-up. The weaknesses include using a sample size of convenience, limited duration of exposure, limited follow-up, and a single proctor who also performed the randomization. A single 30-min episode of video game time may not have been enough exposure to detect a real difference on the impact of Wii versus PS2 on long-term laparoscopic skills over the course of residency training.
Both the Wii and PS2 significantly improved laparoscopic skills in bead transfer as a training tool and may serve as a low-cost alternative to expensive simulators.
|
CT-based multi-regional radiomics model for predicting contrast medium extravasation in patients with tumors: A case-control study | 13d259d9-9184-4315-be2b-f066ee7111d1 | 11893132 | Cardiovascular System[mh] | Contrast-enhanced computed tomography (CECT) is a widely utilized non-invasive diagnostic tool in clinical settings that plays an essential role in oncology by facilitating tumor diagnosis, TNM staging, treatment evaluation, prognosis prediction [ – ], and, most importantly, monitoring treatment response . However, contrast medium (CM) extravasation is a potential complication during CECT when injecting CM under high pressure, with a reported incidence rate of 0.1–1.2% . CM extravasation can result in serious complications, including compartment syndrome, skin ulcers, and tissue necrosis. Extravasated CM may also damage the surrounding vasculature, which induces local inflammatory responses that pose significant health threats . Consequently, careful measures are necessary to mitigate the risk of CM extravasation during CECT procedures in patients with tumors. CM leaks from blood vessels due to vascular injury caused by injection pressure. Therefore, maintaining the integrity of venous and arterial vessels is crucial to avoiding CM extravasation. Chemotherapy drugs, including vinblastine, paclitaxel, anthracyclines, and platinum-based agents, may increase the risk of CM extravasation. They cause vascular endothelial dysfunction, reduce vascular elasticity , and render the vascular walls more fragile during tumor diagnosis and treatment. Consequently, to properly manage the risk of CM extravasation in patients with malignancies, a complete vascular evaluation must be performed prior to a CECT examination. However, the current risk assessment for CM extravasation primarily relies on the subjective experience of nursing staff . While this approach is useful for identifying high-risk patients, it is time-consuming and inefficient. Additionally, factors such as communication barriers, varying education levels, and incomplete medical histories complicate pre-injection evaluations for all patients . Furthermore, these methods are unable to assess extravasation risk based on its fundamental cause—the condition of the vasculature, which is critical for accurate risk assessment. Radiomics can effectively characterize pathophysiological features by extracting high-volume image data beyond visual assessment capabilities . Currently, it is used for tumor diagnosis, differential diagnosis, treatment efficacy evaluation, and survival prediction in oncology [ – ]. Previous studies have utilized radiomics to predict vascular invasion in tumors and to evaluate coronary arteries for assessing the risk of coronary artery disease . Existing vascular radiomics research has primarily targeted CECT, with few studies addressing non-contrast CT for vascular evaluation. However, non-contrast CT images contain numerous microscopic vascular heterogeneities, such as plaques, vascular elasticity, inflammation, and other features that are invisible to the naked eye but can be extracted via radiomics . The European Society of Urogenital Radiology (ESUR) guidelines on contrast agents suggest that documenting extravasation using CT scans of the affected region may aid in the management of CM extravasation . This enables a more intuitive and objective assessment of vessel condition, thereby facilitating the prediction of CM extravasation. Therefore, radiomics on non-contrast CT has the potential to help predict CM extravasation in patients with tumors. In this study, we aimed to establish a prediction model based on multi-regional vascular volumes of interest (VOIs) from non-contrast CT images to assess the risk of CM extravasation in patients with tumors and to evaluate its clinical applicability.
This retrospective study (LS2022079) was approved by the Institutional Review Board of the Affiliated Hospital of Jiangnan University and was conducted in accordance with the principles of the Declaration of Helsinki, the date of approval was September 13th, 2023. The requirement for informed consent was waived by the ethics committee, given that all data were anonymised and aggregated, ensuring the privacy and confidentiality of the participants. Patients Between January 1, 2022, and May 31, 2023, a total of 282 tumor patients were enrolled from two medical institutions. Patients were identified using the case management system, the details of the patients with tumors undergo clinical routine CT follow-up examinations are provided in the . At the first institution, 75 patients with tumors with extravasation were matched with 150 patients with tumors non-contrast media (non-CM) extravasation in a 1:2 ratio based on tumor type ( ). These patients were then divided into a training cohort (n = 157) and a validation cohort (n = 68). Additionally, 57 patients with tumors from the second institution were included as an external test cohort. The data collection and access periods for research purposes began in October 2023.The inclusion criteria were confirmed tumor diagnosis through pathological examination, having undergone CECT examination(s), and comprehensive CT imaging of the neck and chest within 2 weeks before CECT. Patients were excluded if their clinical data were incomplete or if they had undergone CT angiography. The patient selection process sees . According to Infiltration Nurses Society Standard , the diagnostic criteria of CM extravasation include local reactions, such as ulcer, redness, edema, induration, venous cord-like changes and discomfort. Study variables The primary outcome variable of interest was “CM extravasation”, defined as the accidental leakage of injected fluid into the surrounding tissue. The independent variables in this study included radiomics scores (Rad scores) and clinical characteristics of patients. Rad scores are calculated by extracting the image information of the blood vessel to evaluate the state of blood vessels through quantitative analysis and calculations of vascular images. Clinical characteristics variables included gender, age, BMI, puncture site, types of tumors, chemotherapy cycles, hypertension, diabetes. CT data acquisition CT images were collected using 2 scanners: Somatom Sensation 64 (Siemens Medical Solutions, Forchheim, Germany); or Optima CT660 (GE Medical Systems, Milwaukee, WI, USA). Scanning parameters included the following: slice thickness and spacing, 5 mm; tube voltage, 120 kV; Smart tube current; matrix, 512 × 512; and pitch, 0.992:1. The scan was performed from the mandible to the diaphragm. According to the injection speed and the patient’s vascular condition, the radiologist nurses choose 22G high-pressure resistant peripheral short catheter (the maximum tolerance pressure is 350 psi). Before intravenous injection of CM, the radiologist nurses performed routine inspection on all venous access again, pre-injected normal saline to determine whether the catheter was unobstructed and confirmed that the catheter was in the vein. Patients should be told to raise their hands to inform the medical staff if there is sudden pain or other discomfort at the injection site during CT high-pressure injection of CM. For enhanced scanning, iohexol was used as the CM, administered at a dose of 0.75 ml/kg using a high-pressure syringe at a rate of 2.5–3.0 ml/s. CM extravasation data collection and clinical variable evaluation The radiology department’s CM extravasation data from January 2022 to May 2023 were searched using the hospital’s adverse event reporting system and the CM extravasation record table. The records of CM extravasation include the Puncture site, injection rate, extravasation volume, local manifestations, patient complaints, intervention measures, and prognosis. The Picture Archiving and Communication System (PACS) retrieved clinical medical records and non-contrast CT images, which were then inspected and analyzed in accordance with inclusion and exclusion criteria. Two radiologists and one nurse participated in the data collection process. The first radiologist (6 years of experience) outlined VOI of vessels on plain CT images with reference to contrast-enhanced CT images. The second radiologist (15 years of experience) was tasked with verifying and rectifying any discrepancies in the delineated regions of interest. Additionally, a nurse was responsible for gathering clinical information of patients in both the CM group and the non-CM group from the medical record system. VOI segmentation in different regions The open-source software ITK-SNAP (version 3.4.0, www.itksnap.org ) was used for vessel segmentation. A radiologist (ZW) with 7 years’ experience, who was blinded to the clinical data, used ITK-SNAP to outline 5 consecutive layers of the right common carotid artery/right internal jugular vein (RCCA/RIJV), right subclavian artery/vein (RSA/RSV), and thoracic aorta (THA). For each subject, three VOIs were outlined and reviewed by a senior radiologist (SDH, 20 years’ experience). The radiomics process is illustrated in . Radiomics feature extraction For the purpose of feature extraction, VOIs were established using a radiomics module that was implemented into the open-source software program 3D Slicer version 4.9 ( http://www.slicer.org ) and supported by Pyradiomics. Four categories were used to organize the retrieved characteristics: morphological, grayscale statistical, texture, and Gabor wavelet features. These features were employed in the radiomics model to examine the macrovascular state of cancer patients after normalization to a standard range. Feature selection and model building The radiomics model combined features drawn from three groups of large blood vessels. Features possessing nonzero coefficients were chosen to construct the radiomics signature. Multivariate logistic regression was utilized to implement techniques like Maximum Relevance and Minimum Redundancy (mRMR) and Least Absolute Shrinkage and Selection Operator (LASSO) in the primary cohort in order to reduce overfitting and choose the most informative radiomic features for a prediction model. The Rad scores for each patient was determined by fitting the selected features linearly based on their relative coefficients using logistic regression. Predictive performance of the radiomics signature The performance of the radiomics signature, including metrics such as area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, specificity, positive predictive value, negative predictive value, and accuracy, was determined in the training cohort. For consistency, the same optimal cut-off was applied to the validation and external test cohorts. A new radiomics nomogram model that combines clinical factors and the Rad scores was developed in the training cohort. Then the calibration curves were used to assess the alignment between the predicted risk and the observed outcomes of CM extravasation. Finally, the clinical value of the CT-based nomogram was established using decision curve analysis (DCA). Statistical analysis Statistical evaluations were performed using SPSS version 26.0 (IBM Corp, Armonk, NY, USA) and R version 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria; http://www.Rproject.org ). Patient data are expressed as mean values with standard deviation (SD) for continuous variables, and as frequency counts for categorical variables. Chi-squared test (χ²) or Fisher’s exact test was used for comparing categorical variables. Binary logistic regression, odds ratios (OR), and 95% confidence intervals (95% CI) were employed to assess the association between risk factors and the CM extravasation. Based on Rad scores and identified independent predictors of CM extravasation, three models were developed: Model 1 is the clinical model, Model 2 is the radiomics model, and Model 3 is the combined R + C model. Diagnostic capabilities were compared using ROC curve analysis, and AUC was calculated to evaluate model performance. A p < 0.05 was considered statistically significant.
Between January 1, 2022, and May 31, 2023, a total of 282 tumor patients were enrolled from two medical institutions. Patients were identified using the case management system, the details of the patients with tumors undergo clinical routine CT follow-up examinations are provided in the . At the first institution, 75 patients with tumors with extravasation were matched with 150 patients with tumors non-contrast media (non-CM) extravasation in a 1:2 ratio based on tumor type ( ). These patients were then divided into a training cohort (n = 157) and a validation cohort (n = 68). Additionally, 57 patients with tumors from the second institution were included as an external test cohort. The data collection and access periods for research purposes began in October 2023.The inclusion criteria were confirmed tumor diagnosis through pathological examination, having undergone CECT examination(s), and comprehensive CT imaging of the neck and chest within 2 weeks before CECT. Patients were excluded if their clinical data were incomplete or if they had undergone CT angiography. The patient selection process sees . According to Infiltration Nurses Society Standard , the diagnostic criteria of CM extravasation include local reactions, such as ulcer, redness, edema, induration, venous cord-like changes and discomfort.
The primary outcome variable of interest was “CM extravasation”, defined as the accidental leakage of injected fluid into the surrounding tissue. The independent variables in this study included radiomics scores (Rad scores) and clinical characteristics of patients. Rad scores are calculated by extracting the image information of the blood vessel to evaluate the state of blood vessels through quantitative analysis and calculations of vascular images. Clinical characteristics variables included gender, age, BMI, puncture site, types of tumors, chemotherapy cycles, hypertension, diabetes.
CT images were collected using 2 scanners: Somatom Sensation 64 (Siemens Medical Solutions, Forchheim, Germany); or Optima CT660 (GE Medical Systems, Milwaukee, WI, USA). Scanning parameters included the following: slice thickness and spacing, 5 mm; tube voltage, 120 kV; Smart tube current; matrix, 512 × 512; and pitch, 0.992:1. The scan was performed from the mandible to the diaphragm. According to the injection speed and the patient’s vascular condition, the radiologist nurses choose 22G high-pressure resistant peripheral short catheter (the maximum tolerance pressure is 350 psi). Before intravenous injection of CM, the radiologist nurses performed routine inspection on all venous access again, pre-injected normal saline to determine whether the catheter was unobstructed and confirmed that the catheter was in the vein. Patients should be told to raise their hands to inform the medical staff if there is sudden pain or other discomfort at the injection site during CT high-pressure injection of CM. For enhanced scanning, iohexol was used as the CM, administered at a dose of 0.75 ml/kg using a high-pressure syringe at a rate of 2.5–3.0 ml/s.
The radiology department’s CM extravasation data from January 2022 to May 2023 were searched using the hospital’s adverse event reporting system and the CM extravasation record table. The records of CM extravasation include the Puncture site, injection rate, extravasation volume, local manifestations, patient complaints, intervention measures, and prognosis. The Picture Archiving and Communication System (PACS) retrieved clinical medical records and non-contrast CT images, which were then inspected and analyzed in accordance with inclusion and exclusion criteria. Two radiologists and one nurse participated in the data collection process. The first radiologist (6 years of experience) outlined VOI of vessels on plain CT images with reference to contrast-enhanced CT images. The second radiologist (15 years of experience) was tasked with verifying and rectifying any discrepancies in the delineated regions of interest. Additionally, a nurse was responsible for gathering clinical information of patients in both the CM group and the non-CM group from the medical record system.
The open-source software ITK-SNAP (version 3.4.0, www.itksnap.org ) was used for vessel segmentation. A radiologist (ZW) with 7 years’ experience, who was blinded to the clinical data, used ITK-SNAP to outline 5 consecutive layers of the right common carotid artery/right internal jugular vein (RCCA/RIJV), right subclavian artery/vein (RSA/RSV), and thoracic aorta (THA). For each subject, three VOIs were outlined and reviewed by a senior radiologist (SDH, 20 years’ experience). The radiomics process is illustrated in .
For the purpose of feature extraction, VOIs were established using a radiomics module that was implemented into the open-source software program 3D Slicer version 4.9 ( http://www.slicer.org ) and supported by Pyradiomics. Four categories were used to organize the retrieved characteristics: morphological, grayscale statistical, texture, and Gabor wavelet features. These features were employed in the radiomics model to examine the macrovascular state of cancer patients after normalization to a standard range.
The radiomics model combined features drawn from three groups of large blood vessels. Features possessing nonzero coefficients were chosen to construct the radiomics signature. Multivariate logistic regression was utilized to implement techniques like Maximum Relevance and Minimum Redundancy (mRMR) and Least Absolute Shrinkage and Selection Operator (LASSO) in the primary cohort in order to reduce overfitting and choose the most informative radiomic features for a prediction model. The Rad scores for each patient was determined by fitting the selected features linearly based on their relative coefficients using logistic regression.
The performance of the radiomics signature, including metrics such as area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, specificity, positive predictive value, negative predictive value, and accuracy, was determined in the training cohort. For consistency, the same optimal cut-off was applied to the validation and external test cohorts. A new radiomics nomogram model that combines clinical factors and the Rad scores was developed in the training cohort. Then the calibration curves were used to assess the alignment between the predicted risk and the observed outcomes of CM extravasation. Finally, the clinical value of the CT-based nomogram was established using decision curve analysis (DCA).
Statistical evaluations were performed using SPSS version 26.0 (IBM Corp, Armonk, NY, USA) and R version 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria; http://www.Rproject.org ). Patient data are expressed as mean values with standard deviation (SD) for continuous variables, and as frequency counts for categorical variables. Chi-squared test (χ²) or Fisher’s exact test was used for comparing categorical variables. Binary logistic regression, odds ratios (OR), and 95% confidence intervals (95% CI) were employed to assess the association between risk factors and the CM extravasation. Based on Rad scores and identified independent predictors of CM extravasation, three models were developed: Model 1 is the clinical model, Model 2 is the radiomics model, and Model 3 is the combined R + C model. Diagnostic capabilities were compared using ROC curve analysis, and AUC was calculated to evaluate model performance. A p < 0.05 was considered statistically significant.
Characteristics of the patients 282 patients with tumors were included after applying the inclusion and exclusion criteria, there were 180 in the non-CM extravasation group and 102 in the CM extravasation group. The training cohorts comprised 157 patients, the validation cohorts comprised 68 patients, and the external test cohorts comprised 57 patients. There were no significant differences among the three cohorts, except for diabetes ( p < 0.001) and hypertension ( p = 0.034), patient characteristics are shown in . In the univariable analysis, gender (OR: 2.218, p = 0.022), age (OR: 1.07, p < 0.001), BMI (OR: 2.16, p = 0.026), chemotherapy cycles (OR: 4.133, 7.75, 7.922, p = 0.054, p = 0.003, p = 0.002), hypertension (OR: 11.227, p < 0.001), and diabetes (OR: 6.012, p = 0.004) were associated with a higher risk of CM extravasation. In addition, multivariable analysis revealed that gender (OR: 2.932, p = 0.021), age (OR: 1.076, p = 0.002), chemotherapy cycles (OR: 3.380, 7.953, 8.968, p = 0.16, p = 0.01, p = 0.005, p = 0.005), and hypertension (OR: 6.356, p = 0.001) were independent predictor for CM extravasation ( ). Features selection Initially, 4278 radiomic features were extracted. After applying sequential feature elimination and conducting a correlation analysis, this number was reduced to a more manageable subset of 667 informative features. Subsequently, the top 13 features were identified using the mRMR and LASSO analyses. These selected features comprised first-order grey histogram metrics, second-order texture descriptors, and higher-order wavelet coefficients; more specifically, 2 were first-order, 1 was texture, and 10 were wavelet-based features. shows the details on feature selection by LASSO. Prediction model performance and comparison A radiomic signature was constructed using the 13 distinctive features that were identified in the training cohort. In both the validation cohort and the external test cohort, Radiomics models demonstrated excellent predictive performance, with AUC values of 0.866, 0.828, respectively ( ), with good sensitivity (0.826, 0.963, respectively) and accuracy (0.809, 0.772, respectively). In the validation, and external test cohorts, the R + C model achieved good predictive performance, with AUCs of 0.911, and 0.869, as well as a good sensitivity (0.831, 0.852, respectively) and accuracy (0.824, 0.807, respectively). However, the AUC values of the clinical model were only 0.806 and 0.740 in the validation and external test cohorts, respectively. Detailed evaluation of the performance of these models is presented in and . Based on the cut-off value derived from the ROC curve, the radiomic signatures were stratified into high- and low-grade groups. Significant variations in Rad-scores were observed between the low- and high-grade patients across the training ( p < 0.001), validation ( p < 0.001), and external test ( p < 0.001) cohorts ( ). Calibration and clinical utility The clinical and radiomics combined model is presented as a nomogram ( ). Calibration curves were used to evaluate the performance of the proposed model and illustrated the concordance between the probabilities predicted by the model and the actual observed outcomes. In the training cohorts, the calibration curve of the combined model demonstrates a good degree of fit, indicating that the difference between its predicted probabilities and the actual incidence rates is small ( - ). In the training, validation, and external test cohorts, the DCA results supported the clinical usefulness of the combined model. The greatest benefits of the combined model model were obtained when the threshold probability was in the range of 25–80%. The use of the nomogram to predict CM extravasation was more effective than was using only clinical variables or Rad-score ( - ).
282 patients with tumors were included after applying the inclusion and exclusion criteria, there were 180 in the non-CM extravasation group and 102 in the CM extravasation group. The training cohorts comprised 157 patients, the validation cohorts comprised 68 patients, and the external test cohorts comprised 57 patients. There were no significant differences among the three cohorts, except for diabetes ( p < 0.001) and hypertension ( p = 0.034), patient characteristics are shown in . In the univariable analysis, gender (OR: 2.218, p = 0.022), age (OR: 1.07, p < 0.001), BMI (OR: 2.16, p = 0.026), chemotherapy cycles (OR: 4.133, 7.75, 7.922, p = 0.054, p = 0.003, p = 0.002), hypertension (OR: 11.227, p < 0.001), and diabetes (OR: 6.012, p = 0.004) were associated with a higher risk of CM extravasation. In addition, multivariable analysis revealed that gender (OR: 2.932, p = 0.021), age (OR: 1.076, p = 0.002), chemotherapy cycles (OR: 3.380, 7.953, 8.968, p = 0.16, p = 0.01, p = 0.005, p = 0.005), and hypertension (OR: 6.356, p = 0.001) were independent predictor for CM extravasation ( ).
Initially, 4278 radiomic features were extracted. After applying sequential feature elimination and conducting a correlation analysis, this number was reduced to a more manageable subset of 667 informative features. Subsequently, the top 13 features were identified using the mRMR and LASSO analyses. These selected features comprised first-order grey histogram metrics, second-order texture descriptors, and higher-order wavelet coefficients; more specifically, 2 were first-order, 1 was texture, and 10 were wavelet-based features. shows the details on feature selection by LASSO.
A radiomic signature was constructed using the 13 distinctive features that were identified in the training cohort. In both the validation cohort and the external test cohort, Radiomics models demonstrated excellent predictive performance, with AUC values of 0.866, 0.828, respectively ( ), with good sensitivity (0.826, 0.963, respectively) and accuracy (0.809, 0.772, respectively). In the validation, and external test cohorts, the R + C model achieved good predictive performance, with AUCs of 0.911, and 0.869, as well as a good sensitivity (0.831, 0.852, respectively) and accuracy (0.824, 0.807, respectively). However, the AUC values of the clinical model were only 0.806 and 0.740 in the validation and external test cohorts, respectively. Detailed evaluation of the performance of these models is presented in and . Based on the cut-off value derived from the ROC curve, the radiomic signatures were stratified into high- and low-grade groups. Significant variations in Rad-scores were observed between the low- and high-grade patients across the training ( p < 0.001), validation ( p < 0.001), and external test ( p < 0.001) cohorts ( ).
The clinical and radiomics combined model is presented as a nomogram ( ). Calibration curves were used to evaluate the performance of the proposed model and illustrated the concordance between the probabilities predicted by the model and the actual observed outcomes. In the training cohorts, the calibration curve of the combined model demonstrates a good degree of fit, indicating that the difference between its predicted probabilities and the actual incidence rates is small ( - ). In the training, validation, and external test cohorts, the DCA results supported the clinical usefulness of the combined model. The greatest benefits of the combined model model were obtained when the threshold probability was in the range of 25–80%. The use of the nomogram to predict CM extravasation was more effective than was using only clinical variables or Rad-score ( - ).
This study aimed to develop a radiomics model to predict CM extravasation in patients with tumors using multi-region vascular delineation from non-contrast CT scans. The model integrates clinical risk factors and radiomics features, demonstrating high predictive efficiency in both internal validation and external testing cohorts. Additionally, a nomogram based on the combined radiomics-clinical model demonstrated significant predictive utility. DCA further highlights the clinical benefits of integrating imaging and clinical variables to predict the risk of CM extravasation. CECT plays a crucial role in tumor management and is an effective method for assessing treatment efficacy. However, cautious application is required for patients with fragile vasculature and a high risk of CM extravasation. Currently, prevention strategies primarily focus on enhancing assessment and intervention procedures. Traditionally, the injection rate of CM has been a crucial factor in determining the occurrence of CM extravasation. Accurate pre-administration prediction of CM extravasation by radiology technicians could potentially enable a proactive approach to minimize this risk . Specifically, if radiology technicians can accurately determine the probability of CM extravasation prior to CM injection, adjusting the injection rate to a slower pace could significantly reduce the incidence of extravasation events. In this study, we developed a model to predict the risk of CM extravasation by extracting macrovascular features (size, shape, location) and microscopic features (texture, gray-level intensity, fluctuation patterns) from previous non-contrast CT images of patients with tumors. Radiomic analysis outperformed methods based solely on clinical data in predicting the incidence of extravasation within both training and validation cohorts. The combined radiomic-clinical model demonstrated superior predictive accuracy, providing a quantitative method for monitoring CECT extravasation in patients with tumors. Consistent with previous studies, multivariate analysis confirmed that gender, age, chemotherapy cycles, and hypertension are independent predictors of CM extravasation [ , , ]. Females have a higher incidence of CM extravasation than males, which may be connected to their estrogen levels. Blood vessel flexibility and permeability may be impacted by estrogen, increasing the likelihood of blood vessel injury during CM injection and raising the risk of extravasation . Age is considered a significant risk factor for CM extravasation. As individuals age, blood vessels undergo several physiological changes, including atrophy, vascular sclerosis, increased susceptibility, and decreased flexibility, which may elevate the risk of CM extravasation in the elderly . Additionally, the response of elderly patients to noxious stimuli is weakened, which may delay the identification and timely treatment of extravasation events . The European Society of Urogenital Radiology (ESUR) guidelines for contrast agents state that frail or damaged blood vessels increase the likelihood of extravasation . In this study, with the increase of chemotherapy cycle, the risk of CM extravasation in tumor patients also increases. This elevation in risk may be due to the vascular damage induced by chemotherapeutic drugs. Repeated exposure to chemotherapeutic agents can cause direct injury to the blood vessels, leading to changes in vascular integrity. With each chemotherapy cycle, the veins may become more susceptible to damage, reducing their capacity to handle the pressure from CM injections, which can lead to an increased risk of extravasation . Hypertension heightens the risk of CM extravasation by inducing vascular changes like decreased elasticity and increased rigidity . When large volumes of CM are rapidly injected into vessels that are rigid or occluded, the venous system may exceed its tolerance, potentially leading to vascular rupture and CM extravasation. This predictive model serves as a valuable tool for nurses to identify patients with tumors who are at a higher risk of CM extravasation. It facilitates appropriate clinical decisions and reduces the likelihood of extravasation events. Objective risk assessment allows for the optimization of care for at-risk patients undergoing CECT . Compared to existing nursing evaluation methods, our model offers a more convenient and efficient estimation of extravasation risk by eliminating the need for intermediate calculations and reducing provider workload. As a result, the model can be used as a daily assessment tool to promptly identify patients with tumors at risk of CM extravasation before undergoing CECT, without increasing staff workload. While this study provided valuable insights, several limitations need to be acknowledged. First, its retrospective design introduces inherent risks of selection bias, potentially impacting the reliability and representativeness of the results. The retrospective nature of the study leads to potential data being missing or not fully recorded, failing to capture all variables that could affect the outcomes, such as the injection process, patient reactions during examination, materials used for injection, and the technical factors of medical staff, which were not fully evaluated. Second, this study only included patients with lung, gastrointestinal, and breast cancers, which have higher incidence rates, and did not cover all types of tumors that might experience contrast medium extravasation during enhanced CT examinations. This selection may limit the broad applicability of the results. Third, although matching the extravasation group with the non-extravasation group in a 1:2 ratio was used to reduce the impact of data imbalance, while this matching reduced some confounding factors, it could not completely eliminate the potential influence of unconsidered confounding factors on the results. Fourth, the sample size in this study was too small, which may limit the statistical power and generalizability of the findings. Future studies with larger sample sizes and more diverse tumor types are needed to validate and enhance the proposed model.
In this study, an innovative radiomics methodology based on non-contrast CT was developed, which relied on multi-region vascular delineation, to predict CM extravasation during CECT. The radiomics model exhibited significantly superior predictive efficacy when compared to conventional clinical models. Moreover, the risk for CM extravasation was objectively and quantitatively assessed through the utilization of a nomogram.
S1 Fig The flowchart of diagnostic process for cervical lymph node metastasis and hepatic metastases. (TIF) S2 Fig The flowchart of patient data screening and enrollment. (TIF) S1 Data Data file of training cohort. (XLS) S2 Data Data file of validation cohort. (XLS) S3 Data Data file of external test cohort. (XLS) S4 Data Data file of radiomics. (XLS) S1 File Description of methods. (PDF)
|
Transcultural Adaptation, Validation, Psychometric Analysis, and Interpretation of the 22-Item Thai Senior Technology Acceptance Model for Mobile Health Apps: Cross-Sectional Study | ca7924fd-62c2-48b5-876b-4e6db60ad021 | 11937714 | Medicine[mh] | As the global population ages, the integration of technology into the lives of older adults becomes increasingly crucial for enhancing their quality of life, independence, and well-being . An emerging technology that promotes healthy aging is mobile health (mHealth). mHealth refers to medical and public health services facilitated by mobile devices . It can provide individualized care plans for older adults to sustain functional ability and enhance quality of life . Examples of mHealth innovations for older adults include supporting services for age-friendly health and facilitating the establishment of behavioral changes . However, the adoption of technology, for example, mHealth, among older adults remains a complex and multifaceted issue, influenced by various factors such as individual perception and experience, ease of use, technological support, and sociocultural contexts . To address this challenge, numerous theoretical frameworks have been proposed to understand and predict older adults’ acceptance of technology. Assessing technology acceptance is essential for the successful implementation and use of mHealth technologies, as it directly influences user engagement, health outcomes, and health care delivery efficiency. Understanding acceptance helps developers create user-friendly applications , improves health outcomes through better adherence to interventions , and guides implementation strategies to address barriers effectively . It also informs policy makers and administrators, enabling evidence-based decisions on mHealth investments . Therefore, the lack of validated questionnaires for assessing technology acceptance could lead to a limited understanding of user needs and missed opportunities for improvement. Addressing this gap by developing and validating robust assessment is critical for maximizing the benefits of mHealth technologies and ensuring their effective adoption across diverse populations. In the field of mHealth, various instruments and frameworks have been developed to assess adoption, intention to use, and acceptance. Established instruments like the Health Information Technology Usability Evaluation Scale (Health-ITUES) , System Usability Scale (SUS) , and mHealth App Usability Questionnaire (MAUQ) provided insights focusing on user experiences and satisfaction. Broader frameworks include the unified theory of acceptance and use of technology (UTAUT) , which was extended to include additional factors relevant to mHealth, such as trust and perceived reliability, and was used in various studies to predict mHealth acceptance; the Fit between Individuals, Tasks, and Technology (FITT) Framework is another, which was introduced to measure acceptance in clinical environments, emphasizing the alignment between user needs and technology capabilities. Despite their usability, these instruments and frameworks often lack specificity when addressing the unique needs of older adults. The senior technology acceptance model (STAM) stands out due to its tailored approach for older adults, which addresses their unique challenges and enhances the relevance of mHealth technologies for this population, making it more relevant than general models like the technology acceptance model (TAM) or the UTAUT . Furthermore, it emphasizes the role of social influence and support, which are critical for older adults who may rely on family and caregivers for technological adoption and addressing common health conditions in older adults, such as cognitive load and physical limitations. The STAM was first proposed by Chen and Chan in 2014 and has gained prominence for its focus on the unique needs and characteristics of older adults. This model was developed based on a study of 1012 older adults aged 55 years and older in Hong Kong, and it specifically targets older adults as its primary population of interest. The STAM integrates concepts from established technology acceptance frameworks, such as the TAM and the UTAUT , tailored to address the specific considerations of older adults and provides a thorough framework for studying the factors that influence technological adoption in this age group. The study indicated 8 dimensions associated with technology acceptance in older adults, which included gerontechnology self-efficacy, gerontechnology anxiety, facilitating conditions, self-reported health conditions, cognitive ability, social relationships, attitude toward life and satisfaction, and physical functioning. Sociodemographic factors such as age, gender, education, and economic status are taken into account . While the STAM has been used in different cultural contexts in other Asian countries, including Hong Kong and South Korea , its applicability to the Thai population has not been validated. Thailand, like many other countries, is experiencing rapid population aging, emphasizing the urgency of understanding and promoting health technology acceptance among older adults . However, cultural background, social norms, and technological infrastructures specific to Thailand may influence older adults’ perceptions and behaviors toward technology differently than in other contexts. Therefore, this study aimed to adapt, validate, and define the interpretation of the STAM questionnaire for evaluating the acceptance and intent to use mHealth in Thailand.
Study Design and Study Population The cross-sectional study was conducted from August 2022 to July 2023 through a nationwide, web-based survey and a community survey. Eligible criteria for the study were Thai citizens aged 45 years and older on the date of the survey who could read and communicate in the Thai language and had no underlying conditions or diseases that limited their ability to complete the survey or use mHealth apps (eg, dementia, active psychological problems, or severe visual problems). The web-based survey was disseminated through an assortment of social media platforms, including the department websites, Facebook, Line, Twitter (rebranded as X in 2023), and Instagram. The information on community survey setting and recruitment is described in the section below. For the survey data collection, the respondents to both the web-based and community surveys used the Research Electronic Data Capture (REDCap; Vanderbilt University) survey platform to self-complete the questionnaires. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture, (2) audit trails for tracking data manipulation and export procedures, (3) automated export procedures for seamless data downloads to common statistical packages, and (4) procedures for data integration and interoperability with external sources. All study data were collected and managed using REDCap tools hosted at the Faculty of Medicine, Chiang Mai University. All respondents provided their informed consent, which was included in the screening questionnaire and study information sheet, before participating in this survey. The study excluded incomplete respondents or participants who spent less than 2 minutes or more than 60 minutes on the survey. This study was reported in accordance with COSMIN (Consensus-Based Standards for the Selection of Health Status Measurement Instruments) reporting guidelines for studies on measurement properties of patient-reported outcome measures . Community Survey Setting and Recruitment The community survey was distributed by the investigator team, consisting of medical students and health care personnel at primary care units from 10 subdistricts in Chiang Mai province. To identify eligible participants in the target area, officers from the subdistrict primary care units reviewed periodic health survey data for community-dwelling adults aged 45 years and older. Subsequently, patients’ information was verified with the health-promoting hospital databases to exclude individuals with health conditions that impeded survey participation or mHealth use as described above. The subdistrict primary health care teams invited eligible individuals to participate in the study through individual contact by community health care volunteers, community radio announcements, and posters displayed at primary care units. Ethical Considerations The ethical consideration of the human subject research was approved by the Institutional Review Board of the Faculty of Medicine, Chiang Mai University (COM-2565-09079). All respondents provided their informed consent, as outlined in the screening questionnaire and study information page, before participating in this survey. For the web-based survey, respondents remained anonymous, and no identification data were recorded. In the case of the community survey, identification data of eligible participants were used solely for recruitment purposes within each target area and were not recorded in either the survey form or the study database. Participants received 100 Thai Baht (US $3) as compensation for answering the questionnaires. Translation and Adaptation of the Thai mHealth STAM The original, English, 38-item STAM is a 10-point Likert scale consisting of 10 subscales and 38 items that capture the acceptance of general technology use for the older adult population. The total ranges from 38 to 380 points, with a higher score indicating greater acceptance of technology. The validity and reliability of 38-item STAM have been established on a satisfactory scale in 1012 older adults aged 55 years and older in Hong Kong . The construct validity of the STAM was also evaluated with the confirmatory factor analysis (CFA) and revealed a satisfactory model fit with the proposed structure (comparative fit index [CFI]=0.938, root mean square error of approximation [RMSEA]=0.054, and standardized root mean square residual [SRMR]=0.075). The reliability of each subscale with Cronbach α coefficients ranged from 0.67 to 0.95. Translation and adaptation of the Thai mHealth STAM was performed in accordance with the second edition of the International Test Commission (ITC) Guidelines for Translating and Adapting Tests . In accordance with the ITC precondition guidelines, permission from the holder of the intellectual property rights relating to the 38-item STAM was obtained before performing any translation and adaptation of the STAM. The forward and backward translation with an expert reconciliation design was performed as recommended by the ITC test development guidelines. Before beginning the forward translation process, we decided to include a new subscale, perceived barriers, in the Thai STAM version due to the findings from the previous scoping review on adopting mobile apps for health-related interventions among older adults. It revealed that barriers to adopting mHealth apps among older adults were the most common topics identified in the included studies. Insufficient technological skills, perceived lack of capability and time, concerns regarding personal data privacy, and trust in mHealth providers were the four items comprising the perceived barriers subscale. Following the translation protocol, the original, English, 38-item STAM was adapted to specify mHealth apps in all items and then forward translated into Thai by a professional translator to ensure accuracy for the target audience. The expert panel, which included a digital health expert (family physician and epidemiologist), 2 gerontology physicians, and a public health expert in community medicine, reviewed the forward translation of the Thai STAM questionnaire to ensure readability and transcultural adaptation. The backward translation was done by another professional translator into English. Then, the expert panel reconciled the backward translation version with the original STAM version. The investigator’s team resolved any discrepancies by reaching a final consensus through discussions with the expert panel. To ensure the face and content validity of the proposed questionnaire, a literature review, an expert review, and public interviews were incorporated into the adaptation of the Thai mHealth STAM. In total, 15 older adults participated in this phase to complete the pilot 40-item Thai STAM. Participants were subsequently interviewed to assess the following: overall questionnaire readability, clarity of instructions and items/response options, comprehension of the questionnaire, and other feedback on each item. Then, the pilot 40-item Thai STAM was reworded and revised as recommended on input from both participants and expert interviews. Finally, the pilot 40-item Thai mHealth STAM was given to a group of 40 older adults to verify its reliability and scale usability. Sample Size Estimation The sample size was estimated based on three parameters, which are as follows: (1) a stable structure for an exploratory factor analysis (EFA) based on the rule of thumb, which is 10 cases per question; (2) expected CFI for a CFA based on the structural equation modeling; and (3) expected Cronbach α for the internal consistency of the questionnaire. For the first parameter, according to the rule of thumb, at least 440 respondents, accounting for 10% of the dropout rate, were required for an EFA. To achieve the expected CFI of 0.95 for a CFA, at least 459 respondents, accounting for 10% of the dropout rate, were required based on an average factor loading of 0.60 and an average factor correlation of .30 to ensure a .05 α (type I) error and power of 90% . For testing overall reliability, at least 146 total respondents were required based on expected Cronbach α=0.80 (SD .05), a confidence level of 95%, and a dropout rate of 10% . All sample size estimation was performed by the web-based sample size calculator . Finally, the minimal required sample size for this study was 920, which was divided into 460 each for the EFA and CFA, respectively. Statistical Analysis Descriptive Analysis All statistical analyses were conducted using Stata (version 17.0; StataCorp). A P value below .05 indicated statistical significance. Categorical data were presented as frequency and percentage, while continuous data were described using mean (SD). Univariable analysis for comparison was performed as appropriate. The Thai mHealth STAM item scores were summarized with central estimations, measures of variability, floor and ceiling effect, skewness, and kurtosis tests. The overall psychometric properties of the Thai mHealth STAM were evaluated for validity and reliability as follows: Dimensionality To explore and reduce the dimensionality of the proposed questionnaire, an EFA was performed using a principal component analysis (PCA). The selection of PCA over common factor analysis was based on its ability to enhance parsimony and aid in the selection of factors for CFA . Communalities were initially evaluated, and then orthogonal rotation with the varimax criteria and oblique rotation with promax criteria of the component was conducted. The Kaiser-Meyer-Olkin (KMO) measure and the Bartlett test of sphericity were conducted to verify the appropriateness of using factor analysis. A KMO value greater than 0.8 and a Bartlett test with a P value less than .05 are suggested for assessing sample adequacy and the suitability of the data for factor analysis, respectively. Eigenvalues greater than 1, the cumulative percentage of variance, and the scree plot with the number of factors that explained more than 5% of the variance were used to determine the number of factors to be retained . A parallel study was conducted to validate the optimal threshold for the number of included factors . Then, we used the following criteria to evaluate the adequacy of the EFA results. First, each should be saliently loaded with at least three items to ensure reliability and stability. In case a factor contains only 2 items, the expert panel consensus will be reached to ensure that the factor is meaningful based on the context and theoretical basis. Second, each item should load saliently on only 1 factor without complex or cross-loadings. Third, each factor should demonstrate internal consistency reliability ≥0.70. Fourth, all factors should be theoretically meaningful [ , , ]. Construct Validity For a CFA, structural equation modeling using a maximum likelihood estimation was performed to assure the factor structure based on the exploratory factor, as described previously. To determine the appropriateness of the proposed model, the specific fit indices were evaluated as follows: RMSEA<0.100, SRMR<0.100, CFI>0.900, and Tucker-Lewis Index (TLI)>0.900 [ - ]. To establish acceptance of the final structure of the final model, the coefficient of determination ( R 2 ) and item-scale correlation (standardized factor loading) should be at least 0.30 and 0.40, respectively. Finally, a nonparametric item response theory (IRT) analysis was done to confirm that the final Thai mHealth STAM had the unidimensional set for the relationship between the latent trait and the responses to the items . The IRT analysis was assessed based on fundamental assumptions, including unidimensionality, local independence, and monotonicity. Loevinger H coefficients ( H s ) less than 0.3, between 0.3 and 0.4, and greater than 0.4, as determined by the item traces, correspond to poor, medium, and strong scalability properties, respectively. The monotonicity assumption criterion was determined by a critical value of less than 80. Discriminant Validity To determine the discriminant validity of the final questionnaire, the intention to use mHealth, as indicated in the external question, “If there are available mHealth applications for you, do you want to use them? (yes/no),” was used as the anchor-based question. The discriminative indices, including sensitivity, specificity, and area under the receiver operating characteristic (AUROC), were used with the intention of determining the appropriate cutoff scores. The 6 proposed bandings for the Thai mHealth STAM scores are categorized into low, moderate, and high acceptance based on score tertiles. Associations between these bandings and the intention to use mHealth are presented by adjusted odds ratios (aORs) with 95% CI from a multivariable logistic regression adjusted for potential confounders such as age, gender, education, income, and living alone. Reliability To estimate the correlation statistics for reliability, 95% CI using 1000 bootstrap resampling was presented alongside the reported correlation statistics. An internal consistency consisting of Cronbach α and McDonald ω coefficients was calculated for each item of the final questionnaire, as well as the entirety of the final questionnaire, to determine internal consistency, reliability, and the degree to which every item on a scale measures the same construct. The values of at least .70 indicated acceptable reliability of the questionnaire . In addition, the item-total correlations and the corrected item-total correlations between .20 and .80 were considerably acceptable. A subgroup analysis of adults aged 45-59 years and adults aged 60 years and older was also performed, recognizing the importance of understanding the unique health needs and challenges faced by both current older populations and those who will age into this group in the future.
The cross-sectional study was conducted from August 2022 to July 2023 through a nationwide, web-based survey and a community survey. Eligible criteria for the study were Thai citizens aged 45 years and older on the date of the survey who could read and communicate in the Thai language and had no underlying conditions or diseases that limited their ability to complete the survey or use mHealth apps (eg, dementia, active psychological problems, or severe visual problems). The web-based survey was disseminated through an assortment of social media platforms, including the department websites, Facebook, Line, Twitter (rebranded as X in 2023), and Instagram. The information on community survey setting and recruitment is described in the section below. For the survey data collection, the respondents to both the web-based and community surveys used the Research Electronic Data Capture (REDCap; Vanderbilt University) survey platform to self-complete the questionnaires. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture, (2) audit trails for tracking data manipulation and export procedures, (3) automated export procedures for seamless data downloads to common statistical packages, and (4) procedures for data integration and interoperability with external sources. All study data were collected and managed using REDCap tools hosted at the Faculty of Medicine, Chiang Mai University. All respondents provided their informed consent, which was included in the screening questionnaire and study information sheet, before participating in this survey. The study excluded incomplete respondents or participants who spent less than 2 minutes or more than 60 minutes on the survey. This study was reported in accordance with COSMIN (Consensus-Based Standards for the Selection of Health Status Measurement Instruments) reporting guidelines for studies on measurement properties of patient-reported outcome measures .
The community survey was distributed by the investigator team, consisting of medical students and health care personnel at primary care units from 10 subdistricts in Chiang Mai province. To identify eligible participants in the target area, officers from the subdistrict primary care units reviewed periodic health survey data for community-dwelling adults aged 45 years and older. Subsequently, patients’ information was verified with the health-promoting hospital databases to exclude individuals with health conditions that impeded survey participation or mHealth use as described above. The subdistrict primary health care teams invited eligible individuals to participate in the study through individual contact by community health care volunteers, community radio announcements, and posters displayed at primary care units.
The ethical consideration of the human subject research was approved by the Institutional Review Board of the Faculty of Medicine, Chiang Mai University (COM-2565-09079). All respondents provided their informed consent, as outlined in the screening questionnaire and study information page, before participating in this survey. For the web-based survey, respondents remained anonymous, and no identification data were recorded. In the case of the community survey, identification data of eligible participants were used solely for recruitment purposes within each target area and were not recorded in either the survey form or the study database. Participants received 100 Thai Baht (US $3) as compensation for answering the questionnaires.
The original, English, 38-item STAM is a 10-point Likert scale consisting of 10 subscales and 38 items that capture the acceptance of general technology use for the older adult population. The total ranges from 38 to 380 points, with a higher score indicating greater acceptance of technology. The validity and reliability of 38-item STAM have been established on a satisfactory scale in 1012 older adults aged 55 years and older in Hong Kong . The construct validity of the STAM was also evaluated with the confirmatory factor analysis (CFA) and revealed a satisfactory model fit with the proposed structure (comparative fit index [CFI]=0.938, root mean square error of approximation [RMSEA]=0.054, and standardized root mean square residual [SRMR]=0.075). The reliability of each subscale with Cronbach α coefficients ranged from 0.67 to 0.95. Translation and adaptation of the Thai mHealth STAM was performed in accordance with the second edition of the International Test Commission (ITC) Guidelines for Translating and Adapting Tests . In accordance with the ITC precondition guidelines, permission from the holder of the intellectual property rights relating to the 38-item STAM was obtained before performing any translation and adaptation of the STAM. The forward and backward translation with an expert reconciliation design was performed as recommended by the ITC test development guidelines. Before beginning the forward translation process, we decided to include a new subscale, perceived barriers, in the Thai STAM version due to the findings from the previous scoping review on adopting mobile apps for health-related interventions among older adults. It revealed that barriers to adopting mHealth apps among older adults were the most common topics identified in the included studies. Insufficient technological skills, perceived lack of capability and time, concerns regarding personal data privacy, and trust in mHealth providers were the four items comprising the perceived barriers subscale. Following the translation protocol, the original, English, 38-item STAM was adapted to specify mHealth apps in all items and then forward translated into Thai by a professional translator to ensure accuracy for the target audience. The expert panel, which included a digital health expert (family physician and epidemiologist), 2 gerontology physicians, and a public health expert in community medicine, reviewed the forward translation of the Thai STAM questionnaire to ensure readability and transcultural adaptation. The backward translation was done by another professional translator into English. Then, the expert panel reconciled the backward translation version with the original STAM version. The investigator’s team resolved any discrepancies by reaching a final consensus through discussions with the expert panel. To ensure the face and content validity of the proposed questionnaire, a literature review, an expert review, and public interviews were incorporated into the adaptation of the Thai mHealth STAM. In total, 15 older adults participated in this phase to complete the pilot 40-item Thai STAM. Participants were subsequently interviewed to assess the following: overall questionnaire readability, clarity of instructions and items/response options, comprehension of the questionnaire, and other feedback on each item. Then, the pilot 40-item Thai STAM was reworded and revised as recommended on input from both participants and expert interviews. Finally, the pilot 40-item Thai mHealth STAM was given to a group of 40 older adults to verify its reliability and scale usability.
The sample size was estimated based on three parameters, which are as follows: (1) a stable structure for an exploratory factor analysis (EFA) based on the rule of thumb, which is 10 cases per question; (2) expected CFI for a CFA based on the structural equation modeling; and (3) expected Cronbach α for the internal consistency of the questionnaire. For the first parameter, according to the rule of thumb, at least 440 respondents, accounting for 10% of the dropout rate, were required for an EFA. To achieve the expected CFI of 0.95 for a CFA, at least 459 respondents, accounting for 10% of the dropout rate, were required based on an average factor loading of 0.60 and an average factor correlation of .30 to ensure a .05 α (type I) error and power of 90% . For testing overall reliability, at least 146 total respondents were required based on expected Cronbach α=0.80 (SD .05), a confidence level of 95%, and a dropout rate of 10% . All sample size estimation was performed by the web-based sample size calculator . Finally, the minimal required sample size for this study was 920, which was divided into 460 each for the EFA and CFA, respectively.
Descriptive Analysis All statistical analyses were conducted using Stata (version 17.0; StataCorp). A P value below .05 indicated statistical significance. Categorical data were presented as frequency and percentage, while continuous data were described using mean (SD). Univariable analysis for comparison was performed as appropriate. The Thai mHealth STAM item scores were summarized with central estimations, measures of variability, floor and ceiling effect, skewness, and kurtosis tests. The overall psychometric properties of the Thai mHealth STAM were evaluated for validity and reliability as follows: Dimensionality To explore and reduce the dimensionality of the proposed questionnaire, an EFA was performed using a principal component analysis (PCA). The selection of PCA over common factor analysis was based on its ability to enhance parsimony and aid in the selection of factors for CFA . Communalities were initially evaluated, and then orthogonal rotation with the varimax criteria and oblique rotation with promax criteria of the component was conducted. The Kaiser-Meyer-Olkin (KMO) measure and the Bartlett test of sphericity were conducted to verify the appropriateness of using factor analysis. A KMO value greater than 0.8 and a Bartlett test with a P value less than .05 are suggested for assessing sample adequacy and the suitability of the data for factor analysis, respectively. Eigenvalues greater than 1, the cumulative percentage of variance, and the scree plot with the number of factors that explained more than 5% of the variance were used to determine the number of factors to be retained . A parallel study was conducted to validate the optimal threshold for the number of included factors . Then, we used the following criteria to evaluate the adequacy of the EFA results. First, each should be saliently loaded with at least three items to ensure reliability and stability. In case a factor contains only 2 items, the expert panel consensus will be reached to ensure that the factor is meaningful based on the context and theoretical basis. Second, each item should load saliently on only 1 factor without complex or cross-loadings. Third, each factor should demonstrate internal consistency reliability ≥0.70. Fourth, all factors should be theoretically meaningful [ , , ]. Construct Validity For a CFA, structural equation modeling using a maximum likelihood estimation was performed to assure the factor structure based on the exploratory factor, as described previously. To determine the appropriateness of the proposed model, the specific fit indices were evaluated as follows: RMSEA<0.100, SRMR<0.100, CFI>0.900, and Tucker-Lewis Index (TLI)>0.900 [ - ]. To establish acceptance of the final structure of the final model, the coefficient of determination ( R 2 ) and item-scale correlation (standardized factor loading) should be at least 0.30 and 0.40, respectively. Finally, a nonparametric item response theory (IRT) analysis was done to confirm that the final Thai mHealth STAM had the unidimensional set for the relationship between the latent trait and the responses to the items . The IRT analysis was assessed based on fundamental assumptions, including unidimensionality, local independence, and monotonicity. Loevinger H coefficients ( H s ) less than 0.3, between 0.3 and 0.4, and greater than 0.4, as determined by the item traces, correspond to poor, medium, and strong scalability properties, respectively. The monotonicity assumption criterion was determined by a critical value of less than 80. Discriminant Validity To determine the discriminant validity of the final questionnaire, the intention to use mHealth, as indicated in the external question, “If there are available mHealth applications for you, do you want to use them? (yes/no),” was used as the anchor-based question. The discriminative indices, including sensitivity, specificity, and area under the receiver operating characteristic (AUROC), were used with the intention of determining the appropriate cutoff scores. The 6 proposed bandings for the Thai mHealth STAM scores are categorized into low, moderate, and high acceptance based on score tertiles. Associations between these bandings and the intention to use mHealth are presented by adjusted odds ratios (aORs) with 95% CI from a multivariable logistic regression adjusted for potential confounders such as age, gender, education, income, and living alone. Reliability To estimate the correlation statistics for reliability, 95% CI using 1000 bootstrap resampling was presented alongside the reported correlation statistics. An internal consistency consisting of Cronbach α and McDonald ω coefficients was calculated for each item of the final questionnaire, as well as the entirety of the final questionnaire, to determine internal consistency, reliability, and the degree to which every item on a scale measures the same construct. The values of at least .70 indicated acceptable reliability of the questionnaire . In addition, the item-total correlations and the corrected item-total correlations between .20 and .80 were considerably acceptable. A subgroup analysis of adults aged 45-59 years and adults aged 60 years and older was also performed, recognizing the importance of understanding the unique health needs and challenges faced by both current older populations and those who will age into this group in the future.
All statistical analyses were conducted using Stata (version 17.0; StataCorp). A P value below .05 indicated statistical significance. Categorical data were presented as frequency and percentage, while continuous data were described using mean (SD). Univariable analysis for comparison was performed as appropriate. The Thai mHealth STAM item scores were summarized with central estimations, measures of variability, floor and ceiling effect, skewness, and kurtosis tests. The overall psychometric properties of the Thai mHealth STAM were evaluated for validity and reliability as follows:
To explore and reduce the dimensionality of the proposed questionnaire, an EFA was performed using a principal component analysis (PCA). The selection of PCA over common factor analysis was based on its ability to enhance parsimony and aid in the selection of factors for CFA . Communalities were initially evaluated, and then orthogonal rotation with the varimax criteria and oblique rotation with promax criteria of the component was conducted. The Kaiser-Meyer-Olkin (KMO) measure and the Bartlett test of sphericity were conducted to verify the appropriateness of using factor analysis. A KMO value greater than 0.8 and a Bartlett test with a P value less than .05 are suggested for assessing sample adequacy and the suitability of the data for factor analysis, respectively. Eigenvalues greater than 1, the cumulative percentage of variance, and the scree plot with the number of factors that explained more than 5% of the variance were used to determine the number of factors to be retained . A parallel study was conducted to validate the optimal threshold for the number of included factors . Then, we used the following criteria to evaluate the adequacy of the EFA results. First, each should be saliently loaded with at least three items to ensure reliability and stability. In case a factor contains only 2 items, the expert panel consensus will be reached to ensure that the factor is meaningful based on the context and theoretical basis. Second, each item should load saliently on only 1 factor without complex or cross-loadings. Third, each factor should demonstrate internal consistency reliability ≥0.70. Fourth, all factors should be theoretically meaningful [ , , ].
For a CFA, structural equation modeling using a maximum likelihood estimation was performed to assure the factor structure based on the exploratory factor, as described previously. To determine the appropriateness of the proposed model, the specific fit indices were evaluated as follows: RMSEA<0.100, SRMR<0.100, CFI>0.900, and Tucker-Lewis Index (TLI)>0.900 [ - ]. To establish acceptance of the final structure of the final model, the coefficient of determination ( R 2 ) and item-scale correlation (standardized factor loading) should be at least 0.30 and 0.40, respectively. Finally, a nonparametric item response theory (IRT) analysis was done to confirm that the final Thai mHealth STAM had the unidimensional set for the relationship between the latent trait and the responses to the items . The IRT analysis was assessed based on fundamental assumptions, including unidimensionality, local independence, and monotonicity. Loevinger H coefficients ( H s ) less than 0.3, between 0.3 and 0.4, and greater than 0.4, as determined by the item traces, correspond to poor, medium, and strong scalability properties, respectively. The monotonicity assumption criterion was determined by a critical value of less than 80.
To determine the discriminant validity of the final questionnaire, the intention to use mHealth, as indicated in the external question, “If there are available mHealth applications for you, do you want to use them? (yes/no),” was used as the anchor-based question. The discriminative indices, including sensitivity, specificity, and area under the receiver operating characteristic (AUROC), were used with the intention of determining the appropriate cutoff scores. The 6 proposed bandings for the Thai mHealth STAM scores are categorized into low, moderate, and high acceptance based on score tertiles. Associations between these bandings and the intention to use mHealth are presented by adjusted odds ratios (aORs) with 95% CI from a multivariable logistic regression adjusted for potential confounders such as age, gender, education, income, and living alone.
To estimate the correlation statistics for reliability, 95% CI using 1000 bootstrap resampling was presented alongside the reported correlation statistics. An internal consistency consisting of Cronbach α and McDonald ω coefficients was calculated for each item of the final questionnaire, as well as the entirety of the final questionnaire, to determine internal consistency, reliability, and the degree to which every item on a scale measures the same construct. The values of at least .70 indicated acceptable reliability of the questionnaire . In addition, the item-total correlations and the corrected item-total correlations between .20 and .80 were considerably acceptable. A subgroup analysis of adults aged 45-59 years and adults aged 60 years and older was also performed, recognizing the importance of understanding the unique health needs and challenges faced by both current older populations and those who will age into this group in the future.
Findings From the Translation and Adaptation of the Thai mHealth STAM After reviewing the forward translation, the panel of experts decided to remove 2 items from the gerontechnology self-efficacy subscale, as they were redundant with the facilitating condition (FC) subscale (FC1 and FC2). Independent back-translation provided an additional check of the semantic equivalence of the translation. A total of 4 items, including PU2, PEOU2, P4, and P8, were modified based on the backward translation. For face and content validity, we conducted interviews with 15 older adults similar to the target population. Based on participants’ feedback, 4 items (FC1, FC2, C4, and P2) were slightly modified for clarity. In addition, 2 gerontology experts suggested rephrasing 2 items (A1 and A2) regarding attitude to aging and life satisfaction due to the sensitive wording. Finally, the 40-item Thai mHealth STAM in the pilot group of 40 older adults indicated acceptable internal consistency (Cronbach α =0.91). The details of the full 40 items (10 dimensions) of the Thai mHealth STAM are presented in Table S1 in . Participant Characteristics From the total of 1100 participants, the mean age was 62.3 (SD 8.8) years. The majority of participants were female (776/1100, 70.5%). Among the 1100 participants, 360 (32.7%) were adults aged 45-59 years, and 740 (67.3%) were older adults aged 60 years and older. Statistically significant differences in the characteristics between adults and older adults were observed in marital status ( P =.003), education levels ( P <.001), income ( P <.001), underlying diseases ( P <.001), and technology experience ( P <.001). The characteristics of the participants of the study population are presented in . The derived data were randomly divided in a 1:1 ratio into 2 datasets in preparation for the EFA and CFA. The characteristics of the participants involved in the EFA and CFA are described in Table S2 in . Dimensionality According to the item analysis, we excluded 6 items from the physical function subscale (P2, P3, P4, P5, P7, and P8) due to a floor effect or ceiling effect of >80% (Table S3 in ). An EFA was conducted using PCA with 34 remaining items. The Bartlett test of sphericity obtained P <.001, indicating that the correlation matrix was not random . The KMO statistic was 0.875, well above the minimum standard for conducting factor analysis . Therefore, we determined that the input data were appropriate for EFA. Subsequently, the rotation of principal components was performed using both orthogonal rotation (varimax) and oblique rotation (promax) in an attempt to achieve a simple structure. Given the fact that an oblique rotation is generally recommended by measurement specialists to facilitate the emergence of factor intercorrelations [ - ], almost all social sciences measurements exhibit some degree of correlation . In addition, the correlation matrix for the factors with oblique (promax) rotation indicated that the highest correlation was 0.445 (Table S3 in ); we thereby determined that the factors were correlated, and hence, oblique rotation was an appropriate approach. The results of parallel analysis (Table S4 in ) and PCA with or without oblique (promax) rotation all recommended the retention of 7 factors. According to the previous criteria, 2-item factors were identified, including factor 4 (PBR1 and PBR2) and factor 6 (S1 and S2). The internal consistency of seven factors demonstrated Cronbach α of 0.884 with 95% CI (0.875-0.894), which met acceptable thresholds. Within the context and theoretical framework of the STAM and the UTAUT [ , , , ], social factors significantly influence behavioral intentions to use technology, particularly in the use of mHealth. Perceived barriers also play a role in determining intentions to use mHealth, as demonstrated by the aforementioned scoping review . The inclusion of factor 4 and factor 6, which represented perceived barriers and social relationships, was considered appropriate. Based on the priori criteria and consensus of the panel experts, the EFA identified 22 candidate items (ATT1, ATT2, PU1, PU2, PU3, PEOU1, PEOU2, PB1, PB2, ANX1, ANX2, FC2, FC4, FC5, H1, H2, H5, C2, C3, C4, S1, and S2) with factor loadings greater than 0.4 that encompassed the 7 factors. The final EFA result is presented in . Construct Validity From the EFA, the 22 items of the 7-factor Thai mHealth STAM explained 91.45% of the variance. The unidimensionality of each factor (subscale) and the overall models were assessed by analyzing modification indices in the CFA. Of the 7 factors from the EFA, the CFA of each factor (subscale) showed that only 5 factors, consisting of cognitive ability (C2, C3, C4), perceived barriers (PB1 and PB2), facilitating conditions (FC2, FC4, and FC5), self-reported health conditions (H1, H2, and H5), and social relationships (S1 and S2), showed satisfactory information criteria indices of the CFA, as presented . Factor 1, which included items from the attitude toward using (ATT1 and ATT2), perceived usefulness (PU1, PU2, and PU3), and perceived ease of use (PEOU1 and PEOU2), did not meet the CFA criteria due to over-factoring issues (Table S5 in ). Factor 2, combining perceived barriers (PB3 and PB4) and gerontechnology anxiety (ANX1 and ANX2), was unfit according to CFA criteria, with a low CFI (0.799), low TLI (0.698), and high RMSEA (0.306, 90% CI 0.293-0.319). Attempts to combine subscales also did not meet CFA criteria (Table S5 in ). However, when items were separated as in the original STAM, including attitude toward using (ATT1 and ATT2), perceived usefulness (PU1, PU2, and PU3), perceived ease of use (PEOU1 and PEOU2), and gerontechnology anxiety (ANX1 and ANX2), these separated factors showed a good fit with CFA criteria (Table S5 in ). Out of the 9 factors from the single latent factor analysis, 5 were 2-item factors. These were kept in the final CFA model because 3 factors (attitude toward using, perceived ease of use, and gerontechnology anxiety) were originally designed as 2-item factors, similar to the original STAM. The perceived barriers and social relationships were also retained because of their contextual relevance, as described above. Finally, the CFA confirmed 9-dimensional sets of 22 items with satisfactory fit indices, as shown in . The details of the CFAs of evaluated and reevaluated models are described in Table S4 in . A nonparametric IRT analysis also affirmed the unidimensionality, local independence, and monotonicity of the 22-item model with 8 factors (Table S6 in ). For the scalability, all 22 items of the Thai mHealth STAM had H s coefficients over 0.4, which indicates medium to strong scalability properties (Table S6 in ). The correlation among the final 22-item Thai mHealth STAM subscales ranged from 0.040 to 0.685 (Table S7 in ). The final 22-item Thai mHealth STAM questions, along with the English version and modeling indices, are described in . Each item is scored on a 10-point Likert scale from 1 (very unsatisfied or strongly disagree) to 10 (very satisfied or strongly agree), with reverse scaling for perceived barriers and gerontechnology anxiety. Discriminant Validity Considering the absence of a reference standard, it is theoretically reasonable that more participants with higher STAM scores will result in greater acceptance and adoption of technology. The discriminative indices, including sensitivity, specificity, and AUROC, were used to determine the cutoff scores for the proposed questionnaire, considering the intention to use mHealth from the external question. The 6 proposed sets of the final 22-item Thai mHealth STAM bands were classified into low, moderate, and high acceptance, as presented in . The set D of the possible banding was preferred as the optimal 22-item Thai mHealth STAM cutoff score based on the highest sensitivity of 89% (95% CI 86.1%-91.5%) and AUROC of 72.4% (95% CI 70%-74.8%). This finding also confirmed the discrimination performance of the 22-item Thai mHealth STAM in identifying persons with and without the intention to use mHealth. For set D, low, moderate, and high scores are defined as ≤151, 152-180, and ≥181, respectively. In addition, we conducted a subgroup analysis based on age groups: pre-older adults (aged 45-59 years) and older adults (aged 60 years and older). The result revealed that the set D banding had robust discriminant validity in older adults (AUROC 73%, 95% CI 70%-76%), but the discriminant validity decreased in the pre-older adult group (AUROC 67.7%, 95% CI 63.3%-71.9%). The discriminant validity of the 22-item Thai mHealth STAM by the subpopulation cohorts is shown in Table S8 in . Scale Reliability Out of 1100 overall participants, the final 22-item Thai mHealth STAM demonstrated an excellent internal consistency in both the Cronbach α (0.88, 95% CI 0.87-0.89) and the McDonald ω coefficients (0.85, 95% CI 0.83-0.87), as shown in . By subpopulation, the Cronbach α and the McDonald ω coefficients were 0.88 (95% CI 0.86-0.90) and 0.84 (95% CI 0.81- 0.89) for adults aged 45-59 years and 0.88 (95% CI 0.86- 0.89) and 0.83 (95% CI 0.81-0.86) for older adults. All 22 items revealed the corrected item-total correlations ranging from 0.26 to 0.71, achieving a level of acceptance between 0.20 and 0.80 ( ).
After reviewing the forward translation, the panel of experts decided to remove 2 items from the gerontechnology self-efficacy subscale, as they were redundant with the facilitating condition (FC) subscale (FC1 and FC2). Independent back-translation provided an additional check of the semantic equivalence of the translation. A total of 4 items, including PU2, PEOU2, P4, and P8, were modified based on the backward translation. For face and content validity, we conducted interviews with 15 older adults similar to the target population. Based on participants’ feedback, 4 items (FC1, FC2, C4, and P2) were slightly modified for clarity. In addition, 2 gerontology experts suggested rephrasing 2 items (A1 and A2) regarding attitude to aging and life satisfaction due to the sensitive wording. Finally, the 40-item Thai mHealth STAM in the pilot group of 40 older adults indicated acceptable internal consistency (Cronbach α =0.91). The details of the full 40 items (10 dimensions) of the Thai mHealth STAM are presented in Table S1 in .
From the total of 1100 participants, the mean age was 62.3 (SD 8.8) years. The majority of participants were female (776/1100, 70.5%). Among the 1100 participants, 360 (32.7%) were adults aged 45-59 years, and 740 (67.3%) were older adults aged 60 years and older. Statistically significant differences in the characteristics between adults and older adults were observed in marital status ( P =.003), education levels ( P <.001), income ( P <.001), underlying diseases ( P <.001), and technology experience ( P <.001). The characteristics of the participants of the study population are presented in . The derived data were randomly divided in a 1:1 ratio into 2 datasets in preparation for the EFA and CFA. The characteristics of the participants involved in the EFA and CFA are described in Table S2 in .
According to the item analysis, we excluded 6 items from the physical function subscale (P2, P3, P4, P5, P7, and P8) due to a floor effect or ceiling effect of >80% (Table S3 in ). An EFA was conducted using PCA with 34 remaining items. The Bartlett test of sphericity obtained P <.001, indicating that the correlation matrix was not random . The KMO statistic was 0.875, well above the minimum standard for conducting factor analysis . Therefore, we determined that the input data were appropriate for EFA. Subsequently, the rotation of principal components was performed using both orthogonal rotation (varimax) and oblique rotation (promax) in an attempt to achieve a simple structure. Given the fact that an oblique rotation is generally recommended by measurement specialists to facilitate the emergence of factor intercorrelations [ - ], almost all social sciences measurements exhibit some degree of correlation . In addition, the correlation matrix for the factors with oblique (promax) rotation indicated that the highest correlation was 0.445 (Table S3 in ); we thereby determined that the factors were correlated, and hence, oblique rotation was an appropriate approach. The results of parallel analysis (Table S4 in ) and PCA with or without oblique (promax) rotation all recommended the retention of 7 factors. According to the previous criteria, 2-item factors were identified, including factor 4 (PBR1 and PBR2) and factor 6 (S1 and S2). The internal consistency of seven factors demonstrated Cronbach α of 0.884 with 95% CI (0.875-0.894), which met acceptable thresholds. Within the context and theoretical framework of the STAM and the UTAUT [ , , , ], social factors significantly influence behavioral intentions to use technology, particularly in the use of mHealth. Perceived barriers also play a role in determining intentions to use mHealth, as demonstrated by the aforementioned scoping review . The inclusion of factor 4 and factor 6, which represented perceived barriers and social relationships, was considered appropriate. Based on the priori criteria and consensus of the panel experts, the EFA identified 22 candidate items (ATT1, ATT2, PU1, PU2, PU3, PEOU1, PEOU2, PB1, PB2, ANX1, ANX2, FC2, FC4, FC5, H1, H2, H5, C2, C3, C4, S1, and S2) with factor loadings greater than 0.4 that encompassed the 7 factors. The final EFA result is presented in .
From the EFA, the 22 items of the 7-factor Thai mHealth STAM explained 91.45% of the variance. The unidimensionality of each factor (subscale) and the overall models were assessed by analyzing modification indices in the CFA. Of the 7 factors from the EFA, the CFA of each factor (subscale) showed that only 5 factors, consisting of cognitive ability (C2, C3, C4), perceived barriers (PB1 and PB2), facilitating conditions (FC2, FC4, and FC5), self-reported health conditions (H1, H2, and H5), and social relationships (S1 and S2), showed satisfactory information criteria indices of the CFA, as presented . Factor 1, which included items from the attitude toward using (ATT1 and ATT2), perceived usefulness (PU1, PU2, and PU3), and perceived ease of use (PEOU1 and PEOU2), did not meet the CFA criteria due to over-factoring issues (Table S5 in ). Factor 2, combining perceived barriers (PB3 and PB4) and gerontechnology anxiety (ANX1 and ANX2), was unfit according to CFA criteria, with a low CFI (0.799), low TLI (0.698), and high RMSEA (0.306, 90% CI 0.293-0.319). Attempts to combine subscales also did not meet CFA criteria (Table S5 in ). However, when items were separated as in the original STAM, including attitude toward using (ATT1 and ATT2), perceived usefulness (PU1, PU2, and PU3), perceived ease of use (PEOU1 and PEOU2), and gerontechnology anxiety (ANX1 and ANX2), these separated factors showed a good fit with CFA criteria (Table S5 in ). Out of the 9 factors from the single latent factor analysis, 5 were 2-item factors. These were kept in the final CFA model because 3 factors (attitude toward using, perceived ease of use, and gerontechnology anxiety) were originally designed as 2-item factors, similar to the original STAM. The perceived barriers and social relationships were also retained because of their contextual relevance, as described above. Finally, the CFA confirmed 9-dimensional sets of 22 items with satisfactory fit indices, as shown in . The details of the CFAs of evaluated and reevaluated models are described in Table S4 in . A nonparametric IRT analysis also affirmed the unidimensionality, local independence, and monotonicity of the 22-item model with 8 factors (Table S6 in ). For the scalability, all 22 items of the Thai mHealth STAM had H s coefficients over 0.4, which indicates medium to strong scalability properties (Table S6 in ). The correlation among the final 22-item Thai mHealth STAM subscales ranged from 0.040 to 0.685 (Table S7 in ). The final 22-item Thai mHealth STAM questions, along with the English version and modeling indices, are described in . Each item is scored on a 10-point Likert scale from 1 (very unsatisfied or strongly disagree) to 10 (very satisfied or strongly agree), with reverse scaling for perceived barriers and gerontechnology anxiety.
Considering the absence of a reference standard, it is theoretically reasonable that more participants with higher STAM scores will result in greater acceptance and adoption of technology. The discriminative indices, including sensitivity, specificity, and AUROC, were used to determine the cutoff scores for the proposed questionnaire, considering the intention to use mHealth from the external question. The 6 proposed sets of the final 22-item Thai mHealth STAM bands were classified into low, moderate, and high acceptance, as presented in . The set D of the possible banding was preferred as the optimal 22-item Thai mHealth STAM cutoff score based on the highest sensitivity of 89% (95% CI 86.1%-91.5%) and AUROC of 72.4% (95% CI 70%-74.8%). This finding also confirmed the discrimination performance of the 22-item Thai mHealth STAM in identifying persons with and without the intention to use mHealth. For set D, low, moderate, and high scores are defined as ≤151, 152-180, and ≥181, respectively. In addition, we conducted a subgroup analysis based on age groups: pre-older adults (aged 45-59 years) and older adults (aged 60 years and older). The result revealed that the set D banding had robust discriminant validity in older adults (AUROC 73%, 95% CI 70%-76%), but the discriminant validity decreased in the pre-older adult group (AUROC 67.7%, 95% CI 63.3%-71.9%). The discriminant validity of the 22-item Thai mHealth STAM by the subpopulation cohorts is shown in Table S8 in .
Out of 1100 overall participants, the final 22-item Thai mHealth STAM demonstrated an excellent internal consistency in both the Cronbach α (0.88, 95% CI 0.87-0.89) and the McDonald ω coefficients (0.85, 95% CI 0.83-0.87), as shown in . By subpopulation, the Cronbach α and the McDonald ω coefficients were 0.88 (95% CI 0.86-0.90) and 0.84 (95% CI 0.81- 0.89) for adults aged 45-59 years and 0.88 (95% CI 0.86- 0.89) and 0.83 (95% CI 0.81-0.86) for older adults. All 22 items revealed the corrected item-total correlations ranging from 0.26 to 0.71, achieving a level of acceptance between 0.20 and 0.80 ( ).
Principal Findings The study aimed to adapt and validate the STAM questionnaire for assessing mHealth technology acceptance among pre-older and older populations regarding the use of health support. The results confirmed the scale’s factor structure, supported an 8-factor model with 22 items, and showed good discriminant validity in predicting mHealth intention. The optimal version was a 22-item Thai mHealth STAM using the scoring cutoff (≥152). Subgroup analysis indicated no significant difference in discriminant validity between pre-older and older adults. The scale demonstrated strong internal consistency and stability, with reliability confirmed by Cronbach α and McDonald ω coefficients. This adapted 22-item version is more relevant for assessing mHealth intention among older adults and is suitable for public surveys and routine practice, which take less than 15 minutes to complete. Our findings are consistent with the previous study conducted by the owner of the original STAM , which was subsequently developed into a brief form to save administration time and reduce the burden on respondents. The 14-item brief version of the STAM questionnaire consisted of a 4-factor structure: attitudinal beliefs, control beliefs, gerontechnology anxiety, and health. These findings are consistent with ours, reflecting the original STAM model constructs and the age-related health characteristics of older adults. We observed a decrease in discriminant validity within the pre-older adult group, indicating a need for additional factors to explain their behavioral intentions. For example, older adults with different genders, education levels, income, marital status, and ethnicity may have different intentions and purposes to use mHealth for their health . Strengths and Limitations On the strength side, this is the first Thai version of the STAM questionnaire suitable for evaluating technology acceptance in Thai older adults. The 22-item Thai STAM version demonstrates structural balance, reliability, and validity in assessing technology acceptance among older individuals. The evaluation process is time-efficient. In addition, this tool can be used with both pre-older adults and older adults to prepare them for engaging with technology in their future lives. Furthermore, the evaluation takes into account the influence of Thai cultural norms on the adoption and acceptance of mHealth. However, there are some limitations to consider. Although the psychometric properties of the 22-item Thai mHealth STAM are satisfied through transcultural adaptation in terms of validity and reliability in both the pre-older and older populations, this scale can be applied for use in a broad. However, our study participants may not be representative of the overall Thai pre-older and older populations, as almost all of the participants lived in the northern part of Thailand, particularly in Chiang Mai province. In order to address this concern, future studies, including those based on different regions of Thailand and other specific populations (eg, teenagers, vulnerable groups, minorities, and specific groups of patients) that could potentially derive advantages from mHealth usage, are recommended to expand the generalizability and usability of this scale. Finally, the 22-item Thai mHealth STAM was evaluated based on the board’s definition of mHealth. It is possible that the proposed questionnaire may not be compatible with all of the existing mHealth technologies due to the diverse range of mHealth technologies in health care. The patient’s choice may vary depending on several factors, such as health care providers, types of services, or the specific application. Hence, we suggest using this questionnaire to assess their acceptance and intention to use it in conjunction with the designated mHealth technology. Practical Implications of the 22-Item Thai mHealth STAM The 22-item Thai mHealth STAM offers a practical assessment of patients’ acceptability—a crucial factor often overlooked, as evidenced by a recent systematic review of technology acceptability in health care, which revealed that only 10% (142/1219) of the reviewed studies examined patient acceptance . This publicly available questionnaire has the potential to support health care professionals, policy makers, and developers in making informed decisions , particularly regarding the adoption and acceptance of mHealth within Thai cultural norms. This questionnaire can be incorporated into the research and development (R&D) processes of mHealth and used as a questionnaire to define the target population based on levels of acceptability, as well as ascertain the factors that encourage or hinder the adoption of their mHealth technologies . This information is important for informing stakeholders and developers in advance of the mHealth R&D and implementation stages, which necessitate user data for resource allocation and planning in consideration of user requirements and experiences . Conclusion The increasing number of older people, along with their growing adoption of technology, indicates that mHealth technologies might offer a new approach to enhancing the health of older adults with lower health care expenses. Although there are many advantages to using mHealth apps, it is important to consider their acceptance and intention to use them for health-related objectives. We proposed the 22-item Thai mHealth STAM as the questionnaire to evaluate the levels of acceptability and intention to use mHealth in the Thai community of pre-older and older adults. The 22-item Thai mHealth STAM has demonstrated satisfactory psychometric properties in terms of validity and reliability. As a result, it is now feasible to use this questionnaire in a public survey to support stakeholders in making informed decisions. Nevertheless, to improve generalizability and long-term use, further study is needed to investigate the various demographic groups with the specific mHealth interventions.
The study aimed to adapt and validate the STAM questionnaire for assessing mHealth technology acceptance among pre-older and older populations regarding the use of health support. The results confirmed the scale’s factor structure, supported an 8-factor model with 22 items, and showed good discriminant validity in predicting mHealth intention. The optimal version was a 22-item Thai mHealth STAM using the scoring cutoff (≥152). Subgroup analysis indicated no significant difference in discriminant validity between pre-older and older adults. The scale demonstrated strong internal consistency and stability, with reliability confirmed by Cronbach α and McDonald ω coefficients. This adapted 22-item version is more relevant for assessing mHealth intention among older adults and is suitable for public surveys and routine practice, which take less than 15 minutes to complete. Our findings are consistent with the previous study conducted by the owner of the original STAM , which was subsequently developed into a brief form to save administration time and reduce the burden on respondents. The 14-item brief version of the STAM questionnaire consisted of a 4-factor structure: attitudinal beliefs, control beliefs, gerontechnology anxiety, and health. These findings are consistent with ours, reflecting the original STAM model constructs and the age-related health characteristics of older adults. We observed a decrease in discriminant validity within the pre-older adult group, indicating a need for additional factors to explain their behavioral intentions. For example, older adults with different genders, education levels, income, marital status, and ethnicity may have different intentions and purposes to use mHealth for their health .
On the strength side, this is the first Thai version of the STAM questionnaire suitable for evaluating technology acceptance in Thai older adults. The 22-item Thai STAM version demonstrates structural balance, reliability, and validity in assessing technology acceptance among older individuals. The evaluation process is time-efficient. In addition, this tool can be used with both pre-older adults and older adults to prepare them for engaging with technology in their future lives. Furthermore, the evaluation takes into account the influence of Thai cultural norms on the adoption and acceptance of mHealth. However, there are some limitations to consider. Although the psychometric properties of the 22-item Thai mHealth STAM are satisfied through transcultural adaptation in terms of validity and reliability in both the pre-older and older populations, this scale can be applied for use in a broad. However, our study participants may not be representative of the overall Thai pre-older and older populations, as almost all of the participants lived in the northern part of Thailand, particularly in Chiang Mai province. In order to address this concern, future studies, including those based on different regions of Thailand and other specific populations (eg, teenagers, vulnerable groups, minorities, and specific groups of patients) that could potentially derive advantages from mHealth usage, are recommended to expand the generalizability and usability of this scale. Finally, the 22-item Thai mHealth STAM was evaluated based on the board’s definition of mHealth. It is possible that the proposed questionnaire may not be compatible with all of the existing mHealth technologies due to the diverse range of mHealth technologies in health care. The patient’s choice may vary depending on several factors, such as health care providers, types of services, or the specific application. Hence, we suggest using this questionnaire to assess their acceptance and intention to use it in conjunction with the designated mHealth technology.
The 22-item Thai mHealth STAM offers a practical assessment of patients’ acceptability—a crucial factor often overlooked, as evidenced by a recent systematic review of technology acceptability in health care, which revealed that only 10% (142/1219) of the reviewed studies examined patient acceptance . This publicly available questionnaire has the potential to support health care professionals, policy makers, and developers in making informed decisions , particularly regarding the adoption and acceptance of mHealth within Thai cultural norms. This questionnaire can be incorporated into the research and development (R&D) processes of mHealth and used as a questionnaire to define the target population based on levels of acceptability, as well as ascertain the factors that encourage or hinder the adoption of their mHealth technologies . This information is important for informing stakeholders and developers in advance of the mHealth R&D and implementation stages, which necessitate user data for resource allocation and planning in consideration of user requirements and experiences .
The increasing number of older people, along with their growing adoption of technology, indicates that mHealth technologies might offer a new approach to enhancing the health of older adults with lower health care expenses. Although there are many advantages to using mHealth apps, it is important to consider their acceptance and intention to use them for health-related objectives. We proposed the 22-item Thai mHealth STAM as the questionnaire to evaluate the levels of acceptability and intention to use mHealth in the Thai community of pre-older and older adults. The 22-item Thai mHealth STAM has demonstrated satisfactory psychometric properties in terms of validity and reliability. As a result, it is now feasible to use this questionnaire in a public survey to support stakeholders in making informed decisions. Nevertheless, to improve generalizability and long-term use, further study is needed to investigate the various demographic groups with the specific mHealth interventions.
|
A highly multiplexed broad pathogen detection assay for infectious disease diagnostics | 95e10f21-f50d-459e-b88d-0d7f4d83e5a4 | 6245831 | Pathology[mh] | Appropriate diagnostic assay selection for infectious diseases depends on multiple parameters including clinical presentation and endemic pathogens known to circulate within a specific geographic region. Rapid point-of-care PCR and lateral flow immunoassays as well as more complex PCR and laboratory based antigen capture ELISAs can generate a clinically actionable diagnosis in patients presenting with an acute febrile illness. These assays are sensitive, rapid, and relatively inexpensive, making this testing approach ideal for initial diagnostic testing. If these assays are negative, however, additional testing including increasingly multiplexed assays and agnostic next-generation sequencing can be utilized. Multiplexed assays such as the MAGPIX or multiplexed real-time PCR can increase the number of targets being tested. For example, Munro and colleagues described a multiplexed PCR assay with detection on the MAGPIX or Luminex instruments capable of detecting multiple influenza viruses with performance similar to real-time PCR . Similarly, a multiplexed real-time RT-PCR assay, developed by Santiago and colleagues and approved by the FDA as an in vitro diagnostic device, detects all four dengue virus serotypes in a single tube reaction . In cases where the initial testing methods do not result in positive pathogen identification, next-generation sequencing (NGS) is another alternative for clinically actionable infectious disease diagnostics . However, metagenomic sequencing can be challenging due to a large host background, necessitating high sequencing depth to generate sufficient on target reads for pathogen detection. Targeted NGS, in which a specific signature is amplified or enriched from a complex sample using hybridization , can increase pathogen specific reads sufficiently to allow detection on desktop sequencers such as the Ion Torrent or the MiSeq. Using these approaches, however, adds time-to-answer due to library preparation, sequencing, and analysis. A potential solution described here is the use of the NanoString nCounter platform for highly multiplexed pathogen detection. This system utilizes direct hybridization and detection of a nucleic acid target and can be highly multiplexed (up to 800 different targets). Since this technology has been successfully implemented for quantitative gene expression studies , we investigated whether this platform could be used for broad, targeted pathogen detection in a situation where rapid testing (ex. real-time PCR) was negative. In this context, we developed and evaluated a panel containing 195 different assay targets against 164 different viruses, bacteria and parasites. Overall, this panel was not as sensitive as real-time PCR; however, this assay successfully identified multiple pathogens quickly, demonstrating utility as a pathogen screening assay.
Viruses, parasites, and bacteria All organisms used in this study (listed in ) are maintained at United States Army Medical Research Institute of Infectious Diseases (USAMRIID) or were provided by the Unified Culture Collection (UCC) or the American Type Culture Collection (ATCC, Manassas, VA). Samples included bacterial, parasite DNA, cell culture supernatant from virus-infected cells treated with TRIzol LS (ThermoFisher Scientific, Waltham, MA) or gamma irradiation. Total nucleic acid from each unpurified sample was extracted using the EZ1 Virus Mini Kit v2.0 (Qiagen, Valencia, CA) with the EZ1 robot (Qiagen) according to the manufacturer’s instructions. Total nucleic acid was eluted in 90 μl elution buffer. Due to a limited supply, Coxiella burnetii DNA was amplified using the REPLI-g Whole Genome Amplification Kit (Qiagen) according to the manufacturer’s instructions. The number of C . burnetii genome equivalents (GE) was approximated using the genome of C . burnetii RSA493 (GenBank# NC_002971) and the C+G (42.7%) and A+T (57.3%) genome percentages. Based on these calculations, 1 GE is approximately 2.05 fg. The approximate number of GE for Plasmodium falciparum 3D7 DNA (ATCC) was similarly determined to be approximately 23.89 fg. NanoString broad pathogen panel A custom Broad Pathogen Detection Assay (BPDA) targeting a broad panel of medically important viruses, bacteria, and parasites was designed and acquired from NanoString Technologies (Seattle, WA). Using 195 different capture and reporter probes, this assay targeted 164 different pathogens of concern for human health . After initial testing showed lower than desired assay sensitivity, nested primers targets were designed by NanoString using Primer3 software ; see for the sequences. These multiplexed primers were used in a multiplexed target enrichment (MTE) reaction to amplify the capture/reporter target prior to detection. Primer pairs for 4 probe targets could not initially be designed and were redesigned for incorporation into a subsequent MTE iteration. Individual MTE primer pairs for all available pathogens were evaluated for amplicon generation using SuperScript One-Step with Platinum taq (Thermo Fisher Scientific) with the following cycling conditions: 50°C for 15 minutes, 95°C for 5 minutes, 40 cycles of 95°C for 30 seconds, 60°C for 1 minute and 72°C for 1 minute. The final reagent concentrations per 20 μL reaction were: 1X Reaction Mix, 4 mM MgSO 4 , 0.25 mg/mL BSA, 50 μM primers, 0.4 units of Platinum Taq . Amplicon generation was visualized on the 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) using the DNA 1000 Kit (Agilent Technologies). After confirming successful amplification using selected individual primer pairs, all primers were combined into a single 500 nM primer mixture for the MTE reaction. Sample cDNA was generated by adding 4 μL purified total nucleic acid to 1μL of SuperScript VILO MasterMix (Thermo Fisher Scientific) and incubating at 25°C for 10 minutes, 42°C for 60 minutes, and 5 minutes at 85°C. The sample was then added to 5 μL of TaqMan PreAmp Master Mix (Thermo Fisher Scientific) and 1 μL of the 500 nM primer mixture. MTE used the following cycling conditions: 94°C for 10 minutes, then 18 cycles of 94°C for 15 seconds, 60°C for 4 minutes, and a 4°C hold. The entire enriched sample (11 μL) was used for detection on the NanoString nCounter platform. Pathogen nucleic acid (total nucleic acid and MTE amplified nucleic acid) was initially denatured at 95°C for 5 minutes and immediately transferred to ice for 2 minutes. Next, 5 μl of each denatured sample or the entire MTE reaction (11 μl) was added to a 20 μl NanoString master mix containing the BPDA Reporter Codeset plus 130 μL hybridization buffer followed by 5 μl of the BPDA Capture Codeset. Reactions were immediately placed on a thermocycler for an overnight incubation at 65°C (for ~16 hours), loaded onto a sample cartridge using the nCounter Prep Station, and scanned using the NanoString nCounter Digital Analyzer. A sample was called positive for a specific target if the number of counts was greater than the average of the internal negative controls for that target plus three times the standard deviation of the negative controls. MTE amplification assessment The impact of the MTE reaction on sensitivity was assessed using SYBR Green real-time PCR assays using primers internal to the MTE primers [See for Bundibugyo virus (BDBV), Marburg virus (MARV), Ebola virus (EBOV), influenza B virus, and P . falciparum assay information]. Total nucleic acid for each organism was serially diluted and amplified by MTE. The levels of enrichment were measured by real-time RT-PCR using Superscript RT-PCR reagents (Thermo Fisher Scientific) and SYBR Green. The cycling conditions for the SYBR Green RT-PCR assays were: 50°C for 15 minutes, 95°C for 5 minutes, 45 cycles of 95°C for 5 seconds and 60°C for 20 seconds, and a final melt step from 60°C—95°C at a rate of 0.2°C per second. The final reagent concentrations per 20 μL reaction were as follows: 1X Reaction Mix, 3mM MgSO 4 , 0.25 mg/mL BSA, 1 μM primers, 2X SYBR Green, and 1 unit of Platinum Taq . Assay limit of detection and reproducibility Following optimization and characterization of the MTE reaction, assay performance was determined utilizing the MTE reaction and detection on the NanoString platform. A preliminary limit of detection (LOD) for BDBV, MARV, EBOV, influenza B virus, and P . falciparum was determined with and without MTE by serially diluting organism and testing for positive detection. The preliminary LOD was defined as the lowest concentration of organism having all three replicates testing positive. Testing at the preliminary LOD was repeated (ten replicates) without MTE for BDBV and MARV Angola and with MTE for influenza B virus, BDBV, and MARV Angola to show assay reproducibility. Similarly, LODs were conducted using existing real-time RT-PCR assays as previously described . Clinical samples Mock clinical samples were prepared in order to evaluate the ability of the BPDA to detect samples that had been extracted from whole human blood. Dengue virus serotype 3 (DENV-3) in TRIzol LS and gamma-irradiated EBOV were diluted in human whole blood treated with EDTA (Bioreclamation, Westbury, NY) in 200 μl samples. Samples were extracted using the Qiagen EZ1 XL Advanced with the EZ1 Virus Mini kit 2.0 by adding an equal volume of ATL buffer (Qiagen) to each sample prior to being placed on the automated instrument for extraction. Each sample was run in triplicate with and without MTE amplification prior to being tested with the pathogen panel. Ethics statement Since the BPDA assay could be a useful diagnostic tool, de-identified human clinical samples were tested. These samples were acquired through USAMRIID’s Special Pathogens Laboratory. All samples were de-identified prior to use, and all studies were conducted in compliance with United States Department of Defense, federal, and state statutes and regulations relating to the protection of human subjects, and adheres to principles identified in the Belmont Report. All data and human subjects research were gathered and conducted for this publication under an Institutional Review Board approved determination FY17-31 as defined by 32 CFR 219.102(f).This sample set comprised of Chikungunya virus positive and negative samples, as determined by real-time PCR . Total nucleic acid was extracted from each sample using the Qiagen EZ1 XL Advanced with the EZ1 Virus Mini kit 2.0 and tested using the BPDA.
All organisms used in this study (listed in ) are maintained at United States Army Medical Research Institute of Infectious Diseases (USAMRIID) or were provided by the Unified Culture Collection (UCC) or the American Type Culture Collection (ATCC, Manassas, VA). Samples included bacterial, parasite DNA, cell culture supernatant from virus-infected cells treated with TRIzol LS (ThermoFisher Scientific, Waltham, MA) or gamma irradiation. Total nucleic acid from each unpurified sample was extracted using the EZ1 Virus Mini Kit v2.0 (Qiagen, Valencia, CA) with the EZ1 robot (Qiagen) according to the manufacturer’s instructions. Total nucleic acid was eluted in 90 μl elution buffer. Due to a limited supply, Coxiella burnetii DNA was amplified using the REPLI-g Whole Genome Amplification Kit (Qiagen) according to the manufacturer’s instructions. The number of C . burnetii genome equivalents (GE) was approximated using the genome of C . burnetii RSA493 (GenBank# NC_002971) and the C+G (42.7%) and A+T (57.3%) genome percentages. Based on these calculations, 1 GE is approximately 2.05 fg. The approximate number of GE for Plasmodium falciparum 3D7 DNA (ATCC) was similarly determined to be approximately 23.89 fg.
A custom Broad Pathogen Detection Assay (BPDA) targeting a broad panel of medically important viruses, bacteria, and parasites was designed and acquired from NanoString Technologies (Seattle, WA). Using 195 different capture and reporter probes, this assay targeted 164 different pathogens of concern for human health . After initial testing showed lower than desired assay sensitivity, nested primers targets were designed by NanoString using Primer3 software ; see for the sequences. These multiplexed primers were used in a multiplexed target enrichment (MTE) reaction to amplify the capture/reporter target prior to detection. Primer pairs for 4 probe targets could not initially be designed and were redesigned for incorporation into a subsequent MTE iteration. Individual MTE primer pairs for all available pathogens were evaluated for amplicon generation using SuperScript One-Step with Platinum taq (Thermo Fisher Scientific) with the following cycling conditions: 50°C for 15 minutes, 95°C for 5 minutes, 40 cycles of 95°C for 30 seconds, 60°C for 1 minute and 72°C for 1 minute. The final reagent concentrations per 20 μL reaction were: 1X Reaction Mix, 4 mM MgSO 4 , 0.25 mg/mL BSA, 50 μM primers, 0.4 units of Platinum Taq . Amplicon generation was visualized on the 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) using the DNA 1000 Kit (Agilent Technologies). After confirming successful amplification using selected individual primer pairs, all primers were combined into a single 500 nM primer mixture for the MTE reaction. Sample cDNA was generated by adding 4 μL purified total nucleic acid to 1μL of SuperScript VILO MasterMix (Thermo Fisher Scientific) and incubating at 25°C for 10 minutes, 42°C for 60 minutes, and 5 minutes at 85°C. The sample was then added to 5 μL of TaqMan PreAmp Master Mix (Thermo Fisher Scientific) and 1 μL of the 500 nM primer mixture. MTE used the following cycling conditions: 94°C for 10 minutes, then 18 cycles of 94°C for 15 seconds, 60°C for 4 minutes, and a 4°C hold. The entire enriched sample (11 μL) was used for detection on the NanoString nCounter platform. Pathogen nucleic acid (total nucleic acid and MTE amplified nucleic acid) was initially denatured at 95°C for 5 minutes and immediately transferred to ice for 2 minutes. Next, 5 μl of each denatured sample or the entire MTE reaction (11 μl) was added to a 20 μl NanoString master mix containing the BPDA Reporter Codeset plus 130 μL hybridization buffer followed by 5 μl of the BPDA Capture Codeset. Reactions were immediately placed on a thermocycler for an overnight incubation at 65°C (for ~16 hours), loaded onto a sample cartridge using the nCounter Prep Station, and scanned using the NanoString nCounter Digital Analyzer. A sample was called positive for a specific target if the number of counts was greater than the average of the internal negative controls for that target plus three times the standard deviation of the negative controls.
The impact of the MTE reaction on sensitivity was assessed using SYBR Green real-time PCR assays using primers internal to the MTE primers [See for Bundibugyo virus (BDBV), Marburg virus (MARV), Ebola virus (EBOV), influenza B virus, and P . falciparum assay information]. Total nucleic acid for each organism was serially diluted and amplified by MTE. The levels of enrichment were measured by real-time RT-PCR using Superscript RT-PCR reagents (Thermo Fisher Scientific) and SYBR Green. The cycling conditions for the SYBR Green RT-PCR assays were: 50°C for 15 minutes, 95°C for 5 minutes, 45 cycles of 95°C for 5 seconds and 60°C for 20 seconds, and a final melt step from 60°C—95°C at a rate of 0.2°C per second. The final reagent concentrations per 20 μL reaction were as follows: 1X Reaction Mix, 3mM MgSO 4 , 0.25 mg/mL BSA, 1 μM primers, 2X SYBR Green, and 1 unit of Platinum Taq .
Following optimization and characterization of the MTE reaction, assay performance was determined utilizing the MTE reaction and detection on the NanoString platform. A preliminary limit of detection (LOD) for BDBV, MARV, EBOV, influenza B virus, and P . falciparum was determined with and without MTE by serially diluting organism and testing for positive detection. The preliminary LOD was defined as the lowest concentration of organism having all three replicates testing positive. Testing at the preliminary LOD was repeated (ten replicates) without MTE for BDBV and MARV Angola and with MTE for influenza B virus, BDBV, and MARV Angola to show assay reproducibility. Similarly, LODs were conducted using existing real-time RT-PCR assays as previously described .
Mock clinical samples were prepared in order to evaluate the ability of the BPDA to detect samples that had been extracted from whole human blood. Dengue virus serotype 3 (DENV-3) in TRIzol LS and gamma-irradiated EBOV were diluted in human whole blood treated with EDTA (Bioreclamation, Westbury, NY) in 200 μl samples. Samples were extracted using the Qiagen EZ1 XL Advanced with the EZ1 Virus Mini kit 2.0 by adding an equal volume of ATL buffer (Qiagen) to each sample prior to being placed on the automated instrument for extraction. Each sample was run in triplicate with and without MTE amplification prior to being tested with the pathogen panel.
Since the BPDA assay could be a useful diagnostic tool, de-identified human clinical samples were tested. These samples were acquired through USAMRIID’s Special Pathogens Laboratory. All samples were de-identified prior to use, and all studies were conducted in compliance with United States Department of Defense, federal, and state statutes and regulations relating to the protection of human subjects, and adheres to principles identified in the Belmont Report. All data and human subjects research were gathered and conducted for this publication under an Institutional Review Board approved determination FY17-31 as defined by 32 CFR 219.102(f).This sample set comprised of Chikungunya virus positive and negative samples, as determined by real-time PCR . Total nucleic acid was extracted from each sample using the Qiagen EZ1 XL Advanced with the EZ1 Virus Mini kit 2.0 and tested using the BPDA.
Preliminary assay evaluation We developed a Broad Pathogen Detection Assay (BPDA) for use with the NanoString nCounter platform in order to quickly screen a sample for multiple pathogens in a single tube reaction. This assay consisted of 195 detection probes targeting 164 different viral, bacterial, and parasitic pathogens of concern for human health. Initial testing using the highest concentration of organism available showed positive detection of multiple pathogens including EBOV, MARV, P . falciparum , and C . burnetii (see for a selected list and the for a full detection list). However, some pathogens such as Crimean-Congo hemorrhagic fever virus (CCHFV) were not detected even at this high concentration . A multiplexed target enrichment (MTE) step utilizing a complex PCR to amplify the pathogen-specific probe hybridization site was used for use prior to detection with the BPDA to mitigate this issue. Incorporating this upfront target enrichment step increased the assay sensitivity as shown by the now positive detection of CCHFV and increased read counts for almost all pathogens tested . MTE target enrichment Having shown the effectiveness of the MTE step for increasing assay sensitivity, we wanted to further assess the target enrichment capability of this method by comparing the amount of target amplicon present with and without MTE. Comparing real-time PCR results for the MTE amplified and non-amplified reactions, MTE enrichment showed a decrease in Cq values, indicating an increase in the target amplicon . Statistical analysis (two-way ANOVA with Bonferroni correction) identified that all points for each virus were significantly different with the exception of the highest influenza B virus concentration. Analytical performance of the BPDA Assay limit of detection (LOD) studies were conducted with serially diluted organism in order to define assay performance for medical relevant concentrations of the tested organism. Incorporating the MTE enrichment improved detection and lowered LODs . This improvement was most notable for influenza B virus which was undetectable without enrichment but tested positive following MTE . Similarly, the preliminary LOD for MARV improved from 2.2 x 10 6 PFU/ml without MTE to 3.75 x 10 4 PFU/ml after enrichment . Generally, the assays were highly specific for the targeted organism. For example, BPDA showed positive results for only P . falciparum while other Plasmodium species including knowlesi , malariae , ovale , and vivax were called as true negatives. To confirm assay reproducibility at the preliminary LOD, ten replicates of BDBV, influenza B virus, and MARV were tested with and without MTE . For influenza B virus, no replicates tested positive at the highest concentration used; however, MTE incorporation resulted in repeated detection of all viruses. In addition, comparison of BPDA performance to real-time PCR, the current gold standard for molecular based pathogen detection, showed real-time RT-PCR was the more sensitive technique . Clinical sample testing Testing EBOV and DENV-3 spiked into whole blood at three different concentrations showed the clinical applicability of this assay. Application of the MTE did not impact the overall testing results, but MTE increased the number of EBOV-specific counts for all dilutions and replicates . DENV-3 tested positive with all replicates and dilutions; however, MTE did not result in increased DENV-3 counts . Interestingly, the pre-amplification of DENV-3 signatures resulted in a lower number of counts as compared to the same samples without MTE, potentially suggesting suboptimal amplification for that DENV-3 isolate. Further characterization of the clinical utility of the BPDA showed positive detection across 14 de-identified, human clinical samples with potential Chikungunya virus (CHIKV) infections . Both the BPDA and the MTE-BPDA assays correctly identified the 12 real-time RT-PCR positive samples and the two negative samples . Incorporating the MTE component increased the number of CHIKV counts by 1–2 log for all of the positive samples tested.
We developed a Broad Pathogen Detection Assay (BPDA) for use with the NanoString nCounter platform in order to quickly screen a sample for multiple pathogens in a single tube reaction. This assay consisted of 195 detection probes targeting 164 different viral, bacterial, and parasitic pathogens of concern for human health. Initial testing using the highest concentration of organism available showed positive detection of multiple pathogens including EBOV, MARV, P . falciparum , and C . burnetii (see for a selected list and the for a full detection list). However, some pathogens such as Crimean-Congo hemorrhagic fever virus (CCHFV) were not detected even at this high concentration . A multiplexed target enrichment (MTE) step utilizing a complex PCR to amplify the pathogen-specific probe hybridization site was used for use prior to detection with the BPDA to mitigate this issue. Incorporating this upfront target enrichment step increased the assay sensitivity as shown by the now positive detection of CCHFV and increased read counts for almost all pathogens tested .
Having shown the effectiveness of the MTE step for increasing assay sensitivity, we wanted to further assess the target enrichment capability of this method by comparing the amount of target amplicon present with and without MTE. Comparing real-time PCR results for the MTE amplified and non-amplified reactions, MTE enrichment showed a decrease in Cq values, indicating an increase in the target amplicon . Statistical analysis (two-way ANOVA with Bonferroni correction) identified that all points for each virus were significantly different with the exception of the highest influenza B virus concentration.
Assay limit of detection (LOD) studies were conducted with serially diluted organism in order to define assay performance for medical relevant concentrations of the tested organism. Incorporating the MTE enrichment improved detection and lowered LODs . This improvement was most notable for influenza B virus which was undetectable without enrichment but tested positive following MTE . Similarly, the preliminary LOD for MARV improved from 2.2 x 10 6 PFU/ml without MTE to 3.75 x 10 4 PFU/ml after enrichment . Generally, the assays were highly specific for the targeted organism. For example, BPDA showed positive results for only P . falciparum while other Plasmodium species including knowlesi , malariae , ovale , and vivax were called as true negatives. To confirm assay reproducibility at the preliminary LOD, ten replicates of BDBV, influenza B virus, and MARV were tested with and without MTE . For influenza B virus, no replicates tested positive at the highest concentration used; however, MTE incorporation resulted in repeated detection of all viruses. In addition, comparison of BPDA performance to real-time PCR, the current gold standard for molecular based pathogen detection, showed real-time RT-PCR was the more sensitive technique .
Testing EBOV and DENV-3 spiked into whole blood at three different concentrations showed the clinical applicability of this assay. Application of the MTE did not impact the overall testing results, but MTE increased the number of EBOV-specific counts for all dilutions and replicates . DENV-3 tested positive with all replicates and dilutions; however, MTE did not result in increased DENV-3 counts . Interestingly, the pre-amplification of DENV-3 signatures resulted in a lower number of counts as compared to the same samples without MTE, potentially suggesting suboptimal amplification for that DENV-3 isolate. Further characterization of the clinical utility of the BPDA showed positive detection across 14 de-identified, human clinical samples with potential Chikungunya virus (CHIKV) infections . Both the BPDA and the MTE-BPDA assays correctly identified the 12 real-time RT-PCR positive samples and the two negative samples . Incorporating the MTE component increased the number of CHIKV counts by 1–2 log for all of the positive samples tested.
Positively identifying the etiologic agent for an acute febrile illness can be critical for ensuring appropriate administration of treatment and supportive care; however, proper identification can be challenging. Highlighting the importance of broad pathogen screening and appropriately fielded diagnostics, a recent study by Schoepp and colleagues found ~70% of the suspected Lassa fever patients admitted to the Lassa Fever Ward in Kenema, Sierra Leone, were negative for both Lassa virus and the malaria parasite, both hyperendemic pathogens in the region . This study found serological evidence of filovirus infection (EBOV and MARV) in the years prior to the explosive Ebola virus disease outbreak in West Africa , and it is likely that previous infections with EBOV as well as MARV were misidentified as severe malaria or Lassa fever. Accurate testing of these acute febrile patients with multiplexed assays could have identified the risk of an EBOV outbreak earlier. Here, we developed and evaluated a highly multiplexed, broad pathogen detection assay for use following negative detection using singleplex assays (ex. real-time PCR). This assay targets 164 different human pathogens of public health concern and includes viruses, bacteria, and parasites. We included multiple organisms with overlapping clinical presentation, such as Plasmodium , Lassa virus, and Ebola virus. We also included a large number of less common organisms in order to maximize the diagnostic potential of the assay. While the assay performed well without target enrichment, applying MTE prior to running on the NanoString platform greatly improved assay sensitivity. While extensive primer optimization was not conducted in this study, such an optimization would likely improve the overall assay sensitivity and the detection variation we observed. As we were unable to do this optimization, all testing was conducted with and without MTE. Overall, 98 of the 164 pathogens on the panel that we had available for testing were positively identified including endemic pathogens to West Africa such as EBOV, Lassa virus, dengue virus, and the malaria parasite P . falciparum . Assay run time, from start to finish, is approximately 27 hours with approximately one hour (or 30 minutes without the MTE) of hands-on time. Future efforts include the acquisition and testing of the remaining pathogens on the panel. Characterization of the BPDA using mock clinical and clinical samples showed the efficacy of this assay for detecting pathogen in patient samples. Specifically, testing of spiked human serum showed positive detection of EBOV and DENV-3 at clinically relevant concentrations. In addition to mock clinical samples, the assays correctly identified all CHIKV human clinical samples, demonstrating the ability to correctly identify pathogens from natural infections. These studies were in agreement with real-time RT-PCR testing establishing preliminary 2-by-2 testing that would be required for regulatory use. There are a variety of easy to use multiplexed assays described in the literature ; however, there are inherent limitations in multiplexability within a single assay. Other technologies offer higher levels of multiplexing through microarray and next-generation sequencing . However, these assays are highly technical, and the large number of targets makes full validation of each signature highly challenging in both cost and time. Furthermore, clinical validation of these assays would be further complicated by the regulatory requirement to validate each potential organism the assay could detect . Preliminary limit of detection (5 dilutions in triplicate for 164 organisms) and confirmation of the preliminary limit of detection (twenty replicates of 164 organisms) testing alone would require 2,460 reactions at approximately $100 USD/reaction (~$575,000 USD). Mock clinical testing would further expand the testing numbers and cost. Ideally, comparison of primer/primer interactions and a direct comparison of each target to a gold standard (ex. real-time PCR) would be performed for diagnostic applications. Within current regulatory paradigm, this type of validation also remains cost prohibitive similar to other highly multiplexed assays (ex. microarray and next-generation sequencing assays) as this would require independent testing of each target on the panel in a statistically robust manner. However, results presented here provide proof of concept testing results and the framework for such a validation by demonstrating the proof of concept utilization of this technology for infectious disease diagnostics.
S1 File The Supporting Information File includes tables containing: 1) all organisms and strains tested and the BPDA test results; 2) primers utilized in the MTE reaction; and 3) the nested real-time PCR primers. (XLSX) Click here for additional data file.
|
Distinct spatiotemporal patterns between fungal alpha and beta diversity of soil–plant continuum in rubber tree | d5efdc93-621b-404b-b386-c506d8c41bd6 | 11792516 | Microbiology[mh] | Microbial organisms inhabit all biomes of the Earth , and provide a number of life-support functions for their host . Therefore, we must develop a better understanding of the distribution and ecological drivers of aboveground and belowground microbial communities. Recent studies have demonstrated the immense role of plant compartments and environmental factors in driving the assembly of microbiomes . Geographic location and seasonal change have been demonstrated to influence community composition. For example, it was suggested that the assembly of the phyllosphere bacterial and fungal communities is predominantly determined by host compartment (epiphytic and endophytic) and site location . As for soil and rhizosphere, microbiomes are influenced by environmental factors (e.g., site, soil properties, and climate) . However, these studies mainly focused on a single niche or compartment, and a significant knowledge gap exists on how spatial heterogeneity versus time shapes the diversity and structure of microbial communities along the soil–plant continuum. Different scales have varying impacts on plant microorganisms . Moreover, these examples have shown that soil microbial communities are influenced by spatial or temporal change, but understanding of how seasonal changes (e.g., dry and rainy seasons) affect the compositions and diversity of soil–plant continuum microbial communities at the regional scale is still limited. Rubber plantation is the most economically important agro-ecosystem in tropical China, particularly in Hainan Island and Xishuanbanna (abbreviated as Banna below) , which accounts for more than 90% of the total rubber plantation area of China . It is reported rubber plantations have multiplied quickly throughout Southeast Asia over the last two decades . As far as we know, rubber forests account for almost 25% and 40% of the total vegetation area in Hainan Island and Banna , respectively. Previous work in Hainan has shown that seasonal change and site location were the dominant factors resulting in shifts in soil microbial composition at the local and geographic scales, respectively . However, these studies were limited, particularly in the sample size scales used. Given the importance of microbes in functional roles in tropical forest ecosystem, such as nutrient acquisition, disease resistance, and stress tolerance , and the central part of rubber plantation of terrestrial ecosystems both in Hainan and Banna. Moreover, little attention has been paid to the plant-associated microbial communities on the same scale. As a result, it is difficult to describe the overall pattern of the soil–plant continuum. Therefore, it is necessary to study the spatiotemporal pattern and ecological drivers of the rubber tree soil–plant continuum microbial communities of these two locations. In this study, we examine fungal communities across multiple compartments (bulk soils, rhizosphere, rhizoplane, root endosphere, phylloplane, and leaf endosphere) based on field samples of rubber trees during both dry and rainy seasons in two major areas of China . We aimed to (i) determine the distribution pattern of fungal communities along the soil–plant continuum of rubber tree and (ii) identify the relative importance of spatial heterogeneity versus season and dominant drivers for driving fungal community at the regional scale. We hypothesized that (i) both geographic and seasonal factors will influence the assembly of rubber-associated fungal communities and (ii) the spatiotemporal distribution pattern is mediated by the spatiotemporal variation of the driving factors. Study site and sampling The study was conducted in two locations: Hainan Island and Banna. Hainan Island experiences a tropical maritime monsoon climate, characterized by a rainy season from May to October and a dry season from November to April. Rubber plantations in Hainan are located at low latitudes and altitudes, making them tropical island-type plantations. On the other hand, Banna has a warm and humid climate throughout the year, with dry (November to April) and rainy seasons (May to October) similar to Hainan. The rubber plantations in Banna are situated at higher latitudes and altitudes, representing a tropical inland static wind plantation area type with fertile soil . We selected six major plantation sites from Danzhou (DZ), Wanning (WN), and Ledong (LD) Districts in Hainan Island, as well as from Jinghong (JH), Menglun (ML), and Mengpeng (MP) Districts in Xishuangbanna . At each site, we chose three plots separated by distances of 5–15 km for sampling, resulting in a total of 18 plots. The sample collections were carried out from 24 August to 24 September 2019 (rainy season) and 24 November to 23 December 2020 (dry season). Therefore, we obtained 36 soil samples from each plot, resulting in a total of 216 samples, including those from the phyllosphere, leaf endosphere, soil, rhizosphere, rhizoplane, and root endosphere compartments. For each leaf sample, we analyzed water content (WC), pH, leaf phosphorus (P), potassium (K), nitrogen (N), and organic matter (LOM). Additionally, we quantified soil total nitrogen (TN), total phosphorus (TP), total potassium (TK), organic matter (SOM), available potassium (AK), ammonium nitrogen (AN), nitrate nitrogen (NN), and available phosphorus (AP). The detailed methodology of the soil, root, and leaf sampling approach, as well as the analysis of physicochemical properties, are described in Method S1. We collected latitude, longitude, and elevation data for each site. Additionally, data on average monthly precipitation and average monthly temperature for each month were obtained from the National Meteorological Information Center ( https://www.data.cma.cn ). DNA extraction and sequencing Microbial community DNA was extracted from soil, root, and leaf samples using the FastDNA Spin Kit for Soil (MP Biomedicals), following the manufacturer’s instructions. The hypervariable region ITS of the fungal gene was amplified using the primer pairs ITS1F (5′- CTTGGTCATTTAGAGGAAGTAA -3′) and ITS2R (5′- GCTGCGTTCTTCATCGATGC -3′) . Sequencing was conducted on the Illumina MiSeq platform, following standard protocols. All samples were pooled in equimolar concentrations and subjected to paired-end sequencing on the Illumina MiSeq platform. The paired-end sequences were merged to generate single sequences of approximately 300 bp. These sequences were then quality-filtered (maximum expected error = 0.2), and singletons were removed using USEARCH v.10 . Raw reads and operational taxonomic units (OTUs) obtained from the sequencing were deposited into the NCBI Sequence Read Archive (SRA) database (accession number: SRP342019 ) and . Data analysis The raw fastq files were demultiplexed and subjected to quality filtering using QIIME (version 1.17). OTUs were clustered at a 97% similarity cutoff using UPARSE version 7.1 . Chimeric sequences were identified and removed using UCHIME. The taxonomy of each representative sequence of the OTUs was determined using the RDP Classifier version 2.2 against the ITS database with a confidence threshold of 0.7 . Diversity indices were based on resampled sequences using the MOTHUR program . Alpha diversity (observed OTU richness, Shannon, and Evenness) index was calculated for each sample using the vegan package in the R environment (version 4.3.1). We modeled the alpha diversity as a response variable and site location and seasonal change as fixed effects. The relative importance of the location and season for alpha diversity was evaluated by analysis of variance (ANOVA) , with the P values corrected using the false discovery rate method. Principal coordinate analysis was conducted using Bray-Curtis dissimilarity for the 216 samples to explore fungal community compositional differences (beta diversity) in different compartments, seasons, and locations, and then was visualized by using ggplot2 . The relative importance of the locations/sites and two seasons for explaining the variation in environmental variables and the alpha diversity of fungi were evaluated by two-way ANOVA. The permutational multivariate analysis of variance in the R vegan package was used to test the variations in fungal beta diversity as explained by locations and seasons . Random forest (RF) analysis (rfPermute function in rfPermute package in R) was used to identify the main environmental drivers for soil fungal alpha diversity . To reveal the relationship between alpha diversity and environmental factors, a linear (linear least-squares regression analysis) or nonlinear regression was used based on RF results. The Mantel test was performed to evaluate the influence of soil properties, climate factors, and leaf properties upon fungal communities of different compartments, using the mantel function of the ecodist package and vegan package for R and visualized by using the linkET package. We used variation partitioning analysis (VPA) to quantify the relative importance of seasonal change (dry and rainy season), climatic factors (temperature and precipitation), physicochemical properties (soil including soil pH, WC, SOM, TK, TN, TP, AN, NN, AK, and AP; leaf including pH, WC, LOM, K, N, and P) and geographic variables . Latitudinal and longitudinal data for each site was transferred to rectangular data to represent spatial distance by function pcnm of Vegan package, and VPAs were conducted with function varpart in the vegan package for R . The study was conducted in two locations: Hainan Island and Banna. Hainan Island experiences a tropical maritime monsoon climate, characterized by a rainy season from May to October and a dry season from November to April. Rubber plantations in Hainan are located at low latitudes and altitudes, making them tropical island-type plantations. On the other hand, Banna has a warm and humid climate throughout the year, with dry (November to April) and rainy seasons (May to October) similar to Hainan. The rubber plantations in Banna are situated at higher latitudes and altitudes, representing a tropical inland static wind plantation area type with fertile soil . We selected six major plantation sites from Danzhou (DZ), Wanning (WN), and Ledong (LD) Districts in Hainan Island, as well as from Jinghong (JH), Menglun (ML), and Mengpeng (MP) Districts in Xishuangbanna . At each site, we chose three plots separated by distances of 5–15 km for sampling, resulting in a total of 18 plots. The sample collections were carried out from 24 August to 24 September 2019 (rainy season) and 24 November to 23 December 2020 (dry season). Therefore, we obtained 36 soil samples from each plot, resulting in a total of 216 samples, including those from the phyllosphere, leaf endosphere, soil, rhizosphere, rhizoplane, and root endosphere compartments. For each leaf sample, we analyzed water content (WC), pH, leaf phosphorus (P), potassium (K), nitrogen (N), and organic matter (LOM). Additionally, we quantified soil total nitrogen (TN), total phosphorus (TP), total potassium (TK), organic matter (SOM), available potassium (AK), ammonium nitrogen (AN), nitrate nitrogen (NN), and available phosphorus (AP). The detailed methodology of the soil, root, and leaf sampling approach, as well as the analysis of physicochemical properties, are described in Method S1. We collected latitude, longitude, and elevation data for each site. Additionally, data on average monthly precipitation and average monthly temperature for each month were obtained from the National Meteorological Information Center ( https://www.data.cma.cn ). Microbial community DNA was extracted from soil, root, and leaf samples using the FastDNA Spin Kit for Soil (MP Biomedicals), following the manufacturer’s instructions. The hypervariable region ITS of the fungal gene was amplified using the primer pairs ITS1F (5′- CTTGGTCATTTAGAGGAAGTAA -3′) and ITS2R (5′- GCTGCGTTCTTCATCGATGC -3′) . Sequencing was conducted on the Illumina MiSeq platform, following standard protocols. All samples were pooled in equimolar concentrations and subjected to paired-end sequencing on the Illumina MiSeq platform. The paired-end sequences were merged to generate single sequences of approximately 300 bp. These sequences were then quality-filtered (maximum expected error = 0.2), and singletons were removed using USEARCH v.10 . Raw reads and operational taxonomic units (OTUs) obtained from the sequencing were deposited into the NCBI Sequence Read Archive (SRA) database (accession number: SRP342019 ) and . The raw fastq files were demultiplexed and subjected to quality filtering using QIIME (version 1.17). OTUs were clustered at a 97% similarity cutoff using UPARSE version 7.1 . Chimeric sequences were identified and removed using UCHIME. The taxonomy of each representative sequence of the OTUs was determined using the RDP Classifier version 2.2 against the ITS database with a confidence threshold of 0.7 . Diversity indices were based on resampled sequences using the MOTHUR program . Alpha diversity (observed OTU richness, Shannon, and Evenness) index was calculated for each sample using the vegan package in the R environment (version 4.3.1). We modeled the alpha diversity as a response variable and site location and seasonal change as fixed effects. The relative importance of the location and season for alpha diversity was evaluated by analysis of variance (ANOVA) , with the P values corrected using the false discovery rate method. Principal coordinate analysis was conducted using Bray-Curtis dissimilarity for the 216 samples to explore fungal community compositional differences (beta diversity) in different compartments, seasons, and locations, and then was visualized by using ggplot2 . The relative importance of the locations/sites and two seasons for explaining the variation in environmental variables and the alpha diversity of fungi were evaluated by two-way ANOVA. The permutational multivariate analysis of variance in the R vegan package was used to test the variations in fungal beta diversity as explained by locations and seasons . Random forest (RF) analysis (rfPermute function in rfPermute package in R) was used to identify the main environmental drivers for soil fungal alpha diversity . To reveal the relationship between alpha diversity and environmental factors, a linear (linear least-squares regression analysis) or nonlinear regression was used based on RF results. The Mantel test was performed to evaluate the influence of soil properties, climate factors, and leaf properties upon fungal communities of different compartments, using the mantel function of the ecodist package and vegan package for R and visualized by using the linkET package. We used variation partitioning analysis (VPA) to quantify the relative importance of seasonal change (dry and rainy season), climatic factors (temperature and precipitation), physicochemical properties (soil including soil pH, WC, SOM, TK, TN, TP, AN, NN, AK, and AP; leaf including pH, WC, LOM, K, N, and P) and geographic variables . Latitudinal and longitudinal data for each site was transferred to rectangular data to represent spatial distance by function pcnm of Vegan package, and VPAs were conducted with function varpart in the vegan package for R . Community composition and environmental variables A total of 9,036, 7,861, 11,001, 10,261, 9,701, and 3,846 OTUs were detected for the six compartments (phyllosphere, leaf endosphere, soil, rhizosphere, rhizoplane, and root endosphere, respectively). Among these OTUs, the majority were assigned to the classes Dothideomycetes (18.10%), Sordariomycetes (16.51%), Eurotiomycetes (10.64%), Tremellomycetes (9.46%), and Agaricomycetes (8.37%) at the class level . Among the six measured leaf environmental variables, WC, leaf pH, and organic matter (LOM) showed strong seasonal variations, with higher values during the rainy season compared to the dry season. Leaf phosphorus (P) variables, on the other hand, were more influenced by geographical location or site rather than seasonal changes . While there were significant seasonal changes of N in LD, ML, and K in DZ, the ANOVA analysis was not found to be significantly affected by the seasons . In contrast, the 10 measured soil physicochemical properties were primarily explained by geographical locations or sites. Variables, such as AK, soil pH, TN, TK, TP, WC), SOM, NN, and AP, were largely influenced by the specific sites where sampling was conducted. In addition, mean seasonal precipitation and mean seasonal temperature showed significant seasonal variations. Approximately 97.6% ( P < 0.001) of the variation in precipitation and 96.4% ( P < 0.001) of the variation in temperature were explained by sampling seasons. During the rainy season, leaf WC, precipitation, and temperature were significantly higher compared to the dry season . In the rainy season, temperature, precipitation, and AN were significantly higher than the dry season in all sites. And WC in JH, pH in WN, were significantly higher, yet WC in DZ, SOM in LD, and TN in MP were significantly lower than the dry season . Spatiotemporal pattern of the fungal community In terms of alpha diversity, the observed OTU richness in the phyllosphere, rhizosphere, rhizoplane, and soil compartments were roughly equal, while the leaf endosphere showed significantly higher alpha diversity in Banna compared to Hainan, while the root endosphere exhibited the opposite trend . When considering the seasonal effect, the observed OTU richness of fungal communities in the leaf endosphere ( P < 0.01), root endosphere ( P < 0.05), and rhizoplane ( P < 0.01) were significantly higher during the rainy season compared to the dry season . However, when analyzing the seasonal effect separately, distinct patterns were observed. In Banna, the richness of all compartments was significantly higher during the rainy season compared to the dry season . In Hainan, although the richness of the root endosphere and rhizoplane were similar between the rainy and dry seasons, and the leaf endosphere showed significantly higher richness during the rainy season, the phyllosphere, rhizosphere, and soil compartments exhibited significantly higher richness during the dry season . Neither Shannon nor Evenness index observed significant differences in Banna . Only the Shannon diversity of leaf endosphere showed significantly higher, yet soil was significantly lower than the dry season in Hainan . Therefore, differences in Shannon and Evenness index between the rainy and dry soils were not considered in further analyses. Overall, the trend in richness was higher in the dry season compared to the rainy season in Hainan . When analyzing the microbial composition based on the Bray-Curtis distance, we observed that geographical location had a significant effect on the beta diversity of fungal communities in all compartments ( P < 0.01), while seasonal effects were not significant . For instance, geographical location accounted for 6.62%, 7.35%, 9.13%, 8.07%, 17.67%, and 18.78% of the variation in the soil, rhizosphere, rhizoplane, root endosphere, phyllosphere, and leaf endosphere, respectively. Additionally, we observed a much greater effect of site variation compared to seasonal variation in all compartments . In summary, our findings indicate that geographical location/site has a significant impact on fungal composition, while season does not. Drives of environmental factors in shaping rubber tree fungal community Among all environmental factors, leaf WC, temperature, and precipitation were identified as the most important predictors of fungal alpha diversity in different compartments . These findings were further supported by simple linear and nonlinear regression analyses. For instance, significant and positive simple linear regressions were observed between leaf WC, temperature, and fungal alpha diversity in the leaf endosphere and root endosphere compartments. Additionally, a similar mid-peak pattern was observed in four datasets, where richness peaked at mid-temperature in the phyllosphere and rhizosphere compartments, and peaked at mid-precipitation in the soil and rhizoplane compartments. These results indicate that climatic factors play a significant role in driving the alpha diversity of fungi in rubber tree ecosystems. The major physicochemical and climatic factors to the fungal composition were further identified using the Mantel tests. Of the most important environmental factors contributing to the leaf composition, leaf P had the largest observed effect, followed by temperature and precipitation, while soil AK affected the composition most in all root-associated compartments . Specifically, temperature and precipitation were significantly correlated with the fungal communities in different compartments to a certain extent . The contributions of seasonal, environmental, climatic, and geographic variables to the variation in fungal composition were quantified by VPA. Geographic factors were better predictors of fungal composition than seasonal, environmental, and climatic ones , confirming a stronger effect of spatial variation in driving the composition of soil fungi. A total of 9,036, 7,861, 11,001, 10,261, 9,701, and 3,846 OTUs were detected for the six compartments (phyllosphere, leaf endosphere, soil, rhizosphere, rhizoplane, and root endosphere, respectively). Among these OTUs, the majority were assigned to the classes Dothideomycetes (18.10%), Sordariomycetes (16.51%), Eurotiomycetes (10.64%), Tremellomycetes (9.46%), and Agaricomycetes (8.37%) at the class level . Among the six measured leaf environmental variables, WC, leaf pH, and organic matter (LOM) showed strong seasonal variations, with higher values during the rainy season compared to the dry season. Leaf phosphorus (P) variables, on the other hand, were more influenced by geographical location or site rather than seasonal changes . While there were significant seasonal changes of N in LD, ML, and K in DZ, the ANOVA analysis was not found to be significantly affected by the seasons . In contrast, the 10 measured soil physicochemical properties were primarily explained by geographical locations or sites. Variables, such as AK, soil pH, TN, TK, TP, WC), SOM, NN, and AP, were largely influenced by the specific sites where sampling was conducted. In addition, mean seasonal precipitation and mean seasonal temperature showed significant seasonal variations. Approximately 97.6% ( P < 0.001) of the variation in precipitation and 96.4% ( P < 0.001) of the variation in temperature were explained by sampling seasons. During the rainy season, leaf WC, precipitation, and temperature were significantly higher compared to the dry season . In the rainy season, temperature, precipitation, and AN were significantly higher than the dry season in all sites. And WC in JH, pH in WN, were significantly higher, yet WC in DZ, SOM in LD, and TN in MP were significantly lower than the dry season . In terms of alpha diversity, the observed OTU richness in the phyllosphere, rhizosphere, rhizoplane, and soil compartments were roughly equal, while the leaf endosphere showed significantly higher alpha diversity in Banna compared to Hainan, while the root endosphere exhibited the opposite trend . When considering the seasonal effect, the observed OTU richness of fungal communities in the leaf endosphere ( P < 0.01), root endosphere ( P < 0.05), and rhizoplane ( P < 0.01) were significantly higher during the rainy season compared to the dry season . However, when analyzing the seasonal effect separately, distinct patterns were observed. In Banna, the richness of all compartments was significantly higher during the rainy season compared to the dry season . In Hainan, although the richness of the root endosphere and rhizoplane were similar between the rainy and dry seasons, and the leaf endosphere showed significantly higher richness during the rainy season, the phyllosphere, rhizosphere, and soil compartments exhibited significantly higher richness during the dry season . Neither Shannon nor Evenness index observed significant differences in Banna . Only the Shannon diversity of leaf endosphere showed significantly higher, yet soil was significantly lower than the dry season in Hainan . Therefore, differences in Shannon and Evenness index between the rainy and dry soils were not considered in further analyses. Overall, the trend in richness was higher in the dry season compared to the rainy season in Hainan . When analyzing the microbial composition based on the Bray-Curtis distance, we observed that geographical location had a significant effect on the beta diversity of fungal communities in all compartments ( P < 0.01), while seasonal effects were not significant . For instance, geographical location accounted for 6.62%, 7.35%, 9.13%, 8.07%, 17.67%, and 18.78% of the variation in the soil, rhizosphere, rhizoplane, root endosphere, phyllosphere, and leaf endosphere, respectively. Additionally, we observed a much greater effect of site variation compared to seasonal variation in all compartments . In summary, our findings indicate that geographical location/site has a significant impact on fungal composition, while season does not. Among all environmental factors, leaf WC, temperature, and precipitation were identified as the most important predictors of fungal alpha diversity in different compartments . These findings were further supported by simple linear and nonlinear regression analyses. For instance, significant and positive simple linear regressions were observed between leaf WC, temperature, and fungal alpha diversity in the leaf endosphere and root endosphere compartments. Additionally, a similar mid-peak pattern was observed in four datasets, where richness peaked at mid-temperature in the phyllosphere and rhizosphere compartments, and peaked at mid-precipitation in the soil and rhizoplane compartments. These results indicate that climatic factors play a significant role in driving the alpha diversity of fungi in rubber tree ecosystems. The major physicochemical and climatic factors to the fungal composition were further identified using the Mantel tests. Of the most important environmental factors contributing to the leaf composition, leaf P had the largest observed effect, followed by temperature and precipitation, while soil AK affected the composition most in all root-associated compartments . Specifically, temperature and precipitation were significantly correlated with the fungal communities in different compartments to a certain extent . The contributions of seasonal, environmental, climatic, and geographic variables to the variation in fungal composition were quantified by VPA. Geographic factors were better predictors of fungal composition than seasonal, environmental, and climatic ones , confirming a stronger effect of spatial variation in driving the composition of soil fungi. Revealing the spatiotemporal pattern of microbial communities is a fundamental topic in ecology, which have been explored extensively in microbial ecology over the last two decades . Furthermore, the significant effect of the seasonal change in regulating the alpha diversity of soil bacteria and fungi was found at the regional scale (Hainan) in our previous study . Still, very few studies have mapped the temporal and spatial distribution changes of microbial communities of the soil–plant continuum. Here, we explored the spatiotemporal of soil–plant continuum fungal community in the rubber tree across different niches and regions, doing so can help to clarify the drivers of space and seasonal change upon microbial community variations. Our findings provide robust empirical evidence that the spatiotemporal variation of fungal diversity in rubber trees was mainly shaped by seasonal change, but only spatial impacts significantly altered microbial beta diversity. Our results demonstrate that the alpha diversity of soil–plant continuum fungal community is highly dependent on seasonal changes, which is in line with previous studies that have observed significant seasonal variations in microbial diversity . In our study, we provide evidence suggesting that climatic factors play a crucial role in mediating the seasonal variation of fungal alpha diversity. Climatic factors have been identified as the best predictors of soil fungal richness and community composition at a global scale . Importantly, our study further extends this earlier observation at the soil–plant continuum and now provides widespread evidence that climatic factors mediate the alpha diversity of plant microbiome. Among the environmental variables, we found that temperature and precipitation were the dominant predictors for fungal richness, which has also been found in numerous previous studies . Notably, the alpha diversity is sensitive to temperature and precipitation, displaying an unimodal pattern in most compartments, except for the leaf endosphere and root endosphere which showed a linear pattern. Given that temperature and precipitation are generally higher during the rainy season and higher in Hainan compared to Banna, this further supports the observed unimodal pattern in most compartments. Thus, it is not surprising to find that the contrasted seasonal change pattern in Hainan and Banna. Furthermore, we observed a stronger seasonal effect compared to a geographical location effect on some environmental predictors. Specifically, leaf physicochemical factors such as WC and climatic factors such as temperature and precipitation were primarily influenced by seasonal changes, contributing to the mechanisms driving microbial seasonal variations. In contrast, we provide an evidence that the beta diversity of soil–plant continuum only showed a geographical variation pattern. This aligns with previous studies that have shown spatial factors to be more important in shaping soil microbial communities across large spatial scales . The effect of site location was far more influential than the seasonal change in regulating the communities of both soil bacteria and fungi at the regional scale (only in Hainan) in our previous study . Given that biogeographic patterns variation of community similarity indicates the influence of historical factors , compared to our previous studies (research in Hainan soil samples) , the stronger geographical variation pattern suggests that the pronounced impact of historical event on fungal community structure in soil–plant continuum of rubber tree. For fungal beta diversity, P and soil AK were the most important factors in leaf and soil samples, respectively. Besides, major taxa belonging to Dothideomycetes and Eurotiomycetes in leaf samples, according to their life strategies, which have been assigned as copiotrophic fungi, this might explain why leaf samples responded to altered leaf nutrients (e.g., leaf P), which have already been demonstrated . It is worth noting that leaf P and AK which is a highly localized variable, were evidently stronger affected by sampling sites. In all, we found a stronger geographic location effect than seasonal effect upon fungal beta diversity estimates. Furthermore, geographic factors contributed a larger proportion of variation relative to edaphic and climatic factors to the beta diversity of rubber leaf than soils that of soil–plant continuum, indicating a stronger effect of stochastic processes in driving the beta diversity of rubber leaf. These results were also confirmed by the VPA models and Mantel test, and are also consistent with previous observations . Conclusion Our results demonstrate that the alpha diversity is highly dependent on seasonal changes, while beta diversity only showed a geographical variation pattern. Especially, our work showed that spatiotemporal variation in fungal community alpha and beta diversity was mainly driven by climatic factors (temperature and precipitation) and soil properties (e.g., AK), leaf properties (e.g., P), respectively. Moreover, the leaf P and AK were mainly explained by the geographic location effect rather than the seasonal effect, and climatic factors showed opposite pattern. Taken together, our study provides empirical evidence for the distinct spatiotemporal patterns and driving factors between alpha and beta diversity in the fungal community of the soil–plant continuum in rubber trees. Our results demonstrate that the alpha diversity is highly dependent on seasonal changes, while beta diversity only showed a geographical variation pattern. Especially, our work showed that spatiotemporal variation in fungal community alpha and beta diversity was mainly driven by climatic factors (temperature and precipitation) and soil properties (e.g., AK), leaf properties (e.g., P), respectively. Moreover, the leaf P and AK were mainly explained by the geographic location effect rather than the seasonal effect, and climatic factors showed opposite pattern. Taken together, our study provides empirical evidence for the distinct spatiotemporal patterns and driving factors between alpha and beta diversity in the fungal community of the soil–plant continuum in rubber trees. |
Relationship Between Low Education and Poor Periodontal Status among Mexican Adults Aged ≥50 Years | a5013f45-f627-485a-82e4-268ceb92e3ca | 11619841 | Dentistry[mh] | This is a retrospective cross-sectional study. Its research protocol was reviewed and approved by the Ethics Committee of the Iztacala Faculty of Higher Studies at the National Autonomous University of Mexico (CE/FESI/032023/1587), and the study itself was conducted in accordance with the Declaration of Helsinki. Data Collection The periodontal status of the adults participating in the present study was evaluated using data taken from a series of annual reports (2019–2022) by the Sistema de Vigilancia Epidemiológica de Patologías Bucales (SIVEPAB or the Epidemiological Monitoring System for Oral Pathologies), which is administered by the Ministry of Health General Directorate for Epidemiology. Among the responsibilities of SIVEPAB is the collection of data on patients seeking dental care, mainly from “primary care services”, at one of the 442 sentinel units located in the 32 states of Mexico. The present study used non-probability convenience sampling methods. The inclusion criteria for participants in the present study were being a patient 50 years of age or more, of either gender, who did not present missing data in the database. Patients with third molars were excluded. Independent Variable: Level of Education The variable “years of education” was used to compare those adults who had completed ≤ 9 years of formal education with those who had completed more than 9 years, which, in Mexico, corresponds to primary and secondary school combined. The independent variable was dichotomized into ≤ 9 years and > 9 years. Dependent Variable: Periodontal Status The present study used the CPI, which measures the prevalence and severity of periodontal disease, and also indicates the corresponding treatment needs. Moreover, the CPI was used to document probing depth, on a scale of 0 to 4, corresponding to CPI = 0 (healthy), CPI = 1 (bleeding on probing), CPI = 2 (calculus), CPI = 3 (pocket depth of 4 to 5 mm), and CPI = 4 (pocket depth of ≥ 6 mm). A complete examination of the oral cavity was performed. The oral cavity was divided into sextants with the presence of at least two functional teeth. According to the CPI criteria, if no first or second molar was present in a sextant, all the teeth present were examined. All the adult participants were examined by dentists, using a periodontal probe WHO, in a dental chair equipped with a light source. For each participant, the final CPI score corresponded to the most severe score of the CPI readings obtained from the six sextants. Covariables The present study used the following sociodemographic variables, with a model fit for potential confounders: age (in years), categorised into two groups (50–64 years and ≥ 65 years); gender (male/female); smoking, categorised into two groups (never/former smoker and current); diabetes (yes/no); and oral hygiene, evaluated using the Simplified Oral Hygiene Index (OHI-S). The index is used to evaluate plaque and calculus debris on selected tooth surfaces. The vestibular and lingual surfaces of six permanent teeth were examined. Oral hygiene was classified as poor (OHI-S score ≥2) or good (OHI-S score <2). Sample Size The sample size was calculated using the formula for two independent proportions with 80% power, with a 0.17 difference in proportions detected between the two groups, and a bilateral p-value of 0.05. Assuming that 48% of the participants of the population investigated present the factor of interest (CPI 3-4), the present study required a sample size of 204 per group, i.e., a total sample size of 408, assuming equal-sized groups. Statistical Analysis All statistical analyses were carried out using the Stata 15 program (Stata; College Station, TX, USA). Chi-squared tests were used to determine associations among the variables of age, gender, oral hygiene, smoking, level of education, and diabetes for the groups obtained using the CPI. Multinomial regression was used to analyse the association among the independent variables (age, gender, oral hygiene, level of education, smoking, and diabetes) and the dependent variable “periodontal status” (CPI categories), which was expressed as an odds ratio (OR) with 95% confidence intervals (CI). Possible interactions between diabetes and level of education were analysed. The participants were classified into four groups according to their periodontal status: CPI = 0; CPI = 1; CPI = 2; and CPI = 3/4. In all analyses, two-tailed values of p < 0.05 were considered statistically significant.
The periodontal status of the adults participating in the present study was evaluated using data taken from a series of annual reports (2019–2022) by the Sistema de Vigilancia Epidemiológica de Patologías Bucales (SIVEPAB or the Epidemiological Monitoring System for Oral Pathologies), which is administered by the Ministry of Health General Directorate for Epidemiology. Among the responsibilities of SIVEPAB is the collection of data on patients seeking dental care, mainly from “primary care services”, at one of the 442 sentinel units located in the 32 states of Mexico. The present study used non-probability convenience sampling methods. The inclusion criteria for participants in the present study were being a patient 50 years of age or more, of either gender, who did not present missing data in the database. Patients with third molars were excluded.
The variable “years of education” was used to compare those adults who had completed ≤ 9 years of formal education with those who had completed more than 9 years, which, in Mexico, corresponds to primary and secondary school combined. The independent variable was dichotomized into ≤ 9 years and > 9 years.
The present study used the CPI, which measures the prevalence and severity of periodontal disease, and also indicates the corresponding treatment needs. Moreover, the CPI was used to document probing depth, on a scale of 0 to 4, corresponding to CPI = 0 (healthy), CPI = 1 (bleeding on probing), CPI = 2 (calculus), CPI = 3 (pocket depth of 4 to 5 mm), and CPI = 4 (pocket depth of ≥ 6 mm). A complete examination of the oral cavity was performed. The oral cavity was divided into sextants with the presence of at least two functional teeth. According to the CPI criteria, if no first or second molar was present in a sextant, all the teeth present were examined. All the adult participants were examined by dentists, using a periodontal probe WHO, in a dental chair equipped with a light source. For each participant, the final CPI score corresponded to the most severe score of the CPI readings obtained from the six sextants.
The present study used the following sociodemographic variables, with a model fit for potential confounders: age (in years), categorised into two groups (50–64 years and ≥ 65 years); gender (male/female); smoking, categorised into two groups (never/former smoker and current); diabetes (yes/no); and oral hygiene, evaluated using the Simplified Oral Hygiene Index (OHI-S). The index is used to evaluate plaque and calculus debris on selected tooth surfaces. The vestibular and lingual surfaces of six permanent teeth were examined. Oral hygiene was classified as poor (OHI-S score ≥2) or good (OHI-S score <2).
The sample size was calculated using the formula for two independent proportions with 80% power, with a 0.17 difference in proportions detected between the two groups, and a bilateral p-value of 0.05. Assuming that 48% of the participants of the population investigated present the factor of interest (CPI 3-4), the present study required a sample size of 204 per group, i.e., a total sample size of 408, assuming equal-sized groups.
All statistical analyses were carried out using the Stata 15 program (Stata; College Station, TX, USA). Chi-squared tests were used to determine associations among the variables of age, gender, oral hygiene, smoking, level of education, and diabetes for the groups obtained using the CPI. Multinomial regression was used to analyse the association among the independent variables (age, gender, oral hygiene, level of education, smoking, and diabetes) and the dependent variable “periodontal status” (CPI categories), which was expressed as an odds ratio (OR) with 95% confidence intervals (CI). Possible interactions between diabetes and level of education were analysed. The participants were classified into four groups according to their periodontal status: CPI = 0; CPI = 1; CPI = 2; and CPI = 3/4. In all analyses, two-tailed values of p < 0.05 were considered statistically significant.
Characteristics of the Study Population The present study was conducted on 2098 adults ages ≥ 50 years, with a mean age of 62.0 (±9.4) years, of whom 59.5% were women. In terms of the number of years of education, 83.1% had ≤ 9 years and 16.9% > 9 years. According to the OHI-S, 62.1% of the adults had poor oral hygiene. In terms of the worst finding for periodontal status of the participants, 39.9% presented periodontal pockets ≥ 4 mm, 20.8% had calculus, and 12.8% showed bleeding, while only 26.4% were classified as healthy. The prevalence of diabetes in the participants was 23.4% and was higher in women than in men (27.1% vs 17.8%; p < 0.001). The characteristics of the participants, according to their periodontal status, are shown in . The male participants presented a higher percentage of periodontal pockets ≥ 4 mm than did female participants (p = 0.025); smokers with poor oral hygiene and diabetes presented the highest percentage of periodontal pockets ≥ 4 mm. Similarly, participants with a low level of education (≤ 9 years) presented a higher number of periodontal pockets ≥ 4 mm than those with >9 years of education (p < 0.001). shows that in participants with diabetes, the percentage of periodontal pockets ≥ 4 mm was similar in the three categories of educational level among participants with diabetes (p = 0.191). The percentage of periodontal pockets ≥ 4 mm was lowest in participants with > 9 years of education and without diabetes (p < 0.001). shows that participants who currently smoke have more periodontal pockets ≥4 mm compared to participants who have never smoked or were former smokers (p = 0.008). shows the results obtained using the multinomial logistic regression models. Poor oral hygiene, a low level of education (≤ 9 years), and the presence of diabetes were statistically significantly associated with the presence of calculus: OR = 5.04 (3.63 – 7.00), p < 0.001; OR = 4.38 (3.05 – 6.25), p < 0.001, and OR = 1.53 (1.09 – 2.14), p = 0.012,respectively. Other indicators, such as age and gender, were associated with the presence of calculus. Similarly, age ≥ 65 years (OR = 1.33 [1.03 – 1.72], p = 0.025), poor oral hygiene (OR = 6.86 [5.08 – 9.26], p < 0.001), a low level of education (≤ 9 years) (OR = 4.84 [3.51 – 6.66], p < 0.001), smoking (OR = 1.51 [1.05 – 2.18], p = 0.025), and the presence of diabetes (OR = 1.73 [1.28 – 2.32]; p < 0.001) were statistically significantly associated with the presence of periodontal pockets ≥ 4 mm.
The present study was conducted on 2098 adults ages ≥ 50 years, with a mean age of 62.0 (±9.4) years, of whom 59.5% were women. In terms of the number of years of education, 83.1% had ≤ 9 years and 16.9% > 9 years. According to the OHI-S, 62.1% of the adults had poor oral hygiene. In terms of the worst finding for periodontal status of the participants, 39.9% presented periodontal pockets ≥ 4 mm, 20.8% had calculus, and 12.8% showed bleeding, while only 26.4% were classified as healthy. The prevalence of diabetes in the participants was 23.4% and was higher in women than in men (27.1% vs 17.8%; p < 0.001). The characteristics of the participants, according to their periodontal status, are shown in . The male participants presented a higher percentage of periodontal pockets ≥ 4 mm than did female participants (p = 0.025); smokers with poor oral hygiene and diabetes presented the highest percentage of periodontal pockets ≥ 4 mm. Similarly, participants with a low level of education (≤ 9 years) presented a higher number of periodontal pockets ≥ 4 mm than those with >9 years of education (p < 0.001). shows that in participants with diabetes, the percentage of periodontal pockets ≥ 4 mm was similar in the three categories of educational level among participants with diabetes (p = 0.191). The percentage of periodontal pockets ≥ 4 mm was lowest in participants with > 9 years of education and without diabetes (p < 0.001). shows that participants who currently smoke have more periodontal pockets ≥4 mm compared to participants who have never smoked or were former smokers (p = 0.008). shows the results obtained using the multinomial logistic regression models. Poor oral hygiene, a low level of education (≤ 9 years), and the presence of diabetes were statistically significantly associated with the presence of calculus: OR = 5.04 (3.63 – 7.00), p < 0.001; OR = 4.38 (3.05 – 6.25), p < 0.001, and OR = 1.53 (1.09 – 2.14), p = 0.012,respectively. Other indicators, such as age and gender, were associated with the presence of calculus. Similarly, age ≥ 65 years (OR = 1.33 [1.03 – 1.72], p = 0.025), poor oral hygiene (OR = 6.86 [5.08 – 9.26], p < 0.001), a low level of education (≤ 9 years) (OR = 4.84 [3.51 – 6.66], p < 0.001), smoking (OR = 1.51 [1.05 – 2.18], p = 0.025), and the presence of diabetes (OR = 1.73 [1.28 – 2.32]; p < 0.001) were statistically significantly associated with the presence of periodontal pockets ≥ 4 mm.
The results of the present study showed that adults age ≥ 50 years with a low level of education present worse periodontal health (periodontal pockets ≥ 4 mm in depth) than those adults with a higher level of education, after the regression model was fit for possible confounders. Level of education is one of the intermediary social determinants found to be related to the well-being of older adults. In general terms, an individual’s health improves with their level of education, because they develop habits, skills, and resources that enable them to improve their own health. According to the literature, individuals with a high level of education have a high level of personal control and simply more information. They know more about health, tending to adopt a healthy lifestyle and carry out preventive actions to take care of themselves. Even in developed countries, such as the United States, it has been observed that adults with a lower level of education present poorer health than other populations. Level of education plays a significant role in the processes affecting oral health. Yamamoto et al reported that patients with a low level of education have worse periodontal health and a lower number of teeth than those with a high level of education. Similarly, a meta-analysis conducted in this area of research found that a low level of education was associated with a higher risk of periodontitis in adults aged 35 years or over. The present study found that nearly 40% of adults ≥ 50 years of age presented periodontal pockets ≥ 4 mm in depth, with approximately 83.0% of this cohort having ≤ 9 years of education. Possible reasons why Mexican adults aged 50 years and over present more education-related oral health inequalities include, first and foremost, the treatment costs of dental visits, wherein it has been reported that older adults without education are 73% less likely (OR = 0.27; p < 0.001) to visit the dentist in the last year. Second, those with a lower level of education tend to have lower incomes and cannot afford dental treatment. Another possible explanation for the relationship found by the present study is the high levels of inequality in access to and use of oral healthcare services. Sánchez-García et al reported that approximately half of older Mexican adults with social security coverage have used oral healthcare services in the last 12 months. Therefore, the level of education in adults ages ≥ 50 years may affect their access to and use of oral healthcare services due the social inequalities they face. Oral healthcare strategies are required to assist in the diagnosis, prevention, and reduction of oral diseases, via self-care or simple evidence-based measures for the entire population, in order to significantly reduce the burden of disease and the negative impact on quality of life. It should be noted that education helps to promote and maintain healthy lifestyles and positive options, support personal relationships, and improve personal and family well-being, as well as that of the population. The present study found that poor oral hygiene increases the probability of bleeding, calculus, and periodontal pockets ≥ 4 mm in depth in adults ages 50 years or over. Plaque and the presence of calculus have been considered factors in the occurrence of gingivitis and its progression into periodontitis, which is mainly due to inadequate oral hygiene. , Some studies have found that people with periodontitis present a higher percentage of calculus. , The present study observed that 62.1% of adults sampled presented poor hygiene, an association that was significant for the CPI = 2 category (OR = 5.04; p < 0.001), meaning that poor oral hygiene generates a greater progression of the disease. The association between diabetes and severe periodontitis has been widely studied in different populations. , The present study found that, in Mexican adults, the presence of diabetes was related to the presence of calculus and periodontal pockets ≥ 4 mm in depth. Pranckeviciene et al found an association between severe periodontitis and diabetes (OR = 1.83; p = 0.047). The association reported between diabetes and severe periodontitis results from the accumulation of plaque and calculus and a lack of toothbrushing, a lack of periodontal treatment, and a lack of glycemic control in the long-term. It is known that smoking is risk factor for severe periodontitis due to its vasoconstricting effect, an effect that results in a reduction of the efficacy of the defense mechanisms of the gums. Research has demonstrated that smokers are more likely than non-smokers to have bone loss, mobility, and tooth loss, as well as increased pocket probing depth. Other studies have reported an association between smoking and severe periodontitis (OR = 1.40; p = 0.028). , The present study found an association between smoking and the presence of periodontal pockets ≥ 4 mm in depth (OR = 1.51; p = 0.025). Finally, the present study observed that adults aged ≥ 65 years and over were more likely to present periodontal pockets ≥ 4 mm in depth (OR = 1.33; p = 0.025). Other studies have reported that periodontitis is common in people aged over 65 years. , Therefore, a higher prevalence of periodontal disease in adults, combined with the increased proportion of older people in the global population, may have an impact on the need for healthcare and dental services in the coming years. Similarly, long-term planning is required for oral health services to meet the needs of the continually increasing older adult population. The present study has various limitations. First, the results were based on cross-sectional data, which did not allow causal inferences. Second, the data collected is not representative of the general population and could lead to the overestimation of the prevalence of oral diseases, thus minimising the prevalence of periodontal disease in the Mexican population. Third, while six sites on each tooth were examined when the periodontal pockets were probed, periodontal probing was not calibrated and the variations between dentists at the sentinel units were not evaluated. Fourth, selection bias could present due to the fact that the participants of the present study had to attend the dental service offered at each sentinel unit. On the other hand, the present research is one of the few studies to have examined the relationship between level of education and periodontal disease in populations aged 50 years or over.
The present cross-sectional study showed that a low level of education is associated with a worse periodontal status in adults ≥ 50 years of age. Similarly, poor oral hygiene, smoking, diabetes, and being over 65 years old are factors related to periodontal status. These results suggest the importance of periodontal education from an early age onward, as well as the need for effective strategies and interventions for reducing oral health inequalities, in order to thus reduce the gaps in access to oral healthcare over the course of an older adult’s life as they age.
|
Artificial intelligence and deep learning in ophthalmology | 031ce0b1-afe9-4775-bfbd-35cdfad0c7a2 | 6362807 | Ophthalmology[mh] | Artificial intelligence (AI) is the fourth industrial revolution in mankind’s history. Deep learning (DL) is a class of state-of-the-art machine learning techniques that has sparked tremendous global interest in the last few years. DL uses representation-learning methods with multiple levels of abstraction to process input data without the need for manual feature engineering, automatically recognising the intricate structures in high-dimensional data through projection onto a lower dimensional manifold. Compared with conventional techniques, DL has been shown to achieve significantly higher accuracies in many domains, including natural language processing, computer vision and voice recognition. In medicine and healthcare, DL has been primarily applied to medical imaging analysis, in which DL systems have shown robust diagnostic performance in detecting various medical conditions, including tuberculosis from chest X-rays, malignant melanoma on skin photographs and lymph node metastases secondary to breast cancer from tissue sections. DL has similarly been applied to ocular imaging, principally fundus photographs and optical coherence tomography (OCT). Major ophthalmic diseases which DL techniques have been used for include diabetic retinopathy (DR), glaucoma, age-related macular degeneration (AMD) and retinopathy of prematurity (ROP). DL has also been applied to estimate refractive error and cardiovascular risk factors (eg, age, blood pressure, smoking status and body mass index). A primary benefit of DL in ophthalmology could be in screening, such as for DR and ROP, for which well-established guidelines exist. Other conditions, such as glaucoma and AMD, may also require screening and long-term follow-up. However, screening requires tremendous manpower and financial resources from healthcare systems, in both developed countries and in low-income and middle-income countries. The use of DL, coupled with telemedicine, may be a long-term solution to screen and monitor patients within primary eye care settings. This review summarises new DL systems for ophthalmology applications, potential challenges in clinical deployment and potential paths forward.
Diabetic retinopathy Globally, 600 million people will have diabetes by 2040, with a third having DR. A pooled analysis of 22 896 people with diabetes from 35 population-based studies in the USA, Australia, Europe and Asia (between 1980 and 2008) showed that the overall prevalence of any DR (in type 1 and type 2 diabetes) was 34.6%, with 7% vision-threatening diabetic retinopathy. Screening for DR, coupled with timely referral and treatment, is a universally accepted strategy for blindness prevention. DR screening can be performed by different healthcare professionals, including ophthalmologists, optometrists, general practitioners, screening technicians and clinical photographers. The screening methods comprise direct ophthalmoscopy, dilated slit lamp biomicroscopy with a hand-held lens (90 D or 78 D), mydriatic or non-mydriatic retinal photography, teleretinal screening, and retinal video recording. Nonetheless, DR screening programmes are challenged by issues related to implementation, availability of human assessors and long-term financial sustainability. Over the past few years, DL has revolutionised the diagnostic performance in detecting DR. Using this technique, many groups have shown excellent diagnostic performance ( ). Abràmoff et al showed that a DL system was able to achieve an area under the receiver operating characteristic curve (AUC) of 0.980, with sensitivity and specificity of 96.8% and 87.0%, respectively, in the detection of referable DR (defined as moderate non-proliferative DR or worse, including diabetic macular oedema (DMO)) on Messidor-2 data set. Similarly, Gargeya and Leng reported an AUC of 0.97 using cross-validation on the same data set, and 0.94 and 0.95 in two independent test sets (Messidor-2 and E-Ophtha). More recently, Gulshan and colleagues from Google AI Healthcare reported another DL system with excellent diagnostic performance. The DL system was developed using 128 175 retinal images, graded between 3 and 7 times for DR and DMO by a panel of 54 US licensed ophthalmologists and ophthalmology residents between May and December 2015. The test set consisted of approximately 10 000 images retrieved from two publicly available databases (EyePACS-1 and Messidor-2), graded by at least seven US board-certified ophthalmologists with high intragrader consistency. The AUC was 0.991 and 0.990 for EyePACS-1 and Messidor-2, respectively ( ). Although a number of groups have demonstrated good results using DL systems on publicly available data sets, the DL systems were not tested in real-world DR screening programmes. In addition, the generalisability of a DL system to populations of different ethnicities, and retinal images captured using different cameras, still remains uncertain. Ting et al reported a clinically acceptable diagnostic performance of a DL system, developed and tested using the Singapore Integrated Diabetic Retinopathy Programme over a 5-year period, and 10 external data sets recruited from 6 different countries, including Singapore, China, Hong Kong, Mexico, USA and Australia. The DL system, developed using the DL architecture VGG-19, was reported to have AUC, sensitivity and specificity of 0.936, 90.5% and 91.6% in detecting referable DR. For vision-threatening DR, the corresponding statistics were 0.958, 100% and 91.1%. The AUC ranged from 0.889 to 0.983 for the 10 external data sets (n=40 752 images). More recently, the DL system, developed by Abramoff et al , has obtained a US Food and Drug Administration approval for the diagnosis of DR. It was evaluated in a prospective, although observational setting, achieving 87.2% sensitivity and 90.7% specificity. Age-related macular degeneration AMD is a major cause of vision impairment in the elderly population globally. The Age-Related Eye Disease Study (AREDS) classified AMD stages into none, early, intermediate and late AMD. The American Academy of Ophthalmology recommends that people with intermediate AMD should be at least seen once every 2 years. It is projected that 288 million patients may have some forms of AMD by 2040, with approximately 10% having intermediate AMD or worse. With the ageing population, there is an urgent clinical need to have a robust DL system to screen these patients for further evaluation in tertiary eye care centres. Ting et al reported a clinically acceptable DL system diagnostic performance in detecting referable AMD ( ). Specifically, the DL system was trained and tested using 108 558 retinal images from 38 189 patients. Fovea-centred images without macula segmentation were used in this study. Given that this was the DR screening population, there were relatively few patients with referable AMD. For the other two studies, DL systems were developed using the AREDS data set, with a high number of referable AMD (intermediate AMD or worse). Using a fivefold cross-validation, Burlina et al reported a diagnostic accuracy of between 88.4% and 91.6%, with an AUC of between 0.94 and 0.96. Unlike Ting et al , the authors presegmented the macula region prior to training and testing, with an 80/20 split between the training and testing in each fold. In terms of the DL architecture, both AlexNet and OverFeat have been used, with AlexNet yielding a better performance. Using the same AREDS data set, Grassmann et al reported a sensitivity of 84.2% in the detection of any AMD. In this study, the authors used six convolutional neural networks—AlexNet, GoogleNet, VGG, Inception-V3, ResNet and Inception-ResNet-V2—to train different models. Data augmentation was also used to increase the diversity of data set and to reduce the risk of overfitting. For the AREDS data set, all the photographs were captured as analogue photographs and then digitised later. Whether this affects the DL system’s performance remains uncertain. In addition, all three abovementioned studies did not have any results for external validation on the individual DL systems. DM, choroidal neovascularisation and other macular diseases OCT has had a transformative effect on the management of macular diseases, specifically neovascular AMD and DMO. OCT also provides a near-microscopic view of the retina in vivo with quick acquisition protocols revealing structural detail that cannot be seen using other ophthalmic examination techniques. Thus, the number of macular OCTs has grown from 4.3 million in 2012 to 6.4 million in 2016 in the US Medicare population alone, and will most likely continue to grow worldwide. From a DL perspective, macular OCTs possess a number of attractive qualities as a modality for DL. First is the explosive growth in the number of macular OCTs that are routinely collected around the world. This large number of OCTs is required to train DL systems where having many training examples can aid in the convergence of many-layered networks with millions of parameters. Second, macular OCTs have dense three-dimensional structural information that is usually consistently captured. Unlike real-world images or even colour fundus photographs, the field of view of the macula and the foveal fixation is usually consistent from one volume scan to another. This lowers the complexity of the computer vision task significantly and allows networks to reach meaningful performance with smaller data sets. Third, OCTs provide structural detail that is not easily visible using conventional imaging techniques and provide an avenue for uncovering novel biomarkers of the disease. One of the first applications of DL to macular OCTs was in automated classification of AMD. Approximately 100 000 OCT B-scans were used to train a DL classifier based on VGG-16 to achieve an AUC of 0.97 ( ). Few studies used a technique known as transfer learning, where a neural network is pretrained on ImageNet and subsequently then trained on OCT B-scans for retinal disease classification. Of note, these initial studies involve the use of two-dimensional DL models trained on single OCT B-scans rather than three-dimensional models trained on OCT volumes. This may be a barrier to their potential clinical applicability. DL has also had a transformative impact in boundary and feature-level segmentation using neural networks that have been developed for semantic segmentation such as the U-Net. Specifically, these networks have been trained to segment intraretinal fluid cysts and subretinal fluid on OCT B-scans. Deep convolutional networks surpassed traditional methods in the quality of segmentation of retinal anatomical boundaries. Also similar approaches were used to segment en-face OCTA images to segment the foveal avascular zone. More recently, DeepMind and the Moorfields Eye Hospital have combined the power of neural networks for both segmentation and classification tasks using a novel AI framework. In this approach, a segmentation network is first used to delineate a range of 15 different retinal morphological features and OCT acquisition artefacts. The output of this network is then passed to a classification network which makes a referral triage decision from four categories (urgent, semiurgent, routine, observation) and classifies the presence of 10 different OCT pathologies (choroidal neovascularisation (CNV), macular oedema without CNV, drusen, geographic atrophy, epiretinal membrane, vitreomacular traction, full-thickness macular hole, partial thickness macular hole, central serous retinopathy and ‘normal’). Using this approach, the Moorfields-DeepMind system reports a performance on par with experts for these classification tasks (although in a retrospective setting). Moreover, the generation of an intermediate tissue representation by the first, segmentation network means that the framework can be generalised across OCT systems from multiple different vendors without prohibitive requirements for retraining. In the near term, this DL system will be implemented in an existing real-world clinical pathway—the rapid access ‘virtual’ clinics that are now widely used for triaging of macular disease in the UK. In the longer term, the system could be used in triaging patients outside the hospital setting, particularly as OCT systems are increasingly being adopted by optometrists in the community. Glaucoma The global prevalence of glaucoma for people aged 40–80 is 3.4%, and by the year 2040 it is projected there will be approximately 112 million affected individuals worldwide. Clinicians and patients alike would welcome improvements in disease detection, assessment of progressive structural and functional damage, treatment optimisation so as to prevent visual disability, and accurate long-term prognosis. Glaucoma is an optic nerve disease categorised by excavation and erosion of the neuroretinal rim that clinically manifests itself by increased optic nerve head (ONH) cupping. Yet, because the ONH area varies by fivefold, there is virtually no cup to disc ratio (CDR) that defines pathological cupping, hampering disease detection. Li et al and Ting et al trained computer algorithms to detect the glaucoma-like disc, defined as a vertical CDR of 0.7 and 0.8, respectively. Investigators have also applied machine learning methods to distinguish glaucomatous nerve fibre layer damage from normal scans on wide-angle OCTs (9×12 mm). Future opportunities include training a neural network to identify the disc that would be associated with manifest visual field (VF) loss across the spectrum of disc size, as our current treatment strategies are aligned with slowing disease detection. Furthermore, DL could be used to detect progressive structural optic nerve changes in glaucoma. In glaucoma, retinal ganglion cell axons atrophy in a confined space within the ONH and ophthalmologists typically rely on low dimensional psychophysical data to detect the functional consequences of that damage. The outputs from these tests typically provide reliability parameters, age-matched normative comparisons and summary global indices, but more detailed analysis of this functional data is lacking. Elze et al developed an unsupervised computer program to analyse VF that recognises clinically relevant VF loss patterns and assigns a weighting coefficient for each of them ( ). This method has proven useful in the detection of early VF loss from glaucoma. Furthermore, a myriad of computer programs to detect VF progression exist, ranging from assessment of global indices over time to point-wise analyses, to sectoral VF analysis; however, these approaches are often not aligned with clinical ground truth nor with one another. Yousefi et al developed a machine-based algorithm that detected VF progression earlier than these conventional strategies. More machine learning algorithms that provide quantitative information about regional VF progression can be expected in the future. Although intraocular pressure (IOP)-lowering has been shown to be therapeutically effective in delaying glaucoma progression, some demonstrated that disease progression is still inevitable, suggesting that we have not arrived at optimised treatment regimens for the various forms of glaucoma. Kazemian et al developed a clinical forecasting tool that uses tonometric and VF data to project disease trajectories at different target IOPs. Further refinement of this tool that integrates other ophthalmic and non-ophthalmic data would be useful to establish target IOPs and the best strategies to achieve them on a case-by-case basis. Finally, it is documented that patients with newly diagnosed glaucoma harbour fears of going blind ; perhaps, the use of machine learning that incorporates genome-wide data, lifestyle behaviour and medical history into a forecasting algorithm will allow early prognostication regarding the future risk of requiring invasive surgery or losing functional vision from glaucoma. As machine learning algorithms are revised, the practising ophthalmologist will have a host of tools available to diagnose glaucoma, detect disease progression and identify optimised treatment strategies using a precision medicine approaches. In an ideal future scenario, they may also have clinical forecasting tools that inform patients as to their overall prognosis and expected clinical course with or without treatment. Retinopathy of prematurity ROP is a leading cause of childhood blindness worldwide, with an annual incidence of ROP-related blindness of 32 000 worldwide. The regional epidemiology of the disease varies based on a number of factors, including the number of preterm births, neonatal mortality of preterm children and capacity to monitor exposure to oxygen. ROP screening either directly via ophthalmoscopic examination or telemedical evaluation using digital fundus photography can identify the earliest signs of severe ROP, and with timely treatment can prevent most cases of blindness from ROP. Due to the high number of preterm births, reductions in neonatal mortality, and limited capacity for oxygen monitoring and ROP screening, the highest burden of blinding ROP today is in low-income and middle-income countries. There are two main barriers to effective implementation of ROP screening: (1) the diagnosis of ROP is subjective, with significant interexaminer variability in the diagnosis leading to inconsistent application of evidence-based interventions ; and (2) there are too few trained examiners in many regions of the world. Telemedicine has emerged as a viable model to address the latter problem, at least in regions where the cost of a fundus camera is not prohibitive, by allowing a single physician to virtually examine infants over a large geographical area. However, telemedicine itself does not solve the subjectivity problem in ROP diagnosis. Indeed, the acute-phase ROP study found nearly 25% of telemedicine examinations by trained graders required adjudication because the graders disagreed on one of three criteria for clinically significant ROP. There have been a number of early attempts to use DL for automated diagnosis of ROP, which could potentially address both implementation barriers for ROP screening. Most recently, Brown et al reported the results of a fully automated DL system that could diagnose plus disease, the most important feature of severe ROP, with an AUC of 0.98 compared with a consensus reference standard diagnosis combining image-based diagnosis and ophthalmoscopy ( ). When directly compared with the eight international experts in ROP diagnosis, the i-ROP DL system agreed with the consensus diagnosis more frequently than six out of eight experts. Subsequent work found that the i-ROP DL system could also produce a severity score for ROP that demonstrated promise for objective monitoring of disease progression, regression and response to treatment. When compared with the same set of 100 images ranked in order of disease severity by experts, the algorithm had 100% sensitivity an 94% specificity in the detection of pre-plus or worse disease. Potential challenges Despite the high level of accuracy of the AI-based models in many of the diseases in ophthalmology, there are still many clinical and technical challenges for clinical implementation and real-time deployment of these models in clinical practice ( ). These challenges could arise in different stages in both the research and clinical settings. First, many of the studies have used training data sets from relatively homogeneous populations. AI training and testing using retinal images is often subject to numerous variabilities, including width of field, field of view, image magnification, image quality and participant ethnicities. Diversifying the data set, in terms of ethnicities, and image-capture hardware could help to address this challenge. Another challenge in the development of AI models in ophthalmology has been the limited availability of large amounts of data for both the rare diseases (eg, ocular tumours) and for common diseases which are not imaged routinely in clinical practice such as cataracts. Furthermore, there are diseases such as glaucoma and ROP where there will be disagreement and interobserver variability in the definition of the disease phenotype. The algorithm learns from what they are presented with. The software is unlikely to produce accurate outcomes if the training set of images given to the AI tool is too small or not representative of real patient populations. More evidence on ways of getting high-quality ground-truth labels is required for different imaging tools. Krause et al reported that adjudication grades by retina specialists were a more rigorous reference standard, especially to detect artefacts and missed microaneurysms in DR, than a majority decision and improved the algorithm performance. Second, many AI groups have reported robust diagnostic performance for their DL systems, although some papers did not show how the power calculation was performed for the independent data sets. A power calculation should take the following into consideration: the prevalence of the disease, type 1 and 2 errors, CIs, desired precision and so on. It is important to first preset the desired operating threshold on the training set, followed by analysis of performance metrics such as sensitivity and specificity on the test set to assess calibration of the algorithm. Third, large-scale adoption of AI in healthcare is still not on the horizon as clinicians and patients are still concerned about AI and DL being ‘black-boxes’. In healthcare, it is not only the quantitative algorithmic performance, but the underlying features through which the algorithm classifies disease which is important to improve physician acceptance. Generating heat maps highlighting the regions of influence on the image which contributed to the algorithm conclusion may be a first step ( ), although such maps are often challenging to interpret (what does it mean if a map highlights an area of vitreous on an OCT of a patient with drusen?). They may also struggle to deal with negations (what would it mean to highlight the most important part of an ophthalmic image that demonstrates that there is no disease present?). An alternative approach has been used for the DL system developed by the Moorfields Eye Hospital and DeepMind—in this system, the generation of an intermediate tissue representation by a segmentation network is used to highlight for the clinician (and quantify) the relevant areas of retinal pathology ( ). It is also important to highlight that ‘interpretability’ of DL systems may mean different things to a healthcare professional than to a machine learning expert. Although it seems likely that interpretable algorithms will be more readily accepted by ophthalmologists, future applied clinical research will be necessary to determine whether this is the case and whether it leads to tangible benefits for patients in terms of clinical effectiveness. Lastly, the current AI screening systems for DR have been developed and validated using two-dimensional images and lack stereoscopic qualities, thus making identification of elevated lesions like retinal tractions challenging. Incorporating the information from multimodal imaging in future AI algorithms may potentially address this challenge. In addition, the medicolegal aspects and the regulatory approvals vary in different countries and settings, and more work will be needed in these areas. An important challenge to the clinical adoption of AI-based technology is how the patients entrust clinical care to machines. Keel et al evaluated the patient acceptability of AI-based DR screening within endocrinology outpatient setting and reported that 96% of participants were satisfied or very satisfied with the automated screening model. However, in different populations and settings, the patient’s acceptability for AI-based screening may vary and may pose challenge in its implementation.
Globally, 600 million people will have diabetes by 2040, with a third having DR. A pooled analysis of 22 896 people with diabetes from 35 population-based studies in the USA, Australia, Europe and Asia (between 1980 and 2008) showed that the overall prevalence of any DR (in type 1 and type 2 diabetes) was 34.6%, with 7% vision-threatening diabetic retinopathy. Screening for DR, coupled with timely referral and treatment, is a universally accepted strategy for blindness prevention. DR screening can be performed by different healthcare professionals, including ophthalmologists, optometrists, general practitioners, screening technicians and clinical photographers. The screening methods comprise direct ophthalmoscopy, dilated slit lamp biomicroscopy with a hand-held lens (90 D or 78 D), mydriatic or non-mydriatic retinal photography, teleretinal screening, and retinal video recording. Nonetheless, DR screening programmes are challenged by issues related to implementation, availability of human assessors and long-term financial sustainability. Over the past few years, DL has revolutionised the diagnostic performance in detecting DR. Using this technique, many groups have shown excellent diagnostic performance ( ). Abràmoff et al showed that a DL system was able to achieve an area under the receiver operating characteristic curve (AUC) of 0.980, with sensitivity and specificity of 96.8% and 87.0%, respectively, in the detection of referable DR (defined as moderate non-proliferative DR or worse, including diabetic macular oedema (DMO)) on Messidor-2 data set. Similarly, Gargeya and Leng reported an AUC of 0.97 using cross-validation on the same data set, and 0.94 and 0.95 in two independent test sets (Messidor-2 and E-Ophtha). More recently, Gulshan and colleagues from Google AI Healthcare reported another DL system with excellent diagnostic performance. The DL system was developed using 128 175 retinal images, graded between 3 and 7 times for DR and DMO by a panel of 54 US licensed ophthalmologists and ophthalmology residents between May and December 2015. The test set consisted of approximately 10 000 images retrieved from two publicly available databases (EyePACS-1 and Messidor-2), graded by at least seven US board-certified ophthalmologists with high intragrader consistency. The AUC was 0.991 and 0.990 for EyePACS-1 and Messidor-2, respectively ( ). Although a number of groups have demonstrated good results using DL systems on publicly available data sets, the DL systems were not tested in real-world DR screening programmes. In addition, the generalisability of a DL system to populations of different ethnicities, and retinal images captured using different cameras, still remains uncertain. Ting et al reported a clinically acceptable diagnostic performance of a DL system, developed and tested using the Singapore Integrated Diabetic Retinopathy Programme over a 5-year period, and 10 external data sets recruited from 6 different countries, including Singapore, China, Hong Kong, Mexico, USA and Australia. The DL system, developed using the DL architecture VGG-19, was reported to have AUC, sensitivity and specificity of 0.936, 90.5% and 91.6% in detecting referable DR. For vision-threatening DR, the corresponding statistics were 0.958, 100% and 91.1%. The AUC ranged from 0.889 to 0.983 for the 10 external data sets (n=40 752 images). More recently, the DL system, developed by Abramoff et al , has obtained a US Food and Drug Administration approval for the diagnosis of DR. It was evaluated in a prospective, although observational setting, achieving 87.2% sensitivity and 90.7% specificity.
AMD is a major cause of vision impairment in the elderly population globally. The Age-Related Eye Disease Study (AREDS) classified AMD stages into none, early, intermediate and late AMD. The American Academy of Ophthalmology recommends that people with intermediate AMD should be at least seen once every 2 years. It is projected that 288 million patients may have some forms of AMD by 2040, with approximately 10% having intermediate AMD or worse. With the ageing population, there is an urgent clinical need to have a robust DL system to screen these patients for further evaluation in tertiary eye care centres. Ting et al reported a clinically acceptable DL system diagnostic performance in detecting referable AMD ( ). Specifically, the DL system was trained and tested using 108 558 retinal images from 38 189 patients. Fovea-centred images without macula segmentation were used in this study. Given that this was the DR screening population, there were relatively few patients with referable AMD. For the other two studies, DL systems were developed using the AREDS data set, with a high number of referable AMD (intermediate AMD or worse). Using a fivefold cross-validation, Burlina et al reported a diagnostic accuracy of between 88.4% and 91.6%, with an AUC of between 0.94 and 0.96. Unlike Ting et al , the authors presegmented the macula region prior to training and testing, with an 80/20 split between the training and testing in each fold. In terms of the DL architecture, both AlexNet and OverFeat have been used, with AlexNet yielding a better performance. Using the same AREDS data set, Grassmann et al reported a sensitivity of 84.2% in the detection of any AMD. In this study, the authors used six convolutional neural networks—AlexNet, GoogleNet, VGG, Inception-V3, ResNet and Inception-ResNet-V2—to train different models. Data augmentation was also used to increase the diversity of data set and to reduce the risk of overfitting. For the AREDS data set, all the photographs were captured as analogue photographs and then digitised later. Whether this affects the DL system’s performance remains uncertain. In addition, all three abovementioned studies did not have any results for external validation on the individual DL systems.
OCT has had a transformative effect on the management of macular diseases, specifically neovascular AMD and DMO. OCT also provides a near-microscopic view of the retina in vivo with quick acquisition protocols revealing structural detail that cannot be seen using other ophthalmic examination techniques. Thus, the number of macular OCTs has grown from 4.3 million in 2012 to 6.4 million in 2016 in the US Medicare population alone, and will most likely continue to grow worldwide. From a DL perspective, macular OCTs possess a number of attractive qualities as a modality for DL. First is the explosive growth in the number of macular OCTs that are routinely collected around the world. This large number of OCTs is required to train DL systems where having many training examples can aid in the convergence of many-layered networks with millions of parameters. Second, macular OCTs have dense three-dimensional structural information that is usually consistently captured. Unlike real-world images or even colour fundus photographs, the field of view of the macula and the foveal fixation is usually consistent from one volume scan to another. This lowers the complexity of the computer vision task significantly and allows networks to reach meaningful performance with smaller data sets. Third, OCTs provide structural detail that is not easily visible using conventional imaging techniques and provide an avenue for uncovering novel biomarkers of the disease. One of the first applications of DL to macular OCTs was in automated classification of AMD. Approximately 100 000 OCT B-scans were used to train a DL classifier based on VGG-16 to achieve an AUC of 0.97 ( ). Few studies used a technique known as transfer learning, where a neural network is pretrained on ImageNet and subsequently then trained on OCT B-scans for retinal disease classification. Of note, these initial studies involve the use of two-dimensional DL models trained on single OCT B-scans rather than three-dimensional models trained on OCT volumes. This may be a barrier to their potential clinical applicability. DL has also had a transformative impact in boundary and feature-level segmentation using neural networks that have been developed for semantic segmentation such as the U-Net. Specifically, these networks have been trained to segment intraretinal fluid cysts and subretinal fluid on OCT B-scans. Deep convolutional networks surpassed traditional methods in the quality of segmentation of retinal anatomical boundaries. Also similar approaches were used to segment en-face OCTA images to segment the foveal avascular zone. More recently, DeepMind and the Moorfields Eye Hospital have combined the power of neural networks for both segmentation and classification tasks using a novel AI framework. In this approach, a segmentation network is first used to delineate a range of 15 different retinal morphological features and OCT acquisition artefacts. The output of this network is then passed to a classification network which makes a referral triage decision from four categories (urgent, semiurgent, routine, observation) and classifies the presence of 10 different OCT pathologies (choroidal neovascularisation (CNV), macular oedema without CNV, drusen, geographic atrophy, epiretinal membrane, vitreomacular traction, full-thickness macular hole, partial thickness macular hole, central serous retinopathy and ‘normal’). Using this approach, the Moorfields-DeepMind system reports a performance on par with experts for these classification tasks (although in a retrospective setting). Moreover, the generation of an intermediate tissue representation by the first, segmentation network means that the framework can be generalised across OCT systems from multiple different vendors without prohibitive requirements for retraining. In the near term, this DL system will be implemented in an existing real-world clinical pathway—the rapid access ‘virtual’ clinics that are now widely used for triaging of macular disease in the UK. In the longer term, the system could be used in triaging patients outside the hospital setting, particularly as OCT systems are increasingly being adopted by optometrists in the community.
The global prevalence of glaucoma for people aged 40–80 is 3.4%, and by the year 2040 it is projected there will be approximately 112 million affected individuals worldwide. Clinicians and patients alike would welcome improvements in disease detection, assessment of progressive structural and functional damage, treatment optimisation so as to prevent visual disability, and accurate long-term prognosis. Glaucoma is an optic nerve disease categorised by excavation and erosion of the neuroretinal rim that clinically manifests itself by increased optic nerve head (ONH) cupping. Yet, because the ONH area varies by fivefold, there is virtually no cup to disc ratio (CDR) that defines pathological cupping, hampering disease detection. Li et al and Ting et al trained computer algorithms to detect the glaucoma-like disc, defined as a vertical CDR of 0.7 and 0.8, respectively. Investigators have also applied machine learning methods to distinguish glaucomatous nerve fibre layer damage from normal scans on wide-angle OCTs (9×12 mm). Future opportunities include training a neural network to identify the disc that would be associated with manifest visual field (VF) loss across the spectrum of disc size, as our current treatment strategies are aligned with slowing disease detection. Furthermore, DL could be used to detect progressive structural optic nerve changes in glaucoma. In glaucoma, retinal ganglion cell axons atrophy in a confined space within the ONH and ophthalmologists typically rely on low dimensional psychophysical data to detect the functional consequences of that damage. The outputs from these tests typically provide reliability parameters, age-matched normative comparisons and summary global indices, but more detailed analysis of this functional data is lacking. Elze et al developed an unsupervised computer program to analyse VF that recognises clinically relevant VF loss patterns and assigns a weighting coefficient for each of them ( ). This method has proven useful in the detection of early VF loss from glaucoma. Furthermore, a myriad of computer programs to detect VF progression exist, ranging from assessment of global indices over time to point-wise analyses, to sectoral VF analysis; however, these approaches are often not aligned with clinical ground truth nor with one another. Yousefi et al developed a machine-based algorithm that detected VF progression earlier than these conventional strategies. More machine learning algorithms that provide quantitative information about regional VF progression can be expected in the future. Although intraocular pressure (IOP)-lowering has been shown to be therapeutically effective in delaying glaucoma progression, some demonstrated that disease progression is still inevitable, suggesting that we have not arrived at optimised treatment regimens for the various forms of glaucoma. Kazemian et al developed a clinical forecasting tool that uses tonometric and VF data to project disease trajectories at different target IOPs. Further refinement of this tool that integrates other ophthalmic and non-ophthalmic data would be useful to establish target IOPs and the best strategies to achieve them on a case-by-case basis. Finally, it is documented that patients with newly diagnosed glaucoma harbour fears of going blind ; perhaps, the use of machine learning that incorporates genome-wide data, lifestyle behaviour and medical history into a forecasting algorithm will allow early prognostication regarding the future risk of requiring invasive surgery or losing functional vision from glaucoma. As machine learning algorithms are revised, the practising ophthalmologist will have a host of tools available to diagnose glaucoma, detect disease progression and identify optimised treatment strategies using a precision medicine approaches. In an ideal future scenario, they may also have clinical forecasting tools that inform patients as to their overall prognosis and expected clinical course with or without treatment.
ROP is a leading cause of childhood blindness worldwide, with an annual incidence of ROP-related blindness of 32 000 worldwide. The regional epidemiology of the disease varies based on a number of factors, including the number of preterm births, neonatal mortality of preterm children and capacity to monitor exposure to oxygen. ROP screening either directly via ophthalmoscopic examination or telemedical evaluation using digital fundus photography can identify the earliest signs of severe ROP, and with timely treatment can prevent most cases of blindness from ROP. Due to the high number of preterm births, reductions in neonatal mortality, and limited capacity for oxygen monitoring and ROP screening, the highest burden of blinding ROP today is in low-income and middle-income countries. There are two main barriers to effective implementation of ROP screening: (1) the diagnosis of ROP is subjective, with significant interexaminer variability in the diagnosis leading to inconsistent application of evidence-based interventions ; and (2) there are too few trained examiners in many regions of the world. Telemedicine has emerged as a viable model to address the latter problem, at least in regions where the cost of a fundus camera is not prohibitive, by allowing a single physician to virtually examine infants over a large geographical area. However, telemedicine itself does not solve the subjectivity problem in ROP diagnosis. Indeed, the acute-phase ROP study found nearly 25% of telemedicine examinations by trained graders required adjudication because the graders disagreed on one of three criteria for clinically significant ROP. There have been a number of early attempts to use DL for automated diagnosis of ROP, which could potentially address both implementation barriers for ROP screening. Most recently, Brown et al reported the results of a fully automated DL system that could diagnose plus disease, the most important feature of severe ROP, with an AUC of 0.98 compared with a consensus reference standard diagnosis combining image-based diagnosis and ophthalmoscopy ( ). When directly compared with the eight international experts in ROP diagnosis, the i-ROP DL system agreed with the consensus diagnosis more frequently than six out of eight experts. Subsequent work found that the i-ROP DL system could also produce a severity score for ROP that demonstrated promise for objective monitoring of disease progression, regression and response to treatment. When compared with the same set of 100 images ranked in order of disease severity by experts, the algorithm had 100% sensitivity an 94% specificity in the detection of pre-plus or worse disease.
Despite the high level of accuracy of the AI-based models in many of the diseases in ophthalmology, there are still many clinical and technical challenges for clinical implementation and real-time deployment of these models in clinical practice ( ). These challenges could arise in different stages in both the research and clinical settings. First, many of the studies have used training data sets from relatively homogeneous populations. AI training and testing using retinal images is often subject to numerous variabilities, including width of field, field of view, image magnification, image quality and participant ethnicities. Diversifying the data set, in terms of ethnicities, and image-capture hardware could help to address this challenge. Another challenge in the development of AI models in ophthalmology has been the limited availability of large amounts of data for both the rare diseases (eg, ocular tumours) and for common diseases which are not imaged routinely in clinical practice such as cataracts. Furthermore, there are diseases such as glaucoma and ROP where there will be disagreement and interobserver variability in the definition of the disease phenotype. The algorithm learns from what they are presented with. The software is unlikely to produce accurate outcomes if the training set of images given to the AI tool is too small or not representative of real patient populations. More evidence on ways of getting high-quality ground-truth labels is required for different imaging tools. Krause et al reported that adjudication grades by retina specialists were a more rigorous reference standard, especially to detect artefacts and missed microaneurysms in DR, than a majority decision and improved the algorithm performance. Second, many AI groups have reported robust diagnostic performance for their DL systems, although some papers did not show how the power calculation was performed for the independent data sets. A power calculation should take the following into consideration: the prevalence of the disease, type 1 and 2 errors, CIs, desired precision and so on. It is important to first preset the desired operating threshold on the training set, followed by analysis of performance metrics such as sensitivity and specificity on the test set to assess calibration of the algorithm. Third, large-scale adoption of AI in healthcare is still not on the horizon as clinicians and patients are still concerned about AI and DL being ‘black-boxes’. In healthcare, it is not only the quantitative algorithmic performance, but the underlying features through which the algorithm classifies disease which is important to improve physician acceptance. Generating heat maps highlighting the regions of influence on the image which contributed to the algorithm conclusion may be a first step ( ), although such maps are often challenging to interpret (what does it mean if a map highlights an area of vitreous on an OCT of a patient with drusen?). They may also struggle to deal with negations (what would it mean to highlight the most important part of an ophthalmic image that demonstrates that there is no disease present?). An alternative approach has been used for the DL system developed by the Moorfields Eye Hospital and DeepMind—in this system, the generation of an intermediate tissue representation by a segmentation network is used to highlight for the clinician (and quantify) the relevant areas of retinal pathology ( ). It is also important to highlight that ‘interpretability’ of DL systems may mean different things to a healthcare professional than to a machine learning expert. Although it seems likely that interpretable algorithms will be more readily accepted by ophthalmologists, future applied clinical research will be necessary to determine whether this is the case and whether it leads to tangible benefits for patients in terms of clinical effectiveness. Lastly, the current AI screening systems for DR have been developed and validated using two-dimensional images and lack stereoscopic qualities, thus making identification of elevated lesions like retinal tractions challenging. Incorporating the information from multimodal imaging in future AI algorithms may potentially address this challenge. In addition, the medicolegal aspects and the regulatory approvals vary in different countries and settings, and more work will be needed in these areas. An important challenge to the clinical adoption of AI-based technology is how the patients entrust clinical care to machines. Keel et al evaluated the patient acceptability of AI-based DR screening within endocrinology outpatient setting and reported that 96% of participants were satisfied or very satisfied with the automated screening model. However, in different populations and settings, the patient’s acceptability for AI-based screening may vary and may pose challenge in its implementation.
DL is the state-of-the-art AI machine learning technique that has revolutionised the AI field. For ophthalmology, DL has shown clinically acceptable diagnostic performance in detecting many retinal diseases, in particular DR and ROP. Future research is crucial in evaluating the clinical deployment and cost-effectiveness of different DL systems in the clinical practice. To improve clinical acceptance of DL systems, it is important to unravel the ‘black-box’ nature of DL using existing and future methodologies. Although there are challenges ahead, DL will likely impact on the practice of medicine and ophthalmology in the coming decades.
|
Thiocoumarins: From the Synthesis to the Biological Applications | 5987f3d7-4e76-4017-8d94-ffdd856353ec | 9369797 | Pharmacology[mh] | This review describes the preparation and biological activities of a particular family of coumarins, the thiocoumarins. The substitution of an oxygen atom by a sulfur is a common strategy in medicinal chemistry and drug discovery, to modulate the chemical properties of a molecule of interest. In fact, the electron deficient and bivalent sulfur atom presents two different areas of positive electrostatic potential due to the low-lying σ* orbitals of the C–S bond, available to interact with different electron donors as nitrogen or oxygen and, possibly, π-systems . Recent studies discuss the potential of this phenomenon in drug design . Therefore, several sulfur-containing functional groups are present in a broad range of drugs and natural products . However, thiocoumarins have been receiving little attention from the research community compared with the coumarins, which are widely explored as scaffold in medicinal chemistry . For example, searching in Pubmed for the keyword thiocoumarin, the last references are from 2022, and are only four, with one related to 4-sulfanylcoumarins. This may be due to impediments regarding their complex synthesis, scarcity of starting materials, bibliographical references, etc. In the particular case of the coumarin scaffold, two different oxygen atoms can be replaced by sulfur, providing three different new scaffolds: Thiocoumarins, 2-thioxocoumarins, and dithiocoumarins . In this review, special attention will be paid to the first group, which is most explored in the scientific bibliography. In addition, we will provide few highlights on the two other groups, to make this overview as complete as possible. Scientific articles and patents, collected from scientific sources, such as Pubmed, SciFinder, Espacenet, and Mendeley, are included in this review. Moreover, the most relevant synthetic pathways and biological applications are herein highlighted.
2.1. Thiocoumarins In the 1980s, thiocoumarins already had clinical use as haemorrhagic agents, anticoagulants (interfering with vitamin K-dependent coagulation factors), and antiallergic agents . In fact, some thiocoumarin derivatives found important applications as anticoagulant rodenticides pesticides . During the decade of the 1980s, the versatile synthesis from the acrylic and propiolic ortho esters and benzenethiols was described . After those first reports, some other evidences on the interest of thiocoumarins as potential bioactive molecules have been published. However, a deeper understanding of biology and biological functions, and how these molecules really work, has to be acquired before considering thiocoumarins as a trend topic in medicinal chemistry. One of the most recent subjects of interest related to this scaffold is the role of thiocoumarins as carbonic anhydrase inhibitors . The described thiocoumarins, in particular the 6-hydroxy-2-thioxocoumarin, proved to bound to the human carbonic anhydrase II active site in a completely different mode compared with coumarins, commonly hydrolyzed by the esterase carbonic anhydrase to the corresponding 2-hydroxycinnamic acids. The role of thiocoumarins inhibiting carcinogenesis has also been studied over the last decade . Moreover, few indole derivatives were tested as potential drugs for PUVA photochemotherapy . Furthermore, ethers of thiocoumarin, at position 4, have been reported as nitric oxide synthase inhibitors in the nanomolar range . Synthetically, most of the described thiocoumarins are obtained from the 4-hydroxythiocoumarin. Recently, a review collected the relevant information on the utility of this scaffold in organic chemistry . Detailed synthetic methodologies, structures, and chemical properties of the 4-hydroxythiocoumarin were mentioned in this overview, as well as its most interesting bioactivities. In fact, this compound is easily obtained and modified to achieve key educts for the synthesis of heterocyclic systems. Few papers per year describe some new advances in the field, related to both synthesis and biological applications. In 2019, there are only four relevant references on this topic, with the first one being a chemical mechanistic approach on the formation of two-center three-electron bonds by the hydroxyl radical induced reaction of thioesculetin . The second study is focused on the anticonvulsant activity of a new series of synthetic 4-amino-3-nitrothiocoumarins . The traditional synthesis starting from the thiophenol and the malonic acid, in the presence of POCl 3 and AlCl 3 with prolonged heating, is commonly used to afford the 4-hydroxythicoumarin, in low yield . The yields described for the first time for this traditional reaction were between 50 and 90%. However, further reports for other substituted thiocoumarins were lower than these. In this recent manuscript , an alternative two-step method has been proposed by the authors . An intermediate dithiophenyl ester of malonic acid is formed in high yield, using moderate conditions. In a second step, this intermediate undergoes a cyclocondensation in the presence of AlCl 3 as a Lewis acid to give the final product in moderate to high yield. This new method has been described to provide an overall yield of 60%, which is considerably higher than the previous reports. Therefore, this can be considered a step forward in the field. The third paper published in 2019 is focused on the design and synthesis of new anticonvulsants . 3-Nitrocoumarins proved to be more active than the studied thiocoumarins. Finally, the synthesis of 4-sulfonylthiocoumarins was also described in 2019 . Starting from the 4-hydroxythiocoumarin, via DABCO-catalyzed direct sulfonylation of 1-sulfonyl-1,2,3-triazoles, new 4-sulfonylthiocoumarins were efficiently obtained. The main advantage of this method is the lack of transition metal catalysts and extra oxidants. As previously mentioned, most of the traditional synthetic routes to afford the thiocoumarin scaffold involve a Lewis acid, as AlCl 3 ( A) . This reaction may be followed by nickel-catalyzed intramolecular recombination fragment coupling of the thioester to afford the corresponding benzothiophene. To show the applicability of this methodology, the 5-methyl benzothiophene has been obtained via an intramolecular decarbonylative strategy. Previously, preparation of thiocoumarin has been easily achieved via Friedel–Crafts type intramolecular cyclization of the precursor . Complex thiocoumarins have been synthetized from simple ones, and not only 4-hydroxythiocoumarin is used as starting material for the synthesis of more complex thiocoumarins. 4-Chlorothiocoumarin is also used for acylating thiocoumarins to afford γ-ketoenones ( B) . The authors described a one-step organocatalytic synthesis of 4-acylcoumarins from 4-chlorocoumarin, reporting the first examples of nucleophilic substitutions at the β-carbons of enones to afford γ-ketoenones . Few scientific papers described the use of simple thiocoumarins as starting materials for more complex structures. As an example, a novel, base-catalyzed and highly diastereoselective direct Michael addition-isomerization was described for the efficient synthesis of Rauhut–Currier-type adducts. An unexpected α-addition of γ-butyrolactam onto the 3-acylthiocoumarin derivatives was observed rather than the γ-addition, which is more common . This reaction may end with obtaining chalcones. Despite being an interesting alternative, it has not been explored in further works. Another example is the synthesis of (±)-thia-calanolide A, successfully accomplished using 3,5-dimethoxythiophenol as starting material, following a six-step reaction in an overall yield of 4.5% . The key reaction involved a Friedel–Crafts tigloylation of 5,7-dihydroxy-4-thiocoumarin. Microwave was presented as an alternative to include lateral chains at position 3 of the 4-hydroxythiocoumarin, via Mannich reaction, to achieve compounds with antibacterial properties . This reaction is clean, versatile, and described with high yields. 4-Hydroxythiocoumarin is also the starting material for Michael reactions at position 3 of the scaffold ( C) . A primary amine-derived organocatalyst modified with an ionic group for asymmetric Michael reactions of C-nucleophiles with α,β-unsaturated ketones was synthesized. In the presence of this catalyst and an acidic co-catalyst, 4-hydroxythiocoumarin reacted with benzylidene-acetones or cyclohexenones to afford the corresponding Michael adducts in high yields (up to 97%) and with reasonable enantioselectivity (up to 80%). 4-Hydroxythiocoumarin is also the starting material for an easy synthesis of 4-acetylthiocoumarin via very high α-regioselective Heck coupling on tosylates ( D) . This has been described as an efficient and versatile reaction. 4-(4′-Aryloxybut-2′-ynylthio) thiocoumarins were obtained via tosylation of 4-hydroxythiocoumarin followed by mercapto-substitution and condensation with 1-aryloxy-4-chlorobut-2-ynes . Via sequential thermal and catalyzed Claisen rearrangements, and starting with the 4-hydroxythiocoumarin, it is possible to achieve the thiocoumarin-annulated furopyran moiety . Pyrrole derivatives, at positions 2,3 of the thiocoumarins, have been also described, following a tandem [2,3] and [3,3] sigmatropic rearrangement . Thiophenyl derivatives have been described by the same group following a similar approach . The synthesis of 3-aminomethylenethiocoumarins is another case of how to increase a lateral chain in thiocoumarins . In the article, a three-component synthesis by condensation of amines, α-amino acids, ureas, and carbamates with 4-hydroxythiocoumarin, in the presence of tri-ethylorthoformate, is described. Even if there is a temptation to compare these reactions, most of them are specific for some substitution patterns, and may only give specific thiocoumarin derivatives. Therefore, it is interesting to have an arsenal of different options to be able to synthetize more derivatives to establish structure–activity relationships. One of the biggest molecules achieved starting from simple thiocoumarins is the novel 6,6′-arylidene-bis-[5-hydroxy-9-methyl-2,3-diarylthieno[3,2- g ]thiocoumarins]. As the biscoumarins, these molecules were tested for their antimicrobial profile . Similarly, and inspired on a class of furocoumarins, the angelicins, thioangelicins were synthetized and studied for PUVA chemotherapy . These are two cases in which coumarins inspired the study of similar families of thiocoumarins. 7-Hydroxy-4-methylthiocoumarin is the precursor for a pyrano thiocoumarins, reported for their anti-HIV potential . From this series, an excellent derivative, with an EC 50 in the low nanomolar range, was described. 4-Mercaptothiocoumarin is also a versatile starting material. It can be alkylated with different propargylic and allylic halides under phase-transfer-catalyzed conditions in the presence of tetrabutylammonium bromide or benzyl triethylammonium chloride catalyst in dichloromethane/NaOH solution, at room temperature . Then, these 4-thiopropynyl- and thioallylthiocoumarins can be the starting material for different derivatives. A tandem sp3 C–H functionalization followed by decarboxylation of 2-alkylazaarenes with thiocoumarin-3-carboxylic acid has also been used to achieve 4-substituted 3,4-dihydrothiocoumarins . This catalyst-free reaction afforded a series of 29 azaarene-substituted 3,4-dihydrothiocoumarins. Another catalyst-free tandem reaction, this time a Michael addition followed by decarboxylation, was also described to achieve 4-substituted thiocoumarins . This reaction uses the same thiocoumarin-3-carboxylic acid as starting material, together with 2-methylindole, in the same conditions of 1,4-dioxane, at 120 °C, this time to give 3-indolyl-substituted 3,4-dihydrothiocoumarins. Few thiocoumarins have also been explored in the design of novel carrier systems. The development of egg shell-like nanovesicles using the thiocoumarin-3-carboxylate has been described . These drug delivery vehicles have been used as a potential carrier for the bacteriostatic antibiotic sulfamethoxazole. The spectroscopic studies as well as the growth inhibition of E. coli exhibit that this formulation leads to pH responsive sustained release of the drug. Thiocoumarins have been recently prepared using microwave irradiation, in a versatile Lewis acid-catalyzed reaction . Functional groups as pyrimidine and 1,3,4-oxadiazole have been introduced in the scaffold. The technique has been described as efficient, selective, fast, and clean. The series of compounds has been reported for the anti-microbial and antioxidant activities. Thiocoumarins can also be converted to dithiocoumarins using Lawesson’s reagent, in a reaction with ~30–40% yield . This reaction may contribute to deeply exploring these analogues, which are less studied due to the scarcity and rarity of starting materials. As some simple coumarins, thiocoumarins have been recently described as fluorescent probes and sensors. A thiocoumarin, named 7-( N , N -diethylamino)-4-trifluoromethyl-2-thiocoumarin, has been used as sensor for heavy metal pollution and bacterial contamination. Its specific thiocarbonyl scaffold as donor-π-accepter electronic properties, provides intriguing optical properties and several applicable functions . A thiocoumarin-containing ratiometric fluorescent probe has been described for the simultaneous detection of hypochlorite and singlet oxygen . Once again, the thiocarbonyl moiety has been highlighted as an excellent alternative to traditional probes. Moreover, these applications have been described for live-cell imaging, with the thiocoumarin fluorescent probe turned-on once inside the cells . The potential of these molecules for imaging has been recently claimed in a patent . These new applications may revolutionize the field, making thiocoumarins more visible to the academic community. 2.2. 2-Thioxocoumarins As previously described for the thiocoumarins, the 2-thioxocoumarins have been studied as carbonic anhydrase inhibitors . The authors presented the X-ray crystal structure of the 6-hydroxy-2-thioxocoumarin on the human carbonic anhydrase II active site, and these results revealed an unprecedented and unexpected inhibition mechanism for these new inhibitors when compared with isostructural coumarins. It was proved that the exo-sulfur atom can link to a zinc-coordinated water molecule, while other parts of the scaffold can establish important interactions with different amino acids of the binding pocket. Compared with simple coumarins with the same substitution patterns, the inhibition mechanism is totally different. These molecules proved to be hydrolyzed, occluding the entrance of the binding pocket. This is a step forward in the design of efficient carbonic anhydrase inhibitors. 7-Diethylamino-4-hydroxymethyl-thiocoumarin (thio-DEACM) caged molecules are recently attracting attention. The thio-DEACM as a caging group has great properties, such as rapid blue-cyan light responsiveness and avoiding UV irradiation of cells in combination with the absence of toxicity of the released photocage . As several other coumarin derivatives, 4-methylthiocoumarins have been described as a multitarget-directed ligand for the treatment of Alzheimer’s disease . A novel 4-methylthiocoumarin derivative has been studied against acetylcholinesterase, butyrylcholinesterase, BACE1, β-amyloid aggregation, and oxidative stress involved in the pathogenesis of this neurodegenerative disease. 2.3. Dithiocoumarins Dithiocoumarins were more explored more than the above mentioned 2-thioxocoumarins. Few reports on new synthetic routes to achieve this scaffold have been recently reported . Thiocoumarins, presenting a variety of functional groups, were prepared from the corresponding thiophenols and diketene via cyclocondensation . Thionation of the obtained products with Lawesson’s or Davy’s reagent led to the corresponding thiono- or dithiocoumarins. These compounds are also used as starting materials to obtain more complex molecules. For most of the reactions involving dithiocoumarins as starting materials, the 4-hydroxy derivative is the selected molecule. This molecule is being used since the 1980s, when an efficient protocol was described to prepare it . Following this protocol, 2’-chloroacetophenones react with carbon disulfide in the presence of sodium hydride to form 4-hydroxydithiocoumarin anions . Kinetic protonation provides the desired 4-hydroxydithiocoumarins. Alkylation of these precursors may provide S-alkyl derivatives. As previously mentioned, reactions using different dithiocoumarins as starting materials have been described. An oxidative cross-coupling reaction of 4-hydroxydithiocoumarin and amines/thiols using a combination of iodine and tert-butyl hydroperoxide (TBHP) to provide access to lead molecules for biomedical applications was recently reported . This simple and versatile reaction allowed the attainment of a big family of compounds: Sulfonamides, disulfides, and sulfides. The authors claim an unprecedented, fast, mild, and environmentally friendly S–C bond formation, in addition to S–N and S–S bonds, in moderate to excellent yields. A reaction starting from different dithiocoumarins, in the presence of nitroalkenes allowed for the formation of tricyclic molecules . This reaction is catalyzed by potassium carbonate, which allows the closure of the new five members ring. This is an unprecedented and efficient method via a thio[3+2] cyclization reaction of 4-hydroxydithiocoumarins and trans-β-nitrostyrenes. This protocol has been described as faster than the previously reported, following mild conditions, presenting good yields, allowing for the formation of C–C and C–S bonds in a regioselective manner . As for thiocoumarins, 4-hydroxy-dithiocoumarins are the most studied starting materials for obtaining new and more complex molecules. The reactions and yields are similar to the previously described analogues. Few other authors described synthetic routes using different dithiocoumarins as starting materials to obtain more complex molecules (usually presenting extra rings). Few of these reactions follow green chemistry, claimed as more efficient, faster, and regio- and stereoselective . Knoevenagel-hetero-Diels–Alder reaction as catalyst-free reaction was also described . Novel pentacyclic thiochromone-annulated thiopyranocoumarins were obtained starting from 4-hydroxydithiocoumarin and O -acrylated salicylaldehydes. The main advantage is that this reaction is high regio- and stereoselective. Regarding the application of these compounds, the use of optically active devices has been described and patented . 3-Aryl derivatives and hydrophilic compounds are particularly interesting for this application. As described for thiocoumarins and 2-thioxocoumarins, dithiocoumarins have been studied as carbonic anhydrase inhibitors . In fact, the reported review describes the synthetic methodologies and application of coumarins and their bioisosteres as carbonic anhydrases inhibitors. The most important chemical features to increase the activity of these molecules is described. A patent also claims the use of these molecules with this same activity and their application in the treatment of cancer . Finally, the use of dithiocoumarins as herbicidal agents, has been described and patented . 4-Phenoxy derivatives have been closely related to this activity. They were obtained starting from the well-known and used 4-hydroxydithiocoumarins.
In the 1980s, thiocoumarins already had clinical use as haemorrhagic agents, anticoagulants (interfering with vitamin K-dependent coagulation factors), and antiallergic agents . In fact, some thiocoumarin derivatives found important applications as anticoagulant rodenticides pesticides . During the decade of the 1980s, the versatile synthesis from the acrylic and propiolic ortho esters and benzenethiols was described . After those first reports, some other evidences on the interest of thiocoumarins as potential bioactive molecules have been published. However, a deeper understanding of biology and biological functions, and how these molecules really work, has to be acquired before considering thiocoumarins as a trend topic in medicinal chemistry. One of the most recent subjects of interest related to this scaffold is the role of thiocoumarins as carbonic anhydrase inhibitors . The described thiocoumarins, in particular the 6-hydroxy-2-thioxocoumarin, proved to bound to the human carbonic anhydrase II active site in a completely different mode compared with coumarins, commonly hydrolyzed by the esterase carbonic anhydrase to the corresponding 2-hydroxycinnamic acids. The role of thiocoumarins inhibiting carcinogenesis has also been studied over the last decade . Moreover, few indole derivatives were tested as potential drugs for PUVA photochemotherapy . Furthermore, ethers of thiocoumarin, at position 4, have been reported as nitric oxide synthase inhibitors in the nanomolar range . Synthetically, most of the described thiocoumarins are obtained from the 4-hydroxythiocoumarin. Recently, a review collected the relevant information on the utility of this scaffold in organic chemistry . Detailed synthetic methodologies, structures, and chemical properties of the 4-hydroxythiocoumarin were mentioned in this overview, as well as its most interesting bioactivities. In fact, this compound is easily obtained and modified to achieve key educts for the synthesis of heterocyclic systems. Few papers per year describe some new advances in the field, related to both synthesis and biological applications. In 2019, there are only four relevant references on this topic, with the first one being a chemical mechanistic approach on the formation of two-center three-electron bonds by the hydroxyl radical induced reaction of thioesculetin . The second study is focused on the anticonvulsant activity of a new series of synthetic 4-amino-3-nitrothiocoumarins . The traditional synthesis starting from the thiophenol and the malonic acid, in the presence of POCl 3 and AlCl 3 with prolonged heating, is commonly used to afford the 4-hydroxythicoumarin, in low yield . The yields described for the first time for this traditional reaction were between 50 and 90%. However, further reports for other substituted thiocoumarins were lower than these. In this recent manuscript , an alternative two-step method has been proposed by the authors . An intermediate dithiophenyl ester of malonic acid is formed in high yield, using moderate conditions. In a second step, this intermediate undergoes a cyclocondensation in the presence of AlCl 3 as a Lewis acid to give the final product in moderate to high yield. This new method has been described to provide an overall yield of 60%, which is considerably higher than the previous reports. Therefore, this can be considered a step forward in the field. The third paper published in 2019 is focused on the design and synthesis of new anticonvulsants . 3-Nitrocoumarins proved to be more active than the studied thiocoumarins. Finally, the synthesis of 4-sulfonylthiocoumarins was also described in 2019 . Starting from the 4-hydroxythiocoumarin, via DABCO-catalyzed direct sulfonylation of 1-sulfonyl-1,2,3-triazoles, new 4-sulfonylthiocoumarins were efficiently obtained. The main advantage of this method is the lack of transition metal catalysts and extra oxidants. As previously mentioned, most of the traditional synthetic routes to afford the thiocoumarin scaffold involve a Lewis acid, as AlCl 3 ( A) . This reaction may be followed by nickel-catalyzed intramolecular recombination fragment coupling of the thioester to afford the corresponding benzothiophene. To show the applicability of this methodology, the 5-methyl benzothiophene has been obtained via an intramolecular decarbonylative strategy. Previously, preparation of thiocoumarin has been easily achieved via Friedel–Crafts type intramolecular cyclization of the precursor . Complex thiocoumarins have been synthetized from simple ones, and not only 4-hydroxythiocoumarin is used as starting material for the synthesis of more complex thiocoumarins. 4-Chlorothiocoumarin is also used for acylating thiocoumarins to afford γ-ketoenones ( B) . The authors described a one-step organocatalytic synthesis of 4-acylcoumarins from 4-chlorocoumarin, reporting the first examples of nucleophilic substitutions at the β-carbons of enones to afford γ-ketoenones . Few scientific papers described the use of simple thiocoumarins as starting materials for more complex structures. As an example, a novel, base-catalyzed and highly diastereoselective direct Michael addition-isomerization was described for the efficient synthesis of Rauhut–Currier-type adducts. An unexpected α-addition of γ-butyrolactam onto the 3-acylthiocoumarin derivatives was observed rather than the γ-addition, which is more common . This reaction may end with obtaining chalcones. Despite being an interesting alternative, it has not been explored in further works. Another example is the synthesis of (±)-thia-calanolide A, successfully accomplished using 3,5-dimethoxythiophenol as starting material, following a six-step reaction in an overall yield of 4.5% . The key reaction involved a Friedel–Crafts tigloylation of 5,7-dihydroxy-4-thiocoumarin. Microwave was presented as an alternative to include lateral chains at position 3 of the 4-hydroxythiocoumarin, via Mannich reaction, to achieve compounds with antibacterial properties . This reaction is clean, versatile, and described with high yields. 4-Hydroxythiocoumarin is also the starting material for Michael reactions at position 3 of the scaffold ( C) . A primary amine-derived organocatalyst modified with an ionic group for asymmetric Michael reactions of C-nucleophiles with α,β-unsaturated ketones was synthesized. In the presence of this catalyst and an acidic co-catalyst, 4-hydroxythiocoumarin reacted with benzylidene-acetones or cyclohexenones to afford the corresponding Michael adducts in high yields (up to 97%) and with reasonable enantioselectivity (up to 80%). 4-Hydroxythiocoumarin is also the starting material for an easy synthesis of 4-acetylthiocoumarin via very high α-regioselective Heck coupling on tosylates ( D) . This has been described as an efficient and versatile reaction. 4-(4′-Aryloxybut-2′-ynylthio) thiocoumarins were obtained via tosylation of 4-hydroxythiocoumarin followed by mercapto-substitution and condensation with 1-aryloxy-4-chlorobut-2-ynes . Via sequential thermal and catalyzed Claisen rearrangements, and starting with the 4-hydroxythiocoumarin, it is possible to achieve the thiocoumarin-annulated furopyran moiety . Pyrrole derivatives, at positions 2,3 of the thiocoumarins, have been also described, following a tandem [2,3] and [3,3] sigmatropic rearrangement . Thiophenyl derivatives have been described by the same group following a similar approach . The synthesis of 3-aminomethylenethiocoumarins is another case of how to increase a lateral chain in thiocoumarins . In the article, a three-component synthesis by condensation of amines, α-amino acids, ureas, and carbamates with 4-hydroxythiocoumarin, in the presence of tri-ethylorthoformate, is described. Even if there is a temptation to compare these reactions, most of them are specific for some substitution patterns, and may only give specific thiocoumarin derivatives. Therefore, it is interesting to have an arsenal of different options to be able to synthetize more derivatives to establish structure–activity relationships. One of the biggest molecules achieved starting from simple thiocoumarins is the novel 6,6′-arylidene-bis-[5-hydroxy-9-methyl-2,3-diarylthieno[3,2- g ]thiocoumarins]. As the biscoumarins, these molecules were tested for their antimicrobial profile . Similarly, and inspired on a class of furocoumarins, the angelicins, thioangelicins were synthetized and studied for PUVA chemotherapy . These are two cases in which coumarins inspired the study of similar families of thiocoumarins. 7-Hydroxy-4-methylthiocoumarin is the precursor for a pyrano thiocoumarins, reported for their anti-HIV potential . From this series, an excellent derivative, with an EC 50 in the low nanomolar range, was described. 4-Mercaptothiocoumarin is also a versatile starting material. It can be alkylated with different propargylic and allylic halides under phase-transfer-catalyzed conditions in the presence of tetrabutylammonium bromide or benzyl triethylammonium chloride catalyst in dichloromethane/NaOH solution, at room temperature . Then, these 4-thiopropynyl- and thioallylthiocoumarins can be the starting material for different derivatives. A tandem sp3 C–H functionalization followed by decarboxylation of 2-alkylazaarenes with thiocoumarin-3-carboxylic acid has also been used to achieve 4-substituted 3,4-dihydrothiocoumarins . This catalyst-free reaction afforded a series of 29 azaarene-substituted 3,4-dihydrothiocoumarins. Another catalyst-free tandem reaction, this time a Michael addition followed by decarboxylation, was also described to achieve 4-substituted thiocoumarins . This reaction uses the same thiocoumarin-3-carboxylic acid as starting material, together with 2-methylindole, in the same conditions of 1,4-dioxane, at 120 °C, this time to give 3-indolyl-substituted 3,4-dihydrothiocoumarins. Few thiocoumarins have also been explored in the design of novel carrier systems. The development of egg shell-like nanovesicles using the thiocoumarin-3-carboxylate has been described . These drug delivery vehicles have been used as a potential carrier for the bacteriostatic antibiotic sulfamethoxazole. The spectroscopic studies as well as the growth inhibition of E. coli exhibit that this formulation leads to pH responsive sustained release of the drug. Thiocoumarins have been recently prepared using microwave irradiation, in a versatile Lewis acid-catalyzed reaction . Functional groups as pyrimidine and 1,3,4-oxadiazole have been introduced in the scaffold. The technique has been described as efficient, selective, fast, and clean. The series of compounds has been reported for the anti-microbial and antioxidant activities. Thiocoumarins can also be converted to dithiocoumarins using Lawesson’s reagent, in a reaction with ~30–40% yield . This reaction may contribute to deeply exploring these analogues, which are less studied due to the scarcity and rarity of starting materials. As some simple coumarins, thiocoumarins have been recently described as fluorescent probes and sensors. A thiocoumarin, named 7-( N , N -diethylamino)-4-trifluoromethyl-2-thiocoumarin, has been used as sensor for heavy metal pollution and bacterial contamination. Its specific thiocarbonyl scaffold as donor-π-accepter electronic properties, provides intriguing optical properties and several applicable functions . A thiocoumarin-containing ratiometric fluorescent probe has been described for the simultaneous detection of hypochlorite and singlet oxygen . Once again, the thiocarbonyl moiety has been highlighted as an excellent alternative to traditional probes. Moreover, these applications have been described for live-cell imaging, with the thiocoumarin fluorescent probe turned-on once inside the cells . The potential of these molecules for imaging has been recently claimed in a patent . These new applications may revolutionize the field, making thiocoumarins more visible to the academic community.
As previously described for the thiocoumarins, the 2-thioxocoumarins have been studied as carbonic anhydrase inhibitors . The authors presented the X-ray crystal structure of the 6-hydroxy-2-thioxocoumarin on the human carbonic anhydrase II active site, and these results revealed an unprecedented and unexpected inhibition mechanism for these new inhibitors when compared with isostructural coumarins. It was proved that the exo-sulfur atom can link to a zinc-coordinated water molecule, while other parts of the scaffold can establish important interactions with different amino acids of the binding pocket. Compared with simple coumarins with the same substitution patterns, the inhibition mechanism is totally different. These molecules proved to be hydrolyzed, occluding the entrance of the binding pocket. This is a step forward in the design of efficient carbonic anhydrase inhibitors. 7-Diethylamino-4-hydroxymethyl-thiocoumarin (thio-DEACM) caged molecules are recently attracting attention. The thio-DEACM as a caging group has great properties, such as rapid blue-cyan light responsiveness and avoiding UV irradiation of cells in combination with the absence of toxicity of the released photocage . As several other coumarin derivatives, 4-methylthiocoumarins have been described as a multitarget-directed ligand for the treatment of Alzheimer’s disease . A novel 4-methylthiocoumarin derivative has been studied against acetylcholinesterase, butyrylcholinesterase, BACE1, β-amyloid aggregation, and oxidative stress involved in the pathogenesis of this neurodegenerative disease.
Dithiocoumarins were more explored more than the above mentioned 2-thioxocoumarins. Few reports on new synthetic routes to achieve this scaffold have been recently reported . Thiocoumarins, presenting a variety of functional groups, were prepared from the corresponding thiophenols and diketene via cyclocondensation . Thionation of the obtained products with Lawesson’s or Davy’s reagent led to the corresponding thiono- or dithiocoumarins. These compounds are also used as starting materials to obtain more complex molecules. For most of the reactions involving dithiocoumarins as starting materials, the 4-hydroxy derivative is the selected molecule. This molecule is being used since the 1980s, when an efficient protocol was described to prepare it . Following this protocol, 2’-chloroacetophenones react with carbon disulfide in the presence of sodium hydride to form 4-hydroxydithiocoumarin anions . Kinetic protonation provides the desired 4-hydroxydithiocoumarins. Alkylation of these precursors may provide S-alkyl derivatives. As previously mentioned, reactions using different dithiocoumarins as starting materials have been described. An oxidative cross-coupling reaction of 4-hydroxydithiocoumarin and amines/thiols using a combination of iodine and tert-butyl hydroperoxide (TBHP) to provide access to lead molecules for biomedical applications was recently reported . This simple and versatile reaction allowed the attainment of a big family of compounds: Sulfonamides, disulfides, and sulfides. The authors claim an unprecedented, fast, mild, and environmentally friendly S–C bond formation, in addition to S–N and S–S bonds, in moderate to excellent yields. A reaction starting from different dithiocoumarins, in the presence of nitroalkenes allowed for the formation of tricyclic molecules . This reaction is catalyzed by potassium carbonate, which allows the closure of the new five members ring. This is an unprecedented and efficient method via a thio[3+2] cyclization reaction of 4-hydroxydithiocoumarins and trans-β-nitrostyrenes. This protocol has been described as faster than the previously reported, following mild conditions, presenting good yields, allowing for the formation of C–C and C–S bonds in a regioselective manner . As for thiocoumarins, 4-hydroxy-dithiocoumarins are the most studied starting materials for obtaining new and more complex molecules. The reactions and yields are similar to the previously described analogues. Few other authors described synthetic routes using different dithiocoumarins as starting materials to obtain more complex molecules (usually presenting extra rings). Few of these reactions follow green chemistry, claimed as more efficient, faster, and regio- and stereoselective . Knoevenagel-hetero-Diels–Alder reaction as catalyst-free reaction was also described . Novel pentacyclic thiochromone-annulated thiopyranocoumarins were obtained starting from 4-hydroxydithiocoumarin and O -acrylated salicylaldehydes. The main advantage is that this reaction is high regio- and stereoselective. Regarding the application of these compounds, the use of optically active devices has been described and patented . 3-Aryl derivatives and hydrophilic compounds are particularly interesting for this application. As described for thiocoumarins and 2-thioxocoumarins, dithiocoumarins have been studied as carbonic anhydrase inhibitors . In fact, the reported review describes the synthetic methodologies and application of coumarins and their bioisosteres as carbonic anhydrases inhibitors. The most important chemical features to increase the activity of these molecules is described. A patent also claims the use of these molecules with this same activity and their application in the treatment of cancer . Finally, the use of dithiocoumarins as herbicidal agents, has been described and patented . 4-Phenoxy derivatives have been closely related to this activity. They were obtained starting from the well-known and used 4-hydroxydithiocoumarins.
Even if sulfur-containing functional groups are present in a broad range of drugs and natural products, thiocoumarins have been rarely explored as drug candidates. Few examples of synthetic routes, as well as the lack of starting materials, may be responsible for the lack of biological studies related to this family of compounds. Furthermore, coumarins have been extensively explored in different scientific fields, overshadowing their sulfur analogues. The most relevant examples of described thiocoumarins are compiled in this review. The role of thiocoumarins as chemical probes and sensors may provide a new life to these compounds for imaging and diagnosis purposes.
|
Human-correlated genetic models identify precision therapy for liver cancer | 64aee839-8c89-475f-95f8-6cf793b8e737 | 11922762 | Medicine[mh] | Precision medicine for patients with advanced HCC has lagged behind other cancers. This is not because HCC has no discernible subtypes, but because targeting these has proved challenging. Tyrosine kinase inhibitors (TKIs; such as sorafenib and lenvatinib ) were the only first-line treatments for unresectable HCC until 2020. Thereafter, the IMbrave150 study (atezolizumab with bevacizumab) highlighted the potential of combination approaches with immune checkpoint inhibition (ICI) therapy, with enhanced responses for some patients and improved overall survival. Alongside advances in treatment options came an increased appreciation that heterogeneous treatment responses in patients with HCC provide a potential for patient stratification , . The lack of necessity for clinical biopsies in advanced HCC has resulted in a lack of tissue from late-stage disease. This hinders advances in defining clinically relevant stratification biomarkers and mechanistic understanding within subtypes for these patients. Preclinical models offer a biological platform for disease interrogation but, currently, few models faithfully recapitulate the complexity of human disease or have been validated against transcriptomic and phenotypic human HCC profiles , . There is therefore currently a need for human-relevant preclinical models to investigate therapy efficacies, providing guidance on subtype-specific treatments for different patient populations.
To address this need, we first set out to generate a broad range of mouse models guided by the most commonly found genetic drivers of human HCC . Human HCC is thought to evolve from a hepatocytic clonal origin under specific conditions promoting carcinogenesis, in contrast to recently described non-malignant clonal expansion , – . We reproduced this aspect of cancer biology in our models by introducing the genetic alterations into adult mouse hepatocytes using conditional recombination technology and allowing the premalignant clones to evolve to HCC over time. We intravenously injected adult mice with a viral vector encoding Cre recombinase with a hepatocyte tropism due to its thyroxine-binding globulin (TBG) promoter, AAV8.TBG.cre. This drove recombination of endogenous floxed alleles in individual hepatocytes in an immunocompetent environment (Fig. ). AAV8 was titrated to a dose (6.4 × 10 8 genomic copies (GC) per mouse) that resulted in solitary hepatocyte targeting at low frequency (approximately 1%) and was highly hepatocyte specific (Extended Data Fig. ). Recombination occurred primarily in the first 5 days after injection, was observed across all three hepatocyte zones , but was significantly different between male and female mice (Extended Data Fig. ). This led to a lower tumour count and consequently extended survival in female mice after induction of HCC-related oncogenes (Extended Data Fig. ). Furthermore, varying the induction dose or mutational burden affected the tumour occurrence and the speed of progression to the end point (Extended Data Fig. ). We next applied this strategy to a broad range of HCC-relevant oncogene/tumour suppressor genes using a standardized dose in male mice unless otherwise stated. We particularly focused on genes identified by a TCGA study belonging to the WNT pathway, the cell cycle or the RTK–RAS–PI3K pathway growth. These genes were tested in multiple combinations with each other for their potency in tumour induction. (Fig. ). We included models with combinations that co-occur in early disease, such as CTNNB1 + MYC or PTEN + TP53 . However, we also included combinations that tend towards mutually exclusive in early disease but not in late-stage disease, such as CTNNB1 + TP53 (Extended Data Fig. ). We decreased the AAV induction titre in specific instances (cohorts 14, 15, 23 and 24, 1.28 × 10 8 GC per mouse) to reduce the clonal burden, facilitating progression of these more aggressive models to larger individual tumours. Genotyping of end-stage tumours confirmed a high fidelity of recombination in the alleles targeted by the AAV induction (97.4–100%) (Extended Data Fig. ). We monitored 35 genetically distinct models, including models with a whole-body knockout of Cdkn1a or Cdkn2a , for liver nodule growth for a minimum of 230 days after induction (Extended Data Figs. and ). The majority of our models (83%) developed end-stage tumours within the study timeframe and most (69%) showed a tumour penetrance of higher than 50%. Notably, some combinations, such as MYC overexpression + Trp53 alteration, which induced HCC in some but not all previously published models , , had very low to no tumour penetrance using our clonal evolution approach and did not reach end-stage tumours within the observed period. Reflective of human disease, we observed intratumoural haemorrhaging and/or rupture (bleeding) as well as metastatic spread to the lungs, one of the main metastatic sites in human HCC, together with bone and lymph nodes , (Fig. and Supplementary Table ). We observed a negative correlation between an increased number of driver mutations and survival, despite a reduced clonal induction with a lower AAV titre, and a positive correlation between an increased number of driver mutations and tumour proliferation, as well as between mutational burden and lung metastasis in our cohorts (Extended Data Fig. ). Tumour haemorrhage did not correlate significantly with mutational burden but occurred predominantly in cohorts with a mutational pattern showing activated Ctnnb1 and Pten loss without MYC overexpression (Extended Data Fig. ). Macroscopic and microscopic appearances were consistent with human HCC and covered a wide range of histological subtype phenotypes microscopically. This included well-differentiated HCC (for example, cohorts 5 + 19), undifferentiated HCC (such as cohorts 23 + 28), pseudoglandular HCC (for example, cohort 30) and steatotic HCC (for example, cohort 35) (Fig. and Extended Data Fig. ). Lung metastatic lesions reflected primary tumour histopathology (Fig. ). Histopathological assessment of morphological parameters is currently the gold standard for differential diagnosis of liver cancer in patients . They showed strong similarities to human HCC histopathology, including typically observed architectural patterns (trabecular, glandular, solid and nested) and cytological atypia. Different combinations of genetic alterations resulted in distinct morphologies (Fig. ). In summary, we used combinatorial genetic alterations, relevant to human HCC, to drive the development of autochthonous tumours in 27 immunocompetent mouse models. Tumour growth happened progressively over several months with individual hepatocytes as the cell of origin. These models recreate key features characteristic of human HCC biology, including histopathological phenotypes and metastatic spread.
To determine how well our models further represent human HCC, we performed unbiased transcriptional analysis. We included a range of well-established carcinogen-induced (TOX) and orthotopic transplant (OT) HCC mouse models with our genetically engineered mouse models (GEMMs) to make this comparison more comprehensive (Fig. ). Using nonlinear dimensionality reduction (uniform manifold approximation and projection, UMAP ) we mapped mouse end-stage HCC data onto the human HCC data (Fig. ). Individual models, both genetically modified and non-genetically modified, clustered within different regions in the UMAP plot (Extended Data Fig. ). However, mutational status is not always indicative of signalling status , and genomic profiling of human HCC previously showed that mutations are not exclusively prognostic of association with specific subtypes . This is especially relevant for advanced disease stages with a relatively high mutational burden , where different genetic alterations can influence each other. We show that, for example, mutations in CTNNB1 / Ctnnb1 (human/mouse gene) do not always lead to upregulation of expression of downstream pathway targets ( GLUL / Glul , LGR5 / Lgr5 , LECT2 / Lect2 or NOTUM / Notum ) in human or mouse HCC (Extended Data Fig. ). Our mouse data also support the observation that mutational status by itself is not always predictive of the resemblance between cohorts (Extended Data Fig. ). We therefore went on to compare the human and mouse transcriptome data based on functionally and mechanistically relevant pathway enrichment. We used the Louvain method for community detection to identify groups in our human/mouse HCC dataset (Fig. ). We detected four major human/mouse (HuMo) clusters (Fig. ). Genetic mouse models are represented in all four clusters with varying heterogeneity within cohorts, whereas the purely carcinogen-induced models are representative of only HuMo cluster 2 (Fig. ). Pathway enrichment analysis could establish cluster-specific characteristics. HuMo cluster 1 was enriched for pathways linked to metabolism and differentiation, but had negative enrichment for proliferation and inflammatory pathways. HuMo cluster 2 was related to cluster 1 but was distinct particularly through a higher enrichment in pro-inflammatory pathways. HuMo clusters 3 and 4 were both poorly differentiated and highly proliferative, with cluster 4 showing enrichment in epithelial-to-mesenchymal transition (Fig. ). To assess whether the transcriptional clustering corresponded to similar histopathological features in mice and human HCC within the same cluster, we compared our mouse tumours to TCGA tissue . We observed that mouse and human tissue belonging to the same HuMo cluster did indeed have analogous morphological characteristics (Extended Data Fig. ). Tissue from HuMo cluster 1 showed well-differentiated HCC (Extended Data Fig. ). HuMo cluster 2 tissue presented with inflammation, steatosis and steatohepatitis (Fig. and Extended Data Fig. ). HuMo cluster 3 and 4 tissue displayed deposition of extracellular matrix and moderately (cluster 3) to poorly (cluster 4) differentiated HCC (Fig. and Extended Data Fig. ). We next validated our classification in a previously published, independent dataset of human HCC . The patients could all be assigned a HuMo cluster with similar distribution dynamics to the TCGA dataset. Again, HuMo cluster 1 was enriched for immune-evasive signatures, including the immune-excluded subclass , and was de-enriched for ICI response signatures, including the IFNAP signature . Conversely, HuMo cluster 2 had higher inflammatory signalling signatures and was enriched for immune-active tumours , but without WNT–β-catenin activation. HuMo clusters 3 and 4 featured a strong progenitor signature (CK19 mutation signature) consistent with the previously observed histological phenotype of these clusters. Only HuMo cluster 4 was significantly enriched for the inflamed HCC class with an immune-exhaustion signature and characterized by TGFβ and EMT signatures (Fig. and Extended Data Fig. ). When comparing survival across the species, there was general correlation between patients and the respective GEMMs across a range of molecular subtype classifications, including HuMo, Hoshida and Chiang . Importantly, HuMo offers a distinct patient classification. This clustering approach distinguished two patient populations within the Hoshida S3 molecular subclass, namely HuMo clusters 1 and 2. Hoshida et al. implied that S3 might consist of two subpopulations with CTNNB1 as a dividing factor, but did not use this as a factor in their classification . This distinction in our analysis resulted in differences in patient survival that were unappreciated when using the Hoshida classification; patients associated with HuMo cluster 2 had an improved survival probability relative to patients associated with the other HuMo clusters. Furthermore, this distinction separates the immune-excluded (HuMo cluster 1) from the immune-active (HuMo cluster 2) subclasses. It also surpasses previous attempts of comparing mouse and human HCC data in scale and detail , (Extended Data Figs. and and Supplementary Table ). In brief, we identified four distinct clusters, common across human and mouse models, by integrating our mouse transcriptional data with human HCC transcriptional data. Our models recapitulate transcriptionally the full range of human HCC, including within individual clusters. This aligned with similar histopathological features and relative survival within clusters, with specific GEMMs representative of individual subtypes of human HCC. Moreover, our HuMo classification is able to discriminate between HCC with WNT–β-catenin activation (HuMo1) and those without WNT–β-catenin activation (Humo2) within non-proliferative tumours (Hoshida S3).
To examine the translational potential of our models, we investigated response to standard-of-care treatments. We focused on one model in a proof-of-principle set of experiments. Approximately 30% of patients with HCC have mutations leading to activation of the β-catenin signalling pathway . HCC with activated β-catenin signalling has a low enrichment score for immune signatures and has been, in most cases, associated with immune exclusion , . Furthermore, active β-catenin pathway signalling has been linked to ICI resistance in a prospective HCC cohort study , suggesting a need for alternative treatment options for this patient subgroup. In the TCGA dataset, 65% (57 out of 88) of patients with mutations in CTNNB1 were associated with HuMo 1 and made up 47% (57 out of 118) of patients in that cluster (Extended Data Fig. ). Moreover, humans and mice associated with HuMo cluster 1 had immune-cell paucity and a low immune score (Fig. and Extended Data Fig. ). We therefore identified HuMo cluster 1 as the one most likely to correspond to the group of patients with activated β-catenin pathway signalling that would benefit from alternative treatment options. Cohort 5 mice ( Ctnnb1 ex3/WT R26 LSL-MYC/LSL-MYC , hereafter BM mice) were used as a representative model and showed phenotypic resemblance to human CTNNB1 -mutated HCC. We aimed to mimic the treatment of established tumour lesions. We therefore first performed a time-course analysis for tumour onset in the BM mouse model (cohort 5) to determine an appropriate timepoint for the start of treatment. We observed clonal induction of hepatocytes, which evolved over time into microscopic lesions and then macroscopic tumour nodules, with glutamine synthetase (GS) as a marker of β-catenin driven tumour induction (Fig. ). Tumour evolution from single clones led to moderate intertumoural and intermurine transcriptional heterogeneity in end-stage tumours, including activation of pro-tumorigenic pathways such as proliferation or angiogenesis. However, while gene expression in tumours was markedly different to non-tumour tissue, it was also consistently different compared with livers with a global hepatocytic short-term expression of the same oncogenes (Extended Data Fig. ). This implied a consistent trajectory of clonal evolution occurring during tumour progression , . Relevant long-term models in which this evolution can take place are essential for studying HCC in preclinical models. We started drug treatment at day 90, based on 100% of cohort 5 (BM) mice having macroscopic tumours and 96% of cohort 5 (BM) mice surviving past this timepoint (Fig. and Extended Data Fig. ). Cohort 5 mice showed a significant increase in survival after treatment with the TKIs sorafenib and lenvatinib (Fig. ). However, treatment with the ICI agent anti-PD1 or treatment with ICI + VEGFRi (modelling atezolizumab + bevacizumab as first-line HCC systemic therapy ) did not impact the overall survival in this cohort (Fig. and Extended Data Fig. ). These results are similar to the reported drug responses to TKIs and ICI in human patients with activated β-catenin signalling . In mice from the immune-active HuMo cluster 2, ICI + VEGFRi resulted in an improved survival (Extended Data Fig. ). This is also consistent with the transcriptomic signatures predicting ICI response (Extended Data Fig. ). Investigating disease progression after initial response to therapy, we observed changes in macroscopic and microscopic appearances in end-stage tumours of cohort 5 (BM) mice treated with lenvatinib. Tumours were different in colour and stiffer. Microscopic HCC patterns shifted from mostly well-differentiated to a poorer differentiated phenotype with a greater stromal presence (Extended Data Fig. ). Furthermore, more mice in this treatment arm presented with lung metastases compared with vehicle treatment or other treatments (Extended Data Fig. ). Monitoring of tumour growth using magnetic resonance imaging suggested a delayed and decreased tumour growth initially after lenvatinib treatment (Extended Data Fig. ). We also observed a higher metastatic burden in a second model (cohort 23, Ctnnb1 ex3/WT R26 LSL-MYC/LSL-MYC Pten fl/fl Trp53 R172H/WT Cdkn2a KO/KO ) with increased survival after lenvatinib treatment (Extended Data Fig. ). We hypothesized that the increased aggressiveness, manifested by morphological changes and greater metastatic burden, resulted from the extended survival coupled with an altered phenotype associated with acquired resistance to lenvatinib therapy. We therefore investigated livers of cohort 5 (BM) mice after 15 days and 30 days of lenvatinib treatment from day 90 after induction (Fig. ). We observed no differences in tumour morphology, but there was a decreased tumour burden through less proliferation, without increased cell death, at both the 15 and 30 day timepoints in lenvatinib-treated mice (Fig. ). There were no detectable metastases at either timepoint, supporting our hypothesis that the heightened aggressiveness in this model is a late-stage on-treatment event. Overall, treatment responses in this specific GEMM were reminiscent of a distinct, common and difficult-to-treat subtype of HCC, characterized by a transient survival benefit observed in human phase 3 clinical studies , .
After establishing the response to current standard-of-care treatments of mice representative of HuMo cluster 1 (cohort 5, BM), we concentrated on identifying therapeutic options for this difficult-to-treat subgroup. We performed an in vitro high-throughput screen based on GEMM-derived HCC organoids (HCCOs) , with subsequent in vivo validation in the respective GEMM (Extended Data Fig. ). HCCOs recapitulate the transcriptomic profile, histological organization and tumorigenic potential of the primary tumour , and are therefore suited to investigate drug effects on tumour cells. They allow for rapid testing of a large range of drugs and for a side-by-side comparison between mouse-derived and human-derived tumour cells. HCCOs derived from end-stage tumours of cohort 5 (BM) mice expressed β-catenin, its downstream target GS and retained MYC overexpression as well as markers of proliferation (Ki-67) and differentiation (HNF4a), features that are shared with the corresponding primary tumour (Extended Data Fig. ). Despite these similarities, the transcriptional phenotype of HCCOs differed from the original tumours. We propose that this is due to the simplified nature of HCCOs as an epithelial-cell-only model as well as adaptive response to the culture conditions. Overall, there was a convergence of HCCO phenotype arising from diverse GEMMs (Extended Data Fig. ). We tested a comprehensive drug library consisting of the 147 FDA-approved anti-cancer drugs available at the time (June 2019) plus internal controls and analysed their effect on HCCO growth (Fig. and Supplementary Table ). The most efficacious drugs were a group of antimetabolites—nucleobase analogues that interfere with DNA synthesis (Fig. and Extended Data Fig. ). We validated the dose-dependent effect of cladribine, the most effective antimetabolite, in several distinct mouse and human HCCOs. This confirmed the results of the screen and demonstrated the nanomolar potency of cladribine (Fig. ). To establish whether this is a compound-specific effect, we tested a wide variety of antimetabolites and validated the high-throughput screen results. We demonstrated similar efficacy of clofarabine (a second-generation version of cladribine) and cladribine itself, suggesting a drug-specific on-target effect within this subclass of antimetabolites (Extended Data Fig. ). We also tested selected other drugs from our screen. Lenvatinib and sorafenib showed little tumour-epithelial efficacy in both the screen and separate validation, including in combination with cladribine (Extended Data Fig. ). Next, we treated cohort 5 (BM) mice, representing HuMo cluster 1, with either cladribine monotherapy or combination therapy of cladribine and lenvatinib, as a standard-of-care TKI (Fig. ). Cladribine monotherapy led to increased survival, but combination therapy extended survival further (Fig. ). Cladribine monotherapy reduced the number of tumours, but the remaining tumours still progressed to end-stage HCC. Combination therapy with lenvatinib showed a synergistic effect, almost completely eradicating all tumours (Fig. ). Study progression to either clinical tumour end point or study end point (day 270 after induction) was limited in some animals (31% cladribine, 62% cladribine + lenvatinib) due to clinically substantial weight loss (<80%). Treatment with either monotherapy or combination therapy showed a decrease in proliferation in end-stage tumours, but no alteration in apoptotic cell death. Notably, we observed an increase in CD3 + T cell infiltration into the tumour after combination therapy compared with vehicle treatment (Extended Data Fig. ). As the time of end point varied greatly between the different treatments, we analysed tumours at a defined timepoint of 30 days after the treatment start. Mice on monotherapy or combination therapy showed decreased tumour size and number, with a significant decrease in proliferation (Fig. and Extended Data Fig. ). Both healthy and tumour tissue exhibited a greater extent of DNA damage (pH2AX), as expected after treatment with a nucleobase analogue, but this did not alter upregulation of another senescence marker (p53) nor apoptosis (Extended Data Fig. ). Again, we observed increased infiltration of CD3 + T cells into the tumour of mice that were treated with combination therapy (Fig. ). Given the lymphocyte infiltration observed after combination therapy, we tested a ‘priming’ approach with 1 week of combination therapy before ICI therapy (Extended Data Fig. ). This resulted in anti-tumour efficacy and immune infiltration and cytotoxicity (Fig. and Extended Data Fig. ). Finally, we tested whether cladribine, either as monotherapy or in combination with lenvatinib, is equally effective in mouse models representing other HuMo clusters. We treated cohort 23 ( Ctnnb1 ex3/WT R26 LSL-MYC/LSL-MYC Pten fl/fl Trp53 R172H/WT Cdkn2a KO/KO ) mice, representing HuMo cluster 4, and cohort 45 ( R26 LSL-MYC/LSL-MYC Kras G12D/WT ) mice (induced with a higher titre of AAV.TBG.cre than cohort 32 to increase tumour burden to make survival time comparable to cohort 5), representing HuMo cluster 2 (Fig. ). Both monotherapy and combination therapy were effective in prolonging the survival of cohort 23 mice (Fig. ). However, cladribine did not extend the survival in cohort 45 mice (a wild-type (WT) Ctnnb1 model with mutated Kras ), either as a monotherapy or as a combination therapy with lenvatinib (Fig. ). In this proof of concept, we demonstrated the potential of our GEMM platform to identify epithelial-targeting therapies that synergized effectively with standard-of-care treatments, the latter of which mainly targeting the tumour microenvironment. This combination of TKI and a repurposed FDA-approved anti-cancer compound led to highly effective subtype-specific treatment responses and a switch to a targetable immune phenotype.
Using a range of genetic alterations that are frequently associated with human HCC , we developed a suite of immunocompetent mouse models that closely resembles the development and progression of human HCC with hepatocytes as the cell of origin. Our models successfully recreate key molecular and pathophysiological events typical of human HCC, including tumour haemorrhaging and metastasis to the lungs , . They mimic various tumour microenvironments, such as immune-active and immune-desert or high/low stroma tumours. We demonstrated the clinical relevance of our models by integrating mouse data with publicly available human HCC datasets, defining shared subtypes and proving response to standard-of-care treatment. Furthermore, we showed that these models can be used as a preclinical platform, together with HCCOs, for investigating rapid drug repurposing, in addition to studying tumour evolution and mechanisms of drug resistance. We appreciate that not all genetic alterations associated with HCC have been tested in this study. TERT promoter modifications, despite being frequent in human HCC (up to 60% of patients) , are difficult to model appropriately in mice due to biological differences between species. Mice have long telomeres and it would take several generations of crossing mice with Tert deletions before detecting a noticeable effect of reactivating Tert . This is an obstacle that will be difficult to overcome in mouse models of HCC and other means are needed to study TERT promoter mutations and their therapeutic targetability. However, as TERT promoter mutations are so omnipresent in HCC, they might be less relevant for subtyping and we did not identify a specific human TERT group that was separate from our GEMMs. Furthermore, some combinations of genetic alterations showed low/no tumour penetrance in our GEMMs, for example, Trp53 modifications in combination with MYC overexpression, while these showed high penetrance in HCC in previous models using hydrodynamic tail vein injections . Administration of hydrodynamic tail vein injection has been shown to cause apoptosis in the liver , leading to higher inflammation and favourable conditions for tumour development. Moreover, levels of MYC might be a determining factor in a clone progressing to a tumour . Although the majority of our studies were performed in male mice, we found no indication that the results are sex specific. Indeed, when we used the same genetic alterations in female mice, we observed a similar phenotype and cluster association (cohort 5/6, BM, male/female, respectively). However, AAV.TBG.cre induction seems to be less potent in female mice, which is particularly impactful in models with a lower mutational burden. Future experiments are needed to explore further genetic alterations or risk factors predominantly associated with female HCC in patient stratification, such as Bap1 mutations or malignant transformation of hepatocellular adenomas , . In contrast to the GEMMs, human patients usually present with cirrhosis, which probably influences the course of disease establishment and progression and impacts treatment options , . Future research incorporating multifaceted environmental factors in preclinical models, including advanced fibrosis, is needed to better understand HCC biology and potential differences between species. Our models can also be easily combined with environmental liver disease models, such as high-fat diets. Preliminary data from our transcriptomic analyses indicated that genetics dominate cluster association, with the addition of background fibrotic disease having little transcriptomic influence in mice (cohort 5 versus 37). Our models strike a balance between allowing time for tumour evolution while still being time efficient. This enables future detailed investigation of tumour evolution and factors contributing to malignant transformation, especially as not all of the recombined clones expand into tumours. Somatic mutations are poorly clinically actionable in HCC and remain difficult to target therapeutically. In the case of multiple genetic alterations, each individual contribution to tumorigenesis might be difficult to determine – . Our models with their increased complexity of multiple genetic alterations, similar to the mutational burden of late-stage HCC , enable the exploration of alternative targets and might contribute to understanding mutational dominance in different contexts. Moreover, by mimicking clonal evolution, they might help to identify the stage in tumour development—initiation, early nodule growth, malignant transformation—when a drug has an optimal effect. We show that HCCOs are a tractable and rapid platform to identify treatments in combination with efficacy testing in vivo, and promote the principles of the 3Rs (replacement, reduction and refinement) for humane animal research. However, current cell culture conditions limit the translatability of HCCO-based drug response predictions and, therefore, validation in animal models remains essential. Future research in HCCOs needs to overcome the reduced complexity in cell culture, a general issue in organoid culture , and address options for co-culture with cells shaping the tumour microenvironment . Modifying HCCOs with CRISPR technology may also provide useful insights to explore tumour biology and the mechanisms beyond drug vulnerabilities . Importantly, we show that our GEMMs map transcriptionally and histologically to human HCC. Using a computational biology approach has enabled us not only to position our GEMMs, and select carcinogen-induced models, against human HCC, but also to identify four shared subclasses with defining characteristics. Notably, some of our models show a degree of heterogeneity often observed in human HCC , with tumours associated to several HuMo clusters. Our newly developed GEMMs represent all identified subtypes, whereas chemical-carcinogen-induced models included in this study only mapped to one HuMo cluster (cluster 2). Our preclinical platform and classification system can be used as a resource for the HCC research community to streamline preclinical research and increase comparability of different mouse models. Furthermore, linking preclinical models with patient data can aid in stratifying patients to treatment, identifying new therapies and improving the likelihood of translational success. The HCCO screen enabled us to rapidly identify and test an FDA-approved anti-cancer drug, cladribine—not previously linked to HCC—in a clinically relevant model. We could show efficacy and improved survival in vivo together with standard-of-care treatment, which will allow for a swift translation into the clinic. We believe that our approach of linking preclinical models to human data in a subtype-specific manner will also be applicable, cross-referable and advantageous in translational research of other solid cancers.
Mice, diets and treatments All animal experiments were performed in accordance with UK Home Office licences (70/8891, PP0604995, 70/8646, 70/8468 and PP8854860) and in accordance with the UK Animal (Scientific Procedures) Act 1986 and EU direction 2010. They were subject to review by the animal welfare and ethical review board of the University of Glasgow and the University of Newcastle upon Tyne. To minimize pain, suffering and distress to the animals, we used single-use needles and non-adverse handling techniques. Mice were housed under controlled conditions (specific-pathogen free, 12 h–12 h light–dark cycle, 19–22 °C, 45–65% humidity) with access to food and water ad libitum. We added environmental enrichments, in the form of gnawing sticks, plastic tunnels and nesting material to all of the cages. Welfare of animals was defined by clinical symptoms, including visible masses, any degree of reduced mobility/distress, weight loss or evidence of haemorrhage; however no maximal tumour volume end points for intrahepatic tumours were mandated. No mouse exceeded the humane end points stipulated in the Home Office Licenses. Unless otherwise specified, male mice on a mixed background were used. The following transgenic mice strains were used: Gt(Rosa)26Sor tm14(CAG-tdTomato)Hze ( R26 LSL-Tom ) , Ctnnb1 tm1Mmt ( Ctnnb1 ex3 ) , Gt(Rosa)26Sor tm1(MYC)Djmy ( R26 LSL-MYC ) , Trp53 tm1Brn ( Trp53 fl ) , Trp53 tm2Tyj ( Trp53 R172H ) , Cdkn2a tm1.1Brn ( Cdkn2a KO ) , Pten tm2Mak ( Pten fl ) , Gt(Rosa)26Sor tm1(Notch1)Dam ( R26 LSL-NICD ) , Kras tm4Tyj ( Kras G12D ) , Cdkn1a tm1Led ( Cdkn1a KO ) , Axin1 fl (ref. ), Bap1 tm2c(EUCOMM) Hmgu (ref. ). Genotyping was performed by Transnetyx using ear notches taken for identification purposes at weaning (3 weeks of age). Mice were induced between 8 and 12 weeks of age, unless otherwise indicated, with AAV8.TBG.PI.eGFP.WPRE.bGH (AAV8-TBG-GFP) (Addgene, 105535-AAV8), AAV8.TBG.PI.cre.rBG (AAV8-TBG-cre) (Addgene, 107787-AAV8) or AAV8.TBG.PI.Null.bGH (AAV8-TBG-Null) (Addgene, 105536-AAV8). Virus was diluted in 100 µl PBS to the desired concentration and injected through the tail vein. Unless otherwise specified, mice received a dose of 6.4 × 10 8 GC per mouse. For the GEMM + MWD model, 6-week-old mice were kept on a modified western diet (Envigo, TD.120528) plus sugar water (23.1 g l −1 fructose and 18.9 g l −1 glucose) in combination with repeated CCl4 injections (intraperitoneal (i.p.), 0.2 µl g −1 of body weight; vehicle, Cornoil) as previously described and were induced with AAV.TBG.cre at 10 weeks of age. For the DEN/ALIOS model, C57BL/6 WT mice, were injected with a single dose of DEN (80 mg per kg by i.p. injection) at 14 days of age. Mice were fed ALIOS diet (Envigo, TD.110201) and sugar water (23.1 g l −1 fructose and 18.9 g l −1 glucose) from 60 days of age. Mice were collected at day 284. For the MWD + CCl4 model, the mice were kept on a modified western diet (Envigo, TD.120528) plus sugar water (23.1 g l −1 fructose and 18.9 g l −1 glucose) in combination with repeated CCl4 injections (i.p., 0.2 µl g −1 body weight; vehicle, Cornoil) as previously described . For the streptozotocin (STZ) model, male and female C57BL/6J WT mice were injected with a single dose of STZ (200 µg in 0.1 M citrate buffer, pH 4.0) subcutaneously at 2 days of age. Mice were fed a high-fat diet (TestDiet 58R3, 1810835) from 30 days of age. All STZ–HFD-treated livers showed pale yellow colour at 6 weeks, mild swelling at 8 weeks, granular surface at 12 weeks and tumour protrusion at 20 weeks of age . Mice were collected between 17 and 35 weeks of age. For the orthotopic model, Hep-53.4 cells (female C57BL/6J hepatoma cell line; Cytion, LOT-L230232R) were injected intrahepatically into the left lobe of male C57BL/6J mice. The procedure was performed under isoflurane general anaesthesia. Analgesia was given to the mice for pain management. Mice were collected at 28 days after implantation or left to reach an approved humane end point. For therapeutic intervention in the BM model, drugs were given at 90 days after induction or in other models determined by mean cohort survival relative to the BM model survival as indicated in the figures. The following drugs were used: sorafenib (LC Laboratories S8502, daily, oral gavage, 45 mg per kg; vehicle, 50% chremophor/50% ethanol, then, before dosing, 3 parts H 2 O added), lenvatinib (SelleckChem, S1164 (end-point studies); or Eisai (monotherapy timepoint studies); daily; oral gavage, 10 mg per kg, vehicle, 3 mM HCl), anti-PD1 (BioLegend, RMP1-14; twice per week; i.p., 200 µg; vehicle, PBS; control, IgG, BioLegend, RTK2758), cladribine (SelleckChem, S1199; daily; i.p., 20 mg per kg; vehicle, PBS), VEGFRi (AstraZeneca, AZD2171; daily; oral gavage, 3 mg per kg; vehicle, 0.5% (w/v) HPMC, 0.1% Tween-80, in H 2 O). To help with drug-induced weight loss, mice on cladribine treatment received irradiated peanuts and sunflower seeds as diet supplements. If mice reached 83% of their weight at treatment start, cladribine treatment was withheld until they gained weight to at least 90% of weight at treatment start. Mice who dropped below 80% of weight at treatment start were sampled according to licence limitations. Confounding factors (for example, litter mates, induction date) were taken into consideration when allocating mice into groups but mice were not randomized using a specific method. Mice who presented with a visible tumour before treatment start were excluded from the experiments according to a priori established criteria. Animal technicians dosing the mice were blinded to the genotype of the mice. The number of biological replicates was ≥3 mice per cohort for all experiments. Further details are provided in the figure legends and Supplementary Table . Animal tissue collection GEMMs were sampled at specific timepoints or at the end point. The end point was defined as the mouse having reached a liver weight/body weight ratio of >20% or having adverse side effects from the tumour, such as tumour haemorrhaging. Mice who died of tumour haemorrhaging were included in the survival analysis but not in any downstream analysis. Tumours were measured macroscopically using digital callipers, and visible tumours were counted. Images of whole livers were taken using a Canon PowerShot G9X camera with a ruler present in each picture. Tissue was either sampled in neutral buffered saline containing 10% formaldehyde or snap-frozen on dry ice. Histology and immunohistochemistry Liver, tumour and lung tissues were fixed using neutral buffered saline containing 10% formaldehyde, dehydrated and embedded in paraffin, and cut into 4-μm-thick sections. The sections were dewaxed and stained with H&E or Sirius Red using standard protocols. Additional sections were stained immunohistochemically using the primary antibodies listed in Supplementary Table . Primary antibodies were detected by HRP-labelled secondary antibodies and subsequently stained using a peroxidase DAB kit; either Agilent (K3468) or Leica (DS92563) DAB for tissue processed in autostainer or Vector Laboratories (SK-4100) with haematoxylin as a counterstain (immunohistochemistry) or by fluorescent-labelled secondary antibodies (Invitrogen) with DAPI used as counterstain (SouthernBiotech, 0100-20) (immunofluorescence). Microscopy and quantitative analysis of immunohistochemistry Images were obtained on the Zeiss Axiovert 200 microscope using the Zeiss Axiocam MRc camera. For image analysis, stained slides were scanned using the Leica Aperio AT2 slide scanner (Leica Microsystems) at ×20 magnification. Quantification of blinded, stained sections (GS, Ki-67, CC3, CD3, yH2AX, p53) was performed using the HALO image analysis software (v.3.1.1076.363, Indica Labs). Quantification of microscopy tumour area in BM mice was performed based on nodules, independent of GS status. Quantification of Ki-67 + was by percentage/cell number and CC3 + by percentage/tumour area. Lungs were microscopically analysed for the presence of extrahepatic HCC spread by examining H&E and GS sections. Metastasis was scored in a binary manner as detected or not-detected but was not analysed in respect to individual metastasis burden per mouse. Images for tissue comparison to HCCOs were taken on the Zeiss 710 confocal microscope. Tumour genotyping After extraction from whole tumour, DNA was suspended in Transnetyx assay buffer and was analysed by Transnetyx using probes (p53Flox EX, Bap1-2 EX, PTEN-EX, LSL-EX-1, Tg-MYC, Axin1-1 EX, Ctnnb1-16 EX) and was additionally purified and concentrated using the Monarch Genomic DNA purification kit (New England Biolabs, T3010L) according to the manufacturer’s instructions. Generation of amplicons indicating successful recombination of genetic loci was performed by PCR (Eppendorf Mastercycler x50a) using the OneTaq Quick-Load 2× Master Mix with Standard Buffer (M0486S); reactions were set up according to the manufacturer’s instructions, amplification conditions (Supplementary Table ) and primer sequences—β-catenin exon 3 (Supplementary Table ) and KRAS G12D (Supplementary Table ). The resulting PCR reactions were separated by electrophoresis on 1.5% agarose (Melford, A20090-500) gel, using the size marker Quick-Load Purple 1 kb Plus DNA Ladder (New England Biolabs, N0550S) and bands were visualized using SYBR Safe DNA gel stain (Invitrogen, S33102). The Gels were imaged on the Chemi-Doc Imaging System (Bio-Rad). Concordance between CTNNB1 recombination results between the two methods was 100%. Where possible, the samples used in histological comparison were also assessed genotypically. Where not possible, due to DNA contamination/low quality, tumours were replaced by other end-stage tumours from the same cohort to achieve n ≥ 6 per cohort (total n = 4 additional samples). Quantitative analysis of fluorescence immunohistochemistry Fluorescent tiled images were generated on the Opera Phenix High-Content Screening System (Perkin Elmer) at ×20 magnification. Fluorescence was detected using the same settings throughout. Consecutive, non-overlapping fields were analysed blindly using Columbus Image analysis software (v.2.8.0.138890, Perkin Elmer). Positivity gating thresholds were defined using negative controls. For representative images, processing adjustments were performed equally. Multiplex immunofluorescence immunohistochemical staining Mouse liver samples (thickness, 4 µm) were sectioned and placed onto TOMO hydrophilic adhesive microscope slides (Matsunami, 0808228600). After antibody validation, semi-automated multiplex immunofluorescence staining was performed on the Ventana Discovery Ultra platform (Roche Tissue Diagnostics, RUO Discovery Universal v.21.00.0019). Fluorescence detection was performed using an Opal fluorophore tyramide-based signal amplification system (Akoya Biosciences). All primary antibodies were optimized using a pH 9 antigen retrieval solution (CC1, Roche Tissue Diagnostics, 06414575001) at 95 °C for 32 min. A denature step was applied using pH 6 antigen retrieval solution (CC2, Roche Tissue Diagnostics, 05279798001) for 24 min between each Opal detection and primary antibody application. The primary–secondary–opal fluorophore combinations (CD45–HRP–Opal480; CD8–HRP–Opal690; CD4–HRP–Opal620; GranzymeB–HRP–Opal650; GS–HRP–Opal520; MYC–HRP–Opal570) are described in Supplementary Table . ImmPRESS rat and Opal 780 were manually applied in their specific sequences, and the remaining reagents were fully automated on the Ventana DISCOVERY ULTRA platform (Roche Tissue Diagnostics). Three drops of nuclear DAPI counterstain (Roche Tissue Diagnostics, 05268826001, RTU (Ready to Use), 24 min) were applied to each sample for nuclear detection. Multiplex immunofluorescence image acquisition and analysis Whole-slide images were collected on the PhenoImager HT multispectral slide scanner (Akoya Biosciences, v.1.0) using a ×10 objective before the acquisition of each region of interest (ROI) using a ×20 objective. Each ROI was spectrally unmixed using InForm (Akoya Biosciences, v.2.6.0) using a project-specific spectral library created using single-channel dyes and an autofluorescence mouse liver control. Visiopharm was used for all image analysis. For tissue segmentation, a bespoke, in-house-trained deep learning algorithm (v.2024.06.0.19093 ×64) was trained using the deep learning module with a U-Net backbone, to segment each lobe into tumour, stroma, non-tumour GS and background ROIs. Tumour regions that were smaller than 10,000 µm 2 were classified as a ‘clone’ region. Tumour and clonal regions were then dilated to generate ‘peritumoural stroma’ and ‘periclonal’ regions, respectively. Necrotic regions were manually segmented. Tumour regions were eroded to create ‘tumour centre’ and ‘tumour periphery’ regions. T cells were classified with a ‘T cell’ label within these regions using the threshold module (v.2024.07.1.16745 ×64) to threshold CD4, CD8 and CD45 fluorescence channels using the original image features. Each image was verified by a pathologist to confirm regional and T cell label segmentation. Post-processing steps were included to change T cell labels into their respective regional labels, that is, a T cell found within the tumour ROI would be changed to ‘tumour T cell’ and so on. Output variables were then generated and exported for downstream data analysis. Area of entire lobes were generated, and areas for each ROI as well as regional mean pixel intensities of MYC, mean pixel intensities of each marker in each T cell label and x–y coordinates of each T cell label. The Phenoplex Guided workflow was used for T cell phenotyping, which generated a phenotype list that was exported for data analysis. Duplex immunofluorescence immunohistochemical staining For duplex immunofluorescence immunohistochemical staining (Extended Data Fig. (bottom)), 4-µm-thick mouse HCCOs and liver lobe samples were sectioned and placed onto TOMO hydrophilic adhesive microscope slides (Matsunami, 0808228600). After antibody validation, fully automated multiplex immunofluorescence staining was performed on the Ventana DISCOVERY ULTRA platform (Roche Tissue Diagnostics, RUO Discovery Universal v.21.00.0019). Fluorescence detection was performed using an Opal fluorophore tyramide-based signal amplification system (Akoya Biosciences). All primary antibodies were optimized using a pH 9 antigen retrieval solution (CC1, Roche Tissue Diagnostics, 06414575001) at 95 °C for 32 min. A denature step was applied using pH 6 antigen retrieval solution (CC2, Roche Tissue Diagnostics, 05279798001) for 24 min between each Opal detection and primary antibody application. The primary–secondary–opal fluorophore combinations (MYC–HRP–Opal570, GS–HRP–Opal520) are described in Supplementary Table . One drop of nuclear DAPI counterstain (Akoya, 232121) was applied to each sample for nuclear detection. Duplex immunofluorescence image acquisition and analysis Whole-slide images were collected on the PhenoImager HT multispectral slide scanner (Akoya Biosciences, v.1.0) using a ×20 objective using Motif mode. Images were spectrally unmixed using Inform (Akoya, v.2.6.0) using an autofluorescence liver control slide to remove autofluorescence. Tumour scoring H&E-stained sections and tumours were additionally assessed by a consultant liver histopathologist and UK liver pathology External Quality Assessment scheme member (T.J.K.) working at a national liver transplant centre. All assessment was undertaken blind to all other data, including genotype and sampling times. An initial screen of the first available 135 cases was made to identify prominent histological features in lesional and non-lesional liver that could be semi-quantitatively assessed. Accepting the inherent limitations of semi-quantitative subjective histological assessment but using a single observer to remove interobserver considerations, semi-quantitative/ordinal scoring systems were created for lesional and non-lesional features. Slides containing transections of whole lobes from each animal were assessed as a whole, giving an overall score or impression rather than scoring on an individual-lesion basis. Non-lesional liver was scored for steatosis (none, focal, abundant) and lobular inflammation (none, focal, abundant). A minority of slides included insufficient non-lesional liver for assessment. For lesional assessment, the presence of glandular tumour, that is, meriting designation as adenocarcinoma (none, focal, extensive) and undifferentiated carcinoma (none, focal, abundant, exclusive) was assessed first. All hepatocellular neoplastic lesions had the morphological and cytological appearances of malignancy, that is, HCC. In all cases in which there was HCC, the following features were assessed using the categories in parentheses: lesional pattern (any from nested, trabecular, solid), lesional steatosis (none, focal, abundant), lesional cell ballooning (none, focal, abundant), intralesional inflammation (none, focal, abundant), lesional necrosis (none, focal, confluent, extensive), lesional cell apoptosis (none, focal, many), intralesional peliosis (none, focal, abundant), lesional nuclear grade (low, minimal/low pleomorphism; high, highly pleomorphic). Whole-tumour RNA-seq Whole tumour and healthy tissue were snap-frozen and stored at −80 °C. To cover the breadth of our models, for each cohort, tissue from the shortest and longest surviving mouse as well as tissue from mice with survival closest to median cohort survival was chosen. Tissue was homogenized using the Precellys Evolution homogenizer and bulk RNA was isolated using a Trizol (Invitrogen) extraction protocol according to the manufacturer’s instructions. RNA quality and quantity was analysed on the Nanodrop 2000 (Thermo Fisher Scientific) and Agilent 2200 TapeStation (D1000 screentape) systems. Only samples with RIN > 7 were used for library preparation. Libraries were prepared using a Lexogen QuantSeq FWD Kit (disease positioning) or the Illumina TruSeq stranded mRNA kit (tumour heterogeneity). Library quality and quantity were assessed using the 2200 TapeStation (Agilent) and Qubit (Thermo Fisher Scientific) systems. The libraries for the disease positioning were sequenced by Novogene Europe. The libraries for the tumour heterogeneity were run on the Illumina NextSeq 500 system using the high-output 75 cycle kit (2 × 36 cycle paired-end reads). Mapping of RNA-seq expression data Quality checks and trimming on the raw RNA-seq data files were done using FastQC v.0.11.9 ( https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ ), FastP (v.0.20.1) , MultiQC (v.1.9) and FastQ Screen (v. 0.14.0) . RNA-seq single-end reads were mapped to the GRCm39.103 version of the Mus musculus genome and annotated using STAR (v.2.7.8a) . Expression levels were determined by FeatureCounts from the Subread package (v.2.0.1) . Computational disease positioning based on human TCGA data TCGA data were downloaded using the GenomicDataCommons R package (v.1.12.0; https://bioconductor.org/packages/GenomicDataCommons ) , TCGA ‘HTSeq–counts’ and corresponding clinical annotations. TCGA mutational data were downloaded using maftools (v.2.4.2) . Both human and mouse RNA-seq counts were normalized using VST from the DESeq2 (v.1.28.1 and v.1.44.0) package and then centred within a sample. Genes were reduced to those with direct one-to-one gene mapping between human and mouse genomes established by Ensembl, as retrieved from the biomaRt (v.2.56.1) package , . Singular-value decomposition (SVD) of the human data was performed followed by matrix factorization of both the human and mouse data into a 100-rank human space. UMAP of the combined dataset was executed using R package uwot (v.0.1.11; https://CRAN.R-project.org/package=uwot ). An adjacency matrix was constructed from a nearest-neighbours search (RANN package v.2.6.1, https://CRAN.R-project.org/package=RANN ) of the human and mouse SVD objects for clustering analysis. R package igraph (v.1.2.11 and v.2.0.3) was used to construct a graph object and the community structure was determined using Louvain •clustering. Single-sample gene set enrichment analysis (ssGSEA) analysis was performed using the R package corto (v.1.2.4) with the Hallmark gene set , downloaded using msigdbr (v.7.4.1; https://CRAN.R-project.org/package=msigdbr ). Hoshida and Chiang (also downloaded using msigdbr) subclass classification was determined by the highest enriched subclass. Tumour immune cell estimation was performed using ConsensusTME . Visualization of data by a combination of the ComplexHeatmap (v.2.4.3 and v.2.14.0) , ggplot2 (v.3.3.6 and v.3.5.1) , cowplot (v.1.1.1; https://CRAN.R-project.org/package=cowplot ) and viridis packages. Human H&E-stained tissue sections were obtained from the TCGA collection ( https://portal.gdc.cancer.gov/ ). Validation of HuMo clusters in an independent HCC cohort HuMo clusters were validated with the bulk RNA-seq data of an independent cohort of 171 HCC samples from patients undergoing resection collected in the setting of the HCC Genomic Consortium (European Genome-Phenome Archive: EGAS00001005364 ). Fastq files were aligned using STAR (v.2.5.1b) to the hg19 reference genome with gencode annotation v19 and were quantified using featureCounts (v.1.5.2). Raw counts were preprocessed and cluster attribution was performed as described above with the TCGA and mouse data. In the analysis of the transcriptomic data, positivity for previously reported gene signatures was evaluated using the Nearest Template Prediction module from GenePattern (v.3.9) . The ssGSEA projection was performed using previously reported gene signatures as well as the Hallmark gene set downloaded using msigdbr (v.7.4.1). The mutational profile of 144 HCC samples was obtained by whole-exome sequencing . Clinicopathological data (such as vascular invasion, AFP levels (≥400 ng ml −1 )) were originally reported previously . Differential expression analysis for intertumoural heterogeneity Genes were restricted to those with significance in all comparisons (with significance defined as adjusted P < 0.05 and log 2 [FC] > 1). Data were scaled and visualized using the ComplexHeatmap package. Gene Ontology over-representation analysis was performed using the clusterProfiler package (v.3.16.1). Human sample ethical approval The use of consenting patients’ tissues surplus to diagnostic requirements for research purposes was approved by the Newcastle and North Tyneside Regional ethics committee, the Newcastle Academic Health Partners Bioresource (NAHPB) and the Newcastle upon Tyne NHS Foundation Trust Research and Development (R&D) department, in accordance with Health Research Authority guidelines. (10/H0906/41; NAHPB Project 48; REC 12/NE/0395; R&D 6579; Human Tissue Act licence 12534). MRI Magnetic resonance imaging (MRI) scans were performed on liver-tumour-bearing mice using the nanoScan imaging system (Mediso Medical Imaging Systems). The mice were anaesthetized and maintained under inhaled isoflurane anaesthesia (induction, 4–5% (v/v); maintenance, 1.5–2.0% (v/v)) in 95% oxygen during the entire imaging procedure. Whole-body T1-weighted gradient echo (GRE) 3D coronal/sagittal MRI sequences (echo time (TE), 3.8 ms; repetition time (TR), 20 ms; flip angle, 30 degrees; and slice thickness, 0.50 mm) were used to obtain MRI images. For quantification of scans, volumes of interest were manually drawn around the liver region on MRI scans by visual inspection using VivoQuant software (v.4.0, InviCRO). For each scan, separate volumes of interest were prepared to adjust for the position and angle of each mouse on the MRI scanner and their tumour size. Mouse HCCO culture, drug screening and imaging HCCOs were extracted and cultured as previously described , , with the exception that HCCOs from mice with activated β-catenin signalling were cultured in the absence of WNT and RSPO1. All mouse HCCO cultures were regularly tested for mycoplasma. For the high-throughput screen cohort 5 (BM) HCCOs were dissociated with TrypLE and plated at a density of 1 × 10 3 cells in 10 μl BME in prewarmed 384-well plates (Greiner BioOne, 781091) 5 days before adding the drugs. On day 0, a panel of 147 FDA-approved oncology drugs (AOD IX, acquired June 2019, https://dtp.cancer.gov/organization/dscb/obtaining/available_plates.htm ) was added at a final concentration of 10 µM. Staurosporin was used as an internal positive control; DMSO and untreated cells were used as an internal negative control. The medium was changed on day 4 and the compounds were freshly added. Incucyte NucLight Rapid Red (Sartorius, 4717) was added on day 6 and cells were imaged using the Opera Phenix High-Content Screening System (Perkin Elmer) on day 9. Volumes were determined using Icy BioImage software (v.2.0.0.0; https://icy.bioimageanalysis.org ) . The experiment was performed twice (using different passages from one HCCO line) in technical quadruplicates. For the drug dose–response curve screen, HCCOs (1–2 lines per cohort) were dissociated with TrypLE and plated at a density of 1 × 10 3 cells in 10 μl Matrigel (Corning, 356231) in prewarmed 96-well plates (Greiner BioOne, 655098). The treatment schedule was the same as for the HTP screen, except the medium was changed and fresh drugs were added on days 3 and 7. Drugs and concentrations are shown in the figures. Drugs were purchased from Selleckchem, dissolved in DMSO to 10 mM, aliquoted and stored at −20 °C. Cell viability was measured on day 9 using CellTitre-Glo 3D reagent (Promega, G9682) according to the manufacturer’s instructions. Luminescence was measured on the Spark Microplate Reader (Tecan). The results were normalized to the vehicle. Curve fitting and IC 50 calculation were performed using a nonlinear regression equation. All of the experiments were performed in duplicate and at least three times using different passages from one to two HCCO lines per cohort. Images of HCCOs were taken on an Olympus CKX41 using the Qimaging Retiga Exi Fast 1394 camera. For immunofluorescence analysis, HCCOs were washed with ice-cold PBS, fixed with 4% PFA and permeabilized with 0.2% Triton X-100. A list of the antibodies used is provided in Supplementary Table . Images were taken using the Zeiss 710 confocal microscope. Tumour-derived mouse HCCOs (available from all GEMMs) will be shared on reasonable request. Human HCCO culture and drug screening Human HCCOs were derived from liver cancer needle biopsies or liver resections as described before . The following human HCCO lines were used: D386-O and D953-O ( CTNNB1 WT, TP53 WT, MYC WT); D455-O ( CTNNB1 MUT, TP53 WT, MYC AMP); C948-O and C949-O ( CTNNB1 MUT, TP53 WT, MYC AMP); C655-O ( CTNNB1 WT, TP53 MUT, MYC ND); D045-O, D046-O, D803-O and R035-O ( CTNNB1 WT, TP53 MUT, MYC WT); C798-O, C975-O, D324-O, D804-O and D876-O ( CTNNB1 WT, TP53 MUT, MYC AMP); and D359-O ( CTNNB1 MUT, TP53 MUT, MYC WT). For expansion, the human HCCOs were seeded into reduced growth factor BME2 (R&D Systems, 3533-005-02) and cultured in expansion medium (EM): advanced DMEM/F-12 (Gibco, 12634010) supplemented with 1× B-27 (Gibco, 17504001), 1× N-2 (Gibco, 17502001), 10 mM nicotinamide (Sigma-Aldrich, N0636), 1.25 mM N -acetyl- l -cysteine (Sigma-Aldrich, A9165), 10 nM [Leu15]-gastrin (Sigma-Aldrich, G9145), 10 μM forskolin (Tocris, 1099), 5 μM A83-01 (Tocris, 2939), 50 ng ml −1 EGF (Peprotech, AF-100-15), 100 ng ml −1 FGF10 (Peprotech, 100-26), 25 ng ml −1 HGF (Peprotech, 100-39), 10% RSPO1-conditioned medium (v/v, homemade). HCCOs were passaged after dissociation with 0.25% trypsin-EDTA (Gibco). All human HCCOs were regularly tested for mycoplasma contamination using the MycoAlert mycoplasma detection kit (Lonza, LT07-118). Drugs were purchased from ApexBio and Selleckchem, dissolved in DMSO to 10 mM, aliquoted and stored at −20 °C. For the screening, human HCCOs were dissociated with 0.25% trypsin-EDTA (Gibco) to single cells and 1 × 10 3 cells per well were plated in a 384-well plate (Greiner BioOne, 781986) on a layer of BME2 (R&D Systems, 3533-005-02) previously diluted with EM (50:50, v/v). Cells were cultured for 3 days without treatment to allow for organoid formation. At day 3, an eight-point half-log dilution series of each compound (ranging from 10 μM to 0.00316 μM) was added using a Tecan D300e. Cell viability was measured after 5 days of treatment using the CellTiter-Glo 3D reagent (Promega, G9682). Luminescence was measured on the Synergy H1 multi-mode reader (BioTek Instruments). Results were normalized to the vehicle (DMSO). The maximal DMSO concentration was 0.2%. Curve fitting was performed using Prism (GraphPad v.9 GraphPad Software) software and the nonlinear regression equation. Results are shown as mean ± s.e.m. Fluorescent activated cell sorting After mincing into small pieces, 100 mg of healthy liver or liver tumour was digested on the gentleMACS Octo dissociator with heaters using the mouse tumour dissociation Kit (Miltenyi Biotec, 130-096-730). Dissociated cells were resuspended in 0.5% BSA in PBS, filtered through a 70 μm strainer and centrifuged at 400 g for 5 min. Cells were then resuspended in 5 ml RBC lysis buffer (Thermo Fisher Scientific, 00-4300-54) and incubated for 5 min at room temperature and washed with 0.5% BSA in PBS before being resuspended in 0.5% BSA in PBS. Cell suspensions were added to 96-well V-bottom plates (maximum density, 0.5 × 10 6 cells per well). Cells were stimulated for 3 h with complete IMDM medium containing 8% FCS, 50 μM β-mercaptoethanol, 1× penicillin–streptomycin with 1× cell activation cocktail with brefeldin A (BioLegend, 423304) at 37 °C as previously described previously . After stimulation, cells were centrifuged at 800 g for 2 min. Cells were stained in Brilliant stain buffer (BD Biosciences, 566349) containing antibodies for surface antigens for 30 min at 4 °C in the dark. Cells were then washed with PBS/0.5% BSA, centrifuged at 800 g for 2 min, followed by ice-cold PBS and incubated with Zombie NIR Fixable Viability dye (BioLegend, 423106) to stain dead cells for 20 min at 4 °C. After further washing the cells with PBS/0.5% BSA, cells were fixed and permeabilized in FOXP3 transcription factor fixation/permeabilization solution (Thermo Fisher Scientific) for 20 min at 4 °C, according to the manufacturer’s instructions. Intracellular antibodies were prepared in permeabilization buffer and incubated with cells for 30 min at 4 °C before cells were washed with permeabilization buffer, followed by PBS/0.5% BSA and finally resuspended in PBS/0.5% BSA. All of the experiments were performed using a five-laser BD LSRFortessa flow cytometer with DIVA software (BD Biosciences v.8.0.1). Compensation was determined automatically using Ultracomp eBeads (01-2222-42; Thermo Fisher Scientific). Data were analysed using FlowJo Software v.9.9.6. Statistics and reproducibility Statistical analyses were performed using GraphPad Prism, software (v9 GraphPad Software) and R (v.4.0.2 and higher) with statistical tests as indicated in the figure legends. Data were tested for normal distribution. All performed t -tests were two-tailed. P values are displayed in figures. No statistical methods were used to predetermine sample sizes but our sample sizes are similar to those reported in previous publications , , – . For animal experiments, biological replicate sizes were chosen taking into account the variability observed in pilot and previous studies in respective cohorts. For all experiments, animal/sample assignment was matched for age-matched control, and assigned based on randomly assigned mouse identification markings. Batched staining and analyses alongside controls were used throughout. The investigators were not blinded for the in vivo experiments. Technical staff administering therapy were blinded to the mouse genotypes. All subsequent tissue handling and analysis were blinded and/or performed using standardized automated analyses where possible. Quantitative image analysis was performed blinded to the genotype and treatment. The data distribution for normality testing and testing of equal variances was assessed using Prism 9 Software (GraphPad Software). No data were excluded, unless mentioned otherwise, except the following, which were excluded before analysis: one biological replicate failed quality control after transcriptomic sequencing—all other biological replicates from this and other cohorts successfully passed quality control and were included in downstream analysis; two drugs were excluded from the HCCO HTP screen due to microbiological contamination and drug precipitation in multiple replicates, respectively. One sample was excluded from the RFP expression analysis during analysis (total n = 4 biological replicates): testing AAV-mediated recombination of RFP alleles (Extended Data Fig. ), one sample was a notable outlier (4.9% versus 25.7%, 25.1% and 25.8%) which on re-review was caused by inconsistent RFP staining of the section—this outlier was removed from final analysis; details are provided in the figure legend also. Where the tumour number could not be quantified due to tumour rupture, no tumour number is reported (Fig. ). Figures were assembled using Scribus v.1.4.8 ( https://www.scribus.net/ ). Images were processed using Gimp v.2.10.14 ( https://www.gimp.org/ ). Reporting summary Further information on research design is available in the linked to this article.
All animal experiments were performed in accordance with UK Home Office licences (70/8891, PP0604995, 70/8646, 70/8468 and PP8854860) and in accordance with the UK Animal (Scientific Procedures) Act 1986 and EU direction 2010. They were subject to review by the animal welfare and ethical review board of the University of Glasgow and the University of Newcastle upon Tyne. To minimize pain, suffering and distress to the animals, we used single-use needles and non-adverse handling techniques. Mice were housed under controlled conditions (specific-pathogen free, 12 h–12 h light–dark cycle, 19–22 °C, 45–65% humidity) with access to food and water ad libitum. We added environmental enrichments, in the form of gnawing sticks, plastic tunnels and nesting material to all of the cages. Welfare of animals was defined by clinical symptoms, including visible masses, any degree of reduced mobility/distress, weight loss or evidence of haemorrhage; however no maximal tumour volume end points for intrahepatic tumours were mandated. No mouse exceeded the humane end points stipulated in the Home Office Licenses. Unless otherwise specified, male mice on a mixed background were used. The following transgenic mice strains were used: Gt(Rosa)26Sor tm14(CAG-tdTomato)Hze ( R26 LSL-Tom ) , Ctnnb1 tm1Mmt ( Ctnnb1 ex3 ) , Gt(Rosa)26Sor tm1(MYC)Djmy ( R26 LSL-MYC ) , Trp53 tm1Brn ( Trp53 fl ) , Trp53 tm2Tyj ( Trp53 R172H ) , Cdkn2a tm1.1Brn ( Cdkn2a KO ) , Pten tm2Mak ( Pten fl ) , Gt(Rosa)26Sor tm1(Notch1)Dam ( R26 LSL-NICD ) , Kras tm4Tyj ( Kras G12D ) , Cdkn1a tm1Led ( Cdkn1a KO ) , Axin1 fl (ref. ), Bap1 tm2c(EUCOMM) Hmgu (ref. ). Genotyping was performed by Transnetyx using ear notches taken for identification purposes at weaning (3 weeks of age). Mice were induced between 8 and 12 weeks of age, unless otherwise indicated, with AAV8.TBG.PI.eGFP.WPRE.bGH (AAV8-TBG-GFP) (Addgene, 105535-AAV8), AAV8.TBG.PI.cre.rBG (AAV8-TBG-cre) (Addgene, 107787-AAV8) or AAV8.TBG.PI.Null.bGH (AAV8-TBG-Null) (Addgene, 105536-AAV8). Virus was diluted in 100 µl PBS to the desired concentration and injected through the tail vein. Unless otherwise specified, mice received a dose of 6.4 × 10 8 GC per mouse. For the GEMM + MWD model, 6-week-old mice were kept on a modified western diet (Envigo, TD.120528) plus sugar water (23.1 g l −1 fructose and 18.9 g l −1 glucose) in combination with repeated CCl4 injections (intraperitoneal (i.p.), 0.2 µl g −1 of body weight; vehicle, Cornoil) as previously described and were induced with AAV.TBG.cre at 10 weeks of age. For the DEN/ALIOS model, C57BL/6 WT mice, were injected with a single dose of DEN (80 mg per kg by i.p. injection) at 14 days of age. Mice were fed ALIOS diet (Envigo, TD.110201) and sugar water (23.1 g l −1 fructose and 18.9 g l −1 glucose) from 60 days of age. Mice were collected at day 284. For the MWD + CCl4 model, the mice were kept on a modified western diet (Envigo, TD.120528) plus sugar water (23.1 g l −1 fructose and 18.9 g l −1 glucose) in combination with repeated CCl4 injections (i.p., 0.2 µl g −1 body weight; vehicle, Cornoil) as previously described . For the streptozotocin (STZ) model, male and female C57BL/6J WT mice were injected with a single dose of STZ (200 µg in 0.1 M citrate buffer, pH 4.0) subcutaneously at 2 days of age. Mice were fed a high-fat diet (TestDiet 58R3, 1810835) from 30 days of age. All STZ–HFD-treated livers showed pale yellow colour at 6 weeks, mild swelling at 8 weeks, granular surface at 12 weeks and tumour protrusion at 20 weeks of age . Mice were collected between 17 and 35 weeks of age. For the orthotopic model, Hep-53.4 cells (female C57BL/6J hepatoma cell line; Cytion, LOT-L230232R) were injected intrahepatically into the left lobe of male C57BL/6J mice. The procedure was performed under isoflurane general anaesthesia. Analgesia was given to the mice for pain management. Mice were collected at 28 days after implantation or left to reach an approved humane end point. For therapeutic intervention in the BM model, drugs were given at 90 days after induction or in other models determined by mean cohort survival relative to the BM model survival as indicated in the figures. The following drugs were used: sorafenib (LC Laboratories S8502, daily, oral gavage, 45 mg per kg; vehicle, 50% chremophor/50% ethanol, then, before dosing, 3 parts H 2 O added), lenvatinib (SelleckChem, S1164 (end-point studies); or Eisai (monotherapy timepoint studies); daily; oral gavage, 10 mg per kg, vehicle, 3 mM HCl), anti-PD1 (BioLegend, RMP1-14; twice per week; i.p., 200 µg; vehicle, PBS; control, IgG, BioLegend, RTK2758), cladribine (SelleckChem, S1199; daily; i.p., 20 mg per kg; vehicle, PBS), VEGFRi (AstraZeneca, AZD2171; daily; oral gavage, 3 mg per kg; vehicle, 0.5% (w/v) HPMC, 0.1% Tween-80, in H 2 O). To help with drug-induced weight loss, mice on cladribine treatment received irradiated peanuts and sunflower seeds as diet supplements. If mice reached 83% of their weight at treatment start, cladribine treatment was withheld until they gained weight to at least 90% of weight at treatment start. Mice who dropped below 80% of weight at treatment start were sampled according to licence limitations. Confounding factors (for example, litter mates, induction date) were taken into consideration when allocating mice into groups but mice were not randomized using a specific method. Mice who presented with a visible tumour before treatment start were excluded from the experiments according to a priori established criteria. Animal technicians dosing the mice were blinded to the genotype of the mice. The number of biological replicates was ≥3 mice per cohort for all experiments. Further details are provided in the figure legends and Supplementary Table .
GEMMs were sampled at specific timepoints or at the end point. The end point was defined as the mouse having reached a liver weight/body weight ratio of >20% or having adverse side effects from the tumour, such as tumour haemorrhaging. Mice who died of tumour haemorrhaging were included in the survival analysis but not in any downstream analysis. Tumours were measured macroscopically using digital callipers, and visible tumours were counted. Images of whole livers were taken using a Canon PowerShot G9X camera with a ruler present in each picture. Tissue was either sampled in neutral buffered saline containing 10% formaldehyde or snap-frozen on dry ice.
Liver, tumour and lung tissues were fixed using neutral buffered saline containing 10% formaldehyde, dehydrated and embedded in paraffin, and cut into 4-μm-thick sections. The sections were dewaxed and stained with H&E or Sirius Red using standard protocols. Additional sections were stained immunohistochemically using the primary antibodies listed in Supplementary Table . Primary antibodies were detected by HRP-labelled secondary antibodies and subsequently stained using a peroxidase DAB kit; either Agilent (K3468) or Leica (DS92563) DAB for tissue processed in autostainer or Vector Laboratories (SK-4100) with haematoxylin as a counterstain (immunohistochemistry) or by fluorescent-labelled secondary antibodies (Invitrogen) with DAPI used as counterstain (SouthernBiotech, 0100-20) (immunofluorescence).
Images were obtained on the Zeiss Axiovert 200 microscope using the Zeiss Axiocam MRc camera. For image analysis, stained slides were scanned using the Leica Aperio AT2 slide scanner (Leica Microsystems) at ×20 magnification. Quantification of blinded, stained sections (GS, Ki-67, CC3, CD3, yH2AX, p53) was performed using the HALO image analysis software (v.3.1.1076.363, Indica Labs). Quantification of microscopy tumour area in BM mice was performed based on nodules, independent of GS status. Quantification of Ki-67 + was by percentage/cell number and CC3 + by percentage/tumour area. Lungs were microscopically analysed for the presence of extrahepatic HCC spread by examining H&E and GS sections. Metastasis was scored in a binary manner as detected or not-detected but was not analysed in respect to individual metastasis burden per mouse. Images for tissue comparison to HCCOs were taken on the Zeiss 710 confocal microscope.
After extraction from whole tumour, DNA was suspended in Transnetyx assay buffer and was analysed by Transnetyx using probes (p53Flox EX, Bap1-2 EX, PTEN-EX, LSL-EX-1, Tg-MYC, Axin1-1 EX, Ctnnb1-16 EX) and was additionally purified and concentrated using the Monarch Genomic DNA purification kit (New England Biolabs, T3010L) according to the manufacturer’s instructions. Generation of amplicons indicating successful recombination of genetic loci was performed by PCR (Eppendorf Mastercycler x50a) using the OneTaq Quick-Load 2× Master Mix with Standard Buffer (M0486S); reactions were set up according to the manufacturer’s instructions, amplification conditions (Supplementary Table ) and primer sequences—β-catenin exon 3 (Supplementary Table ) and KRAS G12D (Supplementary Table ). The resulting PCR reactions were separated by electrophoresis on 1.5% agarose (Melford, A20090-500) gel, using the size marker Quick-Load Purple 1 kb Plus DNA Ladder (New England Biolabs, N0550S) and bands were visualized using SYBR Safe DNA gel stain (Invitrogen, S33102). The Gels were imaged on the Chemi-Doc Imaging System (Bio-Rad). Concordance between CTNNB1 recombination results between the two methods was 100%. Where possible, the samples used in histological comparison were also assessed genotypically. Where not possible, due to DNA contamination/low quality, tumours were replaced by other end-stage tumours from the same cohort to achieve n ≥ 6 per cohort (total n = 4 additional samples).
Fluorescent tiled images were generated on the Opera Phenix High-Content Screening System (Perkin Elmer) at ×20 magnification. Fluorescence was detected using the same settings throughout. Consecutive, non-overlapping fields were analysed blindly using Columbus Image analysis software (v.2.8.0.138890, Perkin Elmer). Positivity gating thresholds were defined using negative controls. For representative images, processing adjustments were performed equally.
Mouse liver samples (thickness, 4 µm) were sectioned and placed onto TOMO hydrophilic adhesive microscope slides (Matsunami, 0808228600). After antibody validation, semi-automated multiplex immunofluorescence staining was performed on the Ventana Discovery Ultra platform (Roche Tissue Diagnostics, RUO Discovery Universal v.21.00.0019). Fluorescence detection was performed using an Opal fluorophore tyramide-based signal amplification system (Akoya Biosciences). All primary antibodies were optimized using a pH 9 antigen retrieval solution (CC1, Roche Tissue Diagnostics, 06414575001) at 95 °C for 32 min. A denature step was applied using pH 6 antigen retrieval solution (CC2, Roche Tissue Diagnostics, 05279798001) for 24 min between each Opal detection and primary antibody application. The primary–secondary–opal fluorophore combinations (CD45–HRP–Opal480; CD8–HRP–Opal690; CD4–HRP–Opal620; GranzymeB–HRP–Opal650; GS–HRP–Opal520; MYC–HRP–Opal570) are described in Supplementary Table . ImmPRESS rat and Opal 780 were manually applied in their specific sequences, and the remaining reagents were fully automated on the Ventana DISCOVERY ULTRA platform (Roche Tissue Diagnostics). Three drops of nuclear DAPI counterstain (Roche Tissue Diagnostics, 05268826001, RTU (Ready to Use), 24 min) were applied to each sample for nuclear detection.
Whole-slide images were collected on the PhenoImager HT multispectral slide scanner (Akoya Biosciences, v.1.0) using a ×10 objective before the acquisition of each region of interest (ROI) using a ×20 objective. Each ROI was spectrally unmixed using InForm (Akoya Biosciences, v.2.6.0) using a project-specific spectral library created using single-channel dyes and an autofluorescence mouse liver control. Visiopharm was used for all image analysis. For tissue segmentation, a bespoke, in-house-trained deep learning algorithm (v.2024.06.0.19093 ×64) was trained using the deep learning module with a U-Net backbone, to segment each lobe into tumour, stroma, non-tumour GS and background ROIs. Tumour regions that were smaller than 10,000 µm 2 were classified as a ‘clone’ region. Tumour and clonal regions were then dilated to generate ‘peritumoural stroma’ and ‘periclonal’ regions, respectively. Necrotic regions were manually segmented. Tumour regions were eroded to create ‘tumour centre’ and ‘tumour periphery’ regions. T cells were classified with a ‘T cell’ label within these regions using the threshold module (v.2024.07.1.16745 ×64) to threshold CD4, CD8 and CD45 fluorescence channels using the original image features. Each image was verified by a pathologist to confirm regional and T cell label segmentation. Post-processing steps were included to change T cell labels into their respective regional labels, that is, a T cell found within the tumour ROI would be changed to ‘tumour T cell’ and so on. Output variables were then generated and exported for downstream data analysis. Area of entire lobes were generated, and areas for each ROI as well as regional mean pixel intensities of MYC, mean pixel intensities of each marker in each T cell label and x–y coordinates of each T cell label. The Phenoplex Guided workflow was used for T cell phenotyping, which generated a phenotype list that was exported for data analysis.
For duplex immunofluorescence immunohistochemical staining (Extended Data Fig. (bottom)), 4-µm-thick mouse HCCOs and liver lobe samples were sectioned and placed onto TOMO hydrophilic adhesive microscope slides (Matsunami, 0808228600). After antibody validation, fully automated multiplex immunofluorescence staining was performed on the Ventana DISCOVERY ULTRA platform (Roche Tissue Diagnostics, RUO Discovery Universal v.21.00.0019). Fluorescence detection was performed using an Opal fluorophore tyramide-based signal amplification system (Akoya Biosciences). All primary antibodies were optimized using a pH 9 antigen retrieval solution (CC1, Roche Tissue Diagnostics, 06414575001) at 95 °C for 32 min. A denature step was applied using pH 6 antigen retrieval solution (CC2, Roche Tissue Diagnostics, 05279798001) for 24 min between each Opal detection and primary antibody application. The primary–secondary–opal fluorophore combinations (MYC–HRP–Opal570, GS–HRP–Opal520) are described in Supplementary Table . One drop of nuclear DAPI counterstain (Akoya, 232121) was applied to each sample for nuclear detection.
Whole-slide images were collected on the PhenoImager HT multispectral slide scanner (Akoya Biosciences, v.1.0) using a ×20 objective using Motif mode. Images were spectrally unmixed using Inform (Akoya, v.2.6.0) using an autofluorescence liver control slide to remove autofluorescence.
H&E-stained sections and tumours were additionally assessed by a consultant liver histopathologist and UK liver pathology External Quality Assessment scheme member (T.J.K.) working at a national liver transplant centre. All assessment was undertaken blind to all other data, including genotype and sampling times. An initial screen of the first available 135 cases was made to identify prominent histological features in lesional and non-lesional liver that could be semi-quantitatively assessed. Accepting the inherent limitations of semi-quantitative subjective histological assessment but using a single observer to remove interobserver considerations, semi-quantitative/ordinal scoring systems were created for lesional and non-lesional features. Slides containing transections of whole lobes from each animal were assessed as a whole, giving an overall score or impression rather than scoring on an individual-lesion basis. Non-lesional liver was scored for steatosis (none, focal, abundant) and lobular inflammation (none, focal, abundant). A minority of slides included insufficient non-lesional liver for assessment. For lesional assessment, the presence of glandular tumour, that is, meriting designation as adenocarcinoma (none, focal, extensive) and undifferentiated carcinoma (none, focal, abundant, exclusive) was assessed first. All hepatocellular neoplastic lesions had the morphological and cytological appearances of malignancy, that is, HCC. In all cases in which there was HCC, the following features were assessed using the categories in parentheses: lesional pattern (any from nested, trabecular, solid), lesional steatosis (none, focal, abundant), lesional cell ballooning (none, focal, abundant), intralesional inflammation (none, focal, abundant), lesional necrosis (none, focal, confluent, extensive), lesional cell apoptosis (none, focal, many), intralesional peliosis (none, focal, abundant), lesional nuclear grade (low, minimal/low pleomorphism; high, highly pleomorphic).
Whole tumour and healthy tissue were snap-frozen and stored at −80 °C. To cover the breadth of our models, for each cohort, tissue from the shortest and longest surviving mouse as well as tissue from mice with survival closest to median cohort survival was chosen. Tissue was homogenized using the Precellys Evolution homogenizer and bulk RNA was isolated using a Trizol (Invitrogen) extraction protocol according to the manufacturer’s instructions. RNA quality and quantity was analysed on the Nanodrop 2000 (Thermo Fisher Scientific) and Agilent 2200 TapeStation (D1000 screentape) systems. Only samples with RIN > 7 were used for library preparation. Libraries were prepared using a Lexogen QuantSeq FWD Kit (disease positioning) or the Illumina TruSeq stranded mRNA kit (tumour heterogeneity). Library quality and quantity were assessed using the 2200 TapeStation (Agilent) and Qubit (Thermo Fisher Scientific) systems. The libraries for the disease positioning were sequenced by Novogene Europe. The libraries for the tumour heterogeneity were run on the Illumina NextSeq 500 system using the high-output 75 cycle kit (2 × 36 cycle paired-end reads).
Quality checks and trimming on the raw RNA-seq data files were done using FastQC v.0.11.9 ( https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ ), FastP (v.0.20.1) , MultiQC (v.1.9) and FastQ Screen (v. 0.14.0) . RNA-seq single-end reads were mapped to the GRCm39.103 version of the Mus musculus genome and annotated using STAR (v.2.7.8a) . Expression levels were determined by FeatureCounts from the Subread package (v.2.0.1) .
TCGA data were downloaded using the GenomicDataCommons R package (v.1.12.0; https://bioconductor.org/packages/GenomicDataCommons ) , TCGA ‘HTSeq–counts’ and corresponding clinical annotations. TCGA mutational data were downloaded using maftools (v.2.4.2) . Both human and mouse RNA-seq counts were normalized using VST from the DESeq2 (v.1.28.1 and v.1.44.0) package and then centred within a sample. Genes were reduced to those with direct one-to-one gene mapping between human and mouse genomes established by Ensembl, as retrieved from the biomaRt (v.2.56.1) package , . Singular-value decomposition (SVD) of the human data was performed followed by matrix factorization of both the human and mouse data into a 100-rank human space. UMAP of the combined dataset was executed using R package uwot (v.0.1.11; https://CRAN.R-project.org/package=uwot ). An adjacency matrix was constructed from a nearest-neighbours search (RANN package v.2.6.1, https://CRAN.R-project.org/package=RANN ) of the human and mouse SVD objects for clustering analysis. R package igraph (v.1.2.11 and v.2.0.3) was used to construct a graph object and the community structure was determined using Louvain •clustering. Single-sample gene set enrichment analysis (ssGSEA) analysis was performed using the R package corto (v.1.2.4) with the Hallmark gene set , downloaded using msigdbr (v.7.4.1; https://CRAN.R-project.org/package=msigdbr ). Hoshida and Chiang (also downloaded using msigdbr) subclass classification was determined by the highest enriched subclass. Tumour immune cell estimation was performed using ConsensusTME . Visualization of data by a combination of the ComplexHeatmap (v.2.4.3 and v.2.14.0) , ggplot2 (v.3.3.6 and v.3.5.1) , cowplot (v.1.1.1; https://CRAN.R-project.org/package=cowplot ) and viridis packages. Human H&E-stained tissue sections were obtained from the TCGA collection ( https://portal.gdc.cancer.gov/ ).
HuMo clusters were validated with the bulk RNA-seq data of an independent cohort of 171 HCC samples from patients undergoing resection collected in the setting of the HCC Genomic Consortium (European Genome-Phenome Archive: EGAS00001005364 ). Fastq files were aligned using STAR (v.2.5.1b) to the hg19 reference genome with gencode annotation v19 and were quantified using featureCounts (v.1.5.2). Raw counts were preprocessed and cluster attribution was performed as described above with the TCGA and mouse data. In the analysis of the transcriptomic data, positivity for previously reported gene signatures was evaluated using the Nearest Template Prediction module from GenePattern (v.3.9) . The ssGSEA projection was performed using previously reported gene signatures as well as the Hallmark gene set downloaded using msigdbr (v.7.4.1). The mutational profile of 144 HCC samples was obtained by whole-exome sequencing . Clinicopathological data (such as vascular invasion, AFP levels (≥400 ng ml −1 )) were originally reported previously .
Genes were restricted to those with significance in all comparisons (with significance defined as adjusted P < 0.05 and log 2 [FC] > 1). Data were scaled and visualized using the ComplexHeatmap package. Gene Ontology over-representation analysis was performed using the clusterProfiler package (v.3.16.1).
The use of consenting patients’ tissues surplus to diagnostic requirements for research purposes was approved by the Newcastle and North Tyneside Regional ethics committee, the Newcastle Academic Health Partners Bioresource (NAHPB) and the Newcastle upon Tyne NHS Foundation Trust Research and Development (R&D) department, in accordance with Health Research Authority guidelines. (10/H0906/41; NAHPB Project 48; REC 12/NE/0395; R&D 6579; Human Tissue Act licence 12534).
Magnetic resonance imaging (MRI) scans were performed on liver-tumour-bearing mice using the nanoScan imaging system (Mediso Medical Imaging Systems). The mice were anaesthetized and maintained under inhaled isoflurane anaesthesia (induction, 4–5% (v/v); maintenance, 1.5–2.0% (v/v)) in 95% oxygen during the entire imaging procedure. Whole-body T1-weighted gradient echo (GRE) 3D coronal/sagittal MRI sequences (echo time (TE), 3.8 ms; repetition time (TR), 20 ms; flip angle, 30 degrees; and slice thickness, 0.50 mm) were used to obtain MRI images. For quantification of scans, volumes of interest were manually drawn around the liver region on MRI scans by visual inspection using VivoQuant software (v.4.0, InviCRO). For each scan, separate volumes of interest were prepared to adjust for the position and angle of each mouse on the MRI scanner and their tumour size.
HCCOs were extracted and cultured as previously described , , with the exception that HCCOs from mice with activated β-catenin signalling were cultured in the absence of WNT and RSPO1. All mouse HCCO cultures were regularly tested for mycoplasma. For the high-throughput screen cohort 5 (BM) HCCOs were dissociated with TrypLE and plated at a density of 1 × 10 3 cells in 10 μl BME in prewarmed 384-well plates (Greiner BioOne, 781091) 5 days before adding the drugs. On day 0, a panel of 147 FDA-approved oncology drugs (AOD IX, acquired June 2019, https://dtp.cancer.gov/organization/dscb/obtaining/available_plates.htm ) was added at a final concentration of 10 µM. Staurosporin was used as an internal positive control; DMSO and untreated cells were used as an internal negative control. The medium was changed on day 4 and the compounds were freshly added. Incucyte NucLight Rapid Red (Sartorius, 4717) was added on day 6 and cells were imaged using the Opera Phenix High-Content Screening System (Perkin Elmer) on day 9. Volumes were determined using Icy BioImage software (v.2.0.0.0; https://icy.bioimageanalysis.org ) . The experiment was performed twice (using different passages from one HCCO line) in technical quadruplicates. For the drug dose–response curve screen, HCCOs (1–2 lines per cohort) were dissociated with TrypLE and plated at a density of 1 × 10 3 cells in 10 μl Matrigel (Corning, 356231) in prewarmed 96-well plates (Greiner BioOne, 655098). The treatment schedule was the same as for the HTP screen, except the medium was changed and fresh drugs were added on days 3 and 7. Drugs and concentrations are shown in the figures. Drugs were purchased from Selleckchem, dissolved in DMSO to 10 mM, aliquoted and stored at −20 °C. Cell viability was measured on day 9 using CellTitre-Glo 3D reagent (Promega, G9682) according to the manufacturer’s instructions. Luminescence was measured on the Spark Microplate Reader (Tecan). The results were normalized to the vehicle. Curve fitting and IC 50 calculation were performed using a nonlinear regression equation. All of the experiments were performed in duplicate and at least three times using different passages from one to two HCCO lines per cohort. Images of HCCOs were taken on an Olympus CKX41 using the Qimaging Retiga Exi Fast 1394 camera. For immunofluorescence analysis, HCCOs were washed with ice-cold PBS, fixed with 4% PFA and permeabilized with 0.2% Triton X-100. A list of the antibodies used is provided in Supplementary Table . Images were taken using the Zeiss 710 confocal microscope. Tumour-derived mouse HCCOs (available from all GEMMs) will be shared on reasonable request.
Human HCCOs were derived from liver cancer needle biopsies or liver resections as described before . The following human HCCO lines were used: D386-O and D953-O ( CTNNB1 WT, TP53 WT, MYC WT); D455-O ( CTNNB1 MUT, TP53 WT, MYC AMP); C948-O and C949-O ( CTNNB1 MUT, TP53 WT, MYC AMP); C655-O ( CTNNB1 WT, TP53 MUT, MYC ND); D045-O, D046-O, D803-O and R035-O ( CTNNB1 WT, TP53 MUT, MYC WT); C798-O, C975-O, D324-O, D804-O and D876-O ( CTNNB1 WT, TP53 MUT, MYC AMP); and D359-O ( CTNNB1 MUT, TP53 MUT, MYC WT). For expansion, the human HCCOs were seeded into reduced growth factor BME2 (R&D Systems, 3533-005-02) and cultured in expansion medium (EM): advanced DMEM/F-12 (Gibco, 12634010) supplemented with 1× B-27 (Gibco, 17504001), 1× N-2 (Gibco, 17502001), 10 mM nicotinamide (Sigma-Aldrich, N0636), 1.25 mM N -acetyl- l -cysteine (Sigma-Aldrich, A9165), 10 nM [Leu15]-gastrin (Sigma-Aldrich, G9145), 10 μM forskolin (Tocris, 1099), 5 μM A83-01 (Tocris, 2939), 50 ng ml −1 EGF (Peprotech, AF-100-15), 100 ng ml −1 FGF10 (Peprotech, 100-26), 25 ng ml −1 HGF (Peprotech, 100-39), 10% RSPO1-conditioned medium (v/v, homemade). HCCOs were passaged after dissociation with 0.25% trypsin-EDTA (Gibco). All human HCCOs were regularly tested for mycoplasma contamination using the MycoAlert mycoplasma detection kit (Lonza, LT07-118). Drugs were purchased from ApexBio and Selleckchem, dissolved in DMSO to 10 mM, aliquoted and stored at −20 °C. For the screening, human HCCOs were dissociated with 0.25% trypsin-EDTA (Gibco) to single cells and 1 × 10 3 cells per well were plated in a 384-well plate (Greiner BioOne, 781986) on a layer of BME2 (R&D Systems, 3533-005-02) previously diluted with EM (50:50, v/v). Cells were cultured for 3 days without treatment to allow for organoid formation. At day 3, an eight-point half-log dilution series of each compound (ranging from 10 μM to 0.00316 μM) was added using a Tecan D300e. Cell viability was measured after 5 days of treatment using the CellTiter-Glo 3D reagent (Promega, G9682). Luminescence was measured on the Synergy H1 multi-mode reader (BioTek Instruments). Results were normalized to the vehicle (DMSO). The maximal DMSO concentration was 0.2%. Curve fitting was performed using Prism (GraphPad v.9 GraphPad Software) software and the nonlinear regression equation. Results are shown as mean ± s.e.m.
After mincing into small pieces, 100 mg of healthy liver or liver tumour was digested on the gentleMACS Octo dissociator with heaters using the mouse tumour dissociation Kit (Miltenyi Biotec, 130-096-730). Dissociated cells were resuspended in 0.5% BSA in PBS, filtered through a 70 μm strainer and centrifuged at 400 g for 5 min. Cells were then resuspended in 5 ml RBC lysis buffer (Thermo Fisher Scientific, 00-4300-54) and incubated for 5 min at room temperature and washed with 0.5% BSA in PBS before being resuspended in 0.5% BSA in PBS. Cell suspensions were added to 96-well V-bottom plates (maximum density, 0.5 × 10 6 cells per well). Cells were stimulated for 3 h with complete IMDM medium containing 8% FCS, 50 μM β-mercaptoethanol, 1× penicillin–streptomycin with 1× cell activation cocktail with brefeldin A (BioLegend, 423304) at 37 °C as previously described previously . After stimulation, cells were centrifuged at 800 g for 2 min. Cells were stained in Brilliant stain buffer (BD Biosciences, 566349) containing antibodies for surface antigens for 30 min at 4 °C in the dark. Cells were then washed with PBS/0.5% BSA, centrifuged at 800 g for 2 min, followed by ice-cold PBS and incubated with Zombie NIR Fixable Viability dye (BioLegend, 423106) to stain dead cells for 20 min at 4 °C. After further washing the cells with PBS/0.5% BSA, cells were fixed and permeabilized in FOXP3 transcription factor fixation/permeabilization solution (Thermo Fisher Scientific) for 20 min at 4 °C, according to the manufacturer’s instructions. Intracellular antibodies were prepared in permeabilization buffer and incubated with cells for 30 min at 4 °C before cells were washed with permeabilization buffer, followed by PBS/0.5% BSA and finally resuspended in PBS/0.5% BSA. All of the experiments were performed using a five-laser BD LSRFortessa flow cytometer with DIVA software (BD Biosciences v.8.0.1). Compensation was determined automatically using Ultracomp eBeads (01-2222-42; Thermo Fisher Scientific). Data were analysed using FlowJo Software v.9.9.6.
Statistical analyses were performed using GraphPad Prism, software (v9 GraphPad Software) and R (v.4.0.2 and higher) with statistical tests as indicated in the figure legends. Data were tested for normal distribution. All performed t -tests were two-tailed. P values are displayed in figures. No statistical methods were used to predetermine sample sizes but our sample sizes are similar to those reported in previous publications , , – . For animal experiments, biological replicate sizes were chosen taking into account the variability observed in pilot and previous studies in respective cohorts. For all experiments, animal/sample assignment was matched for age-matched control, and assigned based on randomly assigned mouse identification markings. Batched staining and analyses alongside controls were used throughout. The investigators were not blinded for the in vivo experiments. Technical staff administering therapy were blinded to the mouse genotypes. All subsequent tissue handling and analysis were blinded and/or performed using standardized automated analyses where possible. Quantitative image analysis was performed blinded to the genotype and treatment. The data distribution for normality testing and testing of equal variances was assessed using Prism 9 Software (GraphPad Software). No data were excluded, unless mentioned otherwise, except the following, which were excluded before analysis: one biological replicate failed quality control after transcriptomic sequencing—all other biological replicates from this and other cohorts successfully passed quality control and were included in downstream analysis; two drugs were excluded from the HCCO HTP screen due to microbiological contamination and drug precipitation in multiple replicates, respectively. One sample was excluded from the RFP expression analysis during analysis (total n = 4 biological replicates): testing AAV-mediated recombination of RFP alleles (Extended Data Fig. ), one sample was a notable outlier (4.9% versus 25.7%, 25.1% and 25.8%) which on re-review was caused by inconsistent RFP staining of the section—this outlier was removed from final analysis; details are provided in the figure legend also. Where the tumour number could not be quantified due to tumour rupture, no tumour number is reported (Fig. ). Figures were assembled using Scribus v.1.4.8 ( https://www.scribus.net/ ). Images were processed using Gimp v.2.10.14 ( https://www.gimp.org/ ).
Further information on research design is available in the linked to this article.
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41586-025-08585-z.
Supplementary Table 1 Summary of end-point data of each HCC GEMM cohort. Reporting Summary Supplementary Table 2 Statistical comparisons between patient HCC tumour bulk transcriptomic data stratified by HuMo subtypes. Supplementary Table 3 Ranked results from high-throughput drug screen in mouse HuMo1 HCC-derived organoids using FDA-approved anti-cancer therapies. Supplementary Table 4 Details for PCR reactions used for genotyping end-point tumours in the HCC GEMMs. Supplementary Table 5 Details of antibodies used in the studies. Peer Review file
Source Data Figs. 1–5 and Source Data Extended Data Figs. 1, 2, 5, 8, 9, 11 and 12
|
Cisplatin Exposure Dysregulates Insulin Secretion in Male and Female Mice | 983ff5d8-9a34-431f-b6f2-71f34211b185 | 11926276 | Digestive System[mh] | Numerous studies have shown that cancer survivors, irrespective of age, treatment, and type of cancer, have an increased risk of developing new-onset type 2 diabetes (T2D) compared with the general population several months after the end of treatment ( ). Cancer survivors also develop T2D earlier than control populations ( , ). Importantly, Sylow et al. ( ) reported that out of 28,000 cancer survivors, those with new-onset T2D had a 21% higher all-cause mortality compared to cancer survivors without T2D. Cisplatin is a platinum-based chemotherapeutic agent used to treat a wide variety of cancers ( ). Cisplatin can enter cells through passive diffusion or the copper transporter 1 protein ( ), which has not been well characterized in islets. Once cisplatin enters the cell, either one or both chlorine groups are replaced by water, effectively becoming aquated and biologically active ( ). Aquated cisplatin is highly electrophilic and binds to nucleophilic centers on purine residues in nuclear and mitochondrial DNA, causing the cross-linking of DNA ( ). Cisplatin also increases the production of reactive oxygen species (ROS) and leads to oxidative stress and mitochondrial deterioration ( , ); this, in turn, leads to increased cell senescence and cell death ( ). Unfortunately, the cytotoxic effects of cisplatin are not limited to cancerous cells. Cisplatin treatment has been linked to acute toxicities and severe off-target effects, including nephrotoxicity and neurotoxicity ( ). Its effects on pancreatic islets have not been well characterized, despite compelling evidence linking cisplatin treatment with metabolic complications. Patients with testicular cancer treated with cisplatin have increased odds of developing metabolic syndrome compared with patients treated with surgery or radiotherapy and the general population, despite no major differences in BMI ( ). In a group of 219 patients who received cisplatin-based chemotherapy for head and neck cancer, 5% developed diabetes during their treatment period ( ), which was almost twice the global rate of diabetes prevalence at the time of the study ( ). The potential off-target effects of cisplatin on pancreatic islets are currently unknown. Given the established roles of DNA damage, mitochondrial dysfunction, and oxidative stress in β-cell dysfunction during T2D pathogenesis ( , ), we hypothesized that β-cells would be particularly susceptible to the off-target effects of cisplatin. Moreover, we predicted that the limited regenerative capacity of β-cells ( ) would prevent recovery of the β-cell population following injury, contributing to an increased risk of T2D in cancer survivors who receive cisplatin treatment. In our study, we assessed the effects of cisplatin on islet function systemically by exposing male and female mice to cisplatin in vivo and directly by treating isolated male and female mouse islets, as well as female human donor islets, with cisplatin in vitro. We found that in vivo cisplatin exposure decreased plasma insulin levels in both male and female mice and in vitro cisplatin exposure profoundly dysregulated insulin secretion in both isolated mouse and human donor islets.
In Vivo Cisplatin Exposure Protocol Male and female C57BL/6N mice (Charles River Laboratories), ∼16 weeks old, were maintained on a 12-h light/dark cycle with ad libitum access to a standard rodent chow diet (Teklad Diet #2018; Harlan Laboratories) and water. All experiments were approved by the Carleton University Animal Care Committee and performed in accordance with Canadian Council on Animal Care guidelines. Prior to starting experimental protocols, animals were tracked for 2 weeks before being randomly assigned to treatment groups, ensuring body weight and fasting blood glucose levels were consistent between groups. At ∼18 weeks of age, mice received i.p. injections of 0.9% saline (vehicle control) or 2 mg/kg cisplatin (#232120-50MG; Sigma-Aldrich) every other day over the course of 14 days, for a total of seven injections and a cumulative dose of 14 mg/kg. A clinical dose of 50 mg/m 2 cisplatin in patients is equivalent to ∼14 mg/kg cisplatin in mice based on an animal equivalent dosing calculation that accounts for metabolic differences between species ( ). Similar dosing protocols in rodents have shown cisplatin-induced toxicity with <20% body weight loss ( , ). Mice were allowed to recover for 1 week after the end of the exposure period before in vivo metabolic assessments were conducted. Mice were euthanized 2 weeks after cisplatin or vehicle exposure. To determine whether the effects of cisplatin were reproducible in different strains of mice, male SCID-beige and DBA/2 mice were also exposed to cisplatin in vivo as described above, and metabolic assessments were performed. Details of the study protocol are described in the Supplementary Methods . In Vivo Metabolic Assessments All metabolic analyses were performed in conscious, restrained mice, as previously described ( ). Briefly, body weight and blood glucose were measured following a 4-h morning fast twice a week throughout the study. An insulin tolerance test (ITT) was conducted at 1-week postexposure and an i.p. glucose tolerance test (ipGTT) was performed the following week. Blood samples were collected at 0, 15, 30, and 60 min during the ipGTT for measuring plasma insulin levels by ultrasensitive rodent insulin ELISA (#80-INSMSU-E10; ALPCO). All blood glucose measurements were conducted using a handheld glucometer (Medisure). Mouse Islet Isolation and Culture Mice were euthanized by isofluorane overdose followed by cervical dislocation, and pancreata were inflated via common bile duct injection with collagenase (1,000 IU/mL, #C7657; Sigma-Aldrich) dissolved in Hanks’ balanced salt solution (HBSS) (137 mmol/L NaCl, 5.4 mmol/L KCl, 4.2 mmol/L NaH 2 PO 4 , 4.1 mmol/L KH 2 PO 4 , 10 mmol/L HEPES, 1 mmol/L MgCl 2 , 5 mmol/L anhydrous dextrose, pH 7.2). Inflated pancreatic tissues were excised and incubated at 37°C for 10.5 min then vigorously agitated, after which the collagenase reaction was stopped by adding cold HBSS with 1 mmol/L CaCl 2 . The digested pancreatic tissues were washed three times in HBSS with CaCl 2 , and islets were separated from exocrine tissue via a Histopaque gradient (#10771; Sigma-Aldrich). Islets were then filtered through a 70-μm cell strainer, resuspended in RPMI medium (#11875093 [Gibco] or #350-000-CL [Wisent Bioproducts]) supplemented with 10% (v/v) FBS (#F1051-500ML; Sigma-Aldrich) and 1% (v/v) penicillin-streptomycin (#30-002-Cl [Corning] or #15140-122-100 [Gibco]) and handpicked under a dissecting scope to >95% purity. Human Donor Islet Culture Islets from three female human donors (ages 52, 57, and 69 years) were isolated at the Alberta Diabetes Institute IsletCore. Islet isolation was approved by the Human Research Ethics Board at the University of Alberta (Pro00013094). All donors’ families gave informed consent for the use of pancreatic tissue in research. Islets were shipped overnight in Connaught Medical Research Laboratories 1066 medium (#15110CV; Corning) supplemented with 0.5% (v/v) BSA (#BAL62; Equitech Bio Inc.), 1% (v/v) insulin-transferrin-selenium (#25800CR; Corning), 0.5% (v/v) penicillin-streptomycin (#09-757F; Lonza), and 2% (v/v) Gibco GlutaMAX supplement (#35050061; Thermo Fisher Scientific). Upon arrival, islets were handpicked into DMEM (#11-885-084; Thermo Fisher Scientific) supplemented with 10% FBS (Sigma-Aldrich) and 1% penicillin-streptomycin (Gibco) and allowed to rest overnight at 37°C with 5% CO 2 before cisplatin or vehicle exposure. See Supplementary Table 1 for donor characteristics. In Vitro Cisplatin Exposure Protocol The peak intact cisplatin concentration in patients after receiving one dose of 50–100 mg/m 2 cisplatin is ∼10 μmol/L ( ). This, coupled with a review of cisplatin concentrations typically used in cell line studies ( , ), prompted us to use 10 μmol/L for our in vitro studies with isolated islets. For all in vitro experiments, islets were isolated from male and female C57BL/6 mice on a mixed J/N background, as described above. For all end points except patch-clamp recordings, intact mouse and human islets were incubated in complete RPMI medium or DMEM, respectively, overnight at 37°C with 5% CO 2 before being handpicked for in vitro cisplatin or vehicle exposure. One-half of the islets from each biological replicate was transferred to complete RPMI medium or DMEM containing either 10 μmol/L cisplatin (Sigma-Aldrich) or 0.9% saline (vehicle control) and cultured between 6 and 48 h. Media were refreshed after 24 h. For patch-clamp recordings, intact mouse islets were dispersed with enzyme-free Hanks’-based cell dissociation buffer (#13150-016; Gibco) immediately after isolation, plated in a tissue culture–treated dish, and then incubated overnight at 37°C with 5% CO 2 in complete RPMI medium prior to a 4-h exposure to 10 μmol/L cisplatin or vehicle. Glucose-Stimulated Insulin Secretion Assays To assess static glucose-stimulated insulin secretion (GSIS), 25 islets/replicate were handpicked and underwent sequential 1-h incubations in Krebs-ringer bicarbonate HEPES buffer (KRBH) containing 2.8 mmol/L glucose (low glucose [LG]), KRBH with 16.7 mmol/L glucose (high glucose [HG]), and KRBH with 30 mmol/L KCl as previously described ( ). To assess GSIS dynamically, 70 islets/replicate were washed with prewarmed (37°C) PBS (#D8662; Sigma-Aldrich) and then loaded in Perspex microcolumns between two layers of acrylamide-based microbeads (#PERI-BEADS-20; Biorep Technologies). Islets were perifused for 40 min with prewarmed LG KRBH to equilibrate the islets. The islets were then perifused with LG KRBH for 15 min, HG KRBH for 45 min, LG KRBH for 25 min, KCl KRBH for 35 min, and LG KRBH for 25 min. Samples were collected every 5 min at a flow rate of 40 μL/min except for the first 15 min after perifusion with HG and KCl KRBH, during which samples were collected every 2.5 min at a rate of 80 μL/min. Islets and solutions were maintained at 37°C throughout the perifusion, while the collection plate was kept at 4°C using a built-in tray cooling system. Samples were stored at −80°C until analysis. Insulin concentrations of all static GSIS and perifusion samples collected from mouse islets were measured by rodent insulin chemiluminescence ELISA (#80-INSMR-CH10; ALPCO). Insulin concentrations of perifusion samples collected from islets of donor R542 were measured by human insulin chemiluminescence ELISA (#80-INSHU-CH10; ALPCO), while those from donors R551 and R552 were measured by radioimmunoassay (#HI-14K; Millipore). Human insulin concentrations are reported as fold change over basal (5–15 min) insulin secretion. Oxygen Consumption Analysis Islet respiration was quantified using the Seahorse XFe24 Analyzer (Agilent Technologies). A total of 70 mouse islets/replicate were handpicked and washed with Seahorse XF RPMI medium (103576, pH 7.4; Agilent Technologies) supplemented with 2 mmol/L sodium pyruvate, 2 mmol/L l -glutamine, and 1% (v/v) FBS containing either 2.8 or 16.7 mmol/L glucose. The islets were plated in a poly- d -lysine (#P7280; Sigma-Aldrich)–coated 24-well islet capture plate (#103518-100; Agilent Technologies) and incubated at 37°C without CO 2 for 1.5 h. Media were refreshed before loading the plate into the XFe24 Analyzer. The basal oxygen consumption rate was measured for ∼35 min. For islets pre-incubated in RPMI medium with 16.7 mmol/L glucose, wells were exposed to sequential injections of 2.5 μmol/L oligomycin for eight cycles, 3 μmol/L carbonyl cyanide- p -trifluoromethoxyphenylhydrazone (FCCP) for five cycles, and a combination of 3 μmol/L rotenone and 3 μmol/L antimycin A for six cycles. For islets pre-incubated in RPMI medium with 2.8 mmol/L glucose, wells were exposed to 16.7 mmol/L glucose for six cycles prior to the same series of injections described above. With each cycle, the solutions were mixed for 3 min, samples were allowed to rest for 2 min, and oxygen consumption was measured for 3 min. Mitochondrial function parameters were calculated as indicated in Supplementary Table 2 . Cell Viability Assay After 6, 24, and 48 h of cisplatin or vehicle exposure, 25 mouse islets/replicate were handpicked, dispersed, and stained with 0.5 μmol/L Hoescht (#62249; Thermo Fisher Scientific), 1.25 μmol/L calcein (#L3224; Invitrogen), and 0.5 μg/mL propidium iodide (PI) (#P21493; Invitrogen) dyes to assess cell viability as previously described ( ). Cells were imaged with an Axio Observer 7 microscope. Exocytotic Capacity Measurements Exocytotic capacity of dispersed mouse β-cells was measured following a 4-h incubation with cisplatin or vehicle in vitro. Cells were washed with 2.8 mmol/L glucose RPMI medium, preincubated with 2.8 mmol/L glucose RPMI medium for 1 h, and then patch-clamped in bath solution (118 mmol/L NaCl, 20 mmol/L tetraethylammonium chloride, 5.6 mmol/L KCl, 1.2 MgCl 2 , 2.6 mmol/L CaCl 2 , 5 mmol/L HEPES, pH 7.4) containing either 2.8 or 16.7 mmol/L glucose. The patch pipettes were filled with an internal pipette solution (125 mmol/L CsOH, 125 mmol/L glutamate, 10 mmol/L CsCl, 10 mmol/L NaCl, 1 mmol/L MgCl 2 , 0.05 mmol/L EGTA, 5 mmol/L HEPES, 3 mmol/L MgATP, 0.1 mmol/L cAMP, pH 7.1). Whole-cell capacitance was recorded using a sine + DC lock-in function of an EPC10 amplifier and PatchMaster software (HEKA Electronics) ∼1 min after obtaining a whole-cell configuration. Capacitance was measured in response to 10 depolarizations from −70 to 0 mV, each 500 ms long. Changes in capacitance were normalized to initial cell size. Quantitative Real-Time PCR After 6, 24, and 48 h of in vitro cisplatin or vehicle exposure, 70 mouse islets/replicate were handpicked and stored in buffer RLT with 1% β-mercaptoethanol. Similarly, 70 human islets/replicate were handpicked after 48 h of in vitro cisplatin or vehicle exposure and stored in buffer RLT with 1% β-mercaptoethanol. RNA was extracted using the RNeasy Micro Kit (#74004; QIAGEN) per the manufacturer’s instructions. DNase treatment was performed prior to cDNA synthesis with iScript gDNA Clear cDNA Synthesis Kit (#1725035; Bio-Rad). Following this, quantitative PCR (qPCR) was performed using SsoAdvanced Universal SYBR Green Supermix (#1725271; Bio-Rad) and run on a CFX384 (Bio-Rad). All targets were run alongside no reverse transcription and no cDNA template controls. Ppia/PPIA was used as the reference gene given its stable expression under both control and treatment conditions. Data were analyzed using the 2 −ΔΔCt method. Primer sequences are listed in Supplementary Table 3 . TempO-Seq Analysis Following 48-h in vitro cisplatin or vehicle exposure, RNA libraries were prepared from RNA lysate isolated from 70 mouse islets/replicate and processed with the TempO-Seq Mouse Whole Transcriptome (version 1.1) Assay panel (BioSpyder Technologies) according to the manufacturer’s instructions. Individual libraries were pooled and purified using the NucleoSpin Gel and PCR Clean Up Kit (#740609; Macherey-Nagel) according to modified BioSpyder instructions. The pooled library was sequenced using the Illumina NextSeq 2000 with a P1 Reagent Kit and 100-cycle high-throughput flow cell (Illumina). Data were processed using the Omics data analysis framework for regulatory application (R-ODAF; https://github.com/R-ODAF/R-ODAF_Health_Canada ) pipeline ( ). Differentially expressed genes (DEGs) were subjected to overrepresentation analysis using clusterProfiler (version 4.12.0) ( ), querying the gene ontology (GO) ( ) and Kyoto Encyclopedia of Genes and Genomes (KEGG) ( ) databases. Significance was set at an adjusted P < 0.01 and Q < 0.05. Full details on library preparation and data processing are provided in the Supplementary Methods . Statistical Analysis Aside from TempO-Seq, all statistical analyses were conducted using GraphPad Prism 10.1.2 (GraphPad Software). Specific statistical tests and sample sizes are indicated in figure legends. For all analyses, P < 0.05 was considered statistically significant. Data are presented as mean ± SEM. Data and Resource Availability Sequencing data are available from the National Center for Biotechnology Information Gene Expression Omnibus database (accession no. GSE278504). All data that support the findings of this study are available from the corresponding author upon reasonable request.
Male and female C57BL/6N mice (Charles River Laboratories), ∼16 weeks old, were maintained on a 12-h light/dark cycle with ad libitum access to a standard rodent chow diet (Teklad Diet #2018; Harlan Laboratories) and water. All experiments were approved by the Carleton University Animal Care Committee and performed in accordance with Canadian Council on Animal Care guidelines. Prior to starting experimental protocols, animals were tracked for 2 weeks before being randomly assigned to treatment groups, ensuring body weight and fasting blood glucose levels were consistent between groups. At ∼18 weeks of age, mice received i.p. injections of 0.9% saline (vehicle control) or 2 mg/kg cisplatin (#232120-50MG; Sigma-Aldrich) every other day over the course of 14 days, for a total of seven injections and a cumulative dose of 14 mg/kg. A clinical dose of 50 mg/m 2 cisplatin in patients is equivalent to ∼14 mg/kg cisplatin in mice based on an animal equivalent dosing calculation that accounts for metabolic differences between species ( ). Similar dosing protocols in rodents have shown cisplatin-induced toxicity with <20% body weight loss ( , ). Mice were allowed to recover for 1 week after the end of the exposure period before in vivo metabolic assessments were conducted. Mice were euthanized 2 weeks after cisplatin or vehicle exposure. To determine whether the effects of cisplatin were reproducible in different strains of mice, male SCID-beige and DBA/2 mice were also exposed to cisplatin in vivo as described above, and metabolic assessments were performed. Details of the study protocol are described in the Supplementary Methods .
All metabolic analyses were performed in conscious, restrained mice, as previously described ( ). Briefly, body weight and blood glucose were measured following a 4-h morning fast twice a week throughout the study. An insulin tolerance test (ITT) was conducted at 1-week postexposure and an i.p. glucose tolerance test (ipGTT) was performed the following week. Blood samples were collected at 0, 15, 30, and 60 min during the ipGTT for measuring plasma insulin levels by ultrasensitive rodent insulin ELISA (#80-INSMSU-E10; ALPCO). All blood glucose measurements were conducted using a handheld glucometer (Medisure).
Mice were euthanized by isofluorane overdose followed by cervical dislocation, and pancreata were inflated via common bile duct injection with collagenase (1,000 IU/mL, #C7657; Sigma-Aldrich) dissolved in Hanks’ balanced salt solution (HBSS) (137 mmol/L NaCl, 5.4 mmol/L KCl, 4.2 mmol/L NaH 2 PO 4 , 4.1 mmol/L KH 2 PO 4 , 10 mmol/L HEPES, 1 mmol/L MgCl 2 , 5 mmol/L anhydrous dextrose, pH 7.2). Inflated pancreatic tissues were excised and incubated at 37°C for 10.5 min then vigorously agitated, after which the collagenase reaction was stopped by adding cold HBSS with 1 mmol/L CaCl 2 . The digested pancreatic tissues were washed three times in HBSS with CaCl 2 , and islets were separated from exocrine tissue via a Histopaque gradient (#10771; Sigma-Aldrich). Islets were then filtered through a 70-μm cell strainer, resuspended in RPMI medium (#11875093 [Gibco] or #350-000-CL [Wisent Bioproducts]) supplemented with 10% (v/v) FBS (#F1051-500ML; Sigma-Aldrich) and 1% (v/v) penicillin-streptomycin (#30-002-Cl [Corning] or #15140-122-100 [Gibco]) and handpicked under a dissecting scope to >95% purity.
Islets from three female human donors (ages 52, 57, and 69 years) were isolated at the Alberta Diabetes Institute IsletCore. Islet isolation was approved by the Human Research Ethics Board at the University of Alberta (Pro00013094). All donors’ families gave informed consent for the use of pancreatic tissue in research. Islets were shipped overnight in Connaught Medical Research Laboratories 1066 medium (#15110CV; Corning) supplemented with 0.5% (v/v) BSA (#BAL62; Equitech Bio Inc.), 1% (v/v) insulin-transferrin-selenium (#25800CR; Corning), 0.5% (v/v) penicillin-streptomycin (#09-757F; Lonza), and 2% (v/v) Gibco GlutaMAX supplement (#35050061; Thermo Fisher Scientific). Upon arrival, islets were handpicked into DMEM (#11-885-084; Thermo Fisher Scientific) supplemented with 10% FBS (Sigma-Aldrich) and 1% penicillin-streptomycin (Gibco) and allowed to rest overnight at 37°C with 5% CO 2 before cisplatin or vehicle exposure. See Supplementary Table 1 for donor characteristics.
The peak intact cisplatin concentration in patients after receiving one dose of 50–100 mg/m 2 cisplatin is ∼10 μmol/L ( ). This, coupled with a review of cisplatin concentrations typically used in cell line studies ( , ), prompted us to use 10 μmol/L for our in vitro studies with isolated islets. For all in vitro experiments, islets were isolated from male and female C57BL/6 mice on a mixed J/N background, as described above. For all end points except patch-clamp recordings, intact mouse and human islets were incubated in complete RPMI medium or DMEM, respectively, overnight at 37°C with 5% CO 2 before being handpicked for in vitro cisplatin or vehicle exposure. One-half of the islets from each biological replicate was transferred to complete RPMI medium or DMEM containing either 10 μmol/L cisplatin (Sigma-Aldrich) or 0.9% saline (vehicle control) and cultured between 6 and 48 h. Media were refreshed after 24 h. For patch-clamp recordings, intact mouse islets were dispersed with enzyme-free Hanks’-based cell dissociation buffer (#13150-016; Gibco) immediately after isolation, plated in a tissue culture–treated dish, and then incubated overnight at 37°C with 5% CO 2 in complete RPMI medium prior to a 4-h exposure to 10 μmol/L cisplatin or vehicle.
To assess static glucose-stimulated insulin secretion (GSIS), 25 islets/replicate were handpicked and underwent sequential 1-h incubations in Krebs-ringer bicarbonate HEPES buffer (KRBH) containing 2.8 mmol/L glucose (low glucose [LG]), KRBH with 16.7 mmol/L glucose (high glucose [HG]), and KRBH with 30 mmol/L KCl as previously described ( ). To assess GSIS dynamically, 70 islets/replicate were washed with prewarmed (37°C) PBS (#D8662; Sigma-Aldrich) and then loaded in Perspex microcolumns between two layers of acrylamide-based microbeads (#PERI-BEADS-20; Biorep Technologies). Islets were perifused for 40 min with prewarmed LG KRBH to equilibrate the islets. The islets were then perifused with LG KRBH for 15 min, HG KRBH for 45 min, LG KRBH for 25 min, KCl KRBH for 35 min, and LG KRBH for 25 min. Samples were collected every 5 min at a flow rate of 40 μL/min except for the first 15 min after perifusion with HG and KCl KRBH, during which samples were collected every 2.5 min at a rate of 80 μL/min. Islets and solutions were maintained at 37°C throughout the perifusion, while the collection plate was kept at 4°C using a built-in tray cooling system. Samples were stored at −80°C until analysis. Insulin concentrations of all static GSIS and perifusion samples collected from mouse islets were measured by rodent insulin chemiluminescence ELISA (#80-INSMR-CH10; ALPCO). Insulin concentrations of perifusion samples collected from islets of donor R542 were measured by human insulin chemiluminescence ELISA (#80-INSHU-CH10; ALPCO), while those from donors R551 and R552 were measured by radioimmunoassay (#HI-14K; Millipore). Human insulin concentrations are reported as fold change over basal (5–15 min) insulin secretion.
Islet respiration was quantified using the Seahorse XFe24 Analyzer (Agilent Technologies). A total of 70 mouse islets/replicate were handpicked and washed with Seahorse XF RPMI medium (103576, pH 7.4; Agilent Technologies) supplemented with 2 mmol/L sodium pyruvate, 2 mmol/L l -glutamine, and 1% (v/v) FBS containing either 2.8 or 16.7 mmol/L glucose. The islets were plated in a poly- d -lysine (#P7280; Sigma-Aldrich)–coated 24-well islet capture plate (#103518-100; Agilent Technologies) and incubated at 37°C without CO 2 for 1.5 h. Media were refreshed before loading the plate into the XFe24 Analyzer. The basal oxygen consumption rate was measured for ∼35 min. For islets pre-incubated in RPMI medium with 16.7 mmol/L glucose, wells were exposed to sequential injections of 2.5 μmol/L oligomycin for eight cycles, 3 μmol/L carbonyl cyanide- p -trifluoromethoxyphenylhydrazone (FCCP) for five cycles, and a combination of 3 μmol/L rotenone and 3 μmol/L antimycin A for six cycles. For islets pre-incubated in RPMI medium with 2.8 mmol/L glucose, wells were exposed to 16.7 mmol/L glucose for six cycles prior to the same series of injections described above. With each cycle, the solutions were mixed for 3 min, samples were allowed to rest for 2 min, and oxygen consumption was measured for 3 min. Mitochondrial function parameters were calculated as indicated in Supplementary Table 2 .
After 6, 24, and 48 h of cisplatin or vehicle exposure, 25 mouse islets/replicate were handpicked, dispersed, and stained with 0.5 μmol/L Hoescht (#62249; Thermo Fisher Scientific), 1.25 μmol/L calcein (#L3224; Invitrogen), and 0.5 μg/mL propidium iodide (PI) (#P21493; Invitrogen) dyes to assess cell viability as previously described ( ). Cells were imaged with an Axio Observer 7 microscope.
Exocytotic capacity of dispersed mouse β-cells was measured following a 4-h incubation with cisplatin or vehicle in vitro. Cells were washed with 2.8 mmol/L glucose RPMI medium, preincubated with 2.8 mmol/L glucose RPMI medium for 1 h, and then patch-clamped in bath solution (118 mmol/L NaCl, 20 mmol/L tetraethylammonium chloride, 5.6 mmol/L KCl, 1.2 MgCl 2 , 2.6 mmol/L CaCl 2 , 5 mmol/L HEPES, pH 7.4) containing either 2.8 or 16.7 mmol/L glucose. The patch pipettes were filled with an internal pipette solution (125 mmol/L CsOH, 125 mmol/L glutamate, 10 mmol/L CsCl, 10 mmol/L NaCl, 1 mmol/L MgCl 2 , 0.05 mmol/L EGTA, 5 mmol/L HEPES, 3 mmol/L MgATP, 0.1 mmol/L cAMP, pH 7.1). Whole-cell capacitance was recorded using a sine + DC lock-in function of an EPC10 amplifier and PatchMaster software (HEKA Electronics) ∼1 min after obtaining a whole-cell configuration. Capacitance was measured in response to 10 depolarizations from −70 to 0 mV, each 500 ms long. Changes in capacitance were normalized to initial cell size.
After 6, 24, and 48 h of in vitro cisplatin or vehicle exposure, 70 mouse islets/replicate were handpicked and stored in buffer RLT with 1% β-mercaptoethanol. Similarly, 70 human islets/replicate were handpicked after 48 h of in vitro cisplatin or vehicle exposure and stored in buffer RLT with 1% β-mercaptoethanol. RNA was extracted using the RNeasy Micro Kit (#74004; QIAGEN) per the manufacturer’s instructions. DNase treatment was performed prior to cDNA synthesis with iScript gDNA Clear cDNA Synthesis Kit (#1725035; Bio-Rad). Following this, quantitative PCR (qPCR) was performed using SsoAdvanced Universal SYBR Green Supermix (#1725271; Bio-Rad) and run on a CFX384 (Bio-Rad). All targets were run alongside no reverse transcription and no cDNA template controls. Ppia/PPIA was used as the reference gene given its stable expression under both control and treatment conditions. Data were analyzed using the 2 −ΔΔCt method. Primer sequences are listed in Supplementary Table 3 .
Following 48-h in vitro cisplatin or vehicle exposure, RNA libraries were prepared from RNA lysate isolated from 70 mouse islets/replicate and processed with the TempO-Seq Mouse Whole Transcriptome (version 1.1) Assay panel (BioSpyder Technologies) according to the manufacturer’s instructions. Individual libraries were pooled and purified using the NucleoSpin Gel and PCR Clean Up Kit (#740609; Macherey-Nagel) according to modified BioSpyder instructions. The pooled library was sequenced using the Illumina NextSeq 2000 with a P1 Reagent Kit and 100-cycle high-throughput flow cell (Illumina). Data were processed using the Omics data analysis framework for regulatory application (R-ODAF; https://github.com/R-ODAF/R-ODAF_Health_Canada ) pipeline ( ). Differentially expressed genes (DEGs) were subjected to overrepresentation analysis using clusterProfiler (version 4.12.0) ( ), querying the gene ontology (GO) ( ) and Kyoto Encyclopedia of Genes and Genomes (KEGG) ( ) databases. Significance was set at an adjusted P < 0.01 and Q < 0.05. Full details on library preparation and data processing are provided in the Supplementary Methods .
Aside from TempO-Seq, all statistical analyses were conducted using GraphPad Prism 10.1.2 (GraphPad Software). Specific statistical tests and sample sizes are indicated in figure legends. For all analyses, P < 0.05 was considered statistically significant. Data are presented as mean ± SEM.
Sequencing data are available from the National Center for Biotechnology Information Gene Expression Omnibus database (accession no. GSE278504). All data that support the findings of this study are available from the corresponding author upon reasonable request.
In Vivo Cisplatin Exposure Decreases Plasma Insulin Levels in Both Male and Female Mice and Alters Ex Vivo Islet Function in Males We first assessed how cisplatin impacts systemic glucose homeostasis in male and female mice exposed to vehicle or cisplatin over 2 weeks ( ). Cisplatin exposure did not affect body weight ( Supplementary Fig. 1 A and C ), fasting blood glucose levels ( Supplementary Fig. 1 B and D ), or insulin sensitivity at 1 week posttreatment ( ) in either sex. During an ipGTT at 2 weeks postexposure, male mice exhibited no differences in glucose tolerance ( ), but female cisplatin-exposed mice had a slightly lower peak glucose value than vehicle-exposed female mice ( ). Both male and female cisplatin-exposed mice had reduced plasma insulin levels compared with controls during the ipGTT ( ). The reduction of plasma insulin in cisplatin-exposed mice was reproduced in two separate cohorts using DBA/2 and immunocompromised SCID-beige male mice ( Supplementary Fig. 2 ), which are relevant mouse models to metabolic research and patients with cancer. Islets isolated 2 weeks postexposure showed no differences in basal insulin release, GSIS, stimulation index, or insulin content between treatment groups for either sex ( ). However, islets from cisplatin-exposed male mice had reduced KCl-stimulated insulin secretion and reduced oxygen consumption in response to glucose stimulation compared with controls ( ); this was not observed in islets from female mice ( ). This suggests that low plasma insulin levels in cisplatin-exposed mice may originate from intrinsic impairments within the islets but are likely also influenced by defects in peripheral tissues. In Vitro Cisplatin Exposure Impairs Oxygen Consumption and Insulin Secretion in Mouse Islets To determine whether cisplatin directly impairs islet function, islets were isolated from male and female mice and exposed to 10 μmol/L cisplatin or vehicle for 48 h in vitro prior to functional analyses as outlined in . Male and female mouse islets exposed directly to cisplatin showed profoundly impaired oxygen consumption ( ) and insulin secretion ( ) after 48 h. Specifically, cisplatin-exposed mouse islets had trending increases in basal insulin secretion at the beginning and end of the perifusion assay ( ) and had pronounced reductions in both GSIS and KCl-stimulated insulin release ( ) compared with vehicle-exposed islets. Interestingly, despite cisplatin-exposed islets having nearly abolished KCl-stimulated insulin secretion ( ), there was a delayed insulin peak post-KCl perifusion in both sexes ( ). Cisplatin appears to hyperstimulate insulin release under basal glucose conditions, reduce the sensitivity of islets to glucose, and impair the release of insulin independent of glucose metabolism. Notably, the effects of cisplatin on the in vitro metabolic function of mouse islets were not sex-specific. Cisplatin Exposure Does Not Affect Mouse Islet Cell Viability Within 48 h Since the effects of cisplatin on β-cell function in vitro were similar between sexes, we focused on male mouse islets for additional analyses. We first determined whether our in vitro cisplatin dosing protocol affected islet cell viability to assess if the profound decreases in metabolic function and insulin secretion were attributed to cell death. Islets were treated with cisplatin or vehicle for 6, 24, or 48 h. At each timepoint, intact islets were imaged to visualize morphology and then dispersed into a single-cell suspension to measure cell viability via an image-based assay ( ). Based on a qualitative assessment, islets from both treatment groups appeared to be healthy at all timepoints ( ). Cisplatin exposure did not affect the percentage of PI + (i.e., dead/dying) islet cells at any timepoint ( ). These data indicate that exposure to 10 μmol/L cisplatin for 48 h was not overtly cytotoxic to mouse islets. Cisplatin Alters Insulin Release and Exocytotic Capacity of β-Cells in Male Mouse Islets Given that we were not able to recover islets from the perifusion system to assess insulin content, we measured bulk insulin secretion and insulin content in mouse islets using a static GSIS assay after 48 h of cisplatin exposure in vitro ( ). Cisplatin-exposed islets showed significantly elevated basal insulin secretion under LG conditions ( ), a small decrease in stimulation index ( ), and no change in total insulin content ( ). The robust increase in basal insulin secretion in cisplatin-exposed islets aligned with the perifusion data ( ), but the lack of difference under HG conditions in the static assay was surprising. This may be attributed to the longer period for second-phase insulin secretion during a static GSIS, which was not as severely impacted by cisplatin as first-phase insulin secretion ( ). The similarity in insulin content between treatment groups confirms that cisplatin does not cause pronounced β-cell loss or depletion of insulin stores. The altered insulin secretion in cisplatin-exposed islets could be attributed to changes in exocytotic capacity of the β-cells. To explore this, we measured the capacitance responses of male mouse β-cells by whole-cell patch-clamp. Under LG conditions, there was a clear increase in both the average and cumulative exocytotic capacity of cisplatin-exposed β-cells compared with vehicle-exposed β-cells ( ). However, exocytotic capacity of β-cells from both treatment groups was comparable under HG conditions ( ). These results suggest that cisplatin exposure increases exocytosis of insulin granules from β-cells in LG environments but not in HG environments. Thus, additional mechanisms are driving defective GSIS in cisplatin-exposed islets. Cisplatin Exposure Significantly Reduces Oxygen Consumption in Male Mouse Islets To better understand how cisplatin disrupts GSIS ( ), we assessed mitochondrial function in vehicle- and cisplatin-exposed male mouse islets in greater detail. In this experiment, islets were pre-incubated in LG Seahorse XF RPMI medium to allow for assessment of oxygen consumption in response to HG. Cisplatin-exposed islets had significantly impaired oxygen consumption compared with vehicle-exposed islets throughout the assay and showed a significant or trending reduction in all calculated parameters ( ), consistent with our previous experiment ( ). Importantly, cisplatin-exposed islets did not robustly increase oxygen consumption when stimulated with HG ( ). These data indicate that in vitro cisplatin exposure disrupts mitochondrial function in mouse islets, which likely contributes to the impaired GSIS ( and K ). Cisplatin Exposure Alters the Transcription of Genes Regulating Apoptosis, Oxidative Stress, and Insulin Processing Over Time We measured expression of key genes related to islet function and cell stress at 6, 24, and 48 h following vehicle or cisplatin exposure to better understand the temporal effects of cisplatin ( ). We first assessed expression of genes in the Bcl-2 family, key players in the intrinsic apoptosis pathway. Cisplatin-exposed islets showed a sustained, approximately fivefold downregulation of Bcl2 and a progressively increasing upregulation of Bcl2l1 , both anti-apoptotic genes, compared with vehicle-exposed islets between 6 and 48 h ( ). The pro-apoptotic gene Bax was upregulated approximately twofold in cisplatin-exposed islets at both 24 and 48 h ( ). Additionally, Cdkn1a , which encodes for p21, a marker of DNA damage and senescence ( ), was highly upregulated in cisplatin-exposed islets as early as 6 h postexposure ( ). Together, these data suggest that cisplatin induces the intrinsic apoptosis pathway, but the anti-apoptotic gene Bcl2l1 is being activated to promote a pro-survival phenotype. Ppargc1a , a transcriptional coactivator that regulates mitochondrial function ( , ), was modestly reduced in cisplatin-exposed islets at 6 h but not 24 or 48 h ( ). Cisplatin-exposed islets had an approximately twofold upregulation of Nrf2 , a marker of oxidative stress, at all three time points ( ) and increased expression of Nrf2 downstream targets Gpx1 at 24 and 48 h and Hmox1 at 6 h ( ). These changes imply that cisplatin acutely activates oxidative stress responses in mouse islets. Cisplatin did not affect expression of insulin ( Ins1 , Ins2 ) or proprotein convertase genes ( Pcsk1 , Pcsk2 ) at 6 h but reduced expression of these genes at 24 and 48 h compared with vehicle ( ). The most pronounced effect was on Pcsk2 expression, which was reduced approximately fourfold at 24 h and approximately eightfold at 48 h in cisplatin-exposed islets ( ). These results suggest that cisplatin exposure impairs proinsulin production and processing. Genes Related to Insulin Secretion and β-Cell Identity Are Differentially Expressed in Cisplatin- Versus Vehicle-Exposed Mouse Islets To further explore how cisplatin exposure alters islet gene expression, we performed a TempO-Seq analysis on male mouse islets following 48 h of treatment ( and Supplementary Fig. 3 ). In total, 1,022 probes were significantly different between cisplatin- and vehicle-exposed islets ( Supplementary Table 4 ). Pathway analysis of DEGs using both the GO biological process (GO-BP) and the KEGG pathway databases identified insulin secretion as the most significantly enriched gene set ( and Supplementary Fig. 3 A ). In fact, the 20 most significantly enriched gene sets in the GO-BP database were associated with protein/hormone transport and secretion ( ). Most of the enriched genes within each of the 20 gene sets were downregulated, with >75% of genes downregulated in 18 out of 20 gene sets ( and Supplementary Table 5 ). Since insulin secretion was the most enriched term across both databases, we next used hierarchal clustering of DEGs to more closely examine the insulin secretion pathway ( and Supplementary Fig. 3 B ). Out of 41 DEGs (represented by 46 probes), cisplatin exposure upregulated 6 genes and downregulated 35 genes ( ). Of the 35 downregulated DEGs, several were β-cell maturity and identity markers, including Ucn3 , Nkx6.1 , and Mafa ( ). Genes involved in paracrine signaling and potentiation of insulin secretion, including Gcg , Glp1r , and Abcc8 , were also downregulated ( and Supplementary Fig. 3 B ). When visualized, several aspects of the insulin secretion pathway, from initial stimuli reception to insulin granule release, were downregulated ( and Supplementary Fig. 3 C ). Overall, the transcriptomic analysis confirmed that insulin secretion is highly impacted by cisplatin exposure, supporting our previous findings. Cisplatin Exposure Impairs Insulin Secretion and Reduces Expression of β-Cell Identity Genes in Human Islets To begin translating our results in mouse islets to human health, we performed a dynamic insulin secretion assay and measured the expression of key genes related to islet and β-cell function in human islets obtained from three female organ donors ( ). Both GSIS and KCl-stimulated insulin secretion were reduced in cisplatin-exposed human islets compared with vehicle-exposed islets ( ). Cisplatin-exposed human islets also showed a profound downregulation of insulin ( INS ) and proprotein convertase genes ( PCSK1 and PCSK2 ) after 48 h compared with vehicle-exposed islets ( ). We also observed a consistent decrease in β-cell identity genes GCK , UCN3 , and MAFA in cisplatin-exposed human islets ( ). Interestingly, BCL2 expression was downregulated in cisplatin-exposed human islets from all three donors ( ), while two of the donors showed increased expression of both BCL2L1 and CDKN1A in cisplatin-exposed islets ( ), similar to what was observed in mouse islets ( ). Overall, the decrease in GSIS and KCl-stimulated insulin secretion and downregulation of insulin processing and β-cell identity gene expression are consistent with what was observed in mouse islets 48 h after cisplatin-exposure (Figs. C and K , I – L , and C ).
We first assessed how cisplatin impacts systemic glucose homeostasis in male and female mice exposed to vehicle or cisplatin over 2 weeks ( ). Cisplatin exposure did not affect body weight ( Supplementary Fig. 1 A and C ), fasting blood glucose levels ( Supplementary Fig. 1 B and D ), or insulin sensitivity at 1 week posttreatment ( ) in either sex. During an ipGTT at 2 weeks postexposure, male mice exhibited no differences in glucose tolerance ( ), but female cisplatin-exposed mice had a slightly lower peak glucose value than vehicle-exposed female mice ( ). Both male and female cisplatin-exposed mice had reduced plasma insulin levels compared with controls during the ipGTT ( ). The reduction of plasma insulin in cisplatin-exposed mice was reproduced in two separate cohorts using DBA/2 and immunocompromised SCID-beige male mice ( Supplementary Fig. 2 ), which are relevant mouse models to metabolic research and patients with cancer. Islets isolated 2 weeks postexposure showed no differences in basal insulin release, GSIS, stimulation index, or insulin content between treatment groups for either sex ( ). However, islets from cisplatin-exposed male mice had reduced KCl-stimulated insulin secretion and reduced oxygen consumption in response to glucose stimulation compared with controls ( ); this was not observed in islets from female mice ( ). This suggests that low plasma insulin levels in cisplatin-exposed mice may originate from intrinsic impairments within the islets but are likely also influenced by defects in peripheral tissues.
To determine whether cisplatin directly impairs islet function, islets were isolated from male and female mice and exposed to 10 μmol/L cisplatin or vehicle for 48 h in vitro prior to functional analyses as outlined in . Male and female mouse islets exposed directly to cisplatin showed profoundly impaired oxygen consumption ( ) and insulin secretion ( ) after 48 h. Specifically, cisplatin-exposed mouse islets had trending increases in basal insulin secretion at the beginning and end of the perifusion assay ( ) and had pronounced reductions in both GSIS and KCl-stimulated insulin release ( ) compared with vehicle-exposed islets. Interestingly, despite cisplatin-exposed islets having nearly abolished KCl-stimulated insulin secretion ( ), there was a delayed insulin peak post-KCl perifusion in both sexes ( ). Cisplatin appears to hyperstimulate insulin release under basal glucose conditions, reduce the sensitivity of islets to glucose, and impair the release of insulin independent of glucose metabolism. Notably, the effects of cisplatin on the in vitro metabolic function of mouse islets were not sex-specific.
Since the effects of cisplatin on β-cell function in vitro were similar between sexes, we focused on male mouse islets for additional analyses. We first determined whether our in vitro cisplatin dosing protocol affected islet cell viability to assess if the profound decreases in metabolic function and insulin secretion were attributed to cell death. Islets were treated with cisplatin or vehicle for 6, 24, or 48 h. At each timepoint, intact islets were imaged to visualize morphology and then dispersed into a single-cell suspension to measure cell viability via an image-based assay ( ). Based on a qualitative assessment, islets from both treatment groups appeared to be healthy at all timepoints ( ). Cisplatin exposure did not affect the percentage of PI + (i.e., dead/dying) islet cells at any timepoint ( ). These data indicate that exposure to 10 μmol/L cisplatin for 48 h was not overtly cytotoxic to mouse islets.
Given that we were not able to recover islets from the perifusion system to assess insulin content, we measured bulk insulin secretion and insulin content in mouse islets using a static GSIS assay after 48 h of cisplatin exposure in vitro ( ). Cisplatin-exposed islets showed significantly elevated basal insulin secretion under LG conditions ( ), a small decrease in stimulation index ( ), and no change in total insulin content ( ). The robust increase in basal insulin secretion in cisplatin-exposed islets aligned with the perifusion data ( ), but the lack of difference under HG conditions in the static assay was surprising. This may be attributed to the longer period for second-phase insulin secretion during a static GSIS, which was not as severely impacted by cisplatin as first-phase insulin secretion ( ). The similarity in insulin content between treatment groups confirms that cisplatin does not cause pronounced β-cell loss or depletion of insulin stores. The altered insulin secretion in cisplatin-exposed islets could be attributed to changes in exocytotic capacity of the β-cells. To explore this, we measured the capacitance responses of male mouse β-cells by whole-cell patch-clamp. Under LG conditions, there was a clear increase in both the average and cumulative exocytotic capacity of cisplatin-exposed β-cells compared with vehicle-exposed β-cells ( ). However, exocytotic capacity of β-cells from both treatment groups was comparable under HG conditions ( ). These results suggest that cisplatin exposure increases exocytosis of insulin granules from β-cells in LG environments but not in HG environments. Thus, additional mechanisms are driving defective GSIS in cisplatin-exposed islets.
To better understand how cisplatin disrupts GSIS ( ), we assessed mitochondrial function in vehicle- and cisplatin-exposed male mouse islets in greater detail. In this experiment, islets were pre-incubated in LG Seahorse XF RPMI medium to allow for assessment of oxygen consumption in response to HG. Cisplatin-exposed islets had significantly impaired oxygen consumption compared with vehicle-exposed islets throughout the assay and showed a significant or trending reduction in all calculated parameters ( ), consistent with our previous experiment ( ). Importantly, cisplatin-exposed islets did not robustly increase oxygen consumption when stimulated with HG ( ). These data indicate that in vitro cisplatin exposure disrupts mitochondrial function in mouse islets, which likely contributes to the impaired GSIS ( and K ).
We measured expression of key genes related to islet function and cell stress at 6, 24, and 48 h following vehicle or cisplatin exposure to better understand the temporal effects of cisplatin ( ). We first assessed expression of genes in the Bcl-2 family, key players in the intrinsic apoptosis pathway. Cisplatin-exposed islets showed a sustained, approximately fivefold downregulation of Bcl2 and a progressively increasing upregulation of Bcl2l1 , both anti-apoptotic genes, compared with vehicle-exposed islets between 6 and 48 h ( ). The pro-apoptotic gene Bax was upregulated approximately twofold in cisplatin-exposed islets at both 24 and 48 h ( ). Additionally, Cdkn1a , which encodes for p21, a marker of DNA damage and senescence ( ), was highly upregulated in cisplatin-exposed islets as early as 6 h postexposure ( ). Together, these data suggest that cisplatin induces the intrinsic apoptosis pathway, but the anti-apoptotic gene Bcl2l1 is being activated to promote a pro-survival phenotype. Ppargc1a , a transcriptional coactivator that regulates mitochondrial function ( , ), was modestly reduced in cisplatin-exposed islets at 6 h but not 24 or 48 h ( ). Cisplatin-exposed islets had an approximately twofold upregulation of Nrf2 , a marker of oxidative stress, at all three time points ( ) and increased expression of Nrf2 downstream targets Gpx1 at 24 and 48 h and Hmox1 at 6 h ( ). These changes imply that cisplatin acutely activates oxidative stress responses in mouse islets. Cisplatin did not affect expression of insulin ( Ins1 , Ins2 ) or proprotein convertase genes ( Pcsk1 , Pcsk2 ) at 6 h but reduced expression of these genes at 24 and 48 h compared with vehicle ( ). The most pronounced effect was on Pcsk2 expression, which was reduced approximately fourfold at 24 h and approximately eightfold at 48 h in cisplatin-exposed islets ( ). These results suggest that cisplatin exposure impairs proinsulin production and processing.
To further explore how cisplatin exposure alters islet gene expression, we performed a TempO-Seq analysis on male mouse islets following 48 h of treatment ( and Supplementary Fig. 3 ). In total, 1,022 probes were significantly different between cisplatin- and vehicle-exposed islets ( Supplementary Table 4 ). Pathway analysis of DEGs using both the GO biological process (GO-BP) and the KEGG pathway databases identified insulin secretion as the most significantly enriched gene set ( and Supplementary Fig. 3 A ). In fact, the 20 most significantly enriched gene sets in the GO-BP database were associated with protein/hormone transport and secretion ( ). Most of the enriched genes within each of the 20 gene sets were downregulated, with >75% of genes downregulated in 18 out of 20 gene sets ( and Supplementary Table 5 ). Since insulin secretion was the most enriched term across both databases, we next used hierarchal clustering of DEGs to more closely examine the insulin secretion pathway ( and Supplementary Fig. 3 B ). Out of 41 DEGs (represented by 46 probes), cisplatin exposure upregulated 6 genes and downregulated 35 genes ( ). Of the 35 downregulated DEGs, several were β-cell maturity and identity markers, including Ucn3 , Nkx6.1 , and Mafa ( ). Genes involved in paracrine signaling and potentiation of insulin secretion, including Gcg , Glp1r , and Abcc8 , were also downregulated ( and Supplementary Fig. 3 B ). When visualized, several aspects of the insulin secretion pathway, from initial stimuli reception to insulin granule release, were downregulated ( and Supplementary Fig. 3 C ). Overall, the transcriptomic analysis confirmed that insulin secretion is highly impacted by cisplatin exposure, supporting our previous findings.
To begin translating our results in mouse islets to human health, we performed a dynamic insulin secretion assay and measured the expression of key genes related to islet and β-cell function in human islets obtained from three female organ donors ( ). Both GSIS and KCl-stimulated insulin secretion were reduced in cisplatin-exposed human islets compared with vehicle-exposed islets ( ). Cisplatin-exposed human islets also showed a profound downregulation of insulin ( INS ) and proprotein convertase genes ( PCSK1 and PCSK2 ) after 48 h compared with vehicle-exposed islets ( ). We also observed a consistent decrease in β-cell identity genes GCK , UCN3 , and MAFA in cisplatin-exposed human islets ( ). Interestingly, BCL2 expression was downregulated in cisplatin-exposed human islets from all three donors ( ), while two of the donors showed increased expression of both BCL2L1 and CDKN1A in cisplatin-exposed islets ( ), similar to what was observed in mouse islets ( ). Overall, the decrease in GSIS and KCl-stimulated insulin secretion and downregulation of insulin processing and β-cell identity gene expression are consistent with what was observed in mouse islets 48 h after cisplatin-exposure (Figs. C and K , I – L , and C ).
The increased incidence of T2D among cancer survivors after chemotherapy ( , ) prompted us to investigate whether cisplatin treatment elicits off-target effects on pancreatic islets. Our research shows that cisplatin exposure profoundly dysregulates insulin secretion and oxygen consumption in islets from both male and female mice and significantly alters the expression of genes critical to islet function. Importantly, the adverse effects of cisplatin on insulin secretion were seen after both systemic exposure in mice and direct exposure to islets in vitro. Moreover, the acute effects of cisplatin, which disrupt insulin secretion and reduce expression of genes related to insulin processing and β-cell identity, were replicated in human islets from three female organ donors. These data suggest that cisplatin-induced damage to pancreatic islets may contribute to the long-term risks of metabolic dysregulation in cancer survivors. Mouse islets exposed to cisplatin in vitro showed increased insulin release under LG conditions but reduced insulin release following HG or KCl stimulus. As observed during a dynamic GSIS assay, both cisplatin-exposed mouse and human islets displayed a diminished first-phase insulin peak but were still capable of gradual second-phase insulin release when stimulated with HG. This likely explains why there was no difference in total insulin released by mouse islets after 1 h of HG stimulation in a static GSIS assay. Basal hyperinsulinemia and the loss of first-phase GSIS are predictive markers for T2D ( , ). Our mouse study confirmed that some of these β-cell defects translated in vivo as well. The significantly reduced plasma insulin levels in cisplatin-exposed mice during a GTT is a key marker of β-cell dysfunction ( ). It was surprising to find that islets isolated from these mice only exhibited mild defects in GSIS ex vivo, but this could be related to the recovery period provided to mice after cisplatin administration in vivo. Future studies should assess how long defects in GSIS persist after a prolonged washout period following in vitro cisplatin exposure. Additionally, there may be an interplay of both intrinsic defects in the islets and extrinsic effects caused by disruption of peripheral tissues, influencing the changes in plasma insulin observed in vivo. Cisplatin-induced toxicity has been characterized in various organs, including the kidney and brain ( ). However, the effects of cisplatin on glucoregulatory tissues have not been well characterized. Future studies should investigate liver and adipose phenotypes in parallel with β-cell function. Given that there was no change in insulin content from mouse islets isolated after in vivo cisplatin exposure and no change in total insulin content or percentage of PI + islet cells following in vitro cisplatin exposure, we speculate that cisplatin-induced impairments in insulin secretion are not driven by β-cell loss but, rather, through intrinsic defects within islet endocrine cells. For example, cisplatin exposure significantly downregulated the expression of Abcc8 , the gene encoding for sulfonylurea receptor 1 (SUR1), a key component of the K ATP channel in β-cells. Loss-of-function in this gene has been linked to malfunction of the K ATP channel, which leads to accumulation of potassium ions in the β-cell and depolarization without external nutrient stimulation ( ). This could explain the increased basal insulin secretion and exocytotic capacity of β-cells under LG conditions in cisplatin-exposed mouse islets. While no differences were noted in the exocytotic capacity of β-cells under HG conditions, the heightened exocytotic capacity of cisplatin-exposed β-cells in LG conditions could reduce the availability of insulin granules in the readily releasable pool, causing a delayed release of insulin upon stimulation while the pool replenishes itself ( ). Further investigation into the granule docking machinery of β-cells and calcium signaling is required to better understand how cisplatin alters exocytotic capacity of β-cells. Cisplatin is known to inhibit the replication/transcription of mitochondrial DNA ( ), so we assessed mitochondrial function in cisplatin-exposed islets. Islets isolated from cisplatin-exposed male mice had modest defects in glucose-stimulated oxygen consumption at 2 weeks postexposure. Both male and female mouse islets exposed to cisplatin in vitro had substantially decreased basal oxygen consumption and lacked appropriate changes in oxygen consumption rates in response to glucose or electron transport chain modulators. Because the basal oxygen consumption of cisplatin-exposed mouse islets is near maximal respiration, these islets are likely unable to accommodate changes in energetic demand as effectively as their vehicle-exposed counterparts. The reduced mitochondrial ATP production in cisplatin-exposed mouse islets aligns with the observed downregulation of Gck , which encodes for glucokinase, a critical protein in the transformation of glucose to pyruvate. Glucokinase is considered the glucose sensor of the β-cell; individuals with mutations in GCK have monogenic diabetes and, in turn, require greater nutrient stimulation to trigger insulin secretion ( ). The downregulation of Gck and GCK in cisplatin-exposed mouse and human islets, respectively, may explain the reduced insulin secretion observed upon HG stimulation in cisplatin-exposed islets. This, in combination with the inhibited oxygen consumption, likely contributes to dysregulated GSIS in cisplatin-exposed mouse islets. The rapid upregulation of Nrf2 , a master regulator of oxidative stress response pathways, and its downstream targets suggests that cisplatin exposure increases oxidative stress in mouse islets, much like in other tissues ( ). β-Cells are particularly sensitive to oxidative stress, as they express relatively low levels of antioxidant enzymes compared with other cells ( ). Moreover, the downregulation of Ppargc1a in cisplatin-exposed mouse islets may inhibit ROS detoxification ( ), further contributing to cisplatin-induced oxidative stress in islets. Additional investigation is required to understand whether ROS accumulation is involved in cisplatin-induced β-cell dysfunction and to determine whether interventions with antioxidants could protect islets from the adverse effects of cisplatin. Members of the Bcl-2 family are key regulators of β-cell fate ( ). The robust decline in Bcl2 expression, along with the stark upregulation of Bax expression, suggests that cisplatin induces the intrinsic apoptosis pathway in mouse islets. However, we did not observe increased PI + cells or reduced insulin content in cisplatin-exposed mouse islets, so we speculate that the upregulation of Bcl2l1 (and likely other anti-apoptotic factors) contributes to protecting cisplatin-exposed islets from cell death. Fiebig et al. ( ) found that Bcl-x L , the protein product of Bcl2l1 , is more effective at preventing cell death than Bcl-2 in cells treated with etoposide, another chemotherapeutic agent. Other studies have shown that while upregulation of Bcl-x L prevents cell death, it also impairs mitochondrial function, oxygen consumption, and insulin secretion ( , ); thus, the upregulation of Bcl2l1 may contribute to the dysregulation in oxygen consumption and insulin release in cisplatin-exposed mouse islets. Interestingly, decreased Bcl2 expression has also been correlated with elevated levels of basal insulin secretion ( ), which aligns with our findings. Taken together, our data suggest that cisplatin exposure may promote a prosurvival phenotype in islets, which could shift cell fate toward senescence rather than apoptosis. This notion is further supported by the stark upregulation of Cdkn1a observed in mouse islets as early as 6 h postexposure, a characteristic noted in islets from donors with T2D ( ). In human donor islets, we observed similar trends in the expression of in BCL2 , BCL2l1 , and CDKN1A following cisplatin exposure. However, the limited sample size of human donors prevented us from fully characterizing the responses to cisplatin. A much higher number of human donors is required to fully delve into the effects of cisplatin on human islets to account for the biological variability between donors. Studies using senolytics and small molecule agonists/antagonists of the Bcl-2 family will provide additional insight into the role of senescence in cisplatin-induced β-cell dysfunction. The enrichment of insulin secretion in both the GO-BP and KEGG databases, robust downregulation of genes in this pathway, and reduced expression of Ins1 , Ins2 , Pcsk1 , and Pcks2 over time highlight the detrimental effects of cisplatin on β-cell function. This is further supported by the pronounced and consistent downregulation of INS , PCSK1 , and PCSK2 in cisplatin-exposed human islets from all three donors. Interestingly, cisplatin-exposed mouse and human islets showed downregulation of Ucn3 / UCN3 and Mafa / MAFA , genes critical to β-cell identity. The loss of these genes is linked to β-cell dedifferentiation and contributes to β-cell failure in T2D ( ). Compromised β-cell identity ultimately leads to downstream effects in insulin processing and signaling ( ). Future studies should prioritize the collection of intact fixed pancreatic tissue to characterize the effects of in vivo cisplatin exposure on key β-cell identity markers at the protein level. Furthermore, downregulation of Gcg and Glp1r in cisplatin-exposed mouse islets could point to defects in paracrine signaling between endocrine cells. Deficits in paracrine signaling within the islet contribute to impaired insulin secretion in islets from donors with T2D compared with donors without diabetes ( ). Future studies exploring the effects of cisplatin on other pancreatic endocrine cells will shed light on how paracrine signaling could be contributing to cisplatin-induced defects in insulin secretion. Incidences of new-onset metabolic syndrome have been reported in a wide age range of adult cancer survivors between 18 and 50 years old ( ). In our experiments, mice were exposed to cisplatin between 10 and 18 weeks of age, which would be comparable to young adults ( ). We chose to focus on relatively young and healthy mice to assess the effect of cisplatin on islet function without the confounding influence of age. However, it will be important to elucidate the effects of cisplatin on an aged cohort of mice to better understand how age may influence susceptibility to cisplatin-induced islet dysfunction. The development of new-onset diabetes in cancer survivors is well documented ( , ), but the cause of this dysglycemia remains unclear. Our research strongly indicates that cisplatin exposure causes acute defects in islet function, highlighting an urgent need for further investigation. Our initial experiments in human donor islets suggest that our findings in mouse islets are applicable to human health. Longer-term mouse studies are needed to better understand the chronic impacts of this insulin dysregulation following chemotherapy and elucidate mechanisms by which cisplatin, as well as other chemotherapeutic drugs, affect islet function. While our initial findings in human donor islets are compelling, it is crucial to expand upon these findings with a larger sample size to account for the natural heterogeneity in islet function observed in humans. Additionally, clinical studies examining β-cell function in cancer survivors will be vital in translating our results to patient outcomes. Ultimately, this research will provide critical insight for designing targeted interventions to improve long-term metabolic health outcomes in cancer survivors and reduce their risk of developing T2D after treatment.
|
Minimally invasive interventions for intracranial pathologies using tubular retractors in the pediatric population: Safety, efficacy, technical aspects and outcomes | f119b38e-312c-4a0d-897b-515a4f832abf | 11892860 | Surgical Procedures, Operative[mh] | The use of minimally invasive techniques for the surgical treatment of adult neurosurgical conditions, such as intracranial hemorrhage (ICH) , ventricular or deep-seated tumors and vascular malformations continues to gain popularity. Case reports, as well as larger case series , demonstrate that in experienced hands, very complex and deep-seated pathologies can be safely approached, debulked or fully resected, with similar, if not lower, complication rates, coupled with shorter and better tolerated recovery. The inherent advantages of tubular retractors, that include minimizing collateral damage to normal brain and vasculature , and maintenance of a stable working corridor, occasionally larger than the one available using classic neurosurgical approaches (such as interhemispheric, supra cerebellar-infratentorial, etc.), should be counter-balanced by the increased complexity of those approaches, the required learning curve, need for dedicated instruments, and the adoption of a new conceptual approach. The pediatric developing brain is occasionally considered to be more resilient to injury than the adult brain, most likely secondary to the ability for re-modulation and gain-of-function . Nonetheless, the young brain tissue is typically more friable and therefore, more susceptible to collateral injury. Larger craniotomies can be associated with significant bony defects (with associated growth deformities), and, especially in younger patients, classical approaches pose elevated difficulty and risks associated with positioning such as the limited ability to use rigid fixation frames, smaller working corridors, strict limitations on blood loss and other unique factors . Surgical methods that can minimize procedure length and favor shorter post-surgical recovery, in terms of hospitalization interval, as well as reduced need for prolonged rehabilitation are beneficial in the pediatric population, reducing long-term developmental delays, psychological effects such as anxiety, and improving the overall experience. Minimally invasive parafascicular surgery (MIPS) for pediatric lesions has gained popularity in our institution during the last five years, after recognizing the significant benefits and excellent outcomes, in parallel to the very limited complication profile. However, the benefits and best practices, including technical aspects of using MIPS in the pediatric population with varied pathologies, have not been thoroughly characterized. Herein we report the largest pediatric series of MIPS procedures in 22 patients, between 10-month-old to 19-years-old, with a variety of pathologies, including benign conditions, vascular malformations, tumors, colloid cysts, and infections. The feasibility, safety, benefits and post-surgical outcomes associated with MIPS in a pediatric population are described; interdependencies between technical familiarity and surgical experience with MIPS, along with appropriate patient selection, are emphasized.
This single center retrospective review included all pediatric patients undergoing MIPS between 2018 and 2023 at Helen DeVos Children’s Hospital. Institutional review board (IRB) approval was obtained for the retrospective chart review, further collecting the EMR data between 30/01/2022 and 15/03/2024. The IRB waived the requirement for informed consent. Surgical pre-planning included identifying the most favorable patient position to reach the lesion via the identified surgical trajectory. Trajectories were planned using the StealthStation™ S8 neuronavigation system (Medtronic, Minneapolis, MN), with the further goal to minimize disruption to white matter tracts. All surgeries were performed by neurosurgeons experienced with tubular retraction systems. All procedures implemented a transsulcal approach using the BrainPath ® tubular retractor system (NICO Corporation, Indianapolis, IN), with obturators and sheaths measuring from 11x50 mm to 13.5x75 mm (occasionally more than one size during the same procedure), with or without the Myriad ® microdebrider system (NICO Corporation, Indianapolis, IN). BrainPath® retractor length was determined based on the distance from the calvarial surface to the lesion. For most patients, a 13.5x60 mm obturator and sheath were selected. The smaller sheath (11 mm in diameter) was used for specific pathologies, such as complex fenestrations in very young patients, evacuation of abscess or hematoma or for less complex intraventricular pathologies. The tubular sheath required intraoperative replacement from a 13.5x50 mm sheath to a 13.5x60 mm sheath in two cases, both during resection of intraventricular tumors, given the collapse of the CSF spaces and, as the resection progressed and deepened. The procedures were completed under high magnification, using standard microscopy (Zeiss™) or a 3-dimensional exoscope (Aesculap, AEOS ® ). Rigid or flexible endoscopes were used through the sheath in several cases to determine residual lesion, safety of additional resection, including for a better definition of the intraventricular anatomy. Neuromonitoring was extensively used, with similar indications to standard of care. Optical or electromagnetic (EM) neuronavigation (StealthStation™ S8) was used in all cases. For the more recent cases requiring EM neuronavigation, the BrainPath Navigation Probe Adapter ® (NICO Corporation, Indianapolis, IN) was used. Advanced imaging, including functional MRI (fMRI)-based tractography and magnetoencephalography (MEG)-defined functional areas, were implemented in preoperative planning, identifying safe corridors towards the deep located lesions, as well as intraoperatively during the resection through the tubular retractor. Pre-operative 5-Aminolevulinic acid (5-ALA, Gleolan ® ) was administered in two cases. Clinical, demographic, and radiographic data were obtained including age, sex, presenting neurologic deficits, MRI findings including lesion location and size, and previous surgical procedures. Intraoperatively, the following were captured: operative approach or trajectory, BrainPath® endoport length (mm), operative duration (minutes), use of intraoperative MRI; additionally post-operative day discharge (POD) and length of follow up (days) are reported . The extent of postsurgical resection or additional outcomes were estimated based on the immediate post-operative images (MRIs); complications, the need for additional interventions and clinical outcomes are reported. The outcomes were evaluated based on the most recent available clinical follow-up, in all cases accompanied by updated imaging. Descriptive statistics were captured and evaluated.
Between 2018 and 2023, 22 pediatric patients underwent 27 procedures. The patients’ ages ranged from 10-month-old to 19-year-old (mean 9.61 years-old, median 8 years-old). The cohort included 10 females and 12 males, suffering from a variety of lesions: 13 tumors, 5 cavernomas, 1 colloid cyst, 1 cerebellar abscess, 1 traumatic ICH and 1 resection of choroid plexus causing CSF over production. . Patient 6, A-D: A. T1 post-contrast sagittal MRI view demonstrating a solid mass in the roof of the third ventricle with obstructive hydrocephalus. B, C. T2-weighted sagittal and coronal MRI views demonstrating the mass and the associated hydrocephalus with aqueductal stenosis. D. Post-resection, T1 post-contrast sagittal MRI view, with a limited residual mass and re-establishment of CSF outflow path. Patient 10, E-L: E. Pre-operative T1 post-contrast axial MRI view demonstrating a large contrast-enhancing lesion, with solid and cystic components, compatible with an atypical meningioma (WHO II). F-H. Axial, sagittal and coronal T2-weighted images showing the fMRI findings (yellow- language areas; green – corticospinal tract; dark blue – inferior longitudinal fasciculus; white – arcuate fasciculus). I-L. T2-weighted and T1 post-contrast coronal, sagittal and axial MRI views obtained 2 years after the MIPS procedure (and shunting), with no evidence of residual tumor and well decompressed ventricles. Patient 12: M-P: M, N. Pre-operative T1 post-contrast axial and coronal MRI views showing an enhancing lesion abutting the right corticospinal tract with local mass effect on the temporal lobe and mesial structures. O, P. T1 post-contrast axial and coronal MRI views with a minimal residual enhancing rim along the corticospinal tract. Patient 8, A-D: A. Axial MRI gradient echo (GRE) view showing a left thalamic hemorrhagic lesion; B. T1 post-contrast axial MRI view depicting a left thalamic suspected cavernous malformation. C, D. Intra-operative T2-weighted and T1 post-contrast MRI images obtained after the resection of a cavernous malformation. Patient 13, E-J: E-G. T2-weighted, T1 post-contrast and diffusion-weighted axial MRI images showing a new cerebellar abscess in a patient with Streptococcus Intermedius sinusitis. H-J. Post-MIPS axial, coronal and sagittal fast T2-weighted MRI sequence showing no evidence of residual abscess with minimal signal changes along the lateral trans-cerebellar tract. Patient 14, K-N: K, L. T2-weighted axial and coronal MRI images showing post-infectious loculated hydrocephalus. M-N. T2-weighted axial and coronal MRI views obtained over two months after bilateral fenestrations and shunting showing collapsed (communicating) cystic areas with resolved mass effect on the brainstem. The smaller BrainPath sheath (11 mm in diameter) was used in several cases, including: evacuation of a cerebellar abscess (Patient 13), evacuation of an intracranial hemorrhage (ICH, Patient 17) and hydrocephalus related procedures - resection of cerebrospinal fluid (CSF) over-producing choroid plexus (Patient 15) and fenestrations in post-infectious loculated hydrocephalus (Patient 18, bilateral procedures). Pre-operative 5-ALA was administered in two cases (Patients 9 and 12). Patients 10 and 22 required switching to longer BrainPath sheaths during resection of intraventricular tumors due to collapse of CSF spaces and deepening resection. One patient with post-IVH post-infectious cystic hydrocephalus underwent 4 planned procedures for cysts fenestrations. The postoperative follow-up averages 701 days (23.36 months). The average length of the procedure was 213 minutes (skin-to-skin) in all the cohort. When calculating for patients not undergoing intra-operative MRIs, the average length of procedure was 144 minutes (2.4 hours). In 13 cases an intra-operative MRI (iMRI) was obtained; in two cases the MRIs were completed prior to closure, suspecting minimal tumor residual, which was further fully resected (Patients 16, 21). The average length of stay was 10 post-operative days for all patients (per procedure; Patient 8 underwent two procedures during the second hospitalization due to suspected minimal residual cavernoma; Patient 18 underwent two planned bilateral procedures during the same admission). The average length of stay per-procedure for the subgroup of patients with an external ventricular drain (EVD) was 15 post-operative days (median: 13 days, range: 2-44 days), those undergoing a total of 15 procedures, and 6 post-operative days (median: 3 days, range: 1-9 days) for patients without an EVD, undergoing a total of 12 procedures. When removing the severe multi-trauma patient (with an expected prolonged hospitalization), the average stay for patients without an EVD was 3 days. In 7 of 13 patients operated for the resection of tumors, a gross total resection (53.8%) was achieved. In the remaining 6 patients (46.2%), resections ranging from 40% to 80% were obtained. The extent of resection was limited in two patients with large sellar/parasellar WHO I pilocytic astrocytomas (Patients 11 and 14), in one patient with a third ventricular tanycytic astrocytoma (Patient 6, approximately 80% resection) and in one patient with a right thalamic tumor (Patient 12), in which we stopped the resection approximately 2 mm from the cortico-spinal tract based on intraoperative monitoring activations at 2 mAmp. Patient 14 underwent a subsequent additional resection due to progressive growth the astrocytoma (after completion of chemotherapy and over 18 months from the initial MIPS intervention), obtaining a similar resection of approximately 40-50% of the new tumor volume through the existing MIPS tract (“open approach”). To note, the initial MIPS procedure was much better tolerated by the patient, with a significantly shorter recovery and stay, despite the younger age at the time of the intervention. In two patients with H3K27-altered diffuse midline glioma (Patients 19 and 20) we obtained resections of at least 40% of the overall tumor burden: in the first patient over 50% of the enhancing mass, and, in the latter patient, close to 80% of the enhancing mass. The first patient treated for an H3K27-altered diffuse midline glioma was an urgent procedure for decompression and evacuation of presumed actively hemorrhagic tumor with clinical presentation of herniation; this procedure was performed in the middle of the night, without available monitoring or iMRI support, likely limiting the extent of resection (lifesaving procedure). In the second patient, the infiltrative enhancing tumor was considered to involve midline structures, with T2 hyperintense signal extending from the internal capsule to the medulla and to the cranio-cervical junction. Both patients expired from progressive disease, one 2.3 months after the procedure and the second 5.5 months post-surgery. In one case (Patient 11, middle cranial fossa pilocytic astrocytoma, WHO I), MIPS was superior to the previously performed open approach, resulting in an additional 50% resection of the overall tumor volume, compared to the only 30% achieved through a previous extended orbito-zygomatic approach. The recovery after MIPS was also significantly shorter and better tolerated by the patient, without any additional neurological deficits. One patient (Patient 9) underwent an awake procedure with intraoperative monitoring for speech and memory for the resection of a peri-forniceal cavernoma. The intraoperative monitoring allowed us to complete a gross total resection (GTR) with no cognitive or memory deficits. 5-ALA was administered preoperatively in a 4-year-old patient (Patient 12), positively illuminating a WHO 1 pilocytic astrocytoma, assisting in the resection of 70-80% of the right thalamic enhancing mass; the resection of the lesion was stopped after identifying cortico-spinal activations at 2 mA stimulation, thus preventing any adverse outcomes. Patient 8 underwent two re-operations using MIPS for additional resection of a minimally residual cavernous malformation located in the left thalamus. A contralateral right frontal approach was used during all interventions, utilizing the same tract. The iMRI completed at the end of the initial resection did not reveal any residual abnormal tissue, including no evidence of residual cavernoma, however, during the routine follow-up, several months later, we identified a suspected small rebleed that was initially managed conservatively. Approximately 10 months after the first intervention the patient re-presented with headaches and irritability, most likely secondary to additional intracavitary bleeding. A similar approach was used a second time for clot evacuation, as well as for resection of a small solid component consistent with a cavernous malformation. During the same admission, the patient underwent an additional identical procedure due to suspected minimal residual on the post-operative MRI (more conspicuous after the initial evacuation of the hemorrhagic component). The patient was discharged home on POD 3, with no evidence of residual mass or re-bleed during over two years of follow-up. In our treated population, three patients required a new post-MIPS ventriculoperitoneal shunt (VPS) (Patients 10, 14 and 19), the first due to a trapped left temporal horn, after a gross total resection of an atypical meningioma located in the atrium of the left lateral ventricle (delayed presentation of hydrocephalus), with no evidence of tumor recurrence after 26 months of follow-up. The second patient underwent placement of a VPS approximately 15 months post-op due to progressive hydrocephalus secondary to slow tumor growth during chemotherapy. Patient 19, who initially presented with intra-lesional hemorrhage and clinical herniation, subsequently diagnosed with a diffuse midline glioma H3K27-altered, developed radiological evidence of hydrocephalus with very minimal clinical recovery (palliative care, deceased). Patient 15 underwent resection of approximately 90% of the left lateral ventricle choroid plexus. Briefly, the patient presented shortly after birth with severe ventriculomegaly in the setting of Coffin-Siris syndrome. An initially performed endoscopic third ventriculostomy (ETV) failed within several weeks, requiring the placement of a ventriculoperitoneal shunt. The patient did well for several months, however, presented with severe ascites, presumed to be secondary to excessive CSF over production (over 900 mL per day, estimated). Considering the large ventricles and diminished parenchymal mantle, we opted to use a tubular retractor for an extensive resection of the choroid plexus of the left lateral ventricle, with no complications, including no evidence of shunt obstruction (contralateral catheter, not externalized the at the time of the MIPS intervention). After over 12 months of follow-up, the patient continued to do well, requiring gradually increasing shunt pressures (proGAV 2.0 Valve ® , Aesculap, USA) due to over drainage and the initial formation of subdural CSF collections (resolved, valve settings increased from 0 to 13 cmH 2 O). There were no long-term or permanent new neurological deficits attributed to the minimally invasive surgical procedures. Post-operatively, the return to baseline pre-operative neurological condition was immediate, and, subjectively, the MIPS procedures were better tolerated by the patients.
The use of minimally invasive procedures in neurosurgery, both for cranial and spine pathologies, continues to be broadly adopted. The inherent advantages of tubular retractors, associated with minimal collateral damage to healthy tissue (possibly by dispersing retraction forces) , and the reported shorter recovery times, are well recognized . However, the benefits and best practices of using MIPS in pediatric population are not well characterized. The youngest treated patient in our cohort was 10 months-old, having an excellent outcome and no complications. Eleven procedures were completed in patients 5 years of age and younger, all with excellent outcomes and no post-operative complications, demonstrating the feasibility and safety of the approach in this population. In our series we achieved a 53.8% GTR rate for tumors (7/13 patients), sub-total resections in 4 patients (30.7%) and partial resections in 2 patients (15.3%). In a large subgroup of patients, the preliminary frozen pathology results played a role in decision making regarding the intended extent of resection, as the low-grade pathologies in the pediatric population (such as pilocytic infiltrative tumors) have a good outcome even in cases when a GTR cannot be safely obtained, especially considering the possible benefits from novel targeted chemotherapy. The use of a tubular retractor can be highly beneficial when a partial resection is targeted to determine pathology (obtaining large and sufficient volumes of tissue for molecular analysis), resolving obstructive hydrocephalus or decreasing intracranial pressure. The overall GTR rate for all intracranial lesions (including tumors, cavernomas and a colloid cyst) was 66.67% (12/18 patients). In a sub-analysis, the GTR rate for exophytic intraventricular lesions (7 tumors, one cavernoma and one colloid cyst) was 88.8%. When considering cavernomas, the GTR rate was 100%. One colloid cyst was also fully resected using MIPS, after a previously failed endoscopic resection. Eichberg et al. , reported a GTR rate of 71.7% in a series of 113 transtubular resections in the adult population, with a permanent complication rate of 4.4%. The series included different pathologies approached and treated in the adult cohort: metastasis (23.9%), glioblastoma multiforme (23.0%) and cavernous hemangiomas (22.1%), with a mean lesion depth of 4.4 cm. In a literature review by Shapiro et al. , the GTR rate for tumors using the Vycor™ retractor system (Vycor Medical Inc., Boca Raton, Florida, USA), alone or in combination with other retractors, was 63%, with 3 short-term post-operative complications related to the retractor. Considering the differences between the nature and prevalence of different types of lesions encountered in the adult population compared to the pediatric ones (such as metastases in adults versus complex and large pediatric sellar/suprasellar low grade tumors), the average depth of those tumors from the cortical surface, as well as the different oncologic goals of the procedures, one can argue that our reported outcomes are at least similar if not superior to the adult experience. We did not encounter any measurable complications related to the retractors or the surgical procedure. For pediatric patients presenting with unique aspects, such as a more prominent friability of younger brain tissue and possible catastrophic long-term sequela from surgical-related infections, the significance of reduced surgical time is a key MIPS attribute: 55% of the procedures in our series were completed in less the 3 hours, with an average time of 144 minutes (2.4 hours) for procedures without an intra-operative MRI. We performed complex resections in an acute setting, shortening the surgical time, while promptly debulking large lesions, controlling hemorrhage and effectively decreasing ICPs in pediatric patients presenting with various levels of herniation. Despite a very poor outcome in one patient with acute intra-tumoral hemorrhage and clinical herniation, the tubular retractor allowed for immediate control of the hemorrhage and hematoma evacuation, followed by significant tumor debulking of over 50% of the enhancing mass. Despite the lack of neuromonitoring and performing the procedure in the middle of the night, the post-operative MRI demonstrated preserved presumed functional areas, likely due to a fixed working corridor and orientation, based on a navigated trajectory. A second patient, presenting with a similar lesion with radiological evidence of subfalcine herniation and sub-acute progressive hemiparesis, underwent a successful resection of 70-80% of a thalamic enhancing mass, benefiting from improving hemiparesis in the immediate post-operative course. Only one out of the 22 patients (Patient 8) had an unexpected outcome in terms of residual cavernoma (inconspicuous on the post-operative MRI), requiring return to the operating room, using an identical approach and instrumentation. In all other patients the intervention yielded the expected and even possibly superior outcomes. When accounting for non-trauma patients without an EVD, the average length of stay was 3 post-operative days, which is relatively short for a pediatric population. The use of tubular retractors in the pediatric population, even though similar to the implementation in adult patients, carries several unique aspects. The first one is associated with the limitations on rigid head fixation in young children. For patients younger than 3-4 years-old, we performed our earliest cases using the Stealth electromagnetic navigation stylette secured with bone wax to the obturator. This could have been associated with sub-optimal accuracy. As the company introduced the dedicated adaptor for electromagnetic pointers (BrainPath Navigation Probe Adapter ® , ), our workflow and accuracy has improved. An additional significant difference that can impact the safety and success of the procedure is the size of the craniotomy in relation to the thickness of the bone. We found that in young patients with thin bone, a smaller craniotomy was sufficient to allow satisfactory dynamic angulation of the tubular retractor. As the thickness of the bone increases with age, either a larger sized craniotomy must be completed, or a longer tube has to be utilized (with the inherent disadvantages), thus, to avoid the interference between the rim of the sheath and the cranium. Lastly, special care must be given to the more friable pediatric brain, occasionally with a thinner cortical rim due to associated conditions such as hydrocephalus. Considering the additional lower overall intracranial pressures, there is a possibly higher risk of inward pulling of a larger cortical surface if the pia has not been meticulously and sufficiently fenestrated. This can be associated with persistent post-operative extra-axial hygromas. Therefore, the opening in the pia in the younger patients should be slightly more extensive, considering the use of the smaller diameter BrainPath when technically safe for the specific pathology. A. BrainPath sheath and obturator with BrainPath Navigation Probe Adapter ® ; B. Low magnification view of a 13.5 mm tubular retractor in relation to the craniotomy size. C-E. Resection of a left thalamic cavernoma (Patient 8). F-K. Resection of a right basal ganglia pilocytic astrocytoma (Patient 12): F. Early intra-operative view of the lesion. G, H. Positive 5-ALA illumination (H), demonstrating the lesion in the right lateral quadrant (G). I-K. The size of the cortical corridor and its immediate collapse following the removal of the tubular retractor. Technical aspects of the MIPS procedures in the pediatric population The number and variety of pediatric cases completed using this approach, as well as the analysis of outcomes yielded multiple observations and technical insights. One of the benefits of using a tubular system consists of possibly reducing the collateral damage to the healthy brain tissue along the tract . This can be more significant with deeper lesions and when the approach is in close proximity to functional areas and tracts . From a practical perspective, working through a fixed and continuously protected surgical corridor minimized the risks of “deviation” from the determined safe trajectory and injury to or excessive manipulation of uninvolved structures. Previous reports , emphasized the advantage of the small craniotomies required for the completion of this procedure. While appropriate for lesions such as ICH, in our experience, when approaching large tumors, the ability to partially manipulate and changing the angular orientation of the sheath was critical, thus allowing for larger resections. Therefore, we recommend the completion of a wider diameter craniotomy, at least 1.5-2 times larger than the outer diameter of the sheath. Considering the diversity of the treated pathologies, we resected both soft tumors, using mostly the standard suction and bipolar technique, but also intraventricular meningiomas, with significant calcifications. In the latter case, as well as in others, such as a third ventricular tumor and large astrocytomas, the use of the illuminated Myriad ® debrider and/or the ultrasonic aspirator (using a micro-tip to allow better visualization), were extremely helpful . With experience, we found the longer sheaths to be superior, as they allowed maintenance of the corridor even as the brain collapses secondary to CSF drainage and tumor debulking. This requires experience and surgeon’s comfort using longer instruments throughout the entire procedure, and, even though we were able to safely replace the sheath during several interventions, this step can and should be avoided. In our institution, we assembled dedicated surgical trays, that include long and thin instruments (tumor forceps, micro scissors, micro dissectors, etc.), that can be used through the longest sheaths, while minimizing the interference to the line of sight and the amount of deep illumination from the visualization system in use. Availability of long suctions and bipolars is critical. Intraoperative navigation is an intrinsic part contributing to the success of those interventions. A more unique aspect of neuronavigation in the very young pediatric population consists in limitations using a rigid head fixation system, therefore requiring the use of electromagnetic navigation. Earlier procedures completed in the younger patients required adaptation of the Stealth™ electromagnetic (EM) stylet to navigate the BrainPath complex. The more recent availability of the dedicated adaptor for this system (BrainPath Navigation Probe Adapter ® ) improved the workflow and the overall accuracy. We also found that the longer tracer pointer designed for rigid frame-based navigation of shunt catheters (versus the shorter tracker) was more beneficial, and its use should be considered during the registration step. Preoperative trajectory planning is crucial , in parallel to defining accurate and appropriate goals for the intended procedure. Pre-operative identification of safe corridors (based on DTI and basic functional anatomy), as well as the best angle of approach to the lesion, given the limited ability to manipulate the sheath, can improve and optimize the line-of-sight coverage of the entire lesion . Expected good dissection planes versus areas of blood supply or tumor origin should be taken into consideration when planning the trajectory of the tubular retractor. Intra-operative cortical/subcortical mapping was effectively performed when indicated. In our experience, when subcortical mapping was implemented, the extent of resection of a thalamic pilocytic astrocytoma abutting and displacing the corticospinal tract was maximized, with similar expected efficacy compared to standard open approaches. Awake procedures are feasible with MIPS and, possibly, even better tolerated, considering the smaller incision, craniotomy, dural opening and shorter length of procedure. However, extensive mapping is not as feasible with this approach, unless performed cortically and only for determining safe entry areas. In addition, our experience has shown that the use of adjuncts such as 5-ALA was feasible, with proven benefit when identifying positive immunofluorescence, contributing to a safe maximal resection (Patient 12). For Patient 6, we used a flexible endoscope inserted through the tubular retractor to evaluate the intraventricular anatomy, blood supply to the tumor, relationship to major vascular structures and residual mass, providing valuable information used for critical decision making. We recognize that some of the reported procedures could have been completed endoscopically. The authors believe that, given the complexity of several of the treated lesions, a purely endoscopic approach would have yielded a less optimal outcome even in very experienced hands. Moreover, comparing the diameter a typical peel-away tunneler (for example, 19 Fr), a tubular retractor that is only 11 mm in diameter allows the use of “standard” microsurgical instruments and the safe completion of the procedure by a single neurosurgeon, without the need of an experienced assistant. Lastly, in addition to the previously discussed critical aspects of meticulous presurgical planning , the use of dedicated tools and instruments, while still implementing classical neurosurgical adjuncts (monitoring, 5-ALA, etc.), the learning curve of a new surgical approach cannot be underestimated. We recommend starting with relatively simple procedures and pathologies, such as ICH and metastasis, or superficial lesions where a transition to an open approach can be easily obtained. Evaluation of the expected intraoperative complications should be performed, constantly assessing the possibility of critical hemorrhage and the ability to control it through a small and fixed working corridor. As previously mentioned, pre-operative planning in relation to the expected origin of blood supply to the lesion is critical. In this regard, we favor the completion of preoperative neurovascular imaging if a vascular lesion is included in the differential diagnosis, possibly considering alternative surgical approaches. The availability of extra-long and high quality bipolars, as well as constant evaluation of the actual and possible (realistic) line-of-sight cannot be over-emphasized. We recommend working in defined compartments with the widest possible field of view. The use of micro brain patties as well as various hemostants have been highly efficient in our experience. Use of endoscopy for better evaluation of the lesional architecture and possibly controlling remote focal hemorrhagic areas should be considered (and readily available). The sheath length should be carefully considered: a too long tube will significantly increase the instruments manipulation difficulty, and a too short tube will not maintain the surgical corridor, allowing the surrounding tissue to collapse into the surgical field. A team-based approach, shared expertise, as well as slowly and carefully increasing the complexity of the performed procedures, played a significant role in our results. Risk assessment and management, as well as realistic surgical goals, should be constantly taken into consideration, even when limiting the extent of resection.
The number and variety of pediatric cases completed using this approach, as well as the analysis of outcomes yielded multiple observations and technical insights. One of the benefits of using a tubular system consists of possibly reducing the collateral damage to the healthy brain tissue along the tract . This can be more significant with deeper lesions and when the approach is in close proximity to functional areas and tracts . From a practical perspective, working through a fixed and continuously protected surgical corridor minimized the risks of “deviation” from the determined safe trajectory and injury to or excessive manipulation of uninvolved structures. Previous reports , emphasized the advantage of the small craniotomies required for the completion of this procedure. While appropriate for lesions such as ICH, in our experience, when approaching large tumors, the ability to partially manipulate and changing the angular orientation of the sheath was critical, thus allowing for larger resections. Therefore, we recommend the completion of a wider diameter craniotomy, at least 1.5-2 times larger than the outer diameter of the sheath. Considering the diversity of the treated pathologies, we resected both soft tumors, using mostly the standard suction and bipolar technique, but also intraventricular meningiomas, with significant calcifications. In the latter case, as well as in others, such as a third ventricular tumor and large astrocytomas, the use of the illuminated Myriad ® debrider and/or the ultrasonic aspirator (using a micro-tip to allow better visualization), were extremely helpful . With experience, we found the longer sheaths to be superior, as they allowed maintenance of the corridor even as the brain collapses secondary to CSF drainage and tumor debulking. This requires experience and surgeon’s comfort using longer instruments throughout the entire procedure, and, even though we were able to safely replace the sheath during several interventions, this step can and should be avoided. In our institution, we assembled dedicated surgical trays, that include long and thin instruments (tumor forceps, micro scissors, micro dissectors, etc.), that can be used through the longest sheaths, while minimizing the interference to the line of sight and the amount of deep illumination from the visualization system in use. Availability of long suctions and bipolars is critical. Intraoperative navigation is an intrinsic part contributing to the success of those interventions. A more unique aspect of neuronavigation in the very young pediatric population consists in limitations using a rigid head fixation system, therefore requiring the use of electromagnetic navigation. Earlier procedures completed in the younger patients required adaptation of the Stealth™ electromagnetic (EM) stylet to navigate the BrainPath complex. The more recent availability of the dedicated adaptor for this system (BrainPath Navigation Probe Adapter ® ) improved the workflow and the overall accuracy. We also found that the longer tracer pointer designed for rigid frame-based navigation of shunt catheters (versus the shorter tracker) was more beneficial, and its use should be considered during the registration step. Preoperative trajectory planning is crucial , in parallel to defining accurate and appropriate goals for the intended procedure. Pre-operative identification of safe corridors (based on DTI and basic functional anatomy), as well as the best angle of approach to the lesion, given the limited ability to manipulate the sheath, can improve and optimize the line-of-sight coverage of the entire lesion . Expected good dissection planes versus areas of blood supply or tumor origin should be taken into consideration when planning the trajectory of the tubular retractor. Intra-operative cortical/subcortical mapping was effectively performed when indicated. In our experience, when subcortical mapping was implemented, the extent of resection of a thalamic pilocytic astrocytoma abutting and displacing the corticospinal tract was maximized, with similar expected efficacy compared to standard open approaches. Awake procedures are feasible with MIPS and, possibly, even better tolerated, considering the smaller incision, craniotomy, dural opening and shorter length of procedure. However, extensive mapping is not as feasible with this approach, unless performed cortically and only for determining safe entry areas. In addition, our experience has shown that the use of adjuncts such as 5-ALA was feasible, with proven benefit when identifying positive immunofluorescence, contributing to a safe maximal resection (Patient 12). For Patient 6, we used a flexible endoscope inserted through the tubular retractor to evaluate the intraventricular anatomy, blood supply to the tumor, relationship to major vascular structures and residual mass, providing valuable information used for critical decision making. We recognize that some of the reported procedures could have been completed endoscopically. The authors believe that, given the complexity of several of the treated lesions, a purely endoscopic approach would have yielded a less optimal outcome even in very experienced hands. Moreover, comparing the diameter a typical peel-away tunneler (for example, 19 Fr), a tubular retractor that is only 11 mm in diameter allows the use of “standard” microsurgical instruments and the safe completion of the procedure by a single neurosurgeon, without the need of an experienced assistant. Lastly, in addition to the previously discussed critical aspects of meticulous presurgical planning , the use of dedicated tools and instruments, while still implementing classical neurosurgical adjuncts (monitoring, 5-ALA, etc.), the learning curve of a new surgical approach cannot be underestimated. We recommend starting with relatively simple procedures and pathologies, such as ICH and metastasis, or superficial lesions where a transition to an open approach can be easily obtained. Evaluation of the expected intraoperative complications should be performed, constantly assessing the possibility of critical hemorrhage and the ability to control it through a small and fixed working corridor. As previously mentioned, pre-operative planning in relation to the expected origin of blood supply to the lesion is critical. In this regard, we favor the completion of preoperative neurovascular imaging if a vascular lesion is included in the differential diagnosis, possibly considering alternative surgical approaches. The availability of extra-long and high quality bipolars, as well as constant evaluation of the actual and possible (realistic) line-of-sight cannot be over-emphasized. We recommend working in defined compartments with the widest possible field of view. The use of micro brain patties as well as various hemostants have been highly efficient in our experience. Use of endoscopy for better evaluation of the lesional architecture and possibly controlling remote focal hemorrhagic areas should be considered (and readily available). The sheath length should be carefully considered: a too long tube will significantly increase the instruments manipulation difficulty, and a too short tube will not maintain the surgical corridor, allowing the surrounding tissue to collapse into the surgical field. A team-based approach, shared expertise, as well as slowly and carefully increasing the complexity of the performed procedures, played a significant role in our results. Risk assessment and management, as well as realistic surgical goals, should be constantly taken into consideration, even when limiting the extent of resection.
Our reported experience has several limitations. First, we were unable to complete an accurate comparison to similarly heterogeneous pathologies and patients treated using the “standard of care” approaches. Thus, comparing complication rates, length of stay, overall recovery time, rate of re-operation or the need for additional treatments and interventions (such as shunting, chemotherapy, etc.) to traditional procedures was not feasible. Although our cohort is the largest report consisting of pediatric patients who had undergone minimally invasive interventions using this tubular retractor system, comparison of outcomes based on a sub-classification of different pathologies was not attainable due to the small sample size. Additionally, quantifying the extent of injury to healthy tissue and comparing it to the traditional approaches is extremely difficult. Even though we did not encounter complications related to the tubular retractor and surgical technique, temporary or permanent deficits are more difficult to assess in pediatric patients. The length of stay was also significantly affected by the complexity of some of the lesions and the different pediatric management routines, such as the presence of an EVD with a slow weaning process, or interdisciplinary management of multi-trauma patients.
Our initial results and experience in the treatment of a variety of pediatric pathologies using minimally invasive techniques are very promising. As with every aspect of neurosurgical practice, appropriate patient selection and individualized approaches, meticulous presurgical planning, intraoperative flexibility, in lieu of surgeon’s experience, foster the best outcomes and minimize complications. Developing technical familiarity and surgical experience, in addition to the use of appropriate instruments and visualization systems and, as well as the implementation of adjunctive tools, are critical for the success of minimally invasive interventions and contribute to superior outcomes. Additional data on safety and long-term outcomes is required to better define the role of MIPS for pediatric intracranial pathologies.
|
751d73fa-68f9-4c78-8df4-90f5a0f7cb2d | 8596806 | Anatomy[mh] | The tropomyosin receptor kinase (Trk) family consists of three transmembrane neurotrophin receptors, i.e. TrkA, TrkB, and TrkC, which are encoded by the neurotrophic tyrosine kinase receptor genes NTRK1 , NTRK2 , and NTRK3 , respectively. Oncogenic gene fusions involving these genes lead to constitutive activation of Trk receptors and are targetable with small‐molecule inhibitors. Larotrectinib showed significant and durable antitumour activity in patients with NTRK fusion‐positive cancer, regardless of age or tumour type. , , This has led to specific interest in NTRK testing, especially since clinical trials have shifted away from site‐of‐origin and histology‐dependent designs towards basket trials, in which targeted therapy is evaluated in different diseases that share molecular alterations. NTRK fusions have been found at high frequency and to be characteristic for several rare cancer types, including sarcomas (i.e. infantile fibrosarcoma, secretory breast carcinoma, and congenital mesoblastic nephroma). Moreover, there is an emerging group of mesenchymal tumours defined by NTRK fusions that show a wide morphological spectrum, a variable risk of malignancy, and a non‐specific immunoprofile. This also includes CD34‐positive fibrosarcoma of bone, in which NTRK3 fusions were recently described in two cases. In this specific category, NTRK fusions are diagnostic and NTRK fusion detection should be performed. In addition, there is an increased demand for NTRK fusion testing as a predictive biomarker for potential treatment with Trk inhibitors, irrespective of the tumour type. More common cancers have a low but significant frequency of NTRK fusions, and thus represent a sizeable at‐risk patient population that is worth testing for NTRK fusions. For sarcoma patients with locally advanced and unresectable or metastatic disease, the World Sarcoma Network (WSN) advises NTRK fusion testing by the use of pan‐Trk immunohistochemistry prescreening only for those sarcoma types known to harbour complex genomes (e.g. osteosarcoma). In sarcomas with recurrent gene fusions (e.g. Ewing sarcoma) or amplifications as driver alterations, NTRK fusion testing should be restricted to research, because NTRK fusions and other drivers are typically mutually exclusive. As the proposed screening system is mainly based on the current knowledge of NTRK fusions in soft tissue sarcoma, , we aimed to explore the frequency of NTRK fusions in a large series of different bone tumours. According to WSN recommendations, we used immunohistochemistry as a first screening method, followed by molecular analysis with anchored multiplex polymerase chain reaction (AMP)‐based targeted next‐generation sequencing (NGS) for fusions in selected cases.
CASE SELECTION Tissue microarrays (TMAs) of previously published cohorts were used to screen for NTRK fusions, and included conventional chondrosarcoma ( n = 137), dedifferentiated chondrosarcoma ( n = 36), clear cell chondrosarcoma ( n = 20), mesenchymal chondrosarcoma ( n = 19), osteochondroma ( n = 9), enchondroma ( n = 11), osteosarcoma ( n = 123), angiosarcoma ( n = 26), Ewing sarcoma ( n = 20), giant‐cell tumour of bone ( n = 74), and aneurysmal bone cyst ( n = 6). , , , , , Most TMAs contained at least three 1.5‐mm‐diameter cores of each sample to compensate for intratumoral heterogeneity. Samples were handled according to the ethical guidelines described in the ‘Code for Proper Secondary Use of Human Tissue in the Netherlands’ in a coded (pseudonymised) manner, as approved by the Leiden University Medical Centre ethical board (B17.020, B17.036, and B20.064). IMMUNOHISTOCHEMISTRY Immunohistochemistry was performed as described previously. , For titration of the antibody, several dilutions were used on both neural tissue and a molecularly proven NTRK‐ fusion positive tumour of the parotid gland. In our study, a dilution of 1:200 showed the best signal‐to‐noise ratio. All slides were manually stained in one session. Microwave antigen retrieval in Tris‐EDTA (pH 9.0) was performed with deparaffinised sections preincubated with phosphate‐buffered saline (PBS)/1% bovine serum albumin (BSA)/5% non‐fat dry milk, and this was followed by overnight incubation with the pan‐Trk antibody (Abcam, Cambridge, UK; clone ERP17341, rabbit, 1:200) in PBS/1% BSA/5% non‐fat dry milk. Detection with the PowerVision Poly‐HRP Detection System (ImmunoLogic, Duiven, The Netherlands) and visualisation with a DAB+ substrate chromogen system (Dako, Glostrup, Denmark) were then performed. Finally, slides were counterstained with haematoxylin, dehydrated, and mounted. For NTRK expression, a previously published semiquantitative scoring system was used. Immunoreactivity was scored according to the location (cytoplasmic or nuclear), the intensity (1, weak; 2, moderate; or 3, strong), and the percentage of positive cells (1+, 1–25%; 2+, 25–50%; 3+, 50–75%; and 4+, >75%). Positivity of any intensity in ≥1% of cells was considered to be a positive result. All slides were scored by two independent observers (S.W.L. and J.V.G.M.B). FUSION ANALYSIS For selected cases, additional molecular analysis for NTRK fusions was performed with AMP‐based targeted NGS. RNA was isolated from frozen sections with TRizol reagent (Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. The cDNA library was prepared with the Archer FusionPlex comprehensive thyroid and lung panel (Archer, Boulder, CO, USA), which included primers for NTRK1 (exons 1–14 and 16), NTRK2 (exons 4–17), and NTRK3 (exons 1–12 and 14–17), and this was followed by sequencing with the Ion S5 system (Thermo Fisher Scientific, Waltham, Massachusetts, USA) Archer analysis software (version 6.2.3) was used to analyse the produced libraries for the presence of NTRK fusions.
Tissue microarrays (TMAs) of previously published cohorts were used to screen for NTRK fusions, and included conventional chondrosarcoma ( n = 137), dedifferentiated chondrosarcoma ( n = 36), clear cell chondrosarcoma ( n = 20), mesenchymal chondrosarcoma ( n = 19), osteochondroma ( n = 9), enchondroma ( n = 11), osteosarcoma ( n = 123), angiosarcoma ( n = 26), Ewing sarcoma ( n = 20), giant‐cell tumour of bone ( n = 74), and aneurysmal bone cyst ( n = 6). , , , , , Most TMAs contained at least three 1.5‐mm‐diameter cores of each sample to compensate for intratumoral heterogeneity. Samples were handled according to the ethical guidelines described in the ‘Code for Proper Secondary Use of Human Tissue in the Netherlands’ in a coded (pseudonymised) manner, as approved by the Leiden University Medical Centre ethical board (B17.020, B17.036, and B20.064).
Immunohistochemistry was performed as described previously. , For titration of the antibody, several dilutions were used on both neural tissue and a molecularly proven NTRK‐ fusion positive tumour of the parotid gland. In our study, a dilution of 1:200 showed the best signal‐to‐noise ratio. All slides were manually stained in one session. Microwave antigen retrieval in Tris‐EDTA (pH 9.0) was performed with deparaffinised sections preincubated with phosphate‐buffered saline (PBS)/1% bovine serum albumin (BSA)/5% non‐fat dry milk, and this was followed by overnight incubation with the pan‐Trk antibody (Abcam, Cambridge, UK; clone ERP17341, rabbit, 1:200) in PBS/1% BSA/5% non‐fat dry milk. Detection with the PowerVision Poly‐HRP Detection System (ImmunoLogic, Duiven, The Netherlands) and visualisation with a DAB+ substrate chromogen system (Dako, Glostrup, Denmark) were then performed. Finally, slides were counterstained with haematoxylin, dehydrated, and mounted. For NTRK expression, a previously published semiquantitative scoring system was used. Immunoreactivity was scored according to the location (cytoplasmic or nuclear), the intensity (1, weak; 2, moderate; or 3, strong), and the percentage of positive cells (1+, 1–25%; 2+, 25–50%; 3+, 50–75%; and 4+, >75%). Positivity of any intensity in ≥1% of cells was considered to be a positive result. All slides were scored by two independent observers (S.W.L. and J.V.G.M.B).
For selected cases, additional molecular analysis for NTRK fusions was performed with AMP‐based targeted NGS. RNA was isolated from frozen sections with TRizol reagent (Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. The cDNA library was prepared with the Archer FusionPlex comprehensive thyroid and lung panel (Archer, Boulder, CO, USA), which included primers for NTRK1 (exons 1–14 and 16), NTRK2 (exons 4–17), and NTRK3 (exons 1–12 and 14–17), and this was followed by sequencing with the Ion S5 system (Thermo Fisher Scientific, Waltham, Massachusetts, USA) Archer analysis software (version 6.2.3) was used to analyse the produced libraries for the presence of NTRK fusions.
PAN‐Trk IMMUNOHISTOCHEMISTRY Immunohistochemistry was successful in 354 cases. In the remaining cases, TMA cores were lost during processing. Nineteen cases (5%) showed staining of any intensity in ≥1% of the cells; these included Ewing sarcoma ( n = 6, 33%), osteosarcoma ( n = 11, 13%), and giant‐cell tumour of bone ( n = 2, 3%). In all except one case, cytoplasmic staining was observed. Most of the positive cases showed weak staining ( n = 13), five showed moderate staining, and one showed strong staining (Figure ). Staining in only 1–25% of cells was observed in 12 cases, staining in 25–50% of cells was observed in four cases, staining in 50–75% of cells was observed in two cases, and staining in >75% of cells was observed in one case. The remaining 335 cases were negative (Table ). MOLECULAR ANALYSIS FOR NTRK FUSION Molecular analysis was performed in cases with weak staining in >25% of cells and in all cases with moderate or strong staining; this was successful in six cases, which comprised two Ewing sarcomas, three osteosarcomas, and one giant‐cell tumour of bone (Table ). In three cases, suitable material for molecular analysis was absent. All quality criteria were met, the coverage of NTRK1 – NTRK3 was sufficient, and none of the cases showed an NTRK fusion. The relative RNA expression level of NTRK1 – NTRK3 was low. As NTRK fusions were absent in cases with moderate and strong staining, cases with weak staining in <25% of cells were not further analysed.
Immunohistochemistry was successful in 354 cases. In the remaining cases, TMA cores were lost during processing. Nineteen cases (5%) showed staining of any intensity in ≥1% of the cells; these included Ewing sarcoma ( n = 6, 33%), osteosarcoma ( n = 11, 13%), and giant‐cell tumour of bone ( n = 2, 3%). In all except one case, cytoplasmic staining was observed. Most of the positive cases showed weak staining ( n = 13), five showed moderate staining, and one showed strong staining (Figure ). Staining in only 1–25% of cells was observed in 12 cases, staining in 25–50% of cells was observed in four cases, staining in 50–75% of cells was observed in two cases, and staining in >75% of cells was observed in one case. The remaining 335 cases were negative (Table ).
NTRK FUSION Molecular analysis was performed in cases with weak staining in >25% of cells and in all cases with moderate or strong staining; this was successful in six cases, which comprised two Ewing sarcomas, three osteosarcomas, and one giant‐cell tumour of bone (Table ). In three cases, suitable material for molecular analysis was absent. All quality criteria were met, the coverage of NTRK1 – NTRK3 was sufficient, and none of the cases showed an NTRK fusion. The relative RNA expression level of NTRK1 – NTRK3 was low. As NTRK fusions were absent in cases with moderate and strong staining, cases with weak staining in <25% of cells were not further analysed.
This study provides a comprehensive immunohistochemical evaluation of pan‐Trk expression as a surrogate marker for NTRK fusions in a large series of bone tumours, including osteogenic tumours, chondrogenic tumours, and Ewing sarcoma, which are the three most common bone sarcomas. Following WSN recommendations, we used pan‐Trk immunohistochemistry as a screening method for NTRK fusions to explore the frequency of NTRK as a targetable therapeutic option in well‐characterised bone tumours, and showed that NTRK fusions are almost non‐existent. NTRK fusions were not identified in 354 bone tumours after prescreening with immunohistochemistry, which is in line with the low frequency in the literature, which contains only a handful of anecdotal cases. Besides one NTRK ‐fusion positive bone sarcoma that was found among a diverse set of paediatric malignancies (1.1%), the subtype of which was not further specified, two other NTRK ‐fusion positive bone sarcomas were described. These osteosarcoma patients and dedifferentiated chondrosarcoma patients were enrolled in a clinical trial and received larotrectinib. Interestingly, in another study of 113 osteosarcoma patients whose tumours were sequenced, three cases had an NTRK fusion, the chimaeric transcript of which appeared to be non‐functional and probably represented randomly occurring passenger alterations. Several caveats should be considered when pan‐Trk immunohistochemistry is used as a first screening method for NTRK fusions, including variable staining patterns and intensities. Although the antibody appears to have 100% specificity in carcinomas of the colon, lung, and thyroid, the specificity in sarcomas is much lower. False‐positive staining is especially frequent in tumours with smooth muscle and neural differentiation. , In our study, positivity was observed in 5% of all cases, mostly in osteosarcoma and Ewing sarcoma, whereas NTRK fusions were absent in all sequenced tumours. Although positivity for pan‐Trk in osteosarcoma has not been studied by others, false positivity in Ewing sarcoma has been previously described: pan‐Trk expression was often present in tumours in the small blue round cell category, including desmoplastic small round cell tumours (100%), Ewing sarcoma (20–33%), and sarcomas with BCOR genetic abnormalities (60–100%). , For the last of these categories of tumour, it was shown that pan‐Trk expression was caused by NTRK3 up‐regulation. Our cohort included a large proportion of sarcomas with complex genomes (osteosarcoma, high‐grade chondrosarcoma, and dedifferentiated chondrosarcoma), for which the WSN recommends NTRK fusion testing with immunohistochemistry prescreening in patients with advanced disease. Our results indicate that the subgroup of sarcoma patients who may become eligible for NTRK inhibition is extremely small or even non‐existent. However, it should be noted that not all bone tumour types were assessed for pan‐Trk immunohistochemistry, so the frequency of NTRK fusions in these tumours remains unknown. Also, because the reported sensitivity of pan‐Trk immunohistochemistry in sarcoma is 80%, the possibility of false negativity in our series cannot be completely ruled out, as molecular data on NTRK fusions in our cohort are not available. The false‐negative rate may be even higher in tumours with NTRK3 fusions. Another limitation of this study is that rare oncogenic activating splice variants of NTRK1 , which have been described in neuroblastoma and acute myeloid leukaemia could potentially be missed, because the variant‐calling pipeline used for NTRK fusion analysis is not able to pick these up. Finally, the effect of decalcification on pan‐Trk expression was not studied, so false‐negative results due to decalcification cannot be ruled out completely. However, TMAs were shown to generate positive staining in previous studies, , , , , , and cases that were scored as pan‐Trk‐positive were also decalcified. To conclude, the likelihood of finding NTRK fusions in bone tumours in clinical practice, even in tumours with complex genomes lacking driver alterations, such as osteosarcoma, is extremely low. This implies that, if more comprehensive large‐scale molecular studies confirm this, routine predictive NTRK testing in bone tumour patients with advanced disease may be reconsidered.
The authors state that they have no conflicts of interest.
Support was provided by Leiden University Medical Centre.
The study was designed, written and reviewed by S. W. Lam and J. V. M. G. Bovée. All authors contributed to data collection, data analysis, and interpretation. The manuscript was approved by all authors.
|
|
Refuting misconceptions in medical physiology | 7d3671a8-9231-4ccc-9e1b-ae877fd61353 | 7409498 | Physiology[mh] | Physiology plays a central role in understanding human body functions . It is therefore problematic that many medical students find it difficult to acquire accurate knowledge of physiological concepts . This may be partially due to the presence of misconceptions . Misconceptions can be defined as incorrect ideas that do not match current scientific views . Over the last 30 years, misconceptions in science education have been demonstrated repeatedly . Misconceptions are resistant to change as they frequently persist even after direct instruction . In medical education this topic is less well studied and few didactic strategies have been put forward to address misconceptions and promote conceptual change among students. The process of shifting from an incorrect scientific understanding to a correct one is strongly influenced by what a learner already knows . A learner’s prior knowledge should therefore always be engaged while trying to understand new information. Sometimes, the to-be-learned information conflicts with one’s prior knowledge. It then requires reorganization of cognitive schemas in the brain to accommodate novel information . When such cognitive conflicts concern conceptual knowledge, this reorganization process is referred to as conceptual change . By contaminating the learner’s prior knowledge, misconceptions may inhibit rather than facilitate learning of new information . Teachers may not succeed to alleviate misconceptions by simply providing the right answer or explanation to a question . Educational researchers have proposed that teaching strategies should explicitly undermine, i.e. refute, learners’ misconceptions , for example by using refutation texts. Refutation texts typically comprise three components: (1) the common misconception, (2) the refutation which explicitly debunks the misconception, (3) the correct answer . A refutation text for the misconception that blood slows down at a vessel narrowing states: Many people think that the velocity of blood decreases when it enters a constricted section of a vessel, just like cars slow down when the road narrows, ( i.e. misconception) but this notion is false since blood, being a liquid, cannot be compressed ( i.e. refutation). The velocity of blood actually increases because the same blood volume has to pass through a smaller cross-section in the same time frame ( i.e. correct answer). The potential of refutation texts to induce conceptual change has been demonstrated in various science domains including physics and biology . Superiority of this instructional approach is presumably based on the mechanism of coactivation as described in the Knowledge Revision Components framework . Coactivation of the misconception and correct concept appears crucial for establishing learners’ awareness of the existing conflict. Knowing that there is an apparent inconsistency between prior knowledge and new information may lead to an experience of cognitive conflict, followed by attempts to resolve this issue . In the case of refutation texts, the cognitive conflict may be induced by reading the common misconception plus a refutation that debunks this misconception. This conflict may lead to the reorganization of cognitive schemas in the brain . Such reorganizations as induced by refutation texts reflect the process of conceptual change, ideally resulting in accurate conceptual understanding. The cognitive effect of refutation texts on learners’ conceptual understanding depends on a learner’s metacognition. Metacognition is the process of thinking about one’s thinking , and encompasses an important component referred to as metacognitive knowledge . Metacognitive knowledge entails being aware of what you do and do not know, e.g. when reading refutation texts this could entail becoming aware of a cognitive conflict . Only students who can accurately judge, i.e. metacognitively evaluate that their understanding of a concept is insufficient may choose to further study this concept . Building on the importance of metacognition in concept learning, refutation texts may stimulate students’ conceptual change through enhancing their metacognitive knowledge . Various studies in elementary school and higher education support this hypothesis , however, others failed to demonstrate enhancement of students’ metacognitive knowledge after reading refutation texts . So, despite the theoretical link between conceptual change and metacognitive knowledge, research investigating the influence of refutation texts on learners’ metacognitive knowledge remains limited and so far inconclusive. In addition to investigating the influence of refutation texts on students’ conceptual understanding and metacognitive knowledge, it may be of interest to examine which learner characteristics are most productive for facilitating conceptual change. Some studies distinguish between misconceptions held by learners with low-confidence versus misconceptions held with high-confidence . Of note, in some studies the term misconception is reserved exclusively for the latter type, whereas ‘wrong answers’ held with low confidence are referred to as lack of knowledge. Researchers have suggested that high-confidence misconceptions are hardest to correct because they are more strongly represented in memory and they impair the student in accommodation contrasting information . The hypercorrection effect, however, contradicts this hypothesis, stating that corrective feedback induces coactivation which may particularly surprise learners who are highly confident about their misconceptions, thereby increasing their attention and enhancing text comprehension . In this study, we investigated the cognitive effect, metacognitive effect and hypercorrection effect in a refutation text intervention. Firstly, we investigated if reading refutation texts improves actual knowledge (i.e. cognition). Secondly, we studied if reading refutation texts improves self-perceived knowledge (i.e. metacognition). Thirdly, we tested if the hypercorrection effect occurred: we hypothesised that high-confidence incorrect answers would be corrected more frequently than low-confidence incorrect answers. In summary, we presume that research on conceptual change interventions in medical education is needed to improve students’ conceptual understanding. Moreover, equipping students with accurate knowledge about physiological concepts may ultimately result in improved clinical reasoning and decision-making .
Participants and setting This study was conducted in first-year medical students at the Leiden University Medical Center. At the start of the academic year the student cohort was divided into 24 groups, 12 groups (12–15 students/group, total 161 students) were included in this study. The protocol was implemented in a seminar on cardiovascular physiology. This seminar was part of a compulsory, 8-weeks course on integrative cardiovascular, respiratory and kidney physiology at the beginning of the second semester. The course seminars focus on solving clinically-based scenarios in small-group sessions led by an expert. This study was performed during the first course seminar, which focusses on the concepts of flow, pressure and resistance. These concepts were introduced and explained in a plenary lecture a few days before the seminar and the students were instructed to study them prior to the seminar also using a specified section from a medical physiology textbook . This study protocol was approved by the Leiden University Medical Center Educational Research Review Board (ERRB), reference number: OEC/ERRB/20171010/2. Students provided written informed consent to use their responses for scientific analysis and publication. They received no additional credit and they were informed that all data would be anonymised and test performance had no effect on their course grade. They could withdraw their permission at any time. Procedure In our study, half of the groups were assigned to the Refutation text intervention, and the other half received a Standard text control intervention. Allocation to these two experimental groups was arbitrarily except for the aim to have a similar male-to-female ratio (30:70) in all groups. The study was performed in a classroom setting at the beginning of a seminar. All students performed a pre-test, followed by either a Refutation text or Standard text intervention, and a subsequent post-test with near-transfer questions. Both tests were given on paper and consisted of four multiple-choice multi-tier questions (6). All questions were about cardiovascular physiology topics regarding flow, pressure and resistance. In between the pre-test and post-test, each student received either refutation texts or standard texts (see Additional file : Appendix A for examples). The standard texts gave, for each question of the pre-test, the right answer plus an explanation (average 177 words/text). Students had four minutes to answer each question on the pre-test and post-test, and also four minutes for reading each refutation or standard text. The study was teacher-paced, meaning that students had to wait for the next question or text if they finished earlier, resulting in a total time of 48 min. The refutation texts groups followed a similar procedure, except that the texts contained an additional sentence (i.e. refutation element) that presented a common misconception with an explicit refutation of that misconception, before providing the correct answer with the explanation (average 226 words/text). During the tests, students received a summary sheet with all the relevant factual knowledge to reduce the number of incorrect answers merely due to lack of factual knowledge. Materials The questions and explanations were designed by a physiology teacher (P.S.) with longstanding experience in cardiovascular research and teaching, and designing and reviewing exam questions. Each question consisted of three tiers, i.e. an answer tier, an explanation tier, and a confidence tier (see Additional file : Appendix B for examples). In the answer tier, students were asked to provide a binary Yes/No or an ‘Increase/Decrease/No change’ answer. In the explanation tier students could choose one of the suggested explanations that best supported their reasoning underlying their answer. Except for the right explanation, all other explanations were designed to reflect possible misconceptions that students may hold. After the explanation tier, students had to answer the confidence tier: ‘How sure are you that your answer to the previous question was correct?’. Confidence was self-reported using a 5-pt Likert scale: 1: Very unsure (complete guess), 2: Fairly unsure, 3: In doubt, 4: Fairly sure, 5: Very sure (almost 100%). All questions were designed on the ‘apply’ and ‘analyse’ levels of Bloom’s taxonomy, and focused on examining students’ conceptual knowledge . Data analyses IBM SPSS Statistics Version 23.0 (IBM Corp., Armonk, New York, USA) and GraphPad Prism Version 7.02 (GraphPad Software, La Jolla, California, USA) were used for all data analyses and visualizations. Descriptive statistics are provided as means and standard errors of the mean, unless otherwise mentioned. Only answers that consisted of a correct initial answer and a correct explanation were marked correct. Dependent samples t-tests were performed, for Refutation text and Standard text groups separately, to determine whether there was a difference in pre-test versus post-test scores. An analysis of covariance (ANCOVA) was used to determine whether the post-test means, adjusted for pre-test scores, differed between groups. To determine the effects of response accuracy (incorrect or correct answer, i.e. cognitive effect), stage (pre- or post-intervention) and group (standard or refutation text), and their interactions with confidence (i.e. metacognitive effect), we used a multiple linear regression (MLR) model with dummy variables. We used effects coding to avoid multicollinearity. Consequently the coding for the dummy variables for response (R), stage (S) and group (G) was as follows: incorrect answer R = -1, correct answer R = + 1, pre-test S = -1, post-test S = + 1, standard text G = -1, refutation text G = + 1. The MLR model was: Y = B 0 + B R .R + B S .S + B G .G + B RS .R.S + B RG .R.G + B RSG .R.S.G. This model was applied to the individually corrected confidence scores (Y): a student’s average confidence score was subtracted from their confidence scores on each question to remove the between-students variability in average confidence scores. To test the hypercorrection hypothesis, we determined the fraction of initial misconceptions that were changed to a correct answer after intervention and the fraction of initial lack of knowledge that was changed to a correct answer. A hypercorrection effect is found if the fraction corrected misconceptions is higher than the fraction corrected lack of knowledge. Therefore, outcomes were made dichotomous: a confidence score below or equal to 3 was defined as low and a confidence score above 3 as high. This cut-off was chosen because students selecting “3” were still essentially unsure (‘in doubt’) about being correct. We used Hasan’s decision matrix to label the answers (see Fig. ). According to this matrix incorrect answers given with high confidence are considered misconceptions, incorrect answers given with low confidence are considered a lack of knowledge . Correct answers held with low confidence were labelled lucky guesses and correct answers with high confidence were labelled correct knowledge. In these terms, misconceptions and lucky guesses are considered low metacognition and correct knowledge and lack of knowledge high metacognition. Furthermore, a cognitive effect was labelled positive when one changed an incorrect answer to a correct answer. A metacognitive effect was labelled positive when one changed from low metacognition to high metacognition.
This study was conducted in first-year medical students at the Leiden University Medical Center. At the start of the academic year the student cohort was divided into 24 groups, 12 groups (12–15 students/group, total 161 students) were included in this study. The protocol was implemented in a seminar on cardiovascular physiology. This seminar was part of a compulsory, 8-weeks course on integrative cardiovascular, respiratory and kidney physiology at the beginning of the second semester. The course seminars focus on solving clinically-based scenarios in small-group sessions led by an expert. This study was performed during the first course seminar, which focusses on the concepts of flow, pressure and resistance. These concepts were introduced and explained in a plenary lecture a few days before the seminar and the students were instructed to study them prior to the seminar also using a specified section from a medical physiology textbook . This study protocol was approved by the Leiden University Medical Center Educational Research Review Board (ERRB), reference number: OEC/ERRB/20171010/2. Students provided written informed consent to use their responses for scientific analysis and publication. They received no additional credit and they were informed that all data would be anonymised and test performance had no effect on their course grade. They could withdraw their permission at any time.
In our study, half of the groups were assigned to the Refutation text intervention, and the other half received a Standard text control intervention. Allocation to these two experimental groups was arbitrarily except for the aim to have a similar male-to-female ratio (30:70) in all groups. The study was performed in a classroom setting at the beginning of a seminar. All students performed a pre-test, followed by either a Refutation text or Standard text intervention, and a subsequent post-test with near-transfer questions. Both tests were given on paper and consisted of four multiple-choice multi-tier questions (6). All questions were about cardiovascular physiology topics regarding flow, pressure and resistance. In between the pre-test and post-test, each student received either refutation texts or standard texts (see Additional file : Appendix A for examples). The standard texts gave, for each question of the pre-test, the right answer plus an explanation (average 177 words/text). Students had four minutes to answer each question on the pre-test and post-test, and also four minutes for reading each refutation or standard text. The study was teacher-paced, meaning that students had to wait for the next question or text if they finished earlier, resulting in a total time of 48 min. The refutation texts groups followed a similar procedure, except that the texts contained an additional sentence (i.e. refutation element) that presented a common misconception with an explicit refutation of that misconception, before providing the correct answer with the explanation (average 226 words/text). During the tests, students received a summary sheet with all the relevant factual knowledge to reduce the number of incorrect answers merely due to lack of factual knowledge.
The questions and explanations were designed by a physiology teacher (P.S.) with longstanding experience in cardiovascular research and teaching, and designing and reviewing exam questions. Each question consisted of three tiers, i.e. an answer tier, an explanation tier, and a confidence tier (see Additional file : Appendix B for examples). In the answer tier, students were asked to provide a binary Yes/No or an ‘Increase/Decrease/No change’ answer. In the explanation tier students could choose one of the suggested explanations that best supported their reasoning underlying their answer. Except for the right explanation, all other explanations were designed to reflect possible misconceptions that students may hold. After the explanation tier, students had to answer the confidence tier: ‘How sure are you that your answer to the previous question was correct?’. Confidence was self-reported using a 5-pt Likert scale: 1: Very unsure (complete guess), 2: Fairly unsure, 3: In doubt, 4: Fairly sure, 5: Very sure (almost 100%). All questions were designed on the ‘apply’ and ‘analyse’ levels of Bloom’s taxonomy, and focused on examining students’ conceptual knowledge .
IBM SPSS Statistics Version 23.0 (IBM Corp., Armonk, New York, USA) and GraphPad Prism Version 7.02 (GraphPad Software, La Jolla, California, USA) were used for all data analyses and visualizations. Descriptive statistics are provided as means and standard errors of the mean, unless otherwise mentioned. Only answers that consisted of a correct initial answer and a correct explanation were marked correct. Dependent samples t-tests were performed, for Refutation text and Standard text groups separately, to determine whether there was a difference in pre-test versus post-test scores. An analysis of covariance (ANCOVA) was used to determine whether the post-test means, adjusted for pre-test scores, differed between groups. To determine the effects of response accuracy (incorrect or correct answer, i.e. cognitive effect), stage (pre- or post-intervention) and group (standard or refutation text), and their interactions with confidence (i.e. metacognitive effect), we used a multiple linear regression (MLR) model with dummy variables. We used effects coding to avoid multicollinearity. Consequently the coding for the dummy variables for response (R), stage (S) and group (G) was as follows: incorrect answer R = -1, correct answer R = + 1, pre-test S = -1, post-test S = + 1, standard text G = -1, refutation text G = + 1. The MLR model was: Y = B 0 + B R .R + B S .S + B G .G + B RS .R.S + B RG .R.G + B RSG .R.S.G. This model was applied to the individually corrected confidence scores (Y): a student’s average confidence score was subtracted from their confidence scores on each question to remove the between-students variability in average confidence scores. To test the hypercorrection hypothesis, we determined the fraction of initial misconceptions that were changed to a correct answer after intervention and the fraction of initial lack of knowledge that was changed to a correct answer. A hypercorrection effect is found if the fraction corrected misconceptions is higher than the fraction corrected lack of knowledge. Therefore, outcomes were made dichotomous: a confidence score below or equal to 3 was defined as low and a confidence score above 3 as high. This cut-off was chosen because students selecting “3” were still essentially unsure (‘in doubt’) about being correct. We used Hasan’s decision matrix to label the answers (see Fig. ). According to this matrix incorrect answers given with high confidence are considered misconceptions, incorrect answers given with low confidence are considered a lack of knowledge . Correct answers held with low confidence were labelled lucky guesses and correct answers with high confidence were labelled correct knowledge. In these terms, misconceptions and lucky guesses are considered low metacognition and correct knowledge and lack of knowledge high metacognition. Furthermore, a cognitive effect was labelled positive when one changed an incorrect answer to a correct answer. A metacognitive effect was labelled positive when one changed from low metacognition to high metacognition.
Table presents the overall performance and average confidence scores on the pre-test and post-test in both groups. After reading Refutation texts, the overall test performance score increased significantly from 36.3% ± 0.03 to 58.8% ± 0.03 (t (79) = 6976, p < 0.001). For the Standard text group a significant increase from 34.3% ± 0.03 to 57.1% ± 0.03 was found (t (80) = 7198, p < 0.001). There was no significant difference in post-test performance between the group reading Refutation texts and the group reading Standard texts (F (1,644) = 0.095, p = 0.758). The overall confidences scores increased significantly from 3.10 ± 0.06 to 3.52 ± 0.08 in the Refutation text group (t (79) = 6154, p < 0.001). For the Standard text group a significant increase from 3.23 ± 0.06 to 3.58 ± 0.05 was found (t (80) = 6101, p < 0.001). Additionally, there was also no significant difference in post-test confidence scores between the groups (F (1,160) = 0.003, p = 0.954). Figure shows the relationship between students’ performance and confidence scores on individual questions. For each student, confidence scores per question were corrected for their average confidence to remove between-student variability in average confidence (i.e. confidence*, see ). In the refutation text group, the difference in confidence scores between incorrect and correct answers increased from Δ0.419 points pre-intervention to Δ0.643 points post-intervention. The standard text group showed a similar increase from Δ0.382 to Δ0.695 points. A complete overview of the numbers of incorrect and correct answers and related confidence and confidence* in both groups, at both stages is shown in Table . Multiple linear regression analysis was used to determine main and interactive effects of response, stage and group on confidence*, see Table . Response and stage were significant predictors. The significant interaction effect between response and stage indicates that the difference in confidence* between incorrect and correct answers (i.e. the response effect) was significantly higher post-intervention than pre-intervention. This interaction effect, however, was not significantly different between the groups, as indicated by the lack of significance for the response-stage-group interaction term. Figure displays all changes, from pre- to post-intervention, in performance and confidence for the refutation and standard text groups. Both groups showed comparable changes. A positive cognitive effect was indicated if initially incorrect answers (i.e. misconception or lack of knowledge) were changed to correct answers (i.e. lucky guess or correct knowledge). A positive metacognitive effect was indicated if initially low metacognitive accuracy (i.e. misconception or lucky guess) changed to high metacognitive accuracy (i.e. lack of knowledge or correct knowledge). In the refutation text group, an overall positive cognitive effect was measured in 31.9% of cases compared to a negative cognitive effect of 9.4%. The positive metacognitive effect was 23.8% compared to a negative metacognitive effect of 21.3%. In the standard text group, the overall positive and negative cognitive effects were 32.5 and 9.0%, and the metacognitive effects were 25.0 and 20.5% respectively. The hypercorrection hypothesis was tested by comparing the percentages of low versus high confidence incorrect answers that were changed to correct answers post-intervention. For the Refutation text group, 35.8% of the initially incorrect answers that were rated with high confidence (i.e. misconceptions) changed to correct knowledge after intervention. In contrast, initially incorrect answers rated with low confidence (i.e. lack of knowledge) were corrected to the right conception in 61.0% of cases. Similar findings were obtained in the Standard text group: the percentage of misconceptions that was corrected was 40.3% versus 66.0% of the lack of knowledge answers. Thus, these data do not support the hypercorrection hypothesis for either group. Rather, students with lack of knowledge more frequently corrected their answers than students with misconceptions.
With this study we investigated if reading refutation texts benefits conceptual understanding in medical students. Based on previous research we expected refutation texts to have a positive effect on students’ conceptual understanding (i.e. cognition) plus associated awareness of their understanding (i.e. metacognition). Additionally, based on the hypercorrection hypothesis we suggested that an increase in conceptual understanding would be present in students with high-confidence misconceptions in particular. In summary, we found that reading refutation texts improved students’ cognition and metacognition but these effects were not significantly greater than the effects of reading standard texts. Furthermore, we could not find support for the hypercorrection hypothesis as students’ misconceptions were actually found harder to correct than correcting a lack of knowledge. Here, we elaborate on these findings and propose that instructional methods for concept learning should take into account the key role of metacognition, and the nature of misconceptions. Since the cognitive and metacognitive improvements were found in both groups they could not be attributed to the refutation element, contrasting previous studies in higher education . Instead, the increase in both groups could be due to the answer and explanation elements that were present in both texts. According to the Knowledge Revision Components framework , co-activation of learners’ prior knowledge and new information may result in awareness of a possible cognitive conflict. Learners attempt to resolve this issue, leading to enhanced conceptual understanding. In our case, co-activation could have been induced by the answer or explanation element rather than the refutation element per se. Notably, many different misconceptions may be present among learners whereas the refutation element only addressed one of the supposedly most common misconceptions. Consequently, other possible alternative conceptions may have been left unaddressed, thereby limiting co-activation. As indicated by an increase in accurate metacognitive judgements, co-activation seemed to be established to some extent in both groups, although, the absolute metacognitive outcomes remained relatively poor. These relatively poor metacognition scores align with findings from Thiede and colleagues reporting an average correlation of 0.27 between one’s actual performance and one’s self-perceived performance across 57 studies . Since metacognition plays an important role in conceptual change processes, we suggest educators and researchers should pay more attention to the metacognitive component of learning . This also relates to the view of the medical education community that students are expected to engage in their education as self-regulated learners . Self-regulated learning is an umbrella term that covers the cognitive, metacognitive, behavioural, motivational, and affective aspects of learning (for a review see Panadero, 2017) . According to theory and practice, important metacognitive skills to facilitate self-regulated learning include planning, monitoring and evaluating. Optimizing these skills will contribute to effective learning, independent learning, and lifelong learning . This comes with an important task for the medical educator as explicit teaching of these metacognitive skills inevitable; ‘learning how to learn cannot be left to students. It must be taught.’ . The lack of additional benefit of the refutation element may be further explained by the nature of our misconceptions. As described by Chi, one can distinguish three types of knowledge representation: single ideas, mental models and categories . Faulty ideas are suggested to be refuted more easily compared to flawed mental models or complex concepts such as physiological concepts. Regarding the latter, learners must generate inferences by connecting and understanding cause-effect relations . For our physiological misconceptions, refutation texts alone may not have been sufficient to achieve coherent concept representation. Additional educational approaches including diagramming, concept maps, problem-based learning and peer instruction may be needed to establish conceptual change for abstract scientific concepts . Contrary to the hypercorrection hypothesis, our findings showed that incorrect low confidence answers (i.e. lack of knowledge) were corrected more frequently than incorrect high confidence answers (i.e. misconceptions), after the interventions. This finding suggests that misconceptions are harder to correct than lack of knowledge which resonates with Conceptual Change Theory . Again, the nature of misconceptions may play an important role in the ease with which conceptual change can be achieved. Interestingly, a previous study by van Loon et al. showed results similar to our study and suggested that the absence of the hypercorrection effect may also be clarified by the feedback format . Both van Loon et al., and our study provided feedback (through text reading) to students on all questions, whether they held a misconception or not. Contrastingly, previous hypercorrection studies only provided feedback to learners when they made an error which might cause attentional bias towards the misconceptions . Due to the contextual and protocol variations, it remains difficult to compare and generalise results across studies. Therefore, future studies in the specific context of medicine are needed to advance conceptual change research in medical education. This research has limitations that need be considered when interpreting its results. Conceptual change is a gradual process, therefore, a longitudinal design including long-term outcomes may provide additional insights in students’ learning processes. Furthermore, our study was conducted in a real-life seminar setting and therefore comprised a limited number of questions. Additionally, we used a multi-tier approach with multiple choice answers . Regarding students’ cognition, we cannot know if there were other alternative conceptions that were not explicitly stated in the assessment format. Regarding students’ metacognition, we cannot identify the metacognitive processes that occurred during reading as we only measured their confidence after reading. Future research may include open-ended questions or thinking aloud procedures to provide more information on students’ level of conceptual understanding and metacognitive processes.
This study was the first to investigate the effect of refutation texts on conceptual understanding in medical students. Reading refutation texts did not significantly improve students’ cognition and metacognition beyond reading standard texts. Importantly, we found that misconceptions on cardiovascular physiology were robust and the accuracy of metacognitive judgements among medical students was relatively low. These findings have implications for classroom practice, by addressing the critical role of metacognition and the nature of misconceptions in physiological concept learning. Future studies should take into account these cognitive and metacognitive facets involved in students’ learning processes in order to develop effective teaching practices.
Additional file 1. Appendix A, Appendix B; A. Multi-tier question with 3-tiers: Yes/No, Explanation, and Confidence, B. Refutation text with a refutation element, correct answer, and explanation.
|
Systematic review of clinical effectiveness, components, and delivery of pulmonary rehabilitation in low-resource settings | c2c35e34-fa47-4a47-90fd-a0fadb43bea0 | 7677536 | Patient Education as Topic[mh] | The epidemiological transition from communicable to non-communicable disease (NCD) imposes a ‘double burden’ on low- and middle-income countries (LMICs) , which continue to combat infectious diseases but are typically not yet ready to manage NCDs including chronic respiratory diseases (CRDs) . CRDs are common , and disabling – imposing a substantial burden in LMICs. Poor awareness and insufficient resources – in terms of infrastructure for diagnosis, availability of essential drugs, skilled health professionals, and overall healthcare priorities limit management options . Pulmonary rehabilitation (PR) is an effective component of CRD care . PR is a comprehensive, multidisciplinary, individually tailored intervention designed to overcome the deconditioning induced by CRDs . The components of PR include, but are not limited to, exercise programmes, chest physiotherapy, education, and supporting self-management and lifestyle change, after optimising the recommended pharmacotherapy – . PR cost-effectively reduces symptoms, morbidity, hospital admission (and readmission), duration of hospital stay, and emergency medical help and improves functional exercise capacity and health-related quality of life (HRQoL) – . However, most of the evidence is generated from high-income countries (HICs) and is disease specific – (most commonly chronic obstructive pulmonary disease (COPD)), whereas respiratory disease is often much less differentiated in LMICs. In addition, PR services as developed in HICs may not be deliverable in the same format in LMICs , with substantial differences in resources, awareness, culture, healthcare configuration, and profile of diseases , , which may affect overall management strategy. The potential gains to individuals and healthcare economies, however, are large given the burden of disease in LMICs , . Despite well-established effectiveness , , PR services are often unavailable even in HICs – and uptake (by clinicians and patients) is poor particularly in LMICs and especially in rural communities . A strategy is needed to elaborate PR programmes that are deliverable and effective in LMICs. We therefore aimed to systematically search the literature to: (1) assess the impact of PR on HRQoL and exercise capacity, when delivered in low-resource settings for people with CRD, (2) identify the components used in effective interventions, and (3) describe the models of care deliverable in low-resource settings.
Study selection Our systematic review identified 8912 records. We also found an additional 82 records from forward citation. Following the removal of duplicates, 7437 titles and abstracts were screened (Fig. ). Fifty-six articles were reviewed in full text, with 43 articles excluded. Thirteen articles met the review criteria and were included – . No additional papers were identified in the pre-publication update. Total recruitment for the study was 661 individuals with CRD. Attrition was reported in 9 studies; 96 (20%) of the 479 subjects dropped out. Study participants Study participants were COPD patients , – of varying degree of severity in all the trials except one which recruited people with pulmonary impairment after TB (PIAT) . Total number of enrolled participants was 661 of which COPD and PIAT were 83% and 17%, respectively. Geographical area The trials were conducted in Turkey ( n = 4) , , , , Brazil ( n = 3) , , , India ( n = 2) , , Egypt ( n = 1) , Iran ( n = 1), South Africa ( n = 1) , and Venezuela ( n = 1) . Study settings Five studies were conducted at hospital outpatient departments – , , with or without continuation of exercise at home, seven were home-based , , , , , , training with or without telephonic/face-to-face monitoring or supervision, and one trial was conducted in a community centre . Wherever the PR was delivered, all baseline and follow-up data were collected in a hospital/centre setting. Risk of bias (RoB) assessment Overall RoB is shown in the first column of Table and detailed in Supplementary Results . Almost all studies were at overall high RoB, with only two studies , , which concealed randomisation and took steps to avoid other biases, at moderate RoB. Due to the nature of the intervention, blinding of the patients or the personnel delivering the PR was not possible, but only one study explicitly stated that outcome assessment was blind to allocation . Attrition was a problem or was not clear in all but three studies , , . None of the studies had a published protocol, so selective reporting could not be assessed. Effectiveness of intervention (Objective 1) Although 6-min walking test (6-MWT), St George’s Respiratory Questionnaire (SGRQ), and modified Medical Research Council (mMRC) were widely used to assess functional exercise capacity, HRQoL, and breathlessness respectively, only six of the trials presented between-group comparisons , , , , , . The other seven provided within-group differences , , , , , , . In addition, heterogeneity in terms of mode of intervention, duration, setting, comparator, and baseline measurements confirmed our decision that meta-analysis was not appropriate. We therefore undertook a narrative synthesis and illustrated functional exercise capacity, HRQoL, and breathless in a harvest plot (Fig. ). Our interpretation of the study findings and the structured process determining the decisions that underpinned the harvest plot are described in column 5 of Table . Changes in functional exercise capacity were measured in 11 studies – , , . Significant positive changes were found in 10 studies , – , , ; the exception being one of the two studies at moderate RoB . HRQoL was measured in 12 studies , – ; all showing positive changes. Breathlessness was measured in 11 studies – , – , – of which 9 studies , – , – , , showed significant positive changes and 2 studies (1 at moderate RoB) , showed no changes after intervention. None of the studies reported negative effects after the intervention. Components of the intervention (Objective 2) All interventions included exercise and non-exercise components (as per inclusion criteria), though the approach, content, method of delivery, and duration varied. The components are described in Table and their presence are indicated in a matrix in Table . Endurance training was included in all 13 studies. Other common exercises were upper limb exercise – , , , and strength training in seven studies – , , , and stretching exercises in four studies , , , . Although not described in detail, the other common component was breathing exercises included in eight studies , , , – , . Along with the exercise, patient education was provided in ten studies , , – , , and skills (such as inhaler technique and airway clearance) were included in seven studies , , , , , , . Other components in a minority of studies were social support , optimisation of pharmacotherapy , , nutrition , – , coping strategies , , , , , psychological intervention , , , , self-management , and physical activity interventions , , . Smoking cessation support was reported in only two studies , . Models of care (Objective 3) We identified three models of PR service in our included studies according to the settings in which they were delivered (see Table ). Five were based in hospital or rehabilitation centres – , , , and one was based in a community health centre . Only one was delivered completely at home while most home-based programmes , , , , , provided initial training in the hospital or centre and maintained telephone , , or face-to-face supervision , . The programmes typically lasted 8 weeks (range 4–12), with supervised sessions lasting between 30 and 120 min provided 2 or 3 times per week. Home-based programmes promoted more frequent exercise sessions often supported by telephone or face-to-face contacts. Physiotherapists provided the sessions in six studies , – , , with nurses involved in four studies , , , . Adherence to the PR course was poorly reported with no details provided about reasons for non-completion. Inexpensive instruments were often used in the studies, which ensured the wide availability and acceptability to the consumers. Lower limb endurance exercise was conducted by walking as opposed to expensive stationary bicycle with upper limb resistance/strength training conducted using home-made weights, such as water bottles. Breathing exercises were done with similar devices that are used in higher resource setting (e.g. incentive spirometers, tri-flow).
Our systematic review identified 8912 records. We also found an additional 82 records from forward citation. Following the removal of duplicates, 7437 titles and abstracts were screened (Fig. ). Fifty-six articles were reviewed in full text, with 43 articles excluded. Thirteen articles met the review criteria and were included – . No additional papers were identified in the pre-publication update. Total recruitment for the study was 661 individuals with CRD. Attrition was reported in 9 studies; 96 (20%) of the 479 subjects dropped out.
Study participants were COPD patients , – of varying degree of severity in all the trials except one which recruited people with pulmonary impairment after TB (PIAT) . Total number of enrolled participants was 661 of which COPD and PIAT were 83% and 17%, respectively.
The trials were conducted in Turkey ( n = 4) , , , , Brazil ( n = 3) , , , India ( n = 2) , , Egypt ( n = 1) , Iran ( n = 1), South Africa ( n = 1) , and Venezuela ( n = 1) .
Five studies were conducted at hospital outpatient departments – , , with or without continuation of exercise at home, seven were home-based , , , , , , training with or without telephonic/face-to-face monitoring or supervision, and one trial was conducted in a community centre . Wherever the PR was delivered, all baseline and follow-up data were collected in a hospital/centre setting.
Overall RoB is shown in the first column of Table and detailed in Supplementary Results . Almost all studies were at overall high RoB, with only two studies , , which concealed randomisation and took steps to avoid other biases, at moderate RoB. Due to the nature of the intervention, blinding of the patients or the personnel delivering the PR was not possible, but only one study explicitly stated that outcome assessment was blind to allocation . Attrition was a problem or was not clear in all but three studies , , . None of the studies had a published protocol, so selective reporting could not be assessed.
Although 6-min walking test (6-MWT), St George’s Respiratory Questionnaire (SGRQ), and modified Medical Research Council (mMRC) were widely used to assess functional exercise capacity, HRQoL, and breathlessness respectively, only six of the trials presented between-group comparisons , , , , , . The other seven provided within-group differences , , , , , , . In addition, heterogeneity in terms of mode of intervention, duration, setting, comparator, and baseline measurements confirmed our decision that meta-analysis was not appropriate. We therefore undertook a narrative synthesis and illustrated functional exercise capacity, HRQoL, and breathless in a harvest plot (Fig. ). Our interpretation of the study findings and the structured process determining the decisions that underpinned the harvest plot are described in column 5 of Table . Changes in functional exercise capacity were measured in 11 studies – , , . Significant positive changes were found in 10 studies , – , , ; the exception being one of the two studies at moderate RoB . HRQoL was measured in 12 studies , – ; all showing positive changes. Breathlessness was measured in 11 studies – , – , – of which 9 studies , – , – , , showed significant positive changes and 2 studies (1 at moderate RoB) , showed no changes after intervention. None of the studies reported negative effects after the intervention.
All interventions included exercise and non-exercise components (as per inclusion criteria), though the approach, content, method of delivery, and duration varied. The components are described in Table and their presence are indicated in a matrix in Table . Endurance training was included in all 13 studies. Other common exercises were upper limb exercise – , , , and strength training in seven studies – , , , and stretching exercises in four studies , , , . Although not described in detail, the other common component was breathing exercises included in eight studies , , , – , . Along with the exercise, patient education was provided in ten studies , , – , , and skills (such as inhaler technique and airway clearance) were included in seven studies , , , , , , . Other components in a minority of studies were social support , optimisation of pharmacotherapy , , nutrition , – , coping strategies , , , , , psychological intervention , , , , self-management , and physical activity interventions , , . Smoking cessation support was reported in only two studies , .
We identified three models of PR service in our included studies according to the settings in which they were delivered (see Table ). Five were based in hospital or rehabilitation centres – , , , and one was based in a community health centre . Only one was delivered completely at home while most home-based programmes , , , , , provided initial training in the hospital or centre and maintained telephone , , or face-to-face supervision , . The programmes typically lasted 8 weeks (range 4–12), with supervised sessions lasting between 30 and 120 min provided 2 or 3 times per week. Home-based programmes promoted more frequent exercise sessions often supported by telephone or face-to-face contacts. Physiotherapists provided the sessions in six studies , – , , with nurses involved in four studies , , , . Adherence to the PR course was poorly reported with no details provided about reasons for non-completion. Inexpensive instruments were often used in the studies, which ensured the wide availability and acceptability to the consumers. Lower limb endurance exercise was conducted by walking as opposed to expensive stationary bicycle with upper limb resistance/strength training conducted using home-made weights, such as water bottles. Breathing exercises were done with similar devices that are used in higher resource setting (e.g. incentive spirometers, tri-flow).
In summary, our systematic review identified and selected 13 heterogeneous studies from 7 different countries with a total study population of 661 patients. Overall, PR was reported as being effective in terms of improving functional exercise capacity, HRQoL, and breathlessness, though RoB was high in 11 studies. Of the two at moderate RoB, one showed no benefit in any of the outcomes reported . The exercise programmes typically included endurance, interval, upper limb, and resistance/strength training. The commonest additional components were education to improve knowledge and skill acquisition (e.g. inhaler technique) and strategies for coping with breathlessness. Smoking cessation was provided in only two studies. Most PR services were provided in hospital settings or home based, with some describing adaptations to locally acceptable and deliverable approaches. The strength of this systematic review is its broad literature search constructed with the help of a senior librarian and informed by Cochrane’s standard search terms for COPD and LMICs. Nevertheless, we may have missed important studies of PR conducted in low-resource settings. Although we did not specifically search for papers in other languages, we were open to including non-English language papers but none were identified in our searches, perhaps because locally conducted studies or articles in local languages are often not published in indexed journals . We may have missed important information from these studies but lacked resources to extend the search to non-indexed publications and grey literature. We followed rigorous Cochrane methodology duplicating the selection, data extraction, and quality assessment procedures, but confidence in our findings is limited by the high RoB in most of the studies included. We only included controlled trials because we wanted to assess effectiveness. We acknowledge, however, that in LMICs there are many challenges and barriers such as lack of infrastructure, heterogeneity of resources, and poor health literacy, which discourage clinical trials , . Reliable tools for measuring outcomes (e.g. validated questionnaires in local language, well-trained assessors, effective training facilities, etc.) may not be available in low-resource settings reducing accuracy of assessing effectiveness , . We did not search for health economic assessments. All our included studies reported positive outcomes, but the high RoB limits interpretation of this finding. In contrast, the evidence from studies conducted in HICs are mostly at low-to-moderate RoB, so that the Cochrane review was able to conclude confidently that PR was an effective intervention for people with COPD . It is likely that insufficient resources, training, and facilities in LMICs is responsible for the lack of high-quality trials. This is a gap that NIHR-funded initiatives, such as RESPIRE , and RECHARGE aim to address. Compared to high-resource settings, under-diagnosis due to lack of awareness of CRD compounded by limited access to diagnostic tools such as spirometry results in a minority of potentially eligible participants being approached to be enrolled in studies. Poor universal health coverage and ‘catastrophic’ costs of healthcare further limit participation in trials. The lack of diagnostics means that patients recruited as COPD may in fact have a range of undifferentiated CRDs (e.g. pulmonary impairment after tuberculosis or combined obstructive and restrictive disorder ). While this lack of detailed characterisation may impact on findings, offering PR to people with CRD (regardless of specific diagnosis) may be a more appropriate strategy especially in resource-limited settings. There was considerable variation in the clinical status of participants, which might affect outcomes. There was considerable range in severity of functional limitation (see Table ). In addition, some of the patients were stable at enrolment , , , , , while some had been hospitalised for a recent exacerbation , , . Exercise training is the cornerstone of PR and was an inclusion criterion for the studies in our review. Endurance training was included in all the studies in addition to a range of other modalities as per recognised guidelines. Behavioural changes and continuing physical activities are crucial for maintaining effectiveness of PR , but these were not reported in any of the studies. Education on CRD and its treatment was widely provided along with strategies on managing breathlessness, but other components such as self-management support and addressing social care needs were rarely reported, despite evidence of effectiveness in CRDs . In HICs, smoking is the predominant risk factor and cessation support is seen as essential. Surprisingly, only two of the studies in our review reported a smoking cessation component and none reported avoidance of pollution and indoor biomass exposure, which are also important risk factors in LMICs , . The brief descriptions in the papers make it difficult to assess how these and other important educational topics (such as inhaler technique) were addressed. Models of PR delivery depends on who, where, to whom, and how the service is delivered . Different models of PR services were described in the included studies reflecting diversity in the healthcare context and access to PR services; individuals’ health literacy; and background beliefs, attitudes, and preferences, as well as practical factors such as availability of transport and capability of payment . A home-based, inexpensively equipped PR service with minimal attendance at a potentially distant centre may be more suitable model in rural areas with limited resources and poor transport infrastructure , . In home-based models, the cost to the patient is minimised, and people have flexibility in how they invest their time – . Digital technology is a rising paradigm in LMICs, which may be considered in developing a remote model of PR service . Our findings have implications for clinical practice and research. Breathlessness is the principal symptom that drives the patients with CRDs to seek medical help . In LMICs, diagnosis of chronic respiratory symptoms depends on clinical history and physical examination, with limited, or sometimes no, access to spirometry or other investigations . Poor healthcare coverage may mean that tasks regarded as prerequisites to referral in HICs, such as identifying co-morbidities, optimising pharmacotherapy, and exclusion of contraindications, may need to be a component of PR in LMICs . The studies included in this review identified some practical solutions to these challenges, but high-quality evidence of the clinical and cost effectiveness of these pragmatic approaches is urgently needed. In conclusion, recommendations in PR guidelines typically reflect services delivered in high-income settings. Our literature review, although identifying studies with high-to-moderate RoB, highlighted the feasibility of conducting PR in LMICs with positive effects on outcomes such as exercise tolerance, HRQoL, and symptoms improvement. Our findings point to the need for PR services that are effective across a broad range of (potentially poorly differentiated) CRDs, overcoming barriers of cost, distance, and access to healthcare such that they are deliverable and sustainable in low-resource settings with minimal equipment. Only then will the known benefits of PR be available to address the increasing burden of CRDs in LMICs.
Published review protocol The review is registered with PROSPERO [ID: CRD42019125326]. The detailed systematic review protocol is published with salient points described here. We followed the procedures described in the Cochrane Handbook for Systematic Reviews of Interventions . Deviation from published protocol We planned to use Grading of Recommendations Assessment Development and Evaluation (GRADE ) approach to rate the quality of evidence for primary outcomes and the important secondary outcomes; however, there was substantial missing information in the papers, so we were unable to apply the GRADE approach (see Supplementary Results for our limited GRADE exercise). Search strategy Table gives details of the search strategy developed to detect randomised controlled trials (RCTs) and controlled clinical trials of ‘Pulmonary Rehabilitation’ AND ‘COPD or other CRD’ AND ‘LMIC or low-resource settings’ from 1990 (when global COPD guidelines first recommended PR ) to November 2018 with no language restrictions. We searched MEDLINE (Supplementary Methods ) EMBASE, Global Health (CABI), AMED, PubMed, and the Cochrane Database of Controlled Trials (CENTRAL). We did not undertake hand searching as we found no journal that regularly published PR papers in LMICs. Additionally, we conducted forward citations of the included articles. We used EndNote for overall data management. The searches were completed on 28 October 2018, with a pre-publication update on 8 March 2020 using the ‘efficient and effective’ approach of forward citation using Google Scholar, of all included papers, and the Cochrane review . Selection process Details of inclusion and exclusion criteria and definitions used are in Table . In summary, we undertook a duplicate selection process using rules for operationalising the inclusion/exclusion criteria (see protocol for details ). Two trained reviewers (G.M.M.H. and M.N.U.) independently screened titles and abstracts, then full-text papers (G.M.M.H., M.N.U., and K.D.). Disagreements were resolved by discussion, involving H.P. and R.R. or the wider team as necessary. We reported the process in a PRISMA flow diagram (Fig. ) . Outcome measurement Our primary outcomes were between-group difference in functional exercise capacity (e.g. 6-MWT – ) and HRQoL (e.g. SGRQ , ). We also included breathlessness (e.g. mMRC Dyspnoea score ). These are defined, and secondary outcomes are described in Table . Data extraction and RoB Two reviewers (G.M.M.H. and M.N.U. and checked by H.P.) extracted data on a piloted data extraction form (Supplementary Methods ) based on the Cochrane Effective Practice and Organisation of Care guidance ; G.M.M.H. and M.N.U. (checked by H.P.) independently assessed the methodological quality of all the included studies according to the Cochrane RoB tool . Data analysis The analysis addressed our three objectives: Effectiveness of PR in low-resource settings : On the basis of our initial scoping, we anticipated that our included studies would have substantial clinical, methodological, and statistical heterogeneity, and meta-analysis would not be appropriate. We, therefore, conducted a narrative synthesis illustrating the key outcomes on a harvest plot , . In order to ensure transparency of interpretation, the decisions that underpinned the harvest plot are described in Table : column 5. Components used in effective studies : We identified the components that are described in internationally recognised guidelines , , using categories from the American Thoracic Society/European Respiratory Society task force report , British Thoracic Society guidelines for PR , and Lung Foundation of Australia . We then constructed a matrix with the components used in the (effective and ineffective) studies. Models of care used in the PR interventions : We described the models of care used, including PR providers and (if specified) their training, venue and equipment available, number and frequency of training sessions, use of telehealth, and strategies for sustainability.
The review is registered with PROSPERO [ID: CRD42019125326]. The detailed systematic review protocol is published with salient points described here. We followed the procedures described in the Cochrane Handbook for Systematic Reviews of Interventions .
We planned to use Grading of Recommendations Assessment Development and Evaluation (GRADE ) approach to rate the quality of evidence for primary outcomes and the important secondary outcomes; however, there was substantial missing information in the papers, so we were unable to apply the GRADE approach (see Supplementary Results for our limited GRADE exercise).
Table gives details of the search strategy developed to detect randomised controlled trials (RCTs) and controlled clinical trials of ‘Pulmonary Rehabilitation’ AND ‘COPD or other CRD’ AND ‘LMIC or low-resource settings’ from 1990 (when global COPD guidelines first recommended PR ) to November 2018 with no language restrictions. We searched MEDLINE (Supplementary Methods ) EMBASE, Global Health (CABI), AMED, PubMed, and the Cochrane Database of Controlled Trials (CENTRAL). We did not undertake hand searching as we found no journal that regularly published PR papers in LMICs. Additionally, we conducted forward citations of the included articles. We used EndNote for overall data management. The searches were completed on 28 October 2018, with a pre-publication update on 8 March 2020 using the ‘efficient and effective’ approach of forward citation using Google Scholar, of all included papers, and the Cochrane review .
Details of inclusion and exclusion criteria and definitions used are in Table . In summary, we undertook a duplicate selection process using rules for operationalising the inclusion/exclusion criteria (see protocol for details ). Two trained reviewers (G.M.M.H. and M.N.U.) independently screened titles and abstracts, then full-text papers (G.M.M.H., M.N.U., and K.D.). Disagreements were resolved by discussion, involving H.P. and R.R. or the wider team as necessary. We reported the process in a PRISMA flow diagram (Fig. ) .
Our primary outcomes were between-group difference in functional exercise capacity (e.g. 6-MWT – ) and HRQoL (e.g. SGRQ , ). We also included breathlessness (e.g. mMRC Dyspnoea score ). These are defined, and secondary outcomes are described in Table .
Two reviewers (G.M.M.H. and M.N.U. and checked by H.P.) extracted data on a piloted data extraction form (Supplementary Methods ) based on the Cochrane Effective Practice and Organisation of Care guidance ; G.M.M.H. and M.N.U. (checked by H.P.) independently assessed the methodological quality of all the included studies according to the Cochrane RoB tool .
The analysis addressed our three objectives: Effectiveness of PR in low-resource settings : On the basis of our initial scoping, we anticipated that our included studies would have substantial clinical, methodological, and statistical heterogeneity, and meta-analysis would not be appropriate. We, therefore, conducted a narrative synthesis illustrating the key outcomes on a harvest plot , . In order to ensure transparency of interpretation, the decisions that underpinned the harvest plot are described in Table : column 5. Components used in effective studies : We identified the components that are described in internationally recognised guidelines , , using categories from the American Thoracic Society/European Respiratory Society task force report , British Thoracic Society guidelines for PR , and Lung Foundation of Australia . We then constructed a matrix with the components used in the (effective and ineffective) studies. Models of care used in the PR interventions : We described the models of care used, including PR providers and (if specified) their training, venue and equipment available, number and frequency of training sessions, use of telehealth, and strategies for sustainability.
Supplementary Information
|
Inhibition of Atherosclerosis and Liver Steatosis by Agmatine in Western Diet-Fed apoE-Knockout Mice Is Associated with Decrease in Hepatic De Novo Lipogenesis and Reduction in Plasma Triglyceride/High-Density Lipoprotein Cholesterol Ratio | 1d6fb264-0d29-4ef6-ae3d-487eef354bd9 | 8509476 | Anatomy[mh] | Atherosclerosis is a chronic inflammatory disorder of the arterial vessels, often associated with dyslipidemia, characterized by endothelial dysfunction, infiltration of inflammatory cells into the vessel wall, an accumulation of lipids and fibrous elements, and the formation of atherosclerotic plaques . The stability of the atherosclerotic plaque depends on the thickness of the fibrous cap and the degree of local cap inflammation . The probability of plaque rupture, a catastrophic complication of coronary heart disease responsible for its clinical sequelae, acute coronary syndrome, is increased by the cap thinning, which is promoted by the enhanced activity of macrophages, death of vascular smooth muscle cells (VSMCs), and the breakdown of collagen and the extracellular matrix, which is marked by an increase in the volume of the necrotic core . It is recognized that nonalcoholic fatty liver disease (NAFLD) is an early, independent risk factor for atherosclerosis development . This complex liver disorder is prevalent worldwide, and covers a wide spectrum of pathological conditions ranging from simple hepatic steatosis to steatosis with inflammatory response—nonalcoholic steatohepatitis (NASH), which can develop into fibrosis, cirrhosis, and hepatocellular carcinoma . It has been shown that increased caloric intake through the consumption of a high-fat, high-sugar diet, termed a Western diet, significantly contributes to obesity-related NAFLD . From the point of view of metabolism, the excessive accumulation of triglyceride in the cytoplasm of hepatocytes, characteristic for NAFLD, depends on the imbalance between lipid uptake and de novo synthesis vs. fatty acid oxidation (FAO) and the export of lipids in the form of very low-density lipoprotein (VLDL) . Mitochondrial dysfunction has also been shown to play a key role in the development of NAFLD and its progression to NASH . In NAFLD, hepatic uptake and de novo lipogenesis (DNL) are increased, while a compensatory enhancement of mitochondrial fatty acid oxidation is insufficient . Importantly, DNL seems to be the most important pathway that enables the accumulation of fatty acids in the liver under the condition of a Western diet . The major controller of hepatic DNL is sterol regulatory element-binding protein 1c (SREBP-1c), which regulates the expression of all lipogenesis-associated enzymes . Currently, there are no agents licensed to treat NAFLD, so dietary changes and increased physical activity to decrease body fatness remain the mainstream of recommended management strategies. However, such lifestyle changes are difficult to achieve for many patients . Agmatine is an endogenous amine synthesized by decarboxylation of arginine in the presence of mitochondrial arginine decarboxylase (ADC). It can be formed in several mammalian tissues, especially in the brain, kidney or liver . Agmatine is known to possess several interesting pharmacological properties: e.g., it is an agonist for α2-adrenergic and imidazoline receptors as well as an antagonist of NMDA receptors. It has been shown that agmatine has a protective effect on mitochondria along with anti-inflammatory and neuroprotective functions . Such properties make agmatine a potential candidate for an agent to combat NAFLD. Agmatine affects hepatic metabolism including fatty acid oxidation, gluconeogenesis and glycolysis . Our previous studies showed that exogenous agmatine inhibits atherosclerosis in apoE-knockout mice on a chow diet, which was associated with elevation of blood HDL-cholesterol and activation of β-oxidation of fatty acids in the liver . While reports indicate that agmatine ameliorates atherogenesis in cholesterol-fed rabbits and attenuates insulin resistance and in high-fructose-fed rats , its effect on liver steatosis has not been studied so far in animal NAFLD models. Apolipoprotein E-knockout (apoE −/− ) mice have been widely used in atherosclerosis research due to their propensity to spontaneously develop hypercholesterolemia and atherosclerotic plaques on a chow diet . It has been shown that the apoE −/− mice over 6 months of age on a chow diet develop mild hepatic steatosis, which is significantly exacerbated in mice on a high-fat diet (HFD) . In this study we investigated the effects of agmatine on the progression of atherosclerotic lesions and the development of hepatic steatosis in apoE −/− mice fed with a Western high-fat diet, with a particular focus on its effects on the DNL pathway in the liver.
2.1. Body Weight The mean body weight in the control (25.67 ± 1.78 g) and the agmatine-treated (26.22 ± 1.93 g) groups did not differ ( p = 0.55). Additionally, food intake was similar in both groups. 2.2. Effects of Agmatine on Atherosclerotic Plaque To evaluate the impact of agmatine on the development of atherosclerosis, we treated apoE −/− mice fed a Western high-fat diet with agmatine (20 mg per kg of body weight per day) for 16 weeks. The treatment with agmatine resulted in a significant decrease in atherosclerotic lesions in the aortas of apoE-knockout mice. Measured by the cross-section method, the area occupied by atherosclerosis lesions in the agmatine-treated group was 238,498 ± 6672 μm 2 , whereas in the control group it was 295,531 ± 3224 μm 2 ( A–C). Agmatine did not have an influence on necrotic core formation in atherosclerotic plaque, as indicated by HE staining (control 10.1% ± 0.2% vs. agmatine 10.4% ± 0.1%) ( D–F). Agmatine administration significantly reduced the plaque area covered by CD68-immunopositive macrophages compared to the control group (44.5% ± 0.5% vs. 54.6% ± 0.8%) ( A–C), while the content of smooth muscle cells in the fibromuscular cap was similar in both groups (2.0% ± 0.3% vs. 1.9% ± 0.2%) ( D–F). 2.3. Effects of Agmatine on Lipid Profile in Serum Agmatine treatment had an important influence on the lipid profile in apoE −/− mice on Western HFD. The levels of serum TG were significantly lower by about 33% in the agmatine-treated group compared to the control animals. Agmatine showed only the tendency to decrease levels of TC and LDL and increase the levels of HDL (by about 30%). However, the ratio of TG/HDL showed a significant difference between groups ( ). 2.4. Effects of Agmatine on Liver Steatosis Hematoxylin-eosin staining showed changes in liver structure in both the control and agmatine-treated groups. In livers of control mice, the cytoplasm had a granular structure with signs of macrovesicular steatosis of about 30% of hepatocytes present in all three lobular zones. The lobular structure of the liver was still preserved and portal spaces were minimally enlarged, still with no inflammatory or necrotic changes. In livers of agmatine-treated mice, macrovesicular steatosis was found at about 10% of the hepatocytes, mostly in the first lobular area (around portal spaces). The lobular structure of the liver was still preserved and portal spaces were minimally enlarged, still with no inflammatory infiltrate. Small necrotic changes were noticed in the lobular areas ( A–C). Moreover, treatment with agmatine resulted in a significant decrease in TG level of about 30% in the liver of apoE-knockout mice on Western HFD ( D). Plasma AST and ALT levels did not differ significantly between the control and agmatine-treated groups ( E). 2.5. Effects of Agmatine on Lipid Metabolism To investigate the molecular mechanisms responsible for the reduction in hepatic steatosis upon agmatine treatment of apoE −/− mice on Western HFD, we evaluated the mRNA and protein expression of several molecules associated with lipogenesis in the liver. The SREBP-1c , FASN and SCD1 mRNA levels were significantly decreased in the agmatine-treated group (0.72-, 0.70-, and 0.60-fold, respectively) compared with those in the control group ( A). The relative mRNA expression level of ACCA was also lower (0.86-fold) than those in the control group, but did not reach statistical significance ( p = 0.22). Western blotting confirmed that the treatment with agmatine caused a significant decrease in nuclear, mature SREBP-1c and SCD1 protein levels ( B–D). The expression level of precursor SREBP-1c protein was also lower, but showing only a tendency towards significance ( p = 0.15).
The mean body weight in the control (25.67 ± 1.78 g) and the agmatine-treated (26.22 ± 1.93 g) groups did not differ ( p = 0.55). Additionally, food intake was similar in both groups.
To evaluate the impact of agmatine on the development of atherosclerosis, we treated apoE −/− mice fed a Western high-fat diet with agmatine (20 mg per kg of body weight per day) for 16 weeks. The treatment with agmatine resulted in a significant decrease in atherosclerotic lesions in the aortas of apoE-knockout mice. Measured by the cross-section method, the area occupied by atherosclerosis lesions in the agmatine-treated group was 238,498 ± 6672 μm 2 , whereas in the control group it was 295,531 ± 3224 μm 2 ( A–C). Agmatine did not have an influence on necrotic core formation in atherosclerotic plaque, as indicated by HE staining (control 10.1% ± 0.2% vs. agmatine 10.4% ± 0.1%) ( D–F). Agmatine administration significantly reduced the plaque area covered by CD68-immunopositive macrophages compared to the control group (44.5% ± 0.5% vs. 54.6% ± 0.8%) ( A–C), while the content of smooth muscle cells in the fibromuscular cap was similar in both groups (2.0% ± 0.3% vs. 1.9% ± 0.2%) ( D–F).
Agmatine treatment had an important influence on the lipid profile in apoE −/− mice on Western HFD. The levels of serum TG were significantly lower by about 33% in the agmatine-treated group compared to the control animals. Agmatine showed only the tendency to decrease levels of TC and LDL and increase the levels of HDL (by about 30%). However, the ratio of TG/HDL showed a significant difference between groups ( ).
Hematoxylin-eosin staining showed changes in liver structure in both the control and agmatine-treated groups. In livers of control mice, the cytoplasm had a granular structure with signs of macrovesicular steatosis of about 30% of hepatocytes present in all three lobular zones. The lobular structure of the liver was still preserved and portal spaces were minimally enlarged, still with no inflammatory or necrotic changes. In livers of agmatine-treated mice, macrovesicular steatosis was found at about 10% of the hepatocytes, mostly in the first lobular area (around portal spaces). The lobular structure of the liver was still preserved and portal spaces were minimally enlarged, still with no inflammatory infiltrate. Small necrotic changes were noticed in the lobular areas ( A–C). Moreover, treatment with agmatine resulted in a significant decrease in TG level of about 30% in the liver of apoE-knockout mice on Western HFD ( D). Plasma AST and ALT levels did not differ significantly between the control and agmatine-treated groups ( E).
To investigate the molecular mechanisms responsible for the reduction in hepatic steatosis upon agmatine treatment of apoE −/− mice on Western HFD, we evaluated the mRNA and protein expression of several molecules associated with lipogenesis in the liver. The SREBP-1c , FASN and SCD1 mRNA levels were significantly decreased in the agmatine-treated group (0.72-, 0.70-, and 0.60-fold, respectively) compared with those in the control group ( A). The relative mRNA expression level of ACCA was also lower (0.86-fold) than those in the control group, but did not reach statistical significance ( p = 0.22). Western blotting confirmed that the treatment with agmatine caused a significant decrease in nuclear, mature SREBP-1c and SCD1 protein levels ( B–D). The expression level of precursor SREBP-1c protein was also lower, but showing only a tendency towards significance ( p = 0.15).
Atherosclerosis remains the leading cause of death worldwide and NAFLD is associated not only with liver-related mortality per se, but it also poses a significant cardiovascular risk . We have previously reported that prolonged administration of agmatine has antiatherosclerotic activity in apoE −/− mice on a chow diet . In this study, we demonstrated that agmatine significantly inhibited atherosclerosis and attenuated hepatic steatosis in more aggressive models of atherosclerosis and hepatic steatosis—in apoE −/− mice fed a Western HFD. To our knowledge, this is the first report of agmatine inhibition of fatty liver in vivo. NAFLD occurs with excess accumulation of triglycerides in the liver . In this study, we showed that treatment with agmatine caused a 60% reduction in hepatocytes with signs of steatosis and a reduction in liver TG levels by 30%. Importantly, such effects were accompanied by significant decrease in the plasma TG levels. Such action by agmatine is particularly interesting in the context of the growing awareness of increased levels of plasma TG as a determinant of cardiovascular risk . Interestingly, the use of agmatine in Western diet-fed apoE −/− mice revealed differences in its effect on the plasma lipid profile. While the action of agmatine in mice on the chow diet mainly caused an insignificant increase in plasma HDL, under the conditions of a Western diet, its effect on the levels of TG appeared to be much stronger. Such a pattern of action by agmatine most likely depends on the fact that in the HFD models the dominant metabolic changes related to an increase in hepatic DNL and an increase in VLDL-TG secretion from hepatocytes . Our results are also in line with other studies which demonstrate that agmatine treatment in rats fed a high-fructose diet inhibited hepatic steatosis, decreased liver glycogen content and reduced serum TGs, LDL and VLDL levels . It has been proposed that the ratio of plasma TG to HDL may be a reliable indicator of metabolic and cardiovascular risk—its increase is associated with the exacerbation of insulin resistance, the occurrence of hypertension and an increase in residual cardiovascular risk [ , , ]. It may well be that the beneficial influence of agmatine on lipid metabolism and plasma lipid profile represents an important mechanism of its anti-atherosclerotic action and a link between agmatine actions in the liver and vessel wall. In this context, a significant reduction in the TG/HDL ratio by agmatine in our model becomes a strong argument for further translational research. Our results suggest that the effect of agmatine on the development of hepatic steatosis in Western HFD apoE −/− mice is related to inhibition of DNL, as evidenced by the significant decrease expression of key factors and enzymes involved in this lipogenic pathway, such as: SREBP-1c, fatty acid synthase (FASN), and stearoyl-CoA desaturase 1 (SCD1). DNL begins from the conversion of citrate to acetyl-CoA by ATP-citrate lyase (ACLY), which is further turned into malonyl-CoA with the assistance of acetyl-CoA carboxylase (ACCA). Then, malonyl-CoA is condensed by fatty acid synthase (FASN) to produce palmitate, which finally turns to monounsaturated fatty acids, serving as the substrate of stearoyl-CoA desaturase 1 (SCD1) . SREBP-1c plays a key role in lipid homeostasis by regulating the expression of genes involved in DNL and the synthesis of triglyceride . SREBP-1c is a transcription factor which is synthetized in the endoplasmic reticulum as a precursor (125 kDa), which requires post-translational modification to yield its transcriptionally active nuclear form (65 kDa) . In this study, agmatine caused the decrease in SREBP-1c in mRNA and protein expression, especially the active form of the latter. It was shown that activation of SREBP-1 is essential for development of hepatic steatosis . Several studies have shown that SREBP-1 is upregulated in the livers of mice and patients with NAFLD . The triglyceride levels are higher in transgenic mice with overexpression of SREBP-1c . In addition, there is a correlation between SREBP-1c expression and the severity of insulin resistance. Interestingly, Sharawy et al. (2016) have shown that agmatine administration may attenuate insulin resistance through inhibiting SREBP-1c, mammalian target of rapamycin kinase (mTOR) and glucose transporter GLUT-2 in the liver of rats fed a high-fructose diet . It is of note that in our study, agmatine also decreased the activity of the gene encoding the FASN, which was found to be overexpressed in NAFLD patients . FASN catalyzes the last step in the synthesis of palmitate and is believed to be a major determinant of the maximal hepatic capacity to generate fatty acids by DNL . Moreover, agmatine decreased the expression of SCD1 at both the mRNA and protein levels. SCD1 catalyzes the formation of monounsaturated fatty acids (MUFAs) from saturated fatty acids (SFAs). Intriguingly, SCD1 was shown to be involved in regulating diverse processes including inflammation, hormonal signaling, thermogenesis and both lipid synthesis and oxidation . Although SCD1 is expressed ubiquitously, it is predominant in lipogenic tissues, especially hepatocytes and adipocytes . It has been proposed that SCD1 may play an important role in TG accumulation in the hepatocytes; however, its precise role in NAFLD pathogenesis remains unclear . Some studies show that the hepatic SCD1 expression may be increased by a high-fat diet , however in others its expression was decreased . Interestingly, more studies confirm that SCD1 knockout mice are resistant to a high-fat- or high carbohydrate diet-induced steatosis . The SCD1 knockout mice on HFD show decreased liver TG accumulation, reduced TG synthesis and increased fatty acid oxidation . Moreover, MK-8245, an SCD1 inhibitor, has shown promising antidiabetic and antidyslipidemic efficacy in preclinical animal models . Finally, Zhou et al. (2020) have recently suggested a novel function of SCD1 in hepatic lipogenesis, showing that the inhibition of SCD1 ameliorates hepatic steatosis by inducing AMPK-mediated lipophagy . Altogether, our results indicate that agmatine in an animal model of NAFLD—apoE −/− mice fed a Western diet—comprehensively inhibits the DNL pathway in the liver as well as beneficially alters the plasma lipid profile, and may be a promising drug candidate for prevention or treatment of fatty liver and atherosclerosis. Strengths and Limitations Our research has several strengths: we conducted our research in a well-known model of atherosclerosis, of proven value in research on potential antiatherosclerotic drug candidates; we also used a model of fatty liver, based on the Western diet, which is highly caloric in nature, more relevant to the physiopathology of NAFLD than other models, e.g., based on lack of essential nutritional components; finally, we administered the tested compound by the noninvasive oral route at a dose that did not cause any side effects in mice. Nevertheless, our study also has limitations resulting from the need for extreme caution in transferring the results to the human situation, as well as the unquestionable need for further research into possible molecular mechanisms responsible for the action of agmatine, which is known to possess multiple pharmacological properties. It can stimulate α2-adrenergic and imidazoline receptors, block NMDA receptors and can accumulate in mitochondria and exert mitoprotective action. Thus, future studies should address which of the above mechanisms plays a role in the metabolic action of agmatine in the liver. In particular the role of agmatine in improving mitochondrial function needs to be addressed in the context of the pathophysiology of NAFLD.
Our research has several strengths: we conducted our research in a well-known model of atherosclerosis, of proven value in research on potential antiatherosclerotic drug candidates; we also used a model of fatty liver, based on the Western diet, which is highly caloric in nature, more relevant to the physiopathology of NAFLD than other models, e.g., based on lack of essential nutritional components; finally, we administered the tested compound by the noninvasive oral route at a dose that did not cause any side effects in mice. Nevertheless, our study also has limitations resulting from the need for extreme caution in transferring the results to the human situation, as well as the unquestionable need for further research into possible molecular mechanisms responsible for the action of agmatine, which is known to possess multiple pharmacological properties. It can stimulate α2-adrenergic and imidazoline receptors, block NMDA receptors and can accumulate in mitochondria and exert mitoprotective action. Thus, future studies should address which of the above mechanisms plays a role in the metabolic action of agmatine in the liver. In particular the role of agmatine in improving mitochondrial function needs to be addressed in the context of the pathophysiology of NAFLD.
4.1. Experimental Animals Fourteen female apoE-knockout mice from the C57BL/6J background with average age of 6–8 weeks were purchased from Taconic (Ejby, Denmark). All animal procedures which have been performed conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes and have been approved by the Jagiellonian University Ethical Committee on Animal Experiments (Krakow, Poland) (no. 67/2014). 4.2. Experimental Protocol The animals were housed in air-conditioned rooms (22.5 ± 0.5 °C, 50 ± 5% humidity) with 12 h light/12 h dark cycles, with free access to food and water, in the Animal House of Chair of Immunology of JUMC (Krakow, Poland). The mice were randomly divided into the control ( n = 7) and the treatment ( n = 7) groups. The mice in the control group were fed a Western high-fat diet (Western HFD diet, containing 15% fat + 0.25% cholesterol) (Labofeed B high-fat Diet, Wytwórnia Pasz Morawski, Kcynia, Poland). The percentage of energy obtained from this Western HFD compared to a normal chow is presented in . In addition, the mice in the treatment group received a high-fat diet mixed with agmatine at a dose of 20 mg/kg of body weight per day. The dose of agmatine was chosen based on the previous results . After 4 months of treatment, all mice were sacrificed using a carbon dioxide chamber in accordance with AVMA Panel 2007 recommendations and institutional IACUC guidelines. Blood samples were collected to prepare serum. The hearts were dissected, embedded in OCT compound (CellPath, Newtown, UK) and snap-frozen at −80 °C. Livers were removed and cut into 3 parts, including parts for TG determination, histological and real-time PCR analysis. Liver for histology was fixed in 4% formalin and for real-time analysis was stored in RNAlater (Ambion, Austin, TX, USA) at −80 °C. 4.3. Analysis of Atherosclerotic Plaque The hearts with ascending aorta were sectioned (10 μm) for histological and immunohistochemical analysis, as described before . To assess atherosclerotic lesions, sections were stained with Oil Red O (Sigma-Aldrich, St. Louis, MO, USA). The size of the necrotic core was measured on a hematoxylin-eosin (HE) staining according to a standard method. For immunohistochemistry, sections were processed using antibodies against CD68 (marker for macrophages, Serotec, Kidlington, UK) (dilution 1:800) and alpha smooth muscle actin (αSMA, Sigma-Aldrich, St. Louis, MO, USA) (dilution 1:800). Images were registered using an Olympus Camedia DP71 digital camera and analyzed using LSM Image Browser software (Zeiss, Jena, Germany). 4.4. Histology of the Liver For hematoxylin-eosin (HE) staining, formalin fixed liver tissues were embedded into paraffin and cut into 2 μm sections. Samples were assessed microscopically for the presence of steatosis and the type of steatosis: microvesicular or macrovesicular. The mean percentage of steatotic hepatocytes has been specified in each case. Moreover, maintenance of lobular structure of the liver, the presence of inflammatory infiltrate both in lobular areas and in portal tracts and presence of necrotic changes of hepatocytes were described. 4.5. Biochemical Analysis The blood was centrifuged for 10 min, 1000× g at 4 °C and the plasma was harvested and stored in -80 °C until use. The levels of total cholesterol (TC), triglycerides (TG), and low- and high-density lipoproteins (LDL and HDL) were measured using an enzymatic method on a Cobas 8000 analyzer (Roche Diagnostics, Indianapolis, IN, USA). In addition, plasmatic levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were determined by commercially available kits: Reflotron GPT and Reflotron GOT (Roche, Mannheim, Germany), and Reflovet Plus equipment (Roche, Mannheim, Germany). TG levels in the liver were quantified using the Triglyceride Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI, USA), according to the manufacturer’s protocol. 4.6. Real-Time PCR Total RNA was isolated from the liver samples using ReliaPrep™ RNA Miniprep System (Promega, Madison, WI, USA), according to the manufacturer’s instructions. The concentration of RNA was determined by measuring the absorbance in an EPOCH Microplate Spectrophotometer (BioTek Instruments Inc., Winooski, VT, USA). The A260/A280 ratio was greater than 1.9 in all samples. The same amount of total RNA (1 μg) for each sample was reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). Commercially available primers ( SCD1 , SREBP1 , FASN , ACCA and GAPDH ) from Bio-Rad (Hercules, CA, USA) and GoTaq ® qPCR Master Mix (Promega, Madison, WI, USA) were used to carry out the real-time PCR reaction. Analysis of relative gene expression with GAPDH as a reference gene was performed by the CFX96 Touch Real-Time PCR Detection System. 4.7. Western Blot Samples containing equal amounts of total protein were mixed with gel loading buffer (50 mM Tris, 10% SDS, 10% glycerol, 10% 2-mercaptoethanol, 2 mg/mL bromophenol blue) in a ratio of 4:1 ( v/v ) and incubated at 95 °C for 5 min. Samples (30 μg of protein) were separated on SDS-polyacrylamide gels (12%) (Mini Protean II, Bio Rad, Hercules, CA, USA) using the Laemmli buffer system and proteins were semidry transferred to nitrocelulose membranes (GE Healthcare, Chicago, IL, USA). The membranes were blocked for 1 h with 5% ( w/v ) nonfat dried milk in TTBS and incubated overnight at 4 °C with specific primary antibodies: anti-SCD1 (1:500, R&D Systems, Minneapolis, MN, USA), anti-SREBP1 (1:1000, Novusbio, Littleton, CO, USA), anti-β-ACTIN (1:5000, Sigma, Saint-Louis, MO, USA) then for 1 h with HRP-conjugated secondary antibodies (GE Healthcare, Chicago, IL, USA). Bands were developed with the use of ECL-system reagents (GE Healthcare, Chicago, IL, USA). Rainbow markers (Amersham Biosciences, Amersham, UK) were used for molecular weight determinations. Protein pattern images were taken using an ImageQuant Las 500 (GE Healthcare, Chicago, IL, USA). Data analysis was performed using Image Lite Studio software (LI-COR, Lincoln, NE, USA). 4.8. Statistical Analysis Data are expressed as mean ± standard error of measurement (SEM). Statistical analysis and graphing were carried out using GraphPad software Prism 9 (GraphPad Software Inc., San Diego, CA, USA). The equality of variance (F-test) and the normality of the data (Shapiro–Wilk test) were checked and then the nonparametric Mann–Whitney U test or t -test were used for statistical analysis of the data. Statistical significance was considered at p < 0.05.
Fourteen female apoE-knockout mice from the C57BL/6J background with average age of 6–8 weeks were purchased from Taconic (Ejby, Denmark). All animal procedures which have been performed conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes and have been approved by the Jagiellonian University Ethical Committee on Animal Experiments (Krakow, Poland) (no. 67/2014).
The animals were housed in air-conditioned rooms (22.5 ± 0.5 °C, 50 ± 5% humidity) with 12 h light/12 h dark cycles, with free access to food and water, in the Animal House of Chair of Immunology of JUMC (Krakow, Poland). The mice were randomly divided into the control ( n = 7) and the treatment ( n = 7) groups. The mice in the control group were fed a Western high-fat diet (Western HFD diet, containing 15% fat + 0.25% cholesterol) (Labofeed B high-fat Diet, Wytwórnia Pasz Morawski, Kcynia, Poland). The percentage of energy obtained from this Western HFD compared to a normal chow is presented in . In addition, the mice in the treatment group received a high-fat diet mixed with agmatine at a dose of 20 mg/kg of body weight per day. The dose of agmatine was chosen based on the previous results . After 4 months of treatment, all mice were sacrificed using a carbon dioxide chamber in accordance with AVMA Panel 2007 recommendations and institutional IACUC guidelines. Blood samples were collected to prepare serum. The hearts were dissected, embedded in OCT compound (CellPath, Newtown, UK) and snap-frozen at −80 °C. Livers were removed and cut into 3 parts, including parts for TG determination, histological and real-time PCR analysis. Liver for histology was fixed in 4% formalin and for real-time analysis was stored in RNAlater (Ambion, Austin, TX, USA) at −80 °C.
The hearts with ascending aorta were sectioned (10 μm) for histological and immunohistochemical analysis, as described before . To assess atherosclerotic lesions, sections were stained with Oil Red O (Sigma-Aldrich, St. Louis, MO, USA). The size of the necrotic core was measured on a hematoxylin-eosin (HE) staining according to a standard method. For immunohistochemistry, sections were processed using antibodies against CD68 (marker for macrophages, Serotec, Kidlington, UK) (dilution 1:800) and alpha smooth muscle actin (αSMA, Sigma-Aldrich, St. Louis, MO, USA) (dilution 1:800). Images were registered using an Olympus Camedia DP71 digital camera and analyzed using LSM Image Browser software (Zeiss, Jena, Germany).
For hematoxylin-eosin (HE) staining, formalin fixed liver tissues were embedded into paraffin and cut into 2 μm sections. Samples were assessed microscopically for the presence of steatosis and the type of steatosis: microvesicular or macrovesicular. The mean percentage of steatotic hepatocytes has been specified in each case. Moreover, maintenance of lobular structure of the liver, the presence of inflammatory infiltrate both in lobular areas and in portal tracts and presence of necrotic changes of hepatocytes were described.
The blood was centrifuged for 10 min, 1000× g at 4 °C and the plasma was harvested and stored in -80 °C until use. The levels of total cholesterol (TC), triglycerides (TG), and low- and high-density lipoproteins (LDL and HDL) were measured using an enzymatic method on a Cobas 8000 analyzer (Roche Diagnostics, Indianapolis, IN, USA). In addition, plasmatic levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were determined by commercially available kits: Reflotron GPT and Reflotron GOT (Roche, Mannheim, Germany), and Reflovet Plus equipment (Roche, Mannheim, Germany). TG levels in the liver were quantified using the Triglyceride Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI, USA), according to the manufacturer’s protocol.
Total RNA was isolated from the liver samples using ReliaPrep™ RNA Miniprep System (Promega, Madison, WI, USA), according to the manufacturer’s instructions. The concentration of RNA was determined by measuring the absorbance in an EPOCH Microplate Spectrophotometer (BioTek Instruments Inc., Winooski, VT, USA). The A260/A280 ratio was greater than 1.9 in all samples. The same amount of total RNA (1 μg) for each sample was reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). Commercially available primers ( SCD1 , SREBP1 , FASN , ACCA and GAPDH ) from Bio-Rad (Hercules, CA, USA) and GoTaq ® qPCR Master Mix (Promega, Madison, WI, USA) were used to carry out the real-time PCR reaction. Analysis of relative gene expression with GAPDH as a reference gene was performed by the CFX96 Touch Real-Time PCR Detection System.
Samples containing equal amounts of total protein were mixed with gel loading buffer (50 mM Tris, 10% SDS, 10% glycerol, 10% 2-mercaptoethanol, 2 mg/mL bromophenol blue) in a ratio of 4:1 ( v/v ) and incubated at 95 °C for 5 min. Samples (30 μg of protein) were separated on SDS-polyacrylamide gels (12%) (Mini Protean II, Bio Rad, Hercules, CA, USA) using the Laemmli buffer system and proteins were semidry transferred to nitrocelulose membranes (GE Healthcare, Chicago, IL, USA). The membranes were blocked for 1 h with 5% ( w/v ) nonfat dried milk in TTBS and incubated overnight at 4 °C with specific primary antibodies: anti-SCD1 (1:500, R&D Systems, Minneapolis, MN, USA), anti-SREBP1 (1:1000, Novusbio, Littleton, CO, USA), anti-β-ACTIN (1:5000, Sigma, Saint-Louis, MO, USA) then for 1 h with HRP-conjugated secondary antibodies (GE Healthcare, Chicago, IL, USA). Bands were developed with the use of ECL-system reagents (GE Healthcare, Chicago, IL, USA). Rainbow markers (Amersham Biosciences, Amersham, UK) were used for molecular weight determinations. Protein pattern images were taken using an ImageQuant Las 500 (GE Healthcare, Chicago, IL, USA). Data analysis was performed using Image Lite Studio software (LI-COR, Lincoln, NE, USA).
Data are expressed as mean ± standard error of measurement (SEM). Statistical analysis and graphing were carried out using GraphPad software Prism 9 (GraphPad Software Inc., San Diego, CA, USA). The equality of variance (F-test) and the normality of the data (Shapiro–Wilk test) were checked and then the nonparametric Mann–Whitney U test or t -test were used for statistical analysis of the data. Statistical significance was considered at p < 0.05.
In summary, this study proves that agmatine inhibits the progression of atherosclerosis and attenuates hepatic steatosis by reducing the levels of triglyceride in the serum and in the liver of apoE −/− mice on a Western high-fat diet. Such action is associated with the decrease in hepatic expression of genes involved in de novo lipogenesis. It is tempting to speculate that agmatine may provide a novel therapeutic approach to the treatment/prevention of atherosclerosis and fatty liver disease. However, the exact understanding of the mechanisms of the advantageous actions of agmatine require further study. The potential mechanisms of action of agmatine in atherosclerosis and hepatic steatosis are present in .
|
Proteomic analysis identifies deregulated metabolic and oxidative-associated proteins in Italian intrahepatic cholangiocarcinoma patients | c49e29c8-d1da-4b74-95b5-666bdac328b8 | 8317365 | Anatomy[mh] | Cholangiocarcinoma (CCA) is the second most common type of hepatobiliary cancer arising from the ductal epithelium of the biliary tree, either within the liver (intrahepatic cholangiocarcinoma, ICC) or from extrahepatic bile ducts (extrahepatic cholangiocarcinoma, ECC), which included perihilar and distal cholangiocarcinoma . CCA incidence, pathogenesis and etiology differ not only among the subtypes, but also according to ethnicity, with peculiar genetic alterations and risk factors . Globally, its incidence is higher in Asiatic Countries of the Pacific area, especially due to parasite infection, but the number of cases in Europe is increasing in the last 30 years . As it concerns Italy, a retrospective study demonstrated that CCA incidence in the years 1988–2005 displayed an annual increment of 3–6%, with highly increased rate and mortality for ICC compared to ECC . A minority of patients had surgically resectable tumors at diagnosis, but the recurrence rate was higher than 50% within 5 years, since the diagnosis was often delayed . The therapeutic approach for locally advanced or metastatic diseases is chemotherapy; the backbone is represented by Gemcitabine in association with platinum compounds, with a median overall survival of 11.7 months compared to 8.1 months of gemcitabine alone . Unfortunately, patients developed resistance and disease progression occurred, making this pathology highly lethal. The molecular mechanisms and genetic steps underlying the pathogenesis of this tumor remain largely unknown; the heterogeneity of these tumors, the different etiology and risk factors involved in tumor development, complicate the identification of suitable molecular target and treatment options. In the last years, the importance to properly classify CCA emerged, considering the subtypes as different entities. Large-scale technologies, such as whole genome sequencing, RNA-seq, microarray and methylation arrays, highlighted the real need to distinguish either the subtypes or the intra- and inter-tumoral heterogeneity of CCA [ – ]. It is well-known that some mutations such as IDH1, BAP1, ARID1A, and FGFR2 rearrangements are typically enriched in ICC, while KRAS and TP53 in ECC . In general, IDH1 and FGFR2 aberrations are associated with better prognosis, while KRAS and TP53 with worse outcome . These data enforced the real need to treat ICC and ECC with tailored clinical approaches. Indeed, not only mutations, but also distinct patterns of epigenetic alteration profiling may differentiate ICC from ECC . Moreover, recent studies demonstrated that a classification based on etiology and molecular aspects, such as methylation and copy number variations, is complementary and more useful than the subtypes classification alone . Although there are lots of information about transcriptomic and mutational status, the proteomic profile of CCA remains only partially explored and is mainly associated to particular histotypes and/or ethnic origin, in turn strictly associated with risk factors. In a study of Dos Santos and coll. a panel of 39 altered proteins involved in motility and actin cytoskeleton remodeling was found in an ICC case series compared to non-tumoral adjacent liver tissue . In an Asiatic case series, Annexin A2, peroxiredoxin 1 and, ezrin-radixin-moesin–binding phosphoprotein 50 were identified as negative prognostic markers of CCA patients . Kristiansen and coll analyzed a case series of CCA, identifying some deregulated proteins, some of them never been associated with CCA arising and progression . A recent review summarized the uniqueness molecular profile of liver fluke-associated CCA obtained by combining multi-omics approaches. Anti-inflammatory, immunomodulator/immunosuppressor, epidermal growth factor receptor or platelet-derived growth factor receptor inhibitors, multi-targeted tyrosine kinase inhibitor, IL6 antagonist, Nuclear Factor-κB inhibitor, histone modulator, proteasome inhibitor MetAP2 inhibitor, 1,25(OH) 2 D 3 and cyclosporine A are suggested as targets for the treatment of this tumor subtype . Recently, the comparison of 6 tumor and peritumoral ECC tissues identified 233 de-regulated proteins, one of them, the up-regulated PPP3CA, is a strong predictor of poor survival . To date, no proteome profiling has been explored in CCA derived from Italian patients. Here, we selected a homogeneous series of five ICC tumors obtained at the time of surgery. We processed them with 2-dimensional (2D) electrophoresis followed by mass spectrometry. The comparison of the proteomic profile with that obtained from a normal epithelial bile duct cell line provided precious information about the pathways potentially involved in tumor development and progression.
Cell line and tumor samples Normal biliary epithelium cell line HIBEpiC (ScienCell Research Laboratories, Carlsbad, CA) was cultured in RPMI 1640 (Gibco, Carlsbad, CA, USA) medium supplemented with 10% fetal bovine serum (Sigma–Aldrich, St. Louis,MO, USA) and 1% penicillin/streptomycin (Life Technologies Gathersburg, MD) at 37 °C and 5% CO2. Tumor samples, 5 ICC fresh frozen (FF) tissues and 15 ICC formalin fixed, paraffin embedded (FFPE), were collected from ICC patients of Italian origin. Among FF ICC samples, 3 derived from females, 2 from males, with a median age at the time of diagnosis of 69 years. Tumor tissues used for the experiments were macrodissected from surgical samples, avoiding the inner (more necrotic) and the peripheral part of tumors (useful for margins evaluation in diagnosis). Biological material was obtained after informed consent, following institutional review board-approved protocols (001-IRCC-00IIS-10 FPO-IRCCS, Istituto di Ricovero e Cura a Carattere Scientifico Candiolo (TO), Italy). Proteomic analysis by 2 dimensional (2D) electrophoresis and mass spectrometry Five ICC and one normal epithelial bile duct cell line were subjected to proteomic analysis. Cells in culture were harvested and centrifuged at 400 g for 10 min at room temperature and washed once in 0.3 M sucrose. The pellet was collected and treated with 100 μl/1 × 10 6 cells of lysis buffer (7 M Urea, 2 M Thiourea, 4% CHAPS, 1 mM EDTA, 2 mM PMSF, 1 mM NaF, 40 mM Tris, pH 9) containing protease inhibitors (Halt Protease, Thermo Fisher Scientific, Cat. No. 78429). After vortexing 3 times for 10 s, the cells were sonicated at 20 kHz and 4 °C for pulses of 20 s with 20 s rest, for a total processing time of 3 min. The suspension was incubated for 30 min on a rotating wheel and then centrifuged 30 min at 13000 xg and 4 °C to remove particulate material. The supernatant was collected, placed in a clean microcentrifuge tube and centrifuged again 15 min at 13000 xg and 4 °C. At the end of the procedure, the supernatant was collected in a clean microcentrifuge tube and total protein content was determined with the Bradford assay. For tissue samples, the procedure was slightly different. Samples were washed twice with sterile normal saline (0.15 M NaCl) under a laminar flow hood to remove contaminating hemoglobin and minced with a sterile scalpel. The small pieces were then transferred in a tube together with lysis buffer plus protease inhibitors in a weight (g) to volume (ml) ratio of 1:3. The samples were then subjected to homogenization by using a Polytron Homogenizer with brief cycles and on ice bath. At the end of the procedure, samples were transferred in clean microcentrifuge tubes and centrifuged for 30 min at 13000 xg and 4 °C. The supernatants were collected and subjected to another centrifugation at 15 min at 13000 xg and 4 °C. Finally, supernatants were collected and assayed for total protein content with the Bradford assay. All samples were processed separately and subjected to 2D electrophoresis as follow. Tree hundreds μg of proteins were subjected to isoelectric focusing (IEF) and separated on 2D SDS-PAGE on 12.5% polyacrylamide gels as described by Carcoforo and collaborators . Gels were stained with colloidal Coomassie, and scanned with the Molecular Imager PharosFX System. The analysis was then performed using the ProteomWeaver 4 program (Bio-Rad, Hercules, CA, USA). Protein spots were automatically identified and manually adjusted if needed, then merged by Pair Matching or Multi Matching function in the software. Each spot was normalized by the total density of the gel to account for possible differences in stain procedure and amount of protein loaded. Differences in spot intensities between ICC and control were considered significant if the matched spots had a fold change > 2 for the upregulated and < 0.5 for downregulated signals and a p -value < 0.01 in Student’s t-test. Differentially expressed spots were then processed for mass spectrometry-based peptide identification. Briefly, gel fragments were washed in 100 mM ammonium bicarbonate and 50% (v/v) ACN, dehydrated by incubation in 100% (v/v) ACN and rehydrated in 50 mM ammonium bicarbonate containing 4 ng/μL of trypsin; 50 mM ammonium bicarbonate was added following digestion overnight at 37 °C. Tryptic peptides were concentrated with ZipTip mC18 pipette tips (Millipore) and co-eluted onto the MALDI target in 1 μL of α-cyano-4-hydroxycinnamic acid matrix (5 mg/mL in 50% ACN, 0.1% TFA). MALDI-MS and MALDI-MS/MS were carried out with a 5800 MALDI TOF/TOF Analyzer (Sciex, Ontario—Canada) essentially as described by Carcoforo and collaborators and detailed in supplementary methods (Additional file ). Immunohistochemistry Catalase expression was evaluated in 15 ICC tumors. Briefly, tumor sections were deparaffinized and rehydrated with graded of ethanol. The epitope retrieval was obtained using Antigen Retrieval Citrate solution pH 6.0 and exposed to 2 min cycles at 700 W in a microwave oven. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 10 min, followed by treatment with V-block for 30 min (Dako, Santa Clara, CA). Sections were incubated for 2 h in a moist chamber at room temperature with the specific primary antibody for CAT (mouse anti Human CAT, 1:100, Santa Cruz) diluted in TBS 1X. Then, slides were rinsed twice in buffer and then incubated with the detection system solution, a Dextran polymer conjugated to horseradish peroxidase, for 30 min. The final reaction was visualized using 3,3′-diaminobenzidine in a buffer/hydrogen peroxide solution for 3 min. Finally, sections were counterstained with Harris’s hematoxylin, dehydrated, and mounted in DPX (Sigma Aldrich, Saint Louis, MO, USA). Eleven ICC tumors (for five samples, the biological material is not available and one more tumor tissue was added) were stained for PRDX6, SODM, and DBI (Aurogene). Briefly, after rehydration, the endogenous peroxidases were blocked in a solution of methanol and hydrogen peroxides (0.3%) for 30 min. The epitope retrieval was obtained using Antigen Retrieval Citrate solution pH 6.0 (Dako) and exposed to 2 cycles of 5 min each at 850 W in a microwave oven. The saturation of non-specific sites was performed with a solution of 5% Normal Goat Serum (Dako) in TBS-Tween (0.3%)-Triton (0.1%) for 1 h in moist chamber at room temperature. Then, slides were incubated O/N at 4 °C with the appropriate primary antibodies at the following dilutions: 1:50 for rabbit polyclonal PRDX6, 1:200 for mouse monoclonal SODM and rabbit polyclonal DBI in the saturation solution. After rinsing, slides were incubated with anti-Rabbit-HRP (for DBI and PRDX6) and with anti-Mouse-HRP (for SODM) for 1 h at room temperature. The final reaction was visualized using DAB+ Substrate Chromogen System (Dako) for 3 min. Finally, sections were counterstained with Harris’s hematoxylin, dehydrated, and mounted in DPX (Sigma Aldrich). Immunohistochemical results were evaluated by two different pathologists (LD and GDR). For CAT and ACBP/DBI expression, the intensity of the reaction was classified using a three grade system: weak positivity (a weak intensity cytoplasmatic staining observed in < 30% of the cells), intermediate positivity (a moderate intensity cytoplasmatic staining observed in > 30% of the tumor cells), and strong positivity (a strong intensity cytoplasmatic staining observed in > 30% of the tumor cells). For SODM the percentage of positive tumor cells was evaluated on a scale of 0–3 (0 no staining, 1+ < 10%, 2+ 11–30%, 3+ 31–50%, 4+ > 50%). For PRDX6 the staining was scored as the product of the staining intensity (on a scale of 0–2: negative = 0, low = 1, high = 2) and the percentage of cells stained (on a scale of 0–3:0 = zero, 1 = 1–25%, 2 = 26–50%, 3 = 51–100%) resulting in scores on a scale of 0–5. External dataset for gene expression profiling We extrapolated the gene expression data of the 5 ICC tumors used for proteomic analysis from the dataset GSE107102. They were included in a bigger cohort of ICC tumors analyzed in our previous work . Material and methods used are previously deeply described in Peraldo-Neia et al. . Gene Expression Profiling Interactive analysis GEPIA ( http://gepia.cancer-pku.cn ) database and the Human Protein Atlas ( https://www.proteinatlas.org ) database were used to verify the expression at mRNA and protein levels, respectively, of differentially expressed targets selected by proteomic analysis.
Normal biliary epithelium cell line HIBEpiC (ScienCell Research Laboratories, Carlsbad, CA) was cultured in RPMI 1640 (Gibco, Carlsbad, CA, USA) medium supplemented with 10% fetal bovine serum (Sigma–Aldrich, St. Louis,MO, USA) and 1% penicillin/streptomycin (Life Technologies Gathersburg, MD) at 37 °C and 5% CO2. Tumor samples, 5 ICC fresh frozen (FF) tissues and 15 ICC formalin fixed, paraffin embedded (FFPE), were collected from ICC patients of Italian origin. Among FF ICC samples, 3 derived from females, 2 from males, with a median age at the time of diagnosis of 69 years. Tumor tissues used for the experiments were macrodissected from surgical samples, avoiding the inner (more necrotic) and the peripheral part of tumors (useful for margins evaluation in diagnosis). Biological material was obtained after informed consent, following institutional review board-approved protocols (001-IRCC-00IIS-10 FPO-IRCCS, Istituto di Ricovero e Cura a Carattere Scientifico Candiolo (TO), Italy).
Five ICC and one normal epithelial bile duct cell line were subjected to proteomic analysis. Cells in culture were harvested and centrifuged at 400 g for 10 min at room temperature and washed once in 0.3 M sucrose. The pellet was collected and treated with 100 μl/1 × 10 6 cells of lysis buffer (7 M Urea, 2 M Thiourea, 4% CHAPS, 1 mM EDTA, 2 mM PMSF, 1 mM NaF, 40 mM Tris, pH 9) containing protease inhibitors (Halt Protease, Thermo Fisher Scientific, Cat. No. 78429). After vortexing 3 times for 10 s, the cells were sonicated at 20 kHz and 4 °C for pulses of 20 s with 20 s rest, for a total processing time of 3 min. The suspension was incubated for 30 min on a rotating wheel and then centrifuged 30 min at 13000 xg and 4 °C to remove particulate material. The supernatant was collected, placed in a clean microcentrifuge tube and centrifuged again 15 min at 13000 xg and 4 °C. At the end of the procedure, the supernatant was collected in a clean microcentrifuge tube and total protein content was determined with the Bradford assay. For tissue samples, the procedure was slightly different. Samples were washed twice with sterile normal saline (0.15 M NaCl) under a laminar flow hood to remove contaminating hemoglobin and minced with a sterile scalpel. The small pieces were then transferred in a tube together with lysis buffer plus protease inhibitors in a weight (g) to volume (ml) ratio of 1:3. The samples were then subjected to homogenization by using a Polytron Homogenizer with brief cycles and on ice bath. At the end of the procedure, samples were transferred in clean microcentrifuge tubes and centrifuged for 30 min at 13000 xg and 4 °C. The supernatants were collected and subjected to another centrifugation at 15 min at 13000 xg and 4 °C. Finally, supernatants were collected and assayed for total protein content with the Bradford assay. All samples were processed separately and subjected to 2D electrophoresis as follow. Tree hundreds μg of proteins were subjected to isoelectric focusing (IEF) and separated on 2D SDS-PAGE on 12.5% polyacrylamide gels as described by Carcoforo and collaborators . Gels were stained with colloidal Coomassie, and scanned with the Molecular Imager PharosFX System. The analysis was then performed using the ProteomWeaver 4 program (Bio-Rad, Hercules, CA, USA). Protein spots were automatically identified and manually adjusted if needed, then merged by Pair Matching or Multi Matching function in the software. Each spot was normalized by the total density of the gel to account for possible differences in stain procedure and amount of protein loaded. Differences in spot intensities between ICC and control were considered significant if the matched spots had a fold change > 2 for the upregulated and < 0.5 for downregulated signals and a p -value < 0.01 in Student’s t-test. Differentially expressed spots were then processed for mass spectrometry-based peptide identification. Briefly, gel fragments were washed in 100 mM ammonium bicarbonate and 50% (v/v) ACN, dehydrated by incubation in 100% (v/v) ACN and rehydrated in 50 mM ammonium bicarbonate containing 4 ng/μL of trypsin; 50 mM ammonium bicarbonate was added following digestion overnight at 37 °C. Tryptic peptides were concentrated with ZipTip mC18 pipette tips (Millipore) and co-eluted onto the MALDI target in 1 μL of α-cyano-4-hydroxycinnamic acid matrix (5 mg/mL in 50% ACN, 0.1% TFA). MALDI-MS and MALDI-MS/MS were carried out with a 5800 MALDI TOF/TOF Analyzer (Sciex, Ontario—Canada) essentially as described by Carcoforo and collaborators and detailed in supplementary methods (Additional file ).
Catalase expression was evaluated in 15 ICC tumors. Briefly, tumor sections were deparaffinized and rehydrated with graded of ethanol. The epitope retrieval was obtained using Antigen Retrieval Citrate solution pH 6.0 and exposed to 2 min cycles at 700 W in a microwave oven. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 10 min, followed by treatment with V-block for 30 min (Dako, Santa Clara, CA). Sections were incubated for 2 h in a moist chamber at room temperature with the specific primary antibody for CAT (mouse anti Human CAT, 1:100, Santa Cruz) diluted in TBS 1X. Then, slides were rinsed twice in buffer and then incubated with the detection system solution, a Dextran polymer conjugated to horseradish peroxidase, for 30 min. The final reaction was visualized using 3,3′-diaminobenzidine in a buffer/hydrogen peroxide solution for 3 min. Finally, sections were counterstained with Harris’s hematoxylin, dehydrated, and mounted in DPX (Sigma Aldrich, Saint Louis, MO, USA). Eleven ICC tumors (for five samples, the biological material is not available and one more tumor tissue was added) were stained for PRDX6, SODM, and DBI (Aurogene). Briefly, after rehydration, the endogenous peroxidases were blocked in a solution of methanol and hydrogen peroxides (0.3%) for 30 min. The epitope retrieval was obtained using Antigen Retrieval Citrate solution pH 6.0 (Dako) and exposed to 2 cycles of 5 min each at 850 W in a microwave oven. The saturation of non-specific sites was performed with a solution of 5% Normal Goat Serum (Dako) in TBS-Tween (0.3%)-Triton (0.1%) for 1 h in moist chamber at room temperature. Then, slides were incubated O/N at 4 °C with the appropriate primary antibodies at the following dilutions: 1:50 for rabbit polyclonal PRDX6, 1:200 for mouse monoclonal SODM and rabbit polyclonal DBI in the saturation solution. After rinsing, slides were incubated with anti-Rabbit-HRP (for DBI and PRDX6) and with anti-Mouse-HRP (for SODM) for 1 h at room temperature. The final reaction was visualized using DAB+ Substrate Chromogen System (Dako) for 3 min. Finally, sections were counterstained with Harris’s hematoxylin, dehydrated, and mounted in DPX (Sigma Aldrich). Immunohistochemical results were evaluated by two different pathologists (LD and GDR). For CAT and ACBP/DBI expression, the intensity of the reaction was classified using a three grade system: weak positivity (a weak intensity cytoplasmatic staining observed in < 30% of the cells), intermediate positivity (a moderate intensity cytoplasmatic staining observed in > 30% of the tumor cells), and strong positivity (a strong intensity cytoplasmatic staining observed in > 30% of the tumor cells). For SODM the percentage of positive tumor cells was evaluated on a scale of 0–3 (0 no staining, 1+ < 10%, 2+ 11–30%, 3+ 31–50%, 4+ > 50%). For PRDX6 the staining was scored as the product of the staining intensity (on a scale of 0–2: negative = 0, low = 1, high = 2) and the percentage of cells stained (on a scale of 0–3:0 = zero, 1 = 1–25%, 2 = 26–50%, 3 = 51–100%) resulting in scores on a scale of 0–5.
We extrapolated the gene expression data of the 5 ICC tumors used for proteomic analysis from the dataset GSE107102. They were included in a bigger cohort of ICC tumors analyzed in our previous work . Material and methods used are previously deeply described in Peraldo-Neia et al. . Gene Expression Profiling Interactive analysis GEPIA ( http://gepia.cancer-pku.cn ) database and the Human Protein Atlas ( https://www.proteinatlas.org ) database were used to verify the expression at mRNA and protein levels, respectively, of differentially expressed targets selected by proteomic analysis.
Analysis of differentially expressed proteins Five tissue samples of ICC patients and normal control (normal biliary epithelium cell line) were run on 2-D GE to investigate differentially expressed protein in tumor compared to normal tissue. Approximately 580 spots were detected in 2-D GE, as shown in Table , with an average of 218 spots/gel. Proteins of each sample were run on separate gels. Figure showed a representative image of gels for one sample (A) and for the normal bile duct cell line (B), while the images for the other samples run in different gels are collected in Additional file . The analysis indicated that 19 proteins, (one of them, HBB, was identified by two peptides) were differentially expressed within the two groups. In particular, 13 spots were upregulated (> 2 fold) and 6 were downregulated (< 0.5 fold) in ICC samples. Mass Spectrometry was used to identify the selected 19 protein spots, which is reported in Table . A consistent part of upregulated proteins (23.5%) in ICC tissues, is related to redox biology. In particular CAT, PRDX2, PRDX6 and SODM are highly expressed compared to normal biliary epithelium cell line (10.6, 3.7, 3.7 and 3.5 fold, respectively). Other proteins with an increased expression were related to metabolism (Acyl-CoA-binding protein, Aminoacylase-1, Cytochrome b-c1 complex subunit Rieske, mitochondrial), cell structure (cytoplasmic Actin 2), signaling (Retinal dehydrogenase 1), oxygen transport (Hemoglobin subunit beta) and DNA binding (Histone H4). Downregulated proteins (fold change between 0.45–0.18) are involved in cell metabolism (Hydroxymethylglutaryl-CoA synthase, mitochondrial Protein disulfide-isomerase and Formimidoyltransferase-cyclodeaminase), cytoskeleton organization (Tubulin alpha-1B chain and Tropomyosin alpha-3 chain), and heat shock protein (Stress-70 protein, mitochondrial). Immunohistochemistry validation of redox and metabolism processes In order to validate proteomic data, we selected four among the up-regulated proteins found by previous analysis, all associated to overrepresented above mentioned processes; we evaluated CAT ( n = 15), SODM, PRDX6 and DBI/ACBP ( n = 11 for the last three proteins) protein expression by IHC in an independent case series of archival tissue samples derived from ICC Italian patients. Table summarized the score staining for each protein. CAT was detected in all tumor tissue samples with different staining score: 2 out of 15 (13.3%) were weakly positive, 11 out of 15 (73.3%) were positive, and 2 out of 15 (13.3%) were strongly positive. SODM is overexpressed in all tumor tissues, 8 out of 11 tumors (73%) are classified as 4+ and 3 out of 11 tumors (27%) as 3+. PRDX6 is expressed in 8 out 11 samples (73%); of them, 5 are classified as 2+, 2 as 3+, one as 4+ and 3 were negative. For these two proteins, the expression was mainly weak or absent in the normal surrounding bile duct. DBI/ACBP was expressed in all the samples with different intensities, 2 with weak, 5 with intermediate, and 4 strong intensities. In all tested samples, the normal counterpart expressed lower levels of the proteins examined. Figure represents different score staining for CAT in ICC samples, while representative IHC images of tumor sections compared to the normal counterparts for ACBP/DBI, PRDX6, and SODM are shown in Additional file . We exploited the Human Protein Atlas database to retrieve more information about protein expression. CCA is included in the “liver cancer” disease, and for each protein selected, a different number of CCA samples were available. Six out of 8 expressed CAT, one with strong intensity. SODM was highly expressed in 3 out of 4 CCA available, PRDX6 in 6 out of 7 CCAs (four with strong-moderate expression), and for ACBP/DBI only three samples were available, 2 of them with weak expression and one moderate. Additional file summarized data obtained for all the differentially expressed proteins by using Protein Atlas database. Comparison between proteomic and transcriptomic data In a previous work , we analyzed the gene expression profiling of 13 ICC (including the five samples used for proteomic analysis) fresh frozen tumors comparing them with a dataset including normal bile ducts (GSE107102). From this dataset, we extrapolated and reanalyzed data of the five samples of interest to investigate if there was a correspondence in terms of expression between protein and transcriptomic data. As shown in Table , considering protein expression as reference, and applying a logFC threshold of |0.5| for mRNA expression, we have a protein-mRNA expression concordance of 50%, 5 out of 12 among up- and 4 out of 6 among down-regulated targets in tumors compared to normal tissue. We exploited GEPIA database to compare the expression of these targets between tumor and normal tissue at mRNA levels; 36 ICC and 9 normal tissues are included in the analysis (TCGA dataset). Boxplot for each targets were summarized in Additional file and the concordance between TCGA data and our GEP analysis and TCGA data and proteomic analysis was reported. Results are only partially comparable, mainly due to the small number of samples analyzed in each case series. No association between targets expression and survival was found using TCGA dataset (Additional file ), with the exception of PRDX2, whose high expression in associated with poor disease free survival ( p = 0.01) and of ALDH1A1, whose low expression is associated with poor disease free survival ( p = 0.03).
Five tissue samples of ICC patients and normal control (normal biliary epithelium cell line) were run on 2-D GE to investigate differentially expressed protein in tumor compared to normal tissue. Approximately 580 spots were detected in 2-D GE, as shown in Table , with an average of 218 spots/gel. Proteins of each sample were run on separate gels. Figure showed a representative image of gels for one sample (A) and for the normal bile duct cell line (B), while the images for the other samples run in different gels are collected in Additional file . The analysis indicated that 19 proteins, (one of them, HBB, was identified by two peptides) were differentially expressed within the two groups. In particular, 13 spots were upregulated (> 2 fold) and 6 were downregulated (< 0.5 fold) in ICC samples. Mass Spectrometry was used to identify the selected 19 protein spots, which is reported in Table . A consistent part of upregulated proteins (23.5%) in ICC tissues, is related to redox biology. In particular CAT, PRDX2, PRDX6 and SODM are highly expressed compared to normal biliary epithelium cell line (10.6, 3.7, 3.7 and 3.5 fold, respectively). Other proteins with an increased expression were related to metabolism (Acyl-CoA-binding protein, Aminoacylase-1, Cytochrome b-c1 complex subunit Rieske, mitochondrial), cell structure (cytoplasmic Actin 2), signaling (Retinal dehydrogenase 1), oxygen transport (Hemoglobin subunit beta) and DNA binding (Histone H4). Downregulated proteins (fold change between 0.45–0.18) are involved in cell metabolism (Hydroxymethylglutaryl-CoA synthase, mitochondrial Protein disulfide-isomerase and Formimidoyltransferase-cyclodeaminase), cytoskeleton organization (Tubulin alpha-1B chain and Tropomyosin alpha-3 chain), and heat shock protein (Stress-70 protein, mitochondrial).
In order to validate proteomic data, we selected four among the up-regulated proteins found by previous analysis, all associated to overrepresented above mentioned processes; we evaluated CAT ( n = 15), SODM, PRDX6 and DBI/ACBP ( n = 11 for the last three proteins) protein expression by IHC in an independent case series of archival tissue samples derived from ICC Italian patients. Table summarized the score staining for each protein. CAT was detected in all tumor tissue samples with different staining score: 2 out of 15 (13.3%) were weakly positive, 11 out of 15 (73.3%) were positive, and 2 out of 15 (13.3%) were strongly positive. SODM is overexpressed in all tumor tissues, 8 out of 11 tumors (73%) are classified as 4+ and 3 out of 11 tumors (27%) as 3+. PRDX6 is expressed in 8 out 11 samples (73%); of them, 5 are classified as 2+, 2 as 3+, one as 4+ and 3 were negative. For these two proteins, the expression was mainly weak or absent in the normal surrounding bile duct. DBI/ACBP was expressed in all the samples with different intensities, 2 with weak, 5 with intermediate, and 4 strong intensities. In all tested samples, the normal counterpart expressed lower levels of the proteins examined. Figure represents different score staining for CAT in ICC samples, while representative IHC images of tumor sections compared to the normal counterparts for ACBP/DBI, PRDX6, and SODM are shown in Additional file . We exploited the Human Protein Atlas database to retrieve more information about protein expression. CCA is included in the “liver cancer” disease, and for each protein selected, a different number of CCA samples were available. Six out of 8 expressed CAT, one with strong intensity. SODM was highly expressed in 3 out of 4 CCA available, PRDX6 in 6 out of 7 CCAs (four with strong-moderate expression), and for ACBP/DBI only three samples were available, 2 of them with weak expression and one moderate. Additional file summarized data obtained for all the differentially expressed proteins by using Protein Atlas database.
In a previous work , we analyzed the gene expression profiling of 13 ICC (including the five samples used for proteomic analysis) fresh frozen tumors comparing them with a dataset including normal bile ducts (GSE107102). From this dataset, we extrapolated and reanalyzed data of the five samples of interest to investigate if there was a correspondence in terms of expression between protein and transcriptomic data. As shown in Table , considering protein expression as reference, and applying a logFC threshold of |0.5| for mRNA expression, we have a protein-mRNA expression concordance of 50%, 5 out of 12 among up- and 4 out of 6 among down-regulated targets in tumors compared to normal tissue. We exploited GEPIA database to compare the expression of these targets between tumor and normal tissue at mRNA levels; 36 ICC and 9 normal tissues are included in the analysis (TCGA dataset). Boxplot for each targets were summarized in Additional file and the concordance between TCGA data and our GEP analysis and TCGA data and proteomic analysis was reported. Results are only partially comparable, mainly due to the small number of samples analyzed in each case series. No association between targets expression and survival was found using TCGA dataset (Additional file ), with the exception of PRDX2, whose high expression in associated with poor disease free survival ( p = 0.01) and of ALDH1A1, whose low expression is associated with poor disease free survival ( p = 0.03).
In the last decade, the identification of putative targets for CCA treatment has become challenging. To date, mutational and transcriptomic profiles, as well as methylation status and fusions assessment, are deeply investigated and many progresses in terms of classification and identification of prognostic biomarkers have been made. It is well-known that all the above-mentioned alterations are strictly associated to ethnicity and etiology and we assume the same behavior for proteomic profile. This, together with the increasing incidence of ICC in Italy and the lack of effective therapies prompted us to analyze the protein expression profile of a small cohort of ICC derived from Italian patients. The first limit of this study is the small number of patients available, but it can be considered a training analysis, which suggests potential targets suitable for therapies to be tested in a validation set. Nevertheless, this study pointed out the impairment of the antioxidant system, with a subsequent accumulation of free radicals. In particular, the main process in which the deregulated proteins are involved is the redox pathway. It is well established that metabolic processes play a key role in tumor progression. Here, we evidenced an up-regulation of ACBP1, confirmed also in the independent case series by IHC, which is already described in other tumor types, especially in glioblastoma and astrocytoma . In physiological conditions, its role is the maintenance of lipid metabolism, steroidogenesis, and peptide hormone release; when overexpressed, it supports tumor growth by controlling the availability of long chain fatty acids which are processed by mitochondria with a fatty oxidation reaction . ACBP1/DBI silencing induces cell senescence, reduces cell proliferation, delays tumor initiation and prolongs survival in in vivo model. Moreover, due to its role as adaptor to microenvironment changes, it seems to promote the cancer stem cell niche during neurogenesis . HBB is involved in oxygen metabolism. High expression was detected in breast cancer, in particular in bone metastasis . Authors suggested a positive correlation between HBB expression and ability of disseminating tumor cells in other organs, indicating a more aggressive phenotype . This data is also confirmed by the work of Zheng and coll. in which HBB is abundantly expressed in circulating tumor cells of breast and prostate cancer patients and its expression is closely detected in circulating tumor cells (CTC), and not in primary tumors . CAT and SODM, both involved in antioxidant processes, are up-regulated in our cohort of ICC patients. Loilome and collaborators demonstrated, in an O. viverrini hamster cholangiocarcinoma model, that both enzymes are highly expressed during cholangiocarcinogenesis, while there is a decreased expression when tumors are well established . The same group demonstrated that both proteins are expressed at different levels in CCA tissues, but they are also expressed in normal bile ducts and hepatocytes . Interestingly, its activity is dramatically reduced in CCA compared to normal bile ducts. In contrast, high expression and activity of antioxidant enzymes, among them SODM and CAT, are found in a cholangiocytes hydrogen peroxide resistant cell line obtained by gradual and continuous exposition to hydrogen peroxides. This cell line had a higher proliferation rate and a more aggressive phenotype compared to the parental one; thus, it may be a suitable model of cholangiocarcinogenesis . In our validation case series, we found that CAT protein is expressed at different levels in cancer tissues, but not in normal adjacent ones. A recent work demonstrated that the presence of variants in genes associated to oxidative stress pathway may affect the response to chemotherapy. Moreover, CAT overexpression inhibits proliferation in vitro of CCA in vitro models and promotes cisplatin and doxorubicin-induced antitumor activity, while low levels of CAT induce resistance to these chemotherapeutic agents . We analyzed the expression of another protein strictly associated with CAT, SODM, found overexpressed by proteomic data. The same trend was revealed in the independent case series tested by IHC. ACY1 is found up-regulated in different tumor types. Literature data demonstrated that ACY1 knockdown in colorectal cancer cells inhibits proliferation and increases apoptosis, becoming an interesting target to explore . In contrast, ACY1 is a putative tumor suppressor in small cell lung cancer and hepatocarcinoma . The impairment of UQCRFS1, involved in mitochondrial stability, electron transport driving oxidative phosphorylation, expression was described in gastric cancers where it is frequently amplified and associated to tumor progression . Opposite results are shown in clear cell renal carcinoma; UQCRFS1 is downregulated, probably due to a DNA hypermethylation of that region . ACTB expression is high in tumor tissues and cell lines; its deregulation in tumors is associated to loss of polarization and major invasiveness and metastatic potential , also described in metastatic breast cancer . PRDX2 is already described as overexpressed in CCA tissues compared to the normal surrounding ones , while PRDX6 is overexpressed in the inflammation process induced by Clonorchis Siniensis . The up-regulation of both PRDX2 and PRDX6 is described in many tumors and correlated with invasiveness, migration, drug resistance and enhancing stem cell properties, in particular in NSCLC, colorectal cancer, and esophageal carcinoma [ – ]. From IHC analysis, we found that about 73% of ICC expressed higher levels of PRDX6 compared to the normal adjacent tissues, in line with published data. PRDX6 overexpression is also associated with poor prognosis and overall survival in ovarian cancer . HIST1H4A resulted highly up-regulated in exosomes released by NSCLC . A recent study conducted on CCA patients showed that high expression of ALDH1A1 correlated with a more favorable prognosis ; in many studies, ALDH1A1 is a cancer stem cell marker and a suitable target for therapy and only its activity is associated to worse prognosis . Another limit of this study is the use of normal immortalized colangiocytes cell line as control in MS experiments, instead of the most appropriate normal biliary tissue; in fact, the cell line lacks the microenvironment, the cellular components usually present in tumor surrounding tissues, actually weakening and potentially impairing our findings. However, the proteins identified in our study were validated in an independent cohort of ICC tissues comparing their expression with the normal surrounding tissues, albeit on a limited number of cases. Globally, this study, even if conducted on a small number of samples, provided precious information about the role of oxidative and metabolic processes in CCA progression, suggesting also that they may be good targets for therapy in CCA. Combining therapies able to tip the balance towards the anti-cancer activity of these pathways with standard chemotherapy could be an alternative approach in CCA treatment. Recently, it was demonstrated that the administration of metformin in association to Cisplatin enhances the oxidative stress mediated cell death pathway, hence increasing the efficacy of Cisplatin alone . Moreover, these pathways may have a potential as prognostic biomarkers in serum. Uchida and collaborators demonstrated that an increase concentration of reactive oxygen metabolites and a decrease level of anti-oxidative metabolites in serum are associated to poor outcome in CCA patients, suggesting the importance of such processes in tumor progression . The complexity of metabolic and oxidative pathways deserves tailored studies to clarify their role in cancer development, progression and drug resistance.
Additional file 1. Additional methods. MS spectra data acquisition. Additional file 2. 2D gel images of single ICC samples. The molecular weight on the left side corresponds to (from top to bottom): 250 kDa, 150 kDa, 100 kDa, 75 kDa, 50 kDa, 37 kDa, 25 kDa, 20 kDa, 15 kDa, 10 kDa. Additional file 3. Representative images of IHC for DBI, PRDX6, and SODM. DBI staining in tumor tissue A) in normal counterpart B); PRDX6 staining in tumor tissue C) in normal counterpart D); SODM staining in tumor tissue E) in normal counterpart F). All the images are captured with 20X. Additional file 4. Protein expression data obtained from The Protein Atlas database. Additional file 5. Box plots representing the expression at mRNA level of all the targets identified by proteomics obtained using GEPIA. Red box: ICC; grey box: normal tissues. Additional file 6. Kaplan Meier curves (Overall survival and Disease free survival) obtained using GEPIA exploiting mRNA expression data of TCGA.
|
MEArec: A Fast and Customizable Testbench Simulator for Ground-truth Extracellular Spiking Activity | 1cc01083-cba3-4b09-ad97-a9212855e639 | 7782412 | Physiology[mh] | Extracellular neural electrophysiology is one of the most used and important techniques to study brain function. It consists of measuring the electrical activity of neurons from electrodes in the extracellular space, that pick up the electrical activity of surrounding neurons. To communicate with each other, neurons generate action potentials, which can be identified in the recorded signals as fast potential transients called spikes . Since electrodes can record the extracellular activity of several surrounding neurons, a processing step called spike sorting is needed. Historically this has required manual curation of the data, which in addition to being time consuming also introduces human bias to data interpretations. In recent years, several automated spike sorters have been developed to alleviate this problems. Spike sorting algorithms (Rey et al. ; Lefebvre et al. ) attempt to separate spike trains of different neurons (units) from the extracellular mixture of signals using a variety of different approaches. After a pre-processing step that usually involves high-pass filtering and re-referencing of the signals to reduce noise, some algorithms first detect putative spikes above a detection threshold and then cluster the extracted and aligned waveforms in a lower-dimensional space (Quiroga et al. ; Rossant et al. ; Chung et al. ; Hilgen et al. ; Jun et al. ). Another approach consists of finding spike templates, using clustering methods, and then matching the templates recursively to the recordings to find when a certain spike has occurred. The general term for these approaches is template-matching (Pachitariu et al. ; Yger et al. ; Diggelmann et al. ). Other approaches have been explored, including the use of independent component analysis (Jäckel et al. ; Buccino et al. ) and semi-supervised approaches (Lee et al. ). The recent development of high-density silicon probes both for in vitro (Berdondini et al. ; Frey et al. ) and in vivo applications (Neto et al. ; Jun et al. ) poses new challenges for spike sorting (Steinmetz et al. ). The high electrode count calls for fully automatic spike sorting algorithms, as the process of manually curating hundreds or thousands of channels becomes more time consuming and less manageable. Therefore, spike sorting algorithms need to be be capable of dealing with a large number of units and dense probes. To address these requirements, the latest developments in spike sorting software have attempted to make algorithms scalable and hardware-accelerated (Pachitariu et al. ; Jun et al. ; Yger et al. ; Pachitariu et al. ). The evaluation of spike sorting performance is also not trivial. Spike sorting is unsupervised by definition, as the recorded signals are only measured extracellularly with no knowledge of the underlying spiking activity. A few attempts to provide ground-truth datasets, for example by combining extracellular and patch-clamp or juxtacellular recordings (Henze et al. ; Harris et al. ; Neto et al. ; Yger et al. ; Marques-Smith et al. ; Allen et al. ) exist, but the main limitation of this approach is that only one or a few cells can be patched at the same time, providing very limited ground-truth information with respect to the number of neurons that can be recorded simultaneously from extracellular probes. An alternative method consists of adding artificial or previously-sorted and well-isolated spikes in the recordings (hybrid method) (Rossant et al. ; Wouters et al. ). The hybrid approach is convenient as all the characteristics of the underlying recording are kept. However, only a few hybrid units can be added at a time, and this limits the validation capability of this method. Biophysically detailed simulated data provide a powerful alternative and complementary approach to spike sorting validation (Einevoll et al. ). In simulations, recordings can be built from known ground-truth data for all neurons, which allows one to precisely evaluate the performance of spike sorters. Simulators of extracellular activity should be able to replicate important aspects of spiking activity that can be challenging for spike sorting algorithms, including bursting modulation, spatio-temporal overlap of spikes, unit drifts over time, as well as realistic noise models. Moreover, they should allow users to have full control over these features and they should be efficient and fast. While simulated recordings provide ground-truth information of many units at once, it is an open question how realistically they can reproduce real recordings. In the last years, there have been a few projects aiming to develop neural simulators for benchmarking spike sorting methods (Camuñas-Mesa and Quiroga ; Hagen et al. ; Mondragón-González and Burguière ): Camunas et al. developed NeuroCube (Camuñas-Mesa and Quiroga ), a MATLAB-based simulator which combines biophysically detailed cell models and synthetic spike trains to simulate the activity of neurons close to a recording probe, while noise is simulated by the activity of distant neurons. NeuroCube is very easy to use with a simple and intuitive graphical user interface (GUI). The user has direct control of parameters to control the rate of active neurons, their firing rate properties, and the duration of the recordings. The cell models are shipped with the software and recordings can be simulated on a single electrodes or a tetrode. It is relatively fast, but the cell model simulations (using NEURON (Carnevale and Hines )) are re-simulated for every recording. Hagen et al. developed ViSAPy (Hagen et al. ), a Python-based simulator that uses multi-compartment simulation of single neurons to generate spikes, network modeling of point-neurons in NEST (Diesmann and Gewaltig ) to generate synaptic inputs onto the spiking neurons, and experimentally fitted noise. ViSAPy runs a full network simulation in NEURON (Carnevale and Hines ) and computes the extracellular potentials using LFPy (Lindén et al. ; Hagen et al. ). ViSAPy implements a Python application programming interface (API) which allows the user to set multiple parameters for the network simulation providing the synaptic input, the probe design, and the noise model generator. Cell models can be freely chosen and loaded using the LFPy package. Further, 1-dimensional drift can be incorporated in the simulations by shifting the electrodes over time (Franke et al. ). Learning to use the software and, in particular, tailoring the specific properties of the resulting spike trains, for example burstiness, requires some effort by the user. As the running of NEURON simulations with biophysically detailed neurons can be computationally expensive, the use of ViSAPy to generate long-duration spike-sorting benchmarking data is boosted by access to powerful computers. Mondragon et al. developed a Neural Benchmark Simulator (NBS) (Mondragón-González and Burguière ) extending the NeuroCube software. NBS extends the capability of NeuroCube for using user-specific probes, and it combines the spiking activity signals (from NeuroCube), with low-frequency activity signals, and artifacts libraries shipped with the code. The user can set different weight parameters to assemble the spiking, low-frequency, and artifact signals, but these three signal types are not modifiable. Despite the existence of such tools for generating benchmarking data, their use in spike sorting literature has until now been limited, making the benchmarking and validation of spike sorting algorithms non-standardized and unsystematic. A natural question to ask is thus how to best stimulate the use of such benchmarking tools in the spike sorting community. From a spike sorting developer perspective, we argue that an ideal extracellular simulator should be i) fast, ii) controllable, iii) biophysically detailed, and iv) easy to use. A fast simulator would enable spike sorter developers to generate a large and varied set of recordings to test their algorithms against and to improve their spike sorting methods. Controllability refers to the possibility to have direct control of features of the simulated recordings. The ideal extracellular spike simulator should include the possibility to use different cell models and types, to decide the firing properties of the neurons, to control the rate of spatio-temporal spike collisions, to generate recordings on different probe models, and to have full reproducibility of the simulated recordings. A biophysically detailed simulator should be capable of reproducing key physiological aspects of the recordings, including, but not limited to, bursting spikes, drifts between the electrodes and the neurons, and realistic noise profiles. Finally, to maximize the ease of use, the ideal extracellular simulator should be designed as an accessible and easy to learn software package. Preferably, the tool should be implemented with a graphical user interface (GUI), a command line interface (CLI), or with a simple application programming interface (API). With these principles in mind, we present here MEArec, an open-source Python-based simulator. MEArec provides a fast, highly controllable, biophysically detailed, and easy to use framework to generate simulated extracellular recordings. In addition to producing benchmark datasets, we developed MEArec as a powerful tool that can serve as a testbench for optimizing existing and novel spike sorting methods. To facilitate this goal, MEArec allows users to explore how several aspects of recordings affect spike sorting, with full control of challenging features such as bursting activity, drifting, spatio-temporal synchrony, and noise effects, so that spike sorter developers can use it to help their algorithm design. The source code for MEArec is on Github ( https://github.com/alejoe91/MEArec ) and the Python package is on PyPi ( https://pypi.org/project/MEArec/ ). An extensive documentation is available ( https://mearec.readthedocs.io/ ), and the code is tested with a continuous integration platform ( https://travis-ci.org/ ). Moreover, all the datsets generated for this article and used to make figures are available on Zenodo (10.5281/zenodo.3696926). The article is organized as follows: in Section “ ” we introduce the principles of MEArec and we show how to run simulations with the CLI and Python API. In Section “ ” we explain the different features available in MEArec, including the capability of simulating recordings for MEAs, reproducing bursting behavior, controlling spatio-temporal overlaps, reproducing drifts, and replicating biological noise characteristics. In Section “ ” we present the use of MEArec as a testbench for spike sorting development, and its integration with the SpikeInterface framework (Buccino et al. ). In Section “ ” we document the simulation outputs and how to save and load them with the MEArec API. Finally, in Section we discuss the presented software and contextualize it with respect to the state-of-the-art.
We start by describing the principle of the MEArec simulator and showing examples on how to get started with the simulations. The simulation is split in two phases: templates generation (Fig. ) and recordings generation (Fig. ). Templates (or extracellular action potentials - EAPs) are generated using biophysically realistic cell models which are positioned in the surroundings of a probe model. The templates generation phase is further divided into an intracellular and an extracellular simulation. During the intracellular simulation, each cell model is stimulated with a constant current and transmembrane currents of action potentials are computed (using NEURON (Carnevale and Hines )) and stored to disk (the intracellular simulation is the most time consuming part and storing its output to disk enables one to run it only once). The extracellular simulation uses the LFPy package (Lindén et al. ; Hagen et al. ) to compute extracellular potentials generated at the electrodes’ locations using the well-established line-source approximation (see Supplementary Methods – Templates generation – for details). In particular, the cell morphology is loaded and shifted to a random position around the probe. Additionally, the user can add different rotations to the models. When the cell model is shifted and rotated, the previously computed and stored transmembrane currents are loaded and the EAP is computed. This step is repeated several times for each cell model, for different positions and rotations. The templates generation phase outputs a library of a large variety of extracellular templates, which can then be used to build the recordings. The templates generation phase is the most time consuming, but the same template library can be used to generate multiple recordings. It is therefore recommended to simulate many more templates than needed by a single recording, so that the same template library can be used to simulate a virtually infinite number of recordings. MEArec, at installation, comes with 13 layer 5 cortical cell models from the Neocortical Microcircuit Portal (Ramaswamy et al. ). This enables the user to dive into simulations without the need to download and compile cell models. On the other hand, the initial set of cell models can be easily extended as outlined in the Supplementary Methods – Templates generation . To generate 30 extracellular spikes (also referred as templates) per cell model recorded on a shank tetrode probe, the user can simply run this command: The -prb option allows for choosing the probe model, -n controls the number of templates per cell model to generate, and the --seed option is used to ensure reproducibility and if it is not provided, a random seed is chosen. In both cases, the seed is saved in the HDF5 file, so that the same templates can be replicated. Recordings are then generated by combining templates selected with user-defined rules (based on minimum distance between neurons, amplitudes, spatial overlaps, and cell-types) and by simulating spike trains (Supplementary Methods – Recordings generation – for details on spike trains generation and template selection). Selected templates and spike trains are assembled using a customized (or modulated) convolution, which can replicate interesting features of spiking activity such as bursting and drift. After convolution, additive noise is generated and added to the recordings. Finally, the output recordings can be optionally filtered with a band-pass or a high-pass filter. Note that filtering the recordings will affect the shape and amplitude of the spike waveforms, but this is a common procedure in spike sorting to remove lower frequency components. Recordings can be generated with the CLI as follows: The gen-recordings command combines the selected templates from 4 excitatory cells (-ne 4) and 2 inhibitory cells (-ni 2), that usually have a more narrow spike waveform and a higher firing rate, with randomly generated spike trains. The duration of the output recordings is 30 seconds (-d 30). In this case, four random seeds control the spike train random generation (--st-seed 0), the template selection (--temp-seed 1), the noise generation (--noise-seed 2), and the convolution process (--conv-seed 3). Figure shows one second of the generated recordings (A), the extracted waveforms and the mean waveforms for each unit on the electrode with the largest peak (B), and the principal component analysis (PCA) projections of the waveforms on the tetrode channels. MEArec also implements a convenient Python API, which is run internally by the CLI commands. For example, the following snippet of code implements the same commands shown above for generating templates and recordings: Moreover, the Python API implements plotting functions to visually inspect the simulated templates and recordings. For example, Fig. panels were generated using the plot_recordings (A), plot_waveforms (B), and plot_pca_map (C) functions. MEArec is designed to allow for full customization, transparency, and reproducibility of the simulated recordings. Parameters for the templates and recordings generation are accessible by the user and documented, so that different aspects of the simulated signals can be finely tuned (see Supplementary Methods for a list of parameters and their explanation). Moreover, the implemented command line interface (CLI) and simple Python API, enable the user to easily modify parameters, customize, and run simulations. Finally, MEArec permits to manually set several random seeds used by the simulator to make recordings fully reproducible. This feature also enables one to study how separate characteristics of the recordings affect the spike sorting performance. As an example, we will show in the next sections how to simulate a recording sharing all parameters, hence with exactly the same spiking activity, but with different noise levels or drifting velocities.
Generation of realistic Multi-Electrode Array recordings The recent development of Multi-Electrode Arrays (MEAs) enables researchers to record extracellular activity at very high spatio-temporal density both for in vitro (Berdondini et al. ; Frey et al. ) and in vivo applications (Neto et al. ; Jun et al. ). The large number of electrodes and their high density can result in challenges for spike sorting algorithms. It is therefore important to be able to simulate recordings from these kind of neural probes. To deal with different probe designs, MEArec uses another Python package (MEAutility - https://meautility.readthedocs.io/ ), that allows users to easily import several available probe models and to define custom probe designs. Among others, MEAutility include Neuropixels probes (Jun et al. ), Neuronexus commercial probes ( http://neuronexus.com/products/neural-probes/ ), and a wide variety of square MEA designs with different contact densities (the list of available probes can be found using the mearec available-probes command). Similarly to the tetrode example, we first have to generate templates for the probes. These are the commands to generate templates and recordings for a Neuropixels design with 128 electrodes (Neuropixels-128). The recordings contain 60 neurons, 48 excitatory and 12 inhibitory. With similar commands, we generated templates and recordings for a Neuronexus probe with 32 channels (A1x32-Poly3-5mm-25s-177-CM32 - Neuronexus-32) with 20 cells (16 excitatory and 4 inhibitory), and a square 10x10 MEA with 15 μ m inter-electrode-distance (SqMEA-10-15) and 50 cells (40 excitatory and 10 inhibitory). Figure shows the three above-mentioned probes (A), a sample template for each probe design (B), and one-second snippets of the three recordings (C-D-E), with zoomed in windows to highlight spiking activity. While all the recordings shown so far have been simulated with default parameters, several aspects of the spiking activity are critical for spike sorting. In the next sections, we will show how these features, including bursting, spatio-temporal overlapping spikes, drift, and noise assumptions can be explored with MEArec simulations. Bursting modulation of spike amplitude and shape Bursting activity is one of the most complicated features of spiking activity that can compromise the performance of spike sorting algorithms. When a neuron bursts, i.e., fires a rapid train of action potentials with very short inter-spike intervals, the dynamics underlying the generation of the spikes changes over the bursting period (Hay et al. ). While the bursting mechanism has been largely studied with patch-clamp experiments, combined extracellular-juxtacellular recordings (Allen et al. ) and computational studies (Hagen et al. ) suggest that during bursting, extracellular spikes become lower in amplitude and wider in shape. In order to simulate this property of the extracellular waveforms in a fast and efficient manner, templates can be modulated both in amplitude and shape during the convolution operation, depending on the spiking history. To demonstrate how bursting is mimicked, we built a toy example with a constant spike train with 10 ms inter-spike-interval (Fig. ). A modulation value is computed for each spike and it is used to modulate the waveform for that event by scaling its amplitude, and optionally stretching its shape. The blue dots show the default modulation (bursting disabled), in which the modulation values are drawn from a Gaussian distribution with unitary mean to add some physiological variation to the spike waveforms. When bursting is enabled (by setting the bursting parameter to true), the modulation values are computed based on the spike history, and it depends on the number of consecutive spikes in a bursting event and their average inter-spike-intervals (see Supplementary Methods – Recordings generation - Modulated convolution – for details on the modulation values calculation). Bursting events can be either controlled by the maximum number of spikes making a burst (orange dots - 5 spikes per burst; green dots - 10 spikes per burst) or by setting a maximum bursting duration (red dots - maximum 75 ms ). Note that in Fig. the spike train is constant just to illustrate the computation of the modulation values. In actual simulations, instead, the modulation values will depend on the firing rate and the timing between spikes. By default, spikes are only modulated in amplitude. The user can also enable shape modulation by setting the shape_mod parameter to true. The modulation value, computed for each spike, controls both the amplitude scaling and shape modulation of the spike event. For amplitude modulation, the amplitude of the spike is simply multiplied by the modulation value. Additionally, when shape modulation is enabled, the waveform of each spike is also stretched. The shape_stretch parameter controls the overall amount of stretch, but the actual stretch of single waveforms depends on the modulation value computed for each spike. In Fig. , examples of bursting templates are shown. The blue traces display templates only modulated in amplitude, i.e., the amplitude is scaled by the modulation value. The orange and green traces, instead, also present shape modulation, with different values of the shape_stretch parameter (the higher the shape_stretch, the more stretched waveforms will be). We refer to the Supplementary Methods – Recordings generation - Modulated convolution – for further details on amplitude and shape modulation. Figure shows a one-second snippet of the tetrode recording shown previously after bursting modulation is activated. The top panel shows the spike events, the middle one displays the modulation values computed for each spike, and the bottom panel shows the output of the modulated convolution between one of the templates (on the electrode with the largest amplitude) and the spike train. Figures and e show the waveform projections on the first principal component of each channel for the tetrode recording shown in Section with and without bursting enabled, respectively. In this case all neurons are bursting units and this causes a stretch in the PCA space, which is a clear complication for spike sorting algorithms. Note that shape modulation does not affect all neurons by the same amount, since it depends on the spike history and therefore on the firing rate. Controlling spatio-temporal overlaps Another complicated aspect of extracellular spiking activity that can influence spike sorting performance is the occurrence of overlapping spikes. While temporal overlapping of events on spatially separated locations can be solved with feature masking (Rossant et al. ), spatio-temporal overlapping can cause a distortion of the detected waveform, due to the superposition of separate spikes. Some spike sorting approaches, based on template-matching, are designed to tackle this problem (Pachitariu et al. ; Yger et al. ; Diggelmann et al. ). In order to evaluate to what extent spatio-temporal overlap affects spike sorting, MEArec allows the user to set the number of spatially overlapping templates and to modify the synchrony rate of their spike trains. In Fig. we show an example of this on a Neuronexus-32 probe (see Fig. A). The recording was constructed with two excitatory and spatially overlapping neurons, whose templates are shown in Fig. (see Supplementary Methods – Recordings generation - Overlapping spikes and spatio-temporal synchrony – for details on the spatial overlap definition). The spike synchrony rate can be controlled with the sync_rate parameter. If this parameter is not set (Fig. - left), some spatio-temporal overlapping spikes are present (red events). If the synchrony rate is set to 0, those spikes are removed from the spike trains (Fig. - middle). If set to 0.05, i.e., 5% of the spikes will be spatio-temporal collisions, events are added to the spike trains to reach the specified synchrony rate value of spatio-temporal overlap. As shown in Fig. , the occurrence of spatio-temporal overlapping events affects the recorded extracellular waveform: the waveforms of the neurons, in fact, get summed and might be mistaken for a separate unit by spike sorting algorithms when the spikes are overlapping. The possibility of reproducing and controlling this feature of extracellular recordings within MEArec could aid in the development of spike sorters which are robust to spatio-temporal collisions. Generating drifting recordings When extracellular probes are inserted in the brain, especially for acute experiments, the neural tissue might move with respect to the electrodes. This phenomenon is known as drift. Drift can be due to a slow relaxation of the tissue (slow drift) or to fast re-adjustments of the tissue, for example due to an abrupt motion of the tissue (fast drift). These two types of drifts can also be observed in tandem (Pachitariu et al. ). Drifting units are particularly critical for spike sorting, as the waveform shapes change over time due to the relative movement between the neurons and the probe. New spike sorting algorithms have been developed to specifically tackle the drifting problem (e.g. Kilosort2 (Pachitariu et al. ), IronClust (Jun et al. )). In order to simulate drift in the recordings, we first need to generate drifting templates: Drifting templates are generated by choosing an initial and final soma position with user-defined rules (see Supplementary Methods – Template generation - Drifing templates – for details) and by moving the cell along the line connecting the two positions for a defined number of constant drifting steps that span the segment connecting the initial and final positions (30 steps by default). An example of a drifting template is depicted in Fig. , alongside with the drifting neuron’s soma locations for the different drifting steps. Once a library of drifting templates is generated, drifting recordings can be simulated. MEArec allows users to simulate recordings with three types of drift modes: slow , fast , and slow+fast . When slow drift is selected, the drifting template is selected over time depending on the initial position and the drifting velocity (5 μ / m i n by default). If the final drifting position is reached, the drift direction is reversed. For fast drifts, the position of a drifting neuron is shifted abruptly with a user-defined period (every 20 s by default). The new position is chosen so that the difference in waveform amplitude of the drifting neuron on its current maximum channel remains within user-defined limits (5-20 μ V by default), in order to prevent from moving the neuron too far from its previous position. The slow+fast mode combines the slow and fast mechanisms. In Fig. and c we show examples of slow drift and fast drift, respectively. In the top panel the recordings are displayed, with superimposed drifting templates on the electrode with the largest peak. Note that the maximum channel can change over time due to drift. In the bottom panels, instead, the amplitude of the waveforms on the channels with the initial largest peak for each neuron are shown over time. Slow drift causes the amplitude to slowly vary, while for fast drifts we observe more abrupt changes when a fast drift event occurs. In the slow+fast drift mode, these two effects are combined. Modeling experimental noise Spike sorting performance can be greatly affected by noise in the recordings. Many algorithms first use a spike detection step to identify putative spikes. The threshold for spike detection is usually set depending on the noise standard deviation or median absolute deviation (Quiroga et al. ). Clearly, recordings with larger noise levels will result in higher spike detection thresholds, hence making it harder to robustly detect lower amplitude spiking activity. In addition to the noise amplitude, other noise features can affect spike sorting performance: some clustering algorithms, for example, assume that clusters have Gaussian shape, due to the assumption of an additive normal noise to the recordings. Moreover, the noise generated by biological sources can produce spatial correlations in the noise profiles among different channels and it can be modulated in frequency (Camuñas-Mesa and Quiroga ; Rey et al. ). To investigate how the above-mentioned assumptions on noise can affect spike sorting performance, MEArec can generate recordings with several noise models. Figure shows 5-second spiking-free recordings of a tetrode probe for five different noise profiles that can be generated (A - recordings, B - spectrum, C - channel covariance, D - amplitude distribution). The first column shows uncorrelated Gaussian noise, which presents a flat spectrum, a diagonal covariance matrix, and a symmetrical noise amplitude distribution. In the recording in the second column, spatially correlated noise was generated as a multivariate Gaussian noise with a covariance matrix depending on the channel distance. Also in this case, the spectrum (B) presents a flat profile and the amplitude distribution is symmetrical (D), but the covariance matrix shows a correlation depending on the inter-electrode distance. As previous studies showed (Camuñas-Mesa and Quiroga ; Rey et al. ), the frequency content of extracellular noise is not flat, but its spectrum is affected by the spiking activity of distant neurons, which appear in the recordings as below-threshold biological noise. To reproduce the spectrum profile that is observed in experimental data, MEArec allows coloring the noise spectrum of Gaussian noise with a second order infinite impulse response (IIR) filter (see Supplementary Methods – Recordings generation - Noise models and post-processing – for details). Colored noise represents an efficient way of obtaining the desired spectrum, as shown in the third and fourth columns of Fig. , panel B. Distance correlation is maintained (panel C - fourth column), and the distribution of the noise amplitudes is symmetrical. Finally, a last noise model enables one to generate activity of distant neurons. In this case, noise is built as the convolution between many neurons (300 by default) whose template amplitudes are below an amplitude threshold (10 μ V by default). A Gaussian noise floor is then added to the resulting noise, which is scaled to match the user-defined noise level. The far-neurons noise profile is shown in the last column of Fig. . While the spectrum and spatial correlation of this noise profile are similar to the ones generated with a colored, distance-correlated noise (4th column), the shape of the noise distribution is skewed towards negative values (panel D), mainly due to the negative contribution of the action potentials. The capability of MEArec to simulate several noise models enables spike sorter developers to assess how different noise profiles affect their algorithms and to modify their methods to be insensitive to specific noise assumptions.
The recent development of Multi-Electrode Arrays (MEAs) enables researchers to record extracellular activity at very high spatio-temporal density both for in vitro (Berdondini et al. ; Frey et al. ) and in vivo applications (Neto et al. ; Jun et al. ). The large number of electrodes and their high density can result in challenges for spike sorting algorithms. It is therefore important to be able to simulate recordings from these kind of neural probes. To deal with different probe designs, MEArec uses another Python package (MEAutility - https://meautility.readthedocs.io/ ), that allows users to easily import several available probe models and to define custom probe designs. Among others, MEAutility include Neuropixels probes (Jun et al. ), Neuronexus commercial probes ( http://neuronexus.com/products/neural-probes/ ), and a wide variety of square MEA designs with different contact densities (the list of available probes can be found using the mearec available-probes command). Similarly to the tetrode example, we first have to generate templates for the probes. These are the commands to generate templates and recordings for a Neuropixels design with 128 electrodes (Neuropixels-128). The recordings contain 60 neurons, 48 excitatory and 12 inhibitory. With similar commands, we generated templates and recordings for a Neuronexus probe with 32 channels (A1x32-Poly3-5mm-25s-177-CM32 - Neuronexus-32) with 20 cells (16 excitatory and 4 inhibitory), and a square 10x10 MEA with 15 μ m inter-electrode-distance (SqMEA-10-15) and 50 cells (40 excitatory and 10 inhibitory). Figure shows the three above-mentioned probes (A), a sample template for each probe design (B), and one-second snippets of the three recordings (C-D-E), with zoomed in windows to highlight spiking activity. While all the recordings shown so far have been simulated with default parameters, several aspects of the spiking activity are critical for spike sorting. In the next sections, we will show how these features, including bursting, spatio-temporal overlapping spikes, drift, and noise assumptions can be explored with MEArec simulations.
Bursting activity is one of the most complicated features of spiking activity that can compromise the performance of spike sorting algorithms. When a neuron bursts, i.e., fires a rapid train of action potentials with very short inter-spike intervals, the dynamics underlying the generation of the spikes changes over the bursting period (Hay et al. ). While the bursting mechanism has been largely studied with patch-clamp experiments, combined extracellular-juxtacellular recordings (Allen et al. ) and computational studies (Hagen et al. ) suggest that during bursting, extracellular spikes become lower in amplitude and wider in shape. In order to simulate this property of the extracellular waveforms in a fast and efficient manner, templates can be modulated both in amplitude and shape during the convolution operation, depending on the spiking history. To demonstrate how bursting is mimicked, we built a toy example with a constant spike train with 10 ms inter-spike-interval (Fig. ). A modulation value is computed for each spike and it is used to modulate the waveform for that event by scaling its amplitude, and optionally stretching its shape. The blue dots show the default modulation (bursting disabled), in which the modulation values are drawn from a Gaussian distribution with unitary mean to add some physiological variation to the spike waveforms. When bursting is enabled (by setting the bursting parameter to true), the modulation values are computed based on the spike history, and it depends on the number of consecutive spikes in a bursting event and their average inter-spike-intervals (see Supplementary Methods – Recordings generation - Modulated convolution – for details on the modulation values calculation). Bursting events can be either controlled by the maximum number of spikes making a burst (orange dots - 5 spikes per burst; green dots - 10 spikes per burst) or by setting a maximum bursting duration (red dots - maximum 75 ms ). Note that in Fig. the spike train is constant just to illustrate the computation of the modulation values. In actual simulations, instead, the modulation values will depend on the firing rate and the timing between spikes. By default, spikes are only modulated in amplitude. The user can also enable shape modulation by setting the shape_mod parameter to true. The modulation value, computed for each spike, controls both the amplitude scaling and shape modulation of the spike event. For amplitude modulation, the amplitude of the spike is simply multiplied by the modulation value. Additionally, when shape modulation is enabled, the waveform of each spike is also stretched. The shape_stretch parameter controls the overall amount of stretch, but the actual stretch of single waveforms depends on the modulation value computed for each spike. In Fig. , examples of bursting templates are shown. The blue traces display templates only modulated in amplitude, i.e., the amplitude is scaled by the modulation value. The orange and green traces, instead, also present shape modulation, with different values of the shape_stretch parameter (the higher the shape_stretch, the more stretched waveforms will be). We refer to the Supplementary Methods – Recordings generation - Modulated convolution – for further details on amplitude and shape modulation. Figure shows a one-second snippet of the tetrode recording shown previously after bursting modulation is activated. The top panel shows the spike events, the middle one displays the modulation values computed for each spike, and the bottom panel shows the output of the modulated convolution between one of the templates (on the electrode with the largest amplitude) and the spike train. Figures and e show the waveform projections on the first principal component of each channel for the tetrode recording shown in Section with and without bursting enabled, respectively. In this case all neurons are bursting units and this causes a stretch in the PCA space, which is a clear complication for spike sorting algorithms. Note that shape modulation does not affect all neurons by the same amount, since it depends on the spike history and therefore on the firing rate.
Another complicated aspect of extracellular spiking activity that can influence spike sorting performance is the occurrence of overlapping spikes. While temporal overlapping of events on spatially separated locations can be solved with feature masking (Rossant et al. ), spatio-temporal overlapping can cause a distortion of the detected waveform, due to the superposition of separate spikes. Some spike sorting approaches, based on template-matching, are designed to tackle this problem (Pachitariu et al. ; Yger et al. ; Diggelmann et al. ). In order to evaluate to what extent spatio-temporal overlap affects spike sorting, MEArec allows the user to set the number of spatially overlapping templates and to modify the synchrony rate of their spike trains. In Fig. we show an example of this on a Neuronexus-32 probe (see Fig. A). The recording was constructed with two excitatory and spatially overlapping neurons, whose templates are shown in Fig. (see Supplementary Methods – Recordings generation - Overlapping spikes and spatio-temporal synchrony – for details on the spatial overlap definition). The spike synchrony rate can be controlled with the sync_rate parameter. If this parameter is not set (Fig. - left), some spatio-temporal overlapping spikes are present (red events). If the synchrony rate is set to 0, those spikes are removed from the spike trains (Fig. - middle). If set to 0.05, i.e., 5% of the spikes will be spatio-temporal collisions, events are added to the spike trains to reach the specified synchrony rate value of spatio-temporal overlap. As shown in Fig. , the occurrence of spatio-temporal overlapping events affects the recorded extracellular waveform: the waveforms of the neurons, in fact, get summed and might be mistaken for a separate unit by spike sorting algorithms when the spikes are overlapping. The possibility of reproducing and controlling this feature of extracellular recordings within MEArec could aid in the development of spike sorters which are robust to spatio-temporal collisions.
When extracellular probes are inserted in the brain, especially for acute experiments, the neural tissue might move with respect to the electrodes. This phenomenon is known as drift. Drift can be due to a slow relaxation of the tissue (slow drift) or to fast re-adjustments of the tissue, for example due to an abrupt motion of the tissue (fast drift). These two types of drifts can also be observed in tandem (Pachitariu et al. ). Drifting units are particularly critical for spike sorting, as the waveform shapes change over time due to the relative movement between the neurons and the probe. New spike sorting algorithms have been developed to specifically tackle the drifting problem (e.g. Kilosort2 (Pachitariu et al. ), IronClust (Jun et al. )). In order to simulate drift in the recordings, we first need to generate drifting templates: Drifting templates are generated by choosing an initial and final soma position with user-defined rules (see Supplementary Methods – Template generation - Drifing templates – for details) and by moving the cell along the line connecting the two positions for a defined number of constant drifting steps that span the segment connecting the initial and final positions (30 steps by default). An example of a drifting template is depicted in Fig. , alongside with the drifting neuron’s soma locations for the different drifting steps. Once a library of drifting templates is generated, drifting recordings can be simulated. MEArec allows users to simulate recordings with three types of drift modes: slow , fast , and slow+fast . When slow drift is selected, the drifting template is selected over time depending on the initial position and the drifting velocity (5 μ / m i n by default). If the final drifting position is reached, the drift direction is reversed. For fast drifts, the position of a drifting neuron is shifted abruptly with a user-defined period (every 20 s by default). The new position is chosen so that the difference in waveform amplitude of the drifting neuron on its current maximum channel remains within user-defined limits (5-20 μ V by default), in order to prevent from moving the neuron too far from its previous position. The slow+fast mode combines the slow and fast mechanisms. In Fig. and c we show examples of slow drift and fast drift, respectively. In the top panel the recordings are displayed, with superimposed drifting templates on the electrode with the largest peak. Note that the maximum channel can change over time due to drift. In the bottom panels, instead, the amplitude of the waveforms on the channels with the initial largest peak for each neuron are shown over time. Slow drift causes the amplitude to slowly vary, while for fast drifts we observe more abrupt changes when a fast drift event occurs. In the slow+fast drift mode, these two effects are combined.
Spike sorting performance can be greatly affected by noise in the recordings. Many algorithms first use a spike detection step to identify putative spikes. The threshold for spike detection is usually set depending on the noise standard deviation or median absolute deviation (Quiroga et al. ). Clearly, recordings with larger noise levels will result in higher spike detection thresholds, hence making it harder to robustly detect lower amplitude spiking activity. In addition to the noise amplitude, other noise features can affect spike sorting performance: some clustering algorithms, for example, assume that clusters have Gaussian shape, due to the assumption of an additive normal noise to the recordings. Moreover, the noise generated by biological sources can produce spatial correlations in the noise profiles among different channels and it can be modulated in frequency (Camuñas-Mesa and Quiroga ; Rey et al. ). To investigate how the above-mentioned assumptions on noise can affect spike sorting performance, MEArec can generate recordings with several noise models. Figure shows 5-second spiking-free recordings of a tetrode probe for five different noise profiles that can be generated (A - recordings, B - spectrum, C - channel covariance, D - amplitude distribution). The first column shows uncorrelated Gaussian noise, which presents a flat spectrum, a diagonal covariance matrix, and a symmetrical noise amplitude distribution. In the recording in the second column, spatially correlated noise was generated as a multivariate Gaussian noise with a covariance matrix depending on the channel distance. Also in this case, the spectrum (B) presents a flat profile and the amplitude distribution is symmetrical (D), but the covariance matrix shows a correlation depending on the inter-electrode distance. As previous studies showed (Camuñas-Mesa and Quiroga ; Rey et al. ), the frequency content of extracellular noise is not flat, but its spectrum is affected by the spiking activity of distant neurons, which appear in the recordings as below-threshold biological noise. To reproduce the spectrum profile that is observed in experimental data, MEArec allows coloring the noise spectrum of Gaussian noise with a second order infinite impulse response (IIR) filter (see Supplementary Methods – Recordings generation - Noise models and post-processing – for details). Colored noise represents an efficient way of obtaining the desired spectrum, as shown in the third and fourth columns of Fig. , panel B. Distance correlation is maintained (panel C - fourth column), and the distribution of the noise amplitudes is symmetrical. Finally, a last noise model enables one to generate activity of distant neurons. In this case, noise is built as the convolution between many neurons (300 by default) whose template amplitudes are below an amplitude threshold (10 μ V by default). A Gaussian noise floor is then added to the resulting noise, which is scaled to match the user-defined noise level. The far-neurons noise profile is shown in the last column of Fig. . While the spectrum and spatial correlation of this noise profile are similar to the ones generated with a colored, distance-correlated noise (4th column), the shape of the noise distribution is skewed towards negative values (panel D), mainly due to the negative contribution of the action potentials. The capability of MEArec to simulate several noise models enables spike sorter developers to assess how different noise profiles affect their algorithms and to modify their methods to be insensitive to specific noise assumptions.
In the previous sections, we have shown several examples on how MEArec is capable of reproducing several aspects of extracellular recordings which are critical for spike sorting performance, in a fully reproducible way. The proposed design and its integration with a spike sorting evaluation framework called SpikeInterface (Buccino et al. ) enables developers to actively include customized simulations in the spike sorting development phase. Due to its speed and controllability, we see MEArec as a testbench , rather than a benchmark tool. We provide here a couple of examples. In Fig. , we show a one-second section of recordings simulated on a Neuronexus-32 probe with fixed parameters and random seeds regarding template selection and spike train generation, but with four different levels of additive Gaussian noise, with standard deviations of 5, 10, 20, and 30 μ V . The traces show the same underlying spiking activity, so the only variability in spike sorting performance will be due to the varying noise levels. Similarly, in Fig. , 1-minute drifting recordings were simulated with three different drifting velocities. The recordings show that for low drifting speeds the waveform changes are almost not visible (green traces), while for faster drifts (orange and blue traces), the waveform changes over time become more important. The capability of MEArec of reproducing such behaviors in a highly controlled manner could aid in the design of specific tests for measuring and quantifying the ability of a spike sorting software to deal with specific complexities in extracellular recordings. Other examples include simulating a recording with increasing levels of bursting in order to measure to what extent bursting units are correctly clustered, or changing the synchrony rate of spatially overlapping units to assess how much spatio-temporal collisions affect performance. Integration with SpikeInterface We have recently developed SpikeInterface (Buccino et al. ), a Python-based framework for running several spike sorting algorithms, comparing, and validating their results. MEArec can be easily interfaced to SpikeInterface so that simulated recordings can be loaded, spike sorted, and benchmarked with a few lines of code. In the following example, a MEArec recording is loaded, spike sorted with Mountainsort4 (Chung et al. ) and Kilosort2 (Pachitariu et al. ), and benchmarked with respect to the ground-truth spike times available from the MEArec simulation: The get_performance function returns the accuracy, precision, and recall for all the ground-truth units in the MEArec recording. For further details on these metrics and a more extensive characterization of the comparison we refer to the SpikeInterface documentation and article (Buccino et al. ). The combination of MEArec and SpikeInterface represents a powerful tool for systematically testing and comparing spike sorter performances with respect to several complications of extracellular recordings. MEArec simulations, in combination with SpikeInterface, are already being used to benchmark and compare spike sorting algorithms within the SpikeForest project (Magland et al. ). Performance considerations As a testbench tool, the speed requirement has been one of the main design principle of MEArec. In order to achieve high speed, most parts of the simulation process are fully parallelized. As shown in Fig. , the simulations are split in templates and recordings generation. The templates generation phase is the most time consuming, but the same template library can be used to generate several recordings. This phase is further split in two sub-phases: the intracellular and extracellular simulations. The former only needs to be run once, as it generates a set of cell model-specific spikes that are stored and then used for extracellular simulations, which is instead probe specific. We present here run times for the different phases of the templates generation and for the recordings generation. All simulations were run on an Ubuntu 18.04 Intel(R) Core(TM) i7-6600U CPU @ 2.60GHz, with 16 GB of RAM. The intracellular simulation run time for the 13 cell models shipped with the software was [12pt]{minimal}
$ 130$ ∼ 130 seconds ( [12pt]{minimal}
$ 10$ ∼ 10 seconds per cell model). Run times for extracellular simulations for several probe types, number of templates in the library, and drifting templates are shown in the Templates generation section of Table . The run times for this phase mainly depend on the number of templates to be generated (N templates column), on the minimum amplitude of accepted templates (Min. amplitude column, see Supplementary Methods – Templates generation - Extracellular simulation – for further details), and especially on drift (Drifting column). When simulating drifting templates, in fact, the number of actual extracellular spikes for each cell model is N templates times N drift steps. Note that in order to generate the far-neurons noise model, the minimum amplitude should be set to 0, so that low-amplitude templates are not discarded. The number of templates available in the template library will be the specified number of templates (N templates) times the number of cell models (13 by default). Recordings are then generated using the simulated template libraries. In Table , the Recordings generation section shows run times for several recordings with different probes, durations, number of cells, bursting, and drifting options. The main parameter that affects simulation times is the number of cells, as it increases the number of modulated convolutions. Bursting and drifting behavior also increase the run time of the simulations, because of the extra processing required in the convolution step. The simulation run times, however, range from a few seconds to a few minutes. Therefore, the speed of MEArec enables users to generate numerous recordings with different parameters for testing spike sorter performances. Moreover, the software internally uses memory maps to reduce the RAM usage and the simulations can be chunked in time. These features enable users to simulate long recordings on probes with several hundreds of electrodes (e.g. Neuropixels probes) without the need of large-memory nodes or high-performance computing platforms.
We have recently developed SpikeInterface (Buccino et al. ), a Python-based framework for running several spike sorting algorithms, comparing, and validating their results. MEArec can be easily interfaced to SpikeInterface so that simulated recordings can be loaded, spike sorted, and benchmarked with a few lines of code. In the following example, a MEArec recording is loaded, spike sorted with Mountainsort4 (Chung et al. ) and Kilosort2 (Pachitariu et al. ), and benchmarked with respect to the ground-truth spike times available from the MEArec simulation: The get_performance function returns the accuracy, precision, and recall for all the ground-truth units in the MEArec recording. For further details on these metrics and a more extensive characterization of the comparison we refer to the SpikeInterface documentation and article (Buccino et al. ). The combination of MEArec and SpikeInterface represents a powerful tool for systematically testing and comparing spike sorter performances with respect to several complications of extracellular recordings. MEArec simulations, in combination with SpikeInterface, are already being used to benchmark and compare spike sorting algorithms within the SpikeForest project (Magland et al. ).
As a testbench tool, the speed requirement has been one of the main design principle of MEArec. In order to achieve high speed, most parts of the simulation process are fully parallelized. As shown in Fig. , the simulations are split in templates and recordings generation. The templates generation phase is the most time consuming, but the same template library can be used to generate several recordings. This phase is further split in two sub-phases: the intracellular and extracellular simulations. The former only needs to be run once, as it generates a set of cell model-specific spikes that are stored and then used for extracellular simulations, which is instead probe specific. We present here run times for the different phases of the templates generation and for the recordings generation. All simulations were run on an Ubuntu 18.04 Intel(R) Core(TM) i7-6600U CPU @ 2.60GHz, with 16 GB of RAM. The intracellular simulation run time for the 13 cell models shipped with the software was [12pt]{minimal}
$ 130$ ∼ 130 seconds ( [12pt]{minimal}
$ 10$ ∼ 10 seconds per cell model). Run times for extracellular simulations for several probe types, number of templates in the library, and drifting templates are shown in the Templates generation section of Table . The run times for this phase mainly depend on the number of templates to be generated (N templates column), on the minimum amplitude of accepted templates (Min. amplitude column, see Supplementary Methods – Templates generation - Extracellular simulation – for further details), and especially on drift (Drifting column). When simulating drifting templates, in fact, the number of actual extracellular spikes for each cell model is N templates times N drift steps. Note that in order to generate the far-neurons noise model, the minimum amplitude should be set to 0, so that low-amplitude templates are not discarded. The number of templates available in the template library will be the specified number of templates (N templates) times the number of cell models (13 by default). Recordings are then generated using the simulated template libraries. In Table , the Recordings generation section shows run times for several recordings with different probes, durations, number of cells, bursting, and drifting options. The main parameter that affects simulation times is the number of cells, as it increases the number of modulated convolutions. Bursting and drifting behavior also increase the run time of the simulations, because of the extra processing required in the convolution step. The simulation run times, however, range from a few seconds to a few minutes. Therefore, the speed of MEArec enables users to generate numerous recordings with different parameters for testing spike sorter performances. Moreover, the software internally uses memory maps to reduce the RAM usage and the simulations can be chunked in time. These features enable users to simulate long recordings on probes with several hundreds of electrodes (e.g. Neuropixels probes) without the need of large-memory nodes or high-performance computing platforms.
The templates generation outputs a Template Generator object, containing the following fields: templates contains the generated templates – array with shape (n_templates, n_electrodes, n_points) for non-drifting templates or (n_templates, n_drift_steps, n_electrodes, n_points) for drifting ones locations contains the 3D soma locations of the templates – array with shape (n_templates, 3) for non-drifting templates or (n_templates, n_drift_steps, 3) for drifting templates. rotations contains the 3D rotations applied to the cell model before computing the template – array with shape (n_templates, 3) (for drifting templates rotation is fixed) celltypes contains the cell types of the generated templates – array of strings with length (n_templates) info contains a dictionary with parameters used for the simulation (params key) and information about the probe (electrodes key) The recordings generation outputs a Recording Generator object, containing the following fields: recordings contains the generated recordings – array with shape (n_electrodes, n_samples) spiketrains contains the spike trains – list of (n_neurons) neo.Spiketrain objects (Garcia et al. ) templates contains the selected templates – array with shape (n_neurons, n_jitters, n_electrodes, n_templates samples) templates for non-drifting recordings - or (n_neurons, n_drift_steps, n_jitters, n_electrodes, n_neurons) for drifting ones templates_celltypes contains the cell type of the selected templates – array of strings with length (n_neurons) templates_locations contains the 3D soma locations of the selected templates – array with shape (n_neurons, 3) for non-drifting recordings or (n_neurons, n_drift_steps, 3) for drifting ones templates_rotations contains the 3D rotations applied to the selected templates – array with shape (n_neurons, 3) channel_positions contains the 3D positions of the probe electrodes – array with shape (n_electrodes, 3) timestamps contains the timestamps in seconds – array with length (n_samples) voltage_peaks contains the average voltage peaks of the templates on each electrode – array with shape (n_neurons, n_electrodes) spike_traces contains a clean spike trace for each neuron (generated by a clean convolution between the spike train and the template on the electrode with the largest peak) – array with shape (n_neurons, n_samples) info contains a dictionary with parameters used for the simulation When simulating with the Python API, the returned TemplateGenerator and RecordingGenerator can be saved as .h5 files with: The generation using the CLI saves templates and recordings directly. The saved templates and recordings can be loaded in Python as TemplateGenerator and RecordingGenerator objects with:
In this paper we have presented MEArec, a Python package for simulating extracellular recordings for spike sorting development and validation. We first introduced an overview of the software function, consisting in separating the templates and the recordings generation to improve efficiency and simulation speed. We then showed the ease of use of the software, whose command line interface and simple Python API enable users to simulate extracellular recordings with a couple of commands or a few lines of code. We explored the capability of reproducing and controlling several aspects of extracellular recordings which can be critical for spike sorting algorithms, including spikes in a burst with varying spike shapes, spatio-temporal overlaps, drifting units, and noise assumptions. We illustrated two examples of using MEArec, in combination with SpikeInterface (Buccino et al. ), as a testbench platform for developing spike sorting algorithms. Finally, we benchmarked the speed performance of MEArec (Table ). Investigating the validation section of several recently developed spike sorting algorithms (Rossant et al. ; Pachitariu et al. ; Jun et al. ; Hilgen et al. ; Jun et al. ; Lee et al. ; Yger et al. ), it is clear that the neuroscientific community needs a standardized validation framework for spike sorting performance. Some spike sorters are validated using a so called hybrid approach, in which well-identified units from previous experimental recordings are artificially injected in the recordings and used to compute performance metrics (Rossant et al. ; Pachitariu et al. ; Wouters et al. ). The use of templates extracted from previously sorted datasets poses some questions regarding the accuracy of the initial sorting, as well as the complexity of the well-identified units. Alternatively, other spike sorters are validated on experimental paired ground-truth recordings (Chung et al. ; Yger et al. ). While these valuable datasets (Harris et al. ; Henze et al. ; Neto et al. ; Marques-Smith et al. ) can certainly provide useful information, the low count of ground-truth units makes the validation incomplete and could result in biases (for example algorithm-specific parameters could be tuned to reach a higher performance for the recorded ground-truth units). A third validation method consist of using simulated ground-truth recordings (Einevoll et al. ). While this approach is promising, in combination with experimental paired recordings, the current available simulators (Camuñas-Mesa and Quiroga ; Hagen et al. ; Mondragón-González and Burguière ) present some limitations in terms of biological realism, controllability, speed, and/or ease of use (see Introduction). We therefore introduced MEArec, a software package which is computationally efficient, easy to use, highly controllable, and capable of reproducing critical characteristics of extracellular recordings relevant to spike sorting, including bursting modulation, spatio-temporal overlaps, drift of units over time, and various noise profiles. The capability of MEArec to replicate complexities in extracellular recordings which are usually either ignored or not controlled in other simulators, permits the user to include tailored simulations in the spike sorting implementation process, using the simulator as a testbench platform for algorithm development. MEArec simulations could not only be used to test the final product, but specific simulations could be used to help implementing algorithms that are able to cope with drifts, bursting, and spatio-temporal overlap, which are regarded as the most complex aspects for spike sorting performance (Rey et al. ; Yger et al. ). In MEArec, in order to generate extracellular templates, we used a well-established modeling framework for solving the single neuron dynamics (Carnevale and Hines ), and for calculating extracellular fields generated by transmembrane currents (Lindén et al. ; Hagen et al. ). These models have some assumptions that, if warranted, could be addressed with more sophisticated methods, such as finite element methods (FEM). In a recent work (Buccino et al. ), we used FEM simulations and showed that the extracellular probes, especially MEAs, affect the amplitude of the recorded signals. While this finding is definitely interesting for accurately modeling and understanding how the extracellular potential is generated and recorded, it is unclear how it would affect the spike sorting performance. Moreover, when modeling signals on MEAs, we used the method of images (Ness et al. ; Buccino et al. ), which models the probe as a infinite insulating plane and better describes the recorded potentials for large MEA probes (Buccino et al. ). Secondly, during templates generation, the neuron models were randomly moved around and rotated with physiologically acceptable values (Buccino et al. ). In this phase, some dendritic trees might unnaturally cross the probes. We decided to not modify the cell models and allow for this behavior for sake of efficiency of the simulator. The modification of the dendritic trees for each extracellular spike generation would in fact be too computationally intense. However, since the templates generation phase is only run once for each probes, in the future we plan to both to include the probe effect in the simulations and to carefully modify the dendritic positions so that they do not cross the probes’ plane. Another limitation of the proposed modeling approach is in the replication of bursting behavior. We implemented a simplified bursting modulation that attempts to capture the features recorded from extracellular electrodes by modifying the template amplitude and shape depending on the spiking history. However, more advanced aspects of waveform modulation caused by bursting, including morphology-dependent variation of spike shapes, cannot be modeled with the proposed approach, and their replication requires a full multi-compartment simulation (Hagen et al. ). Nevertheless, the suggested simplified model of bursting could be a valuable tool for testing the capability of spike sorters to deal with this phenomenon. Finally, the current version of MEArec only supports cell models from the Neocortical Microcircuit Portal (Markram et al. ; Ramaswamy et al. ), which includes models from juvenile rat somatosensory cortex. The same cell model format is also being used to build a full hippocampus model (Migliore et al. ) and other brain regions, and therefore the integration of new models should be straightforward. However, we also provide a mechanism to use custom cell models. For example, cell models from the Allen Brain Institute database (Gouwens et al. ) , which contains models from mice and humans, can be easily used to simu late templates and recordings, as documented in this notebook: https://github.com/alejoe91/MEArec/blob/master/notebooks/generate_recordings_with_allen_models.ipynb . Other cell models can be used with the same approach. The use of fully-simulated recordings can raise questions on how well the simulations replicate real extracellular recordings. For example, recordings on freely moving animals present several motion artifacts that are complicated to model and incorporate into simulators. For these reasons, we believe that spike sorting validation cannot be solely limited to simulated recordings. In a recent effort for spike sorting validation, named SpikeForest (Magland et al. ), the authors have gathered more than 650 ground-truth recordings belonging to different categories: paired recordings, simulated synthetic recordings (including MEArec-generated datasets), hybrid recordings, and manually sorted data. We think that a systematic benchmark of spike sorting tools will benefit from this larger collection of diverse ground-truth recordings, and in this light, MEArec can provide high-quality simulated datasets to aid this purpose. In conclusion, we introduced MEArec, which is a Python-based simulation framework for extracellular recordings. Thanks to its speed and controllability, we see MEArec to aid both the development and validation spike sorting algorithms and to help understanding the limitation of current methods, to improve their performance, and to generate new software tools for the hard and still partially unsolved spike sorting problem.
The presented software package is available at https://github.com/alejoe91/MEArec and https://github.com/alejoe91/MEAutility (used for probe handling). The packages are also available on pypi: https://pypi.org/project/MEArec/ - https://pypi.org/project/MEAutility/ . All the datsets generated for the paper and used to make figures are available on Zenodo at 10.5281/zenodo.3696926, where instruction to generate figures are also provided.
Below is the link to the electronic supplementary material. (PDF 377 KB)
|
Milano Policlinico ONCOVID modified Score for risk evaluation in oncology during the COVID-19 pandemic: a prospective monocentric study | 114fbe0e-9b14-4e55-a9f8-d03f7fea1555 | 8995143 | Internal Medicine[mh] | The novel coronavirus disease 2019 (COVID-19), first reported in China in December 2019, is a worldwide health threat. It became a public health emergency of international concern because of its severity and rapid spread . According to the World Health Organization (WHO), as of October 1, 2021, there have been 233,503,524 confirmed cases of COVID-19, including 4,777,503 deaths. As of 23 January 2022, over 346 million confirmed cases and over 5.5 million deaths have been reported worldwide. A slower increase in case incidence was observed at the global level, with only half of the regions reporting an increase in the number of new weekly cases, as compared to five out of six regions in the previous week. The Eastern Mediterranean Region reported the largest increase in the number of new cases (39%), followed by the South-East Asia Region (36%) and the European Region (13%). In Tunisia, since the outbreak of the pandemic in March 2020, 788.012 cases of COVID-19 have been recorded, of which 25.803 have died and 704.030 have recovered. According to the same report, as of January 15, 51 new hospitalizations were recorded in public and private health establishments, of which 128 people were in intensive care and 26 were placed on artificial respirators. A total of 544 people with COVID-19 are currently hospitalized, according to the Ministry of Health. Several patients have been reported to be at increased risk of infection with high morbidity and mortality mostly elderly patients, males, and those with comorbidities, such as hypertension, chronic lung disease, diabetes, and cancer . The present article focuses on the Milano Policlinico ONCOVID Score to weigh the risk of COVID-19 in Tunisian patients with cancer. The main issue raised by the pandemic is whether the risk of COVID-19 outweighs that of cancer treatment delay. In the present situation, oncologists need to decide which kind of patient should start (or continue) which kind of treatment and how much will this increase the risk of complications in case of COVID-19 in these vulnerable patients.
Study design and participant A prospective study was conducted at the department of medical oncology of Sfax, from November 2020 to February 2021. This was a consecutive cohort of all patients during this defined period. We had included in our study patients who have been followed for histologically proven cancer and who are undergoing adjuvant or palliative cancer treatment (chemotherapy or targeted therapy), over the age of 18, able to answer the questions, and who agreed to participate in the study. We had excluded from these study patients who do not have cancer, those under 18 years old, those with backwardness or major cognitive impairment, and patients who refused to participate in the study. All procedures performed in this study were in accordance with the ethical standards of the institutional and the local national research committee of Habib Bourguiba and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Our study was approved by the local Habib Bourguiba committee. We used the Milano Policlinico ONCOVID modified Score to assess the risk of COVID-19 infection in patients with cancer. This score includes patient characteristics such as sex, performance status, age, body mass index ( BMI ), comorbidities such as hypertension, cardiovascular disease, diabetes, chronic obstructive pulmonary disease and chronic systemic infections, and concomitant steroid treatment including continuous therapy with a dose of > 10 mg daily of prednisone equivalent, lasting for more than the 1-month period and history of respiratory infection. In addition, this score includes disease characteristics with thoracic tumor and history of thoracic radiotherapy only for patients with extrathoracic tumors. Furthermore, it includes treatment characteristics with the precision of the lines and type of treatment. Finally, the neutrophil–lymphocyte ratio impacted this score. According to the Milano Policlinico ONCOVID Score, we were able to define 3 groups of risk, low (< 4) intermediate (4–6), and high risk (> 6). At the end of the study, we evaluated the prevalence of COVID19 infection in this population (Table ). Statistical study Descriptive statistics (frequencies, percentage) were calculated. The study of the associations between the variables was made by the hypothesis tests. The chi-square chi2 ( χ 2 ) was useful for analyzing such differences in categorical variables, especially those nominals in nature. The comparison of proportions was made by Pearson’s “chi2 (χ 2 )” test when the theoretical size is greater than 5 and by the “Fisher ( F )” exact test when one of the theoretical numbers is less than 5 for independent samples. The correlation between risk factors of COVID-19 infection and the occurrence of COVID-19 infection was also calculated. A p -value < 0.05 was considered statistically significant.
A prospective study was conducted at the department of medical oncology of Sfax, from November 2020 to February 2021. This was a consecutive cohort of all patients during this defined period. We had included in our study patients who have been followed for histologically proven cancer and who are undergoing adjuvant or palliative cancer treatment (chemotherapy or targeted therapy), over the age of 18, able to answer the questions, and who agreed to participate in the study. We had excluded from these study patients who do not have cancer, those under 18 years old, those with backwardness or major cognitive impairment, and patients who refused to participate in the study. All procedures performed in this study were in accordance with the ethical standards of the institutional and the local national research committee of Habib Bourguiba and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Our study was approved by the local Habib Bourguiba committee. We used the Milano Policlinico ONCOVID modified Score to assess the risk of COVID-19 infection in patients with cancer. This score includes patient characteristics such as sex, performance status, age, body mass index ( BMI ), comorbidities such as hypertension, cardiovascular disease, diabetes, chronic obstructive pulmonary disease and chronic systemic infections, and concomitant steroid treatment including continuous therapy with a dose of > 10 mg daily of prednisone equivalent, lasting for more than the 1-month period and history of respiratory infection. In addition, this score includes disease characteristics with thoracic tumor and history of thoracic radiotherapy only for patients with extrathoracic tumors. Furthermore, it includes treatment characteristics with the precision of the lines and type of treatment. Finally, the neutrophil–lymphocyte ratio impacted this score. According to the Milano Policlinico ONCOVID Score, we were able to define 3 groups of risk, low (< 4) intermediate (4–6), and high risk (> 6). At the end of the study, we evaluated the prevalence of COVID19 infection in this population (Table ).
Descriptive statistics (frequencies, percentage) were calculated. The study of the associations between the variables was made by the hypothesis tests. The chi-square chi2 ( χ 2 ) was useful for analyzing such differences in categorical variables, especially those nominals in nature. The comparison of proportions was made by Pearson’s “chi2 (χ 2 )” test when the theoretical size is greater than 5 and by the “Fisher ( F )” exact test when one of the theoretical numbers is less than 5 for independent samples. The correlation between risk factors of COVID-19 infection and the occurrence of COVID-19 infection was also calculated. A p -value < 0.05 was considered statistically significant.
Patients characteristics Among 226 patients included (Fig. ), patients under 70 years presented 85%. The sex ratio was 0.5 with a female predominance. The most common primary tumor subtypes were breast cancer (37%), colorectal cancers (22%), ovarian (7.5%), and lung cancer (5.5%). A metastatic disease was observed in 58% of patients. Comorbidities such as diabetes, hypertension, cardiovascular disease, and chronic obstructive pulmonary disease were observed in 39% of cases (Table ). Among our patients, coronavirus disease was detected in 19 patients (8.4%). The incidence of COVID-19 infection in female was 10.7% and 4% in men (Table ). Two of them had a poor performance status (> 2) and four patients (21%) were obese. Among 34 patients aged more than 70 years old, 6 patients (31%) had confirmed positive COVID-19 test. Comorbidities such as diabetes, hypertension, cardiovascular disease, and chronic obstructive pulmonary disease were observed in 31% of cases ( n = 6). Patients with metastatic tumor presented 63% of all infected patients. In 42% of cases, patients were under chemotherapy. Clinically, 15 patients (79%) presented with symptoms such as fever, dyspnea, caught, myalgia, and ageusia/anosmia. A severe form of COVID19 requiring hospitalization was seen in 4 patients (21%). Factors correlated with COVID-19 infection In our study, among all patients, the risk of COVID-19 infection was estimated as low in 56% of cases and intermediate or high in 44% of patients. Among our patients, coronavirus disease was detected in 19 patients (8.4%). Of 19 patients tested positive for COVID19, 47% had an intermediate and high risk of infection. COVID-19 infection was correlated with age < 70 years ( p = 0.035, χ 2 = 4.437, ddl = 1), chemotherapy treatment ( p = 0.032, χ 2 = 4.613, ddl = 1), and intermediate or high risk ( p = 0.018, χ 2 = 4.892, ddl = 1) (Table ). In multivariable analysis, only the intermediate or high risk were correlated with COVID-19 infection in cancer patient ( p = 0.025). Symptomatic COVID was correlated with stage IV ( p = 0.041, χ 2 = 4.156, ddl = 1), chemotherapy ( p = 0.004, χ 2 = 7.367, ddl = 1), and intermediate or high risk ( p = 0.04, χ 2 = 3.754, ddl = 1). Only the intermediate or high risk were correlated with COVID-19 infection in cancer patient ( p = 0.03).
Through this prospective study, we demonstrated that using the Milano Policlinico ONCOVID modified Score is very helpful for clinicians to identify vulnerable patients and to make the appropriate decision in the management of cancer patients. COVID-19 infection was correlated with age < 70 years, chemotherapy treatment, and intermediate or high risk. However, in multivariable analysis, only the intermediate or high risk were correlated with COVID-19 infection in cancer patient ( p = 0.025). In fact, in case of a high-risk score, treatment schedules can be maintained only if safe administration is guaranteed. Treatment administration should be tailored depending on the type of treatment and disease response. Unnecessary procedures should be avoided to reduce hospital access. In this prospective study, we investigated the prevalence of COVID-19 infection in cancer patients which was found to be 8.4%. The same prevalence was reported by Fillmort et al. who analyzed electronic health records of the US Veterans Affairs Healthcare System. Of 22 914 cancer patients tested for COVID-19, 1794 (7.8%) were positive. In another study including 2152 patients, 190 patients (9%) were tested positive for COVID-19 . This high prevalence may be explained by the susceptibility of patients with cancer to COVID-19 infection. In fact, in several studies, authors reported that patients with cancer were more prone to infection with severe events and an increased risk of death. This might be related to the systemic immunosuppressive state caused directly by tumor growth and indirectly by the effects of anti-cancer treatment . In the current analysis, we used a personalized score to evaluate the risk of infection in cancer patients. In the Policlinico ONCOVID modified Score, clinical, biological, and therapeutic parameters have been analyzed together to define 3 groups of risk. We found that patients with intermediate or high risk were more exposed to infection. Consequently, using such scores may be helpful to identify the vulnerable population among cancer patients. Males and elderly patients have been identified as more exposed to COVID-19 infection, which was not founded in the current study. This could be explained by the fact of the female predominance in the included patients. Paradoxically, the risk of infection was higher in patients aged less than 70 years, probably because they need to regularly visit hospitals for anti-cancer therapy. In terms of cancer characteristics, in a cohort study of 105 cancer patients with COVID-19, Dai et al. reported that lung cancer was the most common cancer histology in affected patients (20.95%), followed by gastrointestinal cancer (12.38%), and breast cancer (10.48%). However, in the present study, breast cancer (53%) and colorectal cancer (16%) were the two most frequent primary tumors found in infected patients unlike lung cancer (5%). This might be explicated by the heterogeneity in the characteristics of included patients. In a cohort study conducted by Zhang et al. , 28 cancer patients with COVID-19 infection were included. It has been founded that a high percentage of infected patients (35.7%) suffered from a metastatic disease. Additionally, in the current study; the percentage of infected patients with a stage IV disease was high and presented 63% of all infected patients which highlights the susceptibility of this subgroup of patients to COVID-19 infection. In COVID-19, clinical manifestations can vary from mild flu-like symptoms to life-threatening respiratory insufficiency . In a cohort study in China, Liang et al. showed an increased risk of severe clinical events in the case of COVID-19 infection in cancer patients. Likewise, in a meta-analysis published in 2021, about 43.26% of cancer patients with COVID-19 experienced severe events . In our study, the majority of patients were symptomatic (79%) and 21% of them required hospitalization due to a severe disease course. We found that symptomatic. COVID-19 infection was correlated to chemotherapy, stage IV disease, and intermediate or high risk according to the score calculated. In this study, most of the patients who tested positive for COVID-19 had active disease and were under chemotherapy (42%). We founded also that receiving chemotherapy was correlated to the risk of infection. Hence, the question of delaying the treatment in patients at risk of infection remains. A meta-analysis involving 15 studies demonstrated that chemotherapy could increase the risk of death from COVID-19 in cancer patients . A recent meta-analysis composed of 52 cohorts involving 9231 cancer patients with COVID-19 was so far the largest scale investigation with respect to the impact of antitumor approaches on clinical outcomes of cancer patients with COVID-19, indicating that cancer patients with recent anti-tumor therapy (especially chemotherapy) were generally susceptible to develop into severe COVID19 or even death . Our study has some limitations with a relatively small sample size, heterogeneity in patient’s characteristics, tumor stages, and cancer types. However, through this prospective study, we demonstrated the utility of such risk scores to further help clinicians in making the appropriate actions to alleviate this risk in the cancer patient.
|
Hybridization chain reaction enables a unified approach to multiplexed, quantitative, high-resolution immunohistochemistry and | 8a0cf280-98a7-4695-bc7e-e3454efef6d8 | 8645210 | Anatomy[mh] | Biological circuits encoded in the genome of each organism direct development, maintain integrity in the face of attacks, control responses to environmental stimuli and sometimes malfunction to cause disease. RNA in situ hybridization (RNA-ISH) methods ( ; ; ) and immunohistochemistry (IHC) methods ( ; ; ) provide biologists, drug developers and pathologists with crucial windows into the spatial organization of this circuitry, enabling imaging of RNA and protein expression in an anatomical context. Although it is desirable to perform multiplexed experiments in which a panel of targets are imaged quantitatively at high resolution in a single specimen, using traditional RNA-ISH and IHC methods in highly autofluorescent samples including whole-mount vertebrate embryos and FFPE tissue sections, multiplexing is cumbersome, staining is non-quantitative and spatial resolution is routinely compromised by diffusion of reporter molecules. These multi-decade technological shortcomings are significant impediments to biological research, as well as to the advancement of drug development and pathology assays, hindering high-dimensional, quantitative, high-resolution analyses of developmental and disease-related regulatory networks in an anatomical context. RNA-ISH methods detect RNA targets using nucleic acid probes and IHC methods detect protein targets using antibody probes. In either case, probes can be directly labeled with reporter molecules ( ; ; ; ; ), but to increase the signal-to-background ratio, are more often used to mediate signal amplification in the vicinity of the probe ( ; ). A variety of in situ amplification approaches have been developed, including immunological methods ( ; ; ), branched DNA methods ( ; ; ; ), in situ PCR methods ( ; ; ) and rolling circle amplification methods ( ; ; ). However, for both RNA-ISH ( ; ; ; ; ; ; ; ; ) and IHC ( ; ; ; ; ), traditional in situ amplification based on enzyme-mediated catalytic reporter deposition (CARD) remains the dominant approach for achieving high signal-to-background in highly autofluorescent samples, including whole-mount vertebrate embryos and FFPE tissue sections. CARD is widely used despite three significant drawbacks: multiplexing is cumbersome due to the lack of orthogonal deposition chemistries, necessitating serial amplification for one target after another ( ; ; ; ; ; ; ; ; ); staining is qualitative rather than quantitative; and spatial resolution is often compromised by diffusion of reporter molecules before deposition ( ; ; ; ; ; ). In the context of RNA-ISH, in situ amplification based on the mechanism of hybridization chain reaction (HCR; A) ( ) overcomes the longstanding shortcomings of CARD to enable multiplexed, quantitative, high-resolution imaging of RNA expression in diverse organisms and sample types, including highly autofluorescent samples ( , , , ; ; ) (e.g. see Table S1 ). To image RNA expression, targets are detected by nucleic acid probes that trigger isothermal enzyme-free chain reactions in which fluorophore-labeled HCR hairpins self-assemble into tethered fluorescent amplification polymers ( B). Orthogonal HCR amplifiers operate independently within the sample so the experimental timeline for multiplexed experiments is independent of the number of target RNAs ( , ). The amplified HCR signal scales approximately linearly with the number of target molecules ( E), enabling accurate and precise RNA relative quantitation with subcellular resolution in the anatomical context of whole-mount vertebrate embryos ( ; ). Amplification polymers remain tethered to their initiating probes, enabling imaging of RNA expression with subcellular or single-molecule resolution as desired ( , , ; ). These properties that make HCR signal amplification well-suited for RNA-ISH appear equally favorable in the context of IHC, suggesting the approach of combining HCR signal amplification with antibody probes ( ; ; ). Here, we extend the benefits of one-step, quantitative, enzyme-free signal amplification from RNA-ISH to IHC, validating multiplexed, quantitative, high-resolution imaging of protein expression with high signal-to-background in highly autofluorescent samples, thus overcoming the longstanding shortcomings of IHC using CARD. Moreover, we establish a unified framework for simultaneous multiplexed, quantitative, high-resolution IHC and RNA-ISH, with one-step HCR signal amplification performed for all targets simultaneously.
For protein imaging with HCR we pursue two complementary approaches. Using HCR 1°IHC, protein targets are detected using primary antibody probes labeled with one or more HCR initiators ( C). For multiplexed experiments, the probes for different targets are labeled with different HCR initiators that trigger orthogonal HCR amplifiers labeled with spectrally distinct fluorophores. Researchers have the flexibility to detect different targets using primary antibody probes raised in the same host species (or a variety of host species, as convenient). On the other hand, each new initiator-labeled primary antibody probe must be validated, as there is the potential for oligo conjugation to interfere with epitope binding in an antibody- or crosslinker-dependent fashion. Using HCR 2°IHC, protein targets are detected using unlabeled primary antibody probes that are in turn detected by secondary antibody probes labeled with one or more HCR initiators ( D). This approach has the advantage that validation of a small library of initiator-labeled secondary antibodies (e.g. five secondaries targeting different host species) enables immediate use of large libraries of primary antibody probes (e.g. 10 5 commercially available primaries) without modification. On the other hand, for multiplexed experiments, each target must be detected using a primary antibody raised in a different host species to enable subsequent detection by an anti-host secondary antibody probe that triggers an orthogonal spectrally distinct HCR amplifier. Hence, depending on the available antibody probes, one may prefer HCR 1°IHC in one instance and HCR 2°IHC in another. Multiplexed protein imaging using HCR 1°IHC or HCR 2°IHC demonstrates multiplexed protein imaging via HCR 1°IHC using initiator-labeled primary antibody probes. demonstrates multiplexed protein imaging via HCR 2°IHC using unlabeled primary antibody probes and initiator-labeled secondary antibody probes. Both methods achieve high signal-to-background for 3-plex protein imaging in mammalian cells and for 4-plex protein imaging in FFPE mouse brain sections. Across 21 protein imaging scenarios (six in mammalian cells, ten in FFPE mouse brain sections, four in FFPE human breast tissue sections and one in whole-mount zebrafish embryos; nine using HCR 1°IHC and 12 using HCR 2°IHC; 11 using
confocal microscopy and ten using epifluorescence microscopy), the estimated signal-to-background ratio for protein targets ranged from 15 to 609 with a median of 90 (see Tables S9 and S10 for additional details). The level of performance demonstrated in and was achieved for all targets simultaneously in 4-channel and 5-channel images (including a DAPI channel in each case) using fluorophores that compete with lower autofluorescence (Alexa647) as well as with higher autofluorescence (Alexa488) and in samples with lower autofluorescence (mammalian cells) and higher autofluorescence (FFPE mouse brain sections). Using HCR signal amplification, the amplification gain corresponds to the number of fluorophore-labeled hairpins per amplification polymer. Hence, we were curious to measure the mean HCR polymer length in the context of HCR 1°IHC and HCR 2°IHC experiments. We can estimate HCR amplification gain by comparing the signal intensity in HCR experiments using h1 and h2 hairpins together (enabling polymerization to proceed as normal) versus using only hairpin h1 (so that each HCR initiator can bind only one HCR hairpin and polymerization cannot proceed). Across four measurement scenarios (two in mammalian cells and two in FFPE mouse brain sections; two using HCR 1°IHC and two using HCR 2°IHC), we observed a median polymer length of ≈180 hairpins ( see section S5.5 in the supplementary information ). It is this amplification gain that boosts the signal above autofluorescence to yield a high signal-to-background ratio even in FFPE tissues and whole-mount vertebrate embryos. qHCR imaging: protein relative quantitation with subcellular resolution in an anatomical context We have previously demonstrated that HCR RNA-ISH overcomes the historical tradeoff between RNA quantitation and anatomical context, enabling mRNA relative quantitation (qHCR imaging) with subcellular resolution within whole-mount vertebrate embryos ( ; ). Here, we demonstrate that HCR IHC enables analogous subcellular quantitation of proteins in an anatomical context. To test protein relative quantitation, we first redundantly detected a target protein using two primary antibody probes that bind different epitopes on the same protein and trigger different spectrally distinct HCR amplifiers ( A; top), yielding a two-channel image ( B; top). If HCR signal scales approximately linearly with the number of target proteins per voxel, a two-channel scatter plot of normalized voxel intensities will yield a tight linear distribution with approximately zero intercept ( ). On the other hand, observing a tight linear distribution with approximately zero intercept ( C; top), we conclude that the HCR signal scales approximately linearly with the number of target proteins per imaging voxel, after first ruling out potential systematic crowding effects that could permit pairwise voxel intensities to slide undetected along a line ( Fig. S24 ). Using one initiator-labeled primary antibody probe per channel, we observe high accuracy (linearity with zero intercept) and precision (scatter around the line) for subcellular 2×2 µm voxels within 5 µm FFPE mouse brain sections using epifluorescence microscopy. This redundant detection experiment provides a conservative characterization of quantitative performance as there is the risk that two antibody probes may interfere with each other to some extent when attempting to bind different epitopes on the same target protein. As a further test of quantitative imaging characteristics, we detected a protein target with unlabeled primary antibody probes that are subsequently detected by two batches of secondary antibody probes that trigger different spectrally distinct HCR amplifiers ( A; bottom). This experiment is testing the accuracy and precision of the secondary antibody probes and HCR signal amplification, but not that of the primary antibody probes. In FFPE human breast tissue sections using confocal microscopy ( B; bottom), a two-channel scatter plot of voxel intensities for subcellular 2.0×2.0×2.5 µm voxels again reveals a tight linear distribution with approximately zero intercept ( C; bottom). Based on these two studies, we conclude that qHCR imaging enables accurate and precise relative quantitation of protein targets in an anatomical context with subcellular resolution, just as it does for mRNA targets ( ; ). Simultaneous multiplexed protein and RNA imaging using HCR 1°IHC + HCR RNA-ISH or HCR 2°IHC + HCR RNA-ISH It is important for biologists, drug developers and pathologists to have the flexibility to image proteins and RNAs simultaneously so as to enable interrogation of both levels of gene expression in the same specimen. Here, we demonstrate that HCR 1°IHC and HCR 2°IHC are both compatible with HCR RNA-ISH, enabling multiplexed quantitative protein and RNA imaging with high signal-to-background. demonstrates HCR 1°IHC + HCR RNA-ISH (2-plex protein + 2-plex RNA) in mammalian cells and FFPE mouse brain sections using initiator-labeled primary antibody probes for protein targets, split-initiator DNA probes for RNA targets, and simultaneous HCR signal amplification for all targets. demonstrates HCR 2°IHC + HCR RNA-ISH (2-plex protein + 2-plex RNA) in mammalian cells and FFPE mouse brain sections using unlabeled primary antibody probes and initiator-labeled secondary antibody probes for protein targets, split-initiator DNA probes for RNA targets, and simultaneous HCR signal amplification for all targets. Across 16 protein and RNA imaging scenarios (eight in mammalian cells and eight in FFPE mouse brain sections; eight using HCR 1°IHC + HCR RNA-ISH and eight using HCR 2°IHC + HCR RNA-ISH; eight using confocal microscopy and eight using epifluorescence microscopy), the estimated signal-to-background ratio for each target protein or RNA ranged from 20 to 700, with a median of 100 (see Tables S9 and S11 for additional details).
demonstrates multiplexed protein imaging via HCR 1°IHC using initiator-labeled primary antibody probes. demonstrates multiplexed protein imaging via HCR 2°IHC using unlabeled primary antibody probes and initiator-labeled secondary antibody probes. Both methods achieve high signal-to-background for 3-plex protein imaging in mammalian cells and for 4-plex protein imaging in FFPE mouse brain sections. Across 21 protein imaging scenarios (six in mammalian cells, ten in FFPE mouse brain sections, four in FFPE human breast tissue sections and one in whole-mount zebrafish embryos; nine using HCR 1°IHC and 12 using HCR 2°IHC; 11 using
confocal microscopy and ten using epifluorescence microscopy), the estimated signal-to-background ratio for protein targets ranged from 15 to 609 with a median of 90 (see Tables S9 and S10 for additional details). The level of performance demonstrated in and was achieved for all targets simultaneously in 4-channel and 5-channel images (including a DAPI channel in each case) using fluorophores that compete with lower autofluorescence (Alexa647) as well as with higher autofluorescence (Alexa488) and in samples with lower autofluorescence (mammalian cells) and higher autofluorescence (FFPE mouse brain sections). Using HCR signal amplification, the amplification gain corresponds to the number of fluorophore-labeled hairpins per amplification polymer. Hence, we were curious to measure the mean HCR polymer length in the context of HCR 1°IHC and HCR 2°IHC experiments. We can estimate HCR amplification gain by comparing the signal intensity in HCR experiments using h1 and h2 hairpins together (enabling polymerization to proceed as normal) versus using only hairpin h1 (so that each HCR initiator can bind only one HCR hairpin and polymerization cannot proceed). Across four measurement scenarios (two in mammalian cells and two in FFPE mouse brain sections; two using HCR 1°IHC and two using HCR 2°IHC), we observed a median polymer length of ≈180 hairpins ( see section S5.5 in the supplementary information ). It is this amplification gain that boosts the signal above autofluorescence to yield a high signal-to-background ratio even in FFPE tissues and whole-mount vertebrate embryos.
We have previously demonstrated that HCR RNA-ISH overcomes the historical tradeoff between RNA quantitation and anatomical context, enabling mRNA relative quantitation (qHCR imaging) with subcellular resolution within whole-mount vertebrate embryos ( ; ). Here, we demonstrate that HCR IHC enables analogous subcellular quantitation of proteins in an anatomical context. To test protein relative quantitation, we first redundantly detected a target protein using two primary antibody probes that bind different epitopes on the same protein and trigger different spectrally distinct HCR amplifiers ( A; top), yielding a two-channel image ( B; top). If HCR signal scales approximately linearly with the number of target proteins per voxel, a two-channel scatter plot of normalized voxel intensities will yield a tight linear distribution with approximately zero intercept ( ). On the other hand, observing a tight linear distribution with approximately zero intercept ( C; top), we conclude that the HCR signal scales approximately linearly with the number of target proteins per imaging voxel, after first ruling out potential systematic crowding effects that could permit pairwise voxel intensities to slide undetected along a line ( Fig. S24 ). Using one initiator-labeled primary antibody probe per channel, we observe high accuracy (linearity with zero intercept) and precision (scatter around the line) for subcellular 2×2 µm voxels within 5 µm FFPE mouse brain sections using epifluorescence microscopy. This redundant detection experiment provides a conservative characterization of quantitative performance as there is the risk that two antibody probes may interfere with each other to some extent when attempting to bind different epitopes on the same target protein. As a further test of quantitative imaging characteristics, we detected a protein target with unlabeled primary antibody probes that are subsequently detected by two batches of secondary antibody probes that trigger different spectrally distinct HCR amplifiers ( A; bottom). This experiment is testing the accuracy and precision of the secondary antibody probes and HCR signal amplification, but not that of the primary antibody probes. In FFPE human breast tissue sections using confocal microscopy ( B; bottom), a two-channel scatter plot of voxel intensities for subcellular 2.0×2.0×2.5 µm voxels again reveals a tight linear distribution with approximately zero intercept ( C; bottom). Based on these two studies, we conclude that qHCR imaging enables accurate and precise relative quantitation of protein targets in an anatomical context with subcellular resolution, just as it does for mRNA targets ( ; ).
It is important for biologists, drug developers and pathologists to have the flexibility to image proteins and RNAs simultaneously so as to enable interrogation of both levels of gene expression in the same specimen. Here, we demonstrate that HCR 1°IHC and HCR 2°IHC are both compatible with HCR RNA-ISH, enabling multiplexed quantitative protein and RNA imaging with high signal-to-background. demonstrates HCR 1°IHC + HCR RNA-ISH (2-plex protein + 2-plex RNA) in mammalian cells and FFPE mouse brain sections using initiator-labeled primary antibody probes for protein targets, split-initiator DNA probes for RNA targets, and simultaneous HCR signal amplification for all targets. demonstrates HCR 2°IHC + HCR RNA-ISH (2-plex protein + 2-plex RNA) in mammalian cells and FFPE mouse brain sections using unlabeled primary antibody probes and initiator-labeled secondary antibody probes for protein targets, split-initiator DNA probes for RNA targets, and simultaneous HCR signal amplification for all targets. Across 16 protein and RNA imaging scenarios (eight in mammalian cells and eight in FFPE mouse brain sections; eight using HCR 1°IHC + HCR RNA-ISH and eight using HCR 2°IHC + HCR RNA-ISH; eight using confocal microscopy and eight using epifluorescence microscopy), the estimated signal-to-background ratio for each target protein or RNA ranged from 20 to 700, with a median of 100 (see Tables S9 and S11 for additional details).
qHCR imaging enables a unified approach to multiplexed quantitative IHC and RNA-ISH. A single experiment yields accurate and precise relative quantitation of both protein and RNA targets with subcellular resolution in the anatomical context of highly autofluorescent samples. No extra work is necessary to perform quantitative imaging – it is a natural property of HCR signal amplification. Here, we validated two complementary approaches for HCR IHC. Using HCR 1°IHC (initiator-labeled primary antibody probes), each target protein in a multiplexed experiment can be detected with antibodies raised in the same host species, which is often convenient based on available antibody libraries. However, antibody-initiator conjugation must be validated for each new primary antibody probe. Alternatively, using HCR 2°IHC (unlabeled primary antibody probes and initiator-labeled secondary antibody probes), each target protein in a multiplexed experiment must be detected with primary antibodies raised in different host species, thus enabling subsequent binding by initiator-labeled secondary antibodies that react with those different host species. This approach has the benefit that a small library of initiator-labeled secondary antibodies can be validated a priori and then used with large libraries of (unmodified) validated primary antibodies, enabling a plug-and-play approach using validated reagents. For simultaneous protein and RNA imaging: during the protein detection stage, M target proteins are detected in parallel; during the RNA detection stage, N target RNAs are detected in parallel; and during the amplification stage, one-step quantitative HCR signal amplification is performed for all M+N protein and RNA targets simultaneously. In 4-plex experiments in FFPE tissue sections, protein and RNA targets are simultaneously imaged with high signal-to-background in all four channels using fluorophores that compete with varying degrees of autofluorescence. For protein imaging using HCR 1°IHC or HCR 2°IHC, we favor protocols with two overnight incubations ( B and B), and for simultaneous protein and RNA imaging using HCR 1°IHC + HCR RNA-ISH or HCR 2°IHC + HCR RNA-ISH, we favor protocols with three overnight incubations ( A and A), allowing researchers to maintain a normal sleep schedule. HCR RNA-ISH provides automatic background suppression throughout the protocol, ensuring that reagents will not generate amplified background even if they bind non-specifically within the sample ( ). During the detection stage, each RNA target is detected by a probe set comprising one or more pairs of split-initiator probes, each carrying a fraction of HCR initiator i1 ( B). For a given probe pair, probes that hybridize specifically to their adjacent binding sites on the target RNA colocalize full initiator i1, enabling cooperative initiation of HCR signal amplification. Meanwhile, any individual probes that bind non-specifically in the sample do not colocalize full initiator i1, do not trigger HCR and thus suppress generation of amplified background. During the amplification stage, automatic background suppression is inherent to HCR hairpins because polymerization is conditional on the presence of the initiator i1; individual h1 or h2 hairpins that bind non-specifically in the sample do not trigger formation of an amplification polymer. For HCR IHC, during the detection stage, each target protein is detected using primary or secondary antibody probes carrying one or more full i1 initiators ( C,D). Hence, if an antibody probe binds non-specifically in the sample, initiator i1 will nonetheless trigger HCR, generating amplified background. As a result, it is important to use antibody probes that are highly selective for their targets, and to wash unused antibody probes from the sample. Nonetheless, during the amplification stage, kinetically trapped HCR hairpins provide automatic background suppression for protein targets just as they do for RNA targets, ensuring that any hairpins that bind non-specifically in the sample do not trigger growth of an HCR amplification polymer. For experiments using HCR IHC + HCR RNA-ISH to image protein and RNA targets simultaneously, RNA targets enjoy automatic background suppression throughout the protocol, whereas protein targets rely on selective antibody binding to suppress background during the detection stage, combined with automatic background suppression during the amplification stage. For RNA targets, we have previously shown that multiplexed qHCR imaging enables bi-directional quantitative discovery ( ): read-out from anatomical space to expression space to discover co-expression relationships in selected regions of the sample; read-in from expression space to anatomical space to discover those anatomical locations in which selected gene co-expression relationships occur. Here, by validating high-accuracy, high-precision, high-resolution qHCR imaging for protein targets, read-out/read-in analyses can now be performed for RNA and protein targets simultaneously, offering biologists, drug developers and pathologists a significantly expanded window for analyzing biological circuits in an anatomical context.
Probes, amplifiers and buffers Details on the probes, amplifiers and buffers for each experiment are displayed in Table S2 for HCR 1°IHC, in Table S3 for HCR 2°IHC and in Table S4 for HCR RNA-ISH. HCR initiators were conjugated to antibody probes using the Antibody-Oligonucleotide All-in-One Conjugation Kit (Vector Laboratories, A-9202) according to the manufacturer's instructions. HCR IHC with/without HCR RNA-ISH HCR 1°IHC with/without HCR RNA-ISH was performed using the protocols detailed in section S3 in the supplementary information . HCR 2°IHC with/without HCR RNA-ISH was performed using the protocols detailed in section S4 in the supplementary information . These IHC protocols with/without HCR RNA-ISH were developed starting from HCR RNA-ISH protocols ( ). The optional autofluorescence bleaching protocol for FFPE mouse brain tissue sections, combining photo- ( ) and chemical ( ) bleaching, was used only for the HCR IHC + HCR RNA-ISH studies of and , and the associated replicates in Figs S35, S36, S43 and S44 . Strictly speaking, the cultured cell studies represent immunocytochemistry (ICC) rather than IHC; for notational simplicity, we use the term IHC uniformly in the main text but denote protocols for cultured cells as ICC in the supplementary information . For five-channel imaging of HeLa cells ( B, S33, S34 , B, S41, S42 ) the above protocols were modified as follows to enable imaging on an upright confocal microscope: cells were grown on a chambered slide with removable chambers (Ibidi, 81201); prior to imaging, the silicone chambers were removed and cells were mounted with ProLong glass antifade mountant with NucBlue (Thermo Fisher Scientific, P36981) according to the manufacturer's instructions. Experiments were performed in HeLa cells (ATCC, CRM-CCL-2), FFPE C57BL/6 mouse brain sections (coronal; thickness 5 µm, Acepix Biosciences 7011-0120), FFPE human breast tissue sections (thickness 5 µm; Acepix Biosciences, 7310-0620) and whole-mount zebrafish embryos (wildtype Danio rerio strain AB; fixed at 27 hpf). Procedures for the care and use of zebrafish embryos were approved by the Caltech IACUC. Confocal microscopy Confocal microscopy was performed using a Zeiss LSM 800 inverted confocal microscope or a Zeiss LSM 880 with Fast Airyscan upright confocal microscope. All confocal images are displayed without background subtraction. See Table S5 for details on the microscope, objective, excitation lasers, beam splitters and emission bandpass filters used for each experiment. Epifluorescence microscopy Epifluorescence microscopy was performed using a Leica THUNDER Imager 3D cell culture epifluorescence microscope equipped with a Leica LED8 multi-LED light source and sCMOS camera (Leica DFC9000 GTC). All epifluorescence images were acquired without THUNDER computational clearing and are displayed with instrument noise subtracted but without background subtraction. See Table S6 for details on the objective, excitation wavelengths and filters used for each experiment. Image analysis Image analysis was performed as detailed in section S2.6 of the supplementary information, including: definition of raw pixel intensities; measurement of signal, background and signal-to-background; measurement of background components and calculation of normalized subcellular voxel intensities for qHCR imaging.
Details on the probes, amplifiers and buffers for each experiment are displayed in Table S2 for HCR 1°IHC, in Table S3 for HCR 2°IHC and in Table S4 for HCR RNA-ISH. HCR initiators were conjugated to antibody probes using the Antibody-Oligonucleotide All-in-One Conjugation Kit (Vector Laboratories, A-9202) according to the manufacturer's instructions.
HCR 1°IHC with/without HCR RNA-ISH was performed using the protocols detailed in section S3 in the supplementary information . HCR 2°IHC with/without HCR RNA-ISH was performed using the protocols detailed in section S4 in the supplementary information . These IHC protocols with/without HCR RNA-ISH were developed starting from HCR RNA-ISH protocols ( ). The optional autofluorescence bleaching protocol for FFPE mouse brain tissue sections, combining photo- ( ) and chemical ( ) bleaching, was used only for the HCR IHC + HCR RNA-ISH studies of and , and the associated replicates in Figs S35, S36, S43 and S44 . Strictly speaking, the cultured cell studies represent immunocytochemistry (ICC) rather than IHC; for notational simplicity, we use the term IHC uniformly in the main text but denote protocols for cultured cells as ICC in the supplementary information . For five-channel imaging of HeLa cells ( B, S33, S34 , B, S41, S42 ) the above protocols were modified as follows to enable imaging on an upright confocal microscope: cells were grown on a chambered slide with removable chambers (Ibidi, 81201); prior to imaging, the silicone chambers were removed and cells were mounted with ProLong glass antifade mountant with NucBlue (Thermo Fisher Scientific, P36981) according to the manufacturer's instructions. Experiments were performed in HeLa cells (ATCC, CRM-CCL-2), FFPE C57BL/6 mouse brain sections (coronal; thickness 5 µm, Acepix Biosciences 7011-0120), FFPE human breast tissue sections (thickness 5 µm; Acepix Biosciences, 7310-0620) and whole-mount zebrafish embryos (wildtype Danio rerio strain AB; fixed at 27 hpf). Procedures for the care and use of zebrafish embryos were approved by the Caltech IACUC.
Confocal microscopy was performed using a Zeiss LSM 800 inverted confocal microscope or a Zeiss LSM 880 with Fast Airyscan upright confocal microscope. All confocal images are displayed without background subtraction. See Table S5 for details on the microscope, objective, excitation lasers, beam splitters and emission bandpass filters used for each experiment.
Epifluorescence microscopy was performed using a Leica THUNDER Imager 3D cell culture epifluorescence microscope equipped with a Leica LED8 multi-LED light source and sCMOS camera (Leica DFC9000 GTC). All epifluorescence images were acquired without THUNDER computational clearing and are displayed with instrument noise subtracted but without background subtraction. See Table S6 for details on the objective, excitation wavelengths and filters used for each experiment.
Image analysis was performed as detailed in section S2.6 of the supplementary information, including: definition of raw pixel intensities; measurement of signal, background and signal-to-background; measurement of background components and calculation of normalized subcellular voxel intensities for qHCR imaging.
Supplementary information
|
DegS regulates the aerobic metabolism of | 2412fedd-356d-4659-b6de-020b14e90d7c | 11564185 | Biochemistry[mh] | Vibrio cholerae is a facultative anaerobic bacterium capable of both aerobic and anaerobic respiration ( ). Bacteria produce chemical energy through aerobic-mediated energy metabolism, which is stored in the form of ATP to power the cellular processes required for growth. Both aerobic and anaerobic metabolisms are essential for the growth of V. cholerae in vivo ( ; ). Aerobic respiration acts as a powerful driver of replication during infection with the V. cholerae gastrointestinal pathogen ( ). In one study, V. cholerae incapable of aerobic respiration was strongly attenuated (10 5 times) in young mice, whereas strains lacking anaerobic respiration showed no colonization defects ( ). In a suckling mouse model, a related study reported that defects in the pyruvate dehydrogenase aerobic respiration gene of V. cholerae resulted in a significant decrease in colonization rates ( ). In vitro , V. cholerae undergoes aerobic respiration, which produces the metabolic intermediates succinate and pyruvate, resulting in increased motility ( ). In addition, aerobic respiration promotes the transcription of the virulence factor toxT in V. cholerae during pathogenesis. Consequently, aerobic respiration plays a vital role in the pathogenicity of V. cholerae ( ; ). Concerning the control of cholera, it would be desirable to identify mechanisms that regulate aerobic respiration in V. cholerae and establish methods that can attenuate its effects. The tricarboxylic acid (TCA) cycle is an important intermediate link in aerobic respiration and is regarded as the energy-generating engine of aerobic respiration for ATP synthesis. This function connects glycolysis and the electron transport chain and is a central part of cellular energy metabolism ( ). Aerobic respiration control A (ArcA) is a response factor in the two-component Arc system that acts as a global inhibitor of the aerobic respiratory pathway (particularly the TCA cycle), thereby promoting the bacterial fermentation pathway ( ). Recently, a study demonstrated that overexpression of ArcA under aerobic conditions leads to downregulation of the respiratory pathway in E. coli ( ). Hence, it is important to investigate the regulatory mechanisms of aerobic respiration in V. cholerae in terms of the global inhibitors of the TCA cycle. Serine protease DegS is commonly recognized as an initiator of the σ E ( rpoE ) stress response pathway ( ), which affects V. cholerae motility, chemotaxis and antioxidant capacity ( ; ). Our previous results by RNA sequencing (RNA-seq) showed that the knockout of degS resulted in the downregulation of genes associated with aerobic respiration, which are focused on the TCA cycle ( ), but the mechanisms involved are not clear. In the current study, metabolomics analysis revealed that the differential metabolites of the ΔdegS mutant were mainly enriched in purine metabolism and glutathione metabolism associated with aerobic respiration, suggesting that DegS may regulate aerobic respiration in V. cholerae. This study investigated the influence of DegS on the key products of aerobic respiration, NADH, and ATP. Our results suggest that DegS affects isocitrate dehydrogenase (ICDH) expression through the regulation of ArcA, thereby affecting aerobic respiration in V. cholerae , which in turn affects NADH and ATP production.
Bacterial strains and growth conditions Non-O1/non-O139 V. cholerae HN375 from the China Center for Type Culture Collection (CCTCCAB209168) was used as the wild-type (WT) strain ( ). Cloning was carried out using Escherichia coli DH5 and DH5-λpir, and conjugation were carried out using WM3064. Each strain was grown on Luria-Bertani (LB) medium at 37°C until the stationary phase was achieved unless otherwise indicated. The culture medium was modified by adding 0.1% arabinose and 100 g/mL ampicillin depending on the situation. Details of all plasmids and strains used are presented in . DNA manipulations and genetic techniques From the WT HN375 strain, deletion mutants were constructed with pWM91, a suicide plasmid ( ). A list of the primers used can be found in . To construct complementary mutants, the entire arcA encoding region by cloning into the pBAD24 plasmid vector, which was then transformed into ΔdegSΔarcA by electroporation to obtain ΔdegSΔarcA:arcA . A similar method was used to construct ΔdegS -overexpressed icdh strains ( ΔdegS+icdh ). As previously described, we used the pBAD24- arcA plasmid as a template and constructed a point mutation in D54E of arcA using site-directed mutagenesis ( ). The pBAD24- arcA D54E plasmid vector was then transformed into ΔdegSΔarcA by electroporation to obtain ΔdegSΔarcA:arcA D54E . The complement and overexpression strains were grown in LB liquid medium with 0.1% arabinose for gene expression induction. Untargeted metabolomic analysis The WT and ΔdegS strains were added into sterile LB liquid medium and shaken at 220 rpm and incubated at 37°C until the logarithmic growth phase (optical density at 600 nm [OD 600 ] = 0.6). Both cultures were centrifuged at 10,000 g during 10 min at 4°C. After collection, the pellets were washed twice with 50 mM PBS, and used for untargeted metabolomic analysis. This analysis was performed by Biotech-Pack Scientific Co., Ltd. (Beijing, China). The Analysis Base File (ABF) converter software was used to convert the liquid chromatography-mass spectrometry (LC-MS) raw data into the ABF format ( ). The ABF format file was imported into MS-DIAL 4.10 software for preprocessing ( ), including peak extraction, noise removal, inverse convolution, and alignment. The three-dimensional (3D) data matrix was exported in the CSV format (raw data matrix). Finally, the extracted peak information was searched against the MassBank, Respect, and Global Natural Product Social Molecular Networking (GNPS), for a full library comparison. Quantitative RT-PCR All strains were grown to the stationary phase (OD 600 = 1.2) in an LB liquid medium. Bacterial cultures were collected via centrifugation at 8000 rpm for 5 min. Total RNA was extracted with TRIzol reagent and reverse-transcribed into cDNA. qRT-PCR was performed using the TB Green Premix Ex TaqII (TaKaRa Bio, Shiga, Japan) ( ). The 2 -ΔΔct method was used to calculate mRNA levels relative to each other ( ). For each qRT-PCR, two independent experiments were performed, each with three technical replicates. Assay of bacterial NADH levels To examine the NADH levels of bacteria grown in LB liquid medium or M9 liquid medium (containing 0.4% glucose) to stationary phase (OD 600 = 1.2), the concentration of bacteria was adjusted to approximately 1 × 10 8 CFU/mL. The bacteria were ultrasonically lysed. NADH levels were measured using the Amplite™ Colorimetric NADH Assay Kit (AAT Bioquest, Pleasanton, CA, USA). Briefly, equal volumes of bacterial suspension and Amplite™ Colorimetric NADH Assay Kit working solution were mixed and dispensed in wells of a clear 96-well plate, followed by incubation at 26°C for 15 min to 2 h. Read the absorbance at 460 nm using an enzyme marker and construct a standard curve using the kit’s NADH standard ( ). The experiment was repeated three times. Bacterial ATP levels assay ATP levels were determined using an ATP Assay Kit (Beyotime, Shanghai, China) following the manufacturer’s instructions. Briefly, the bacteria were cultured to stationary phase in LB liquid medium or M9 liquid medium (containing 0.4% glucose) and the bacterial concentration was adjusted to 1 × 10 8 CFU/mL. The bacteria were ultrasonically lysed. Equal volumes of the bacterial suspension and the working solution in the ATP reagent were mixed in a black opaque 96-well plate and incubated at 26°C for 5 min ( ). Luminescence was detected using a multifunctional enzyme labeler (Thermo Fisher Scientific Inc, Waltham, MA, USA). A standard curve was plotted using the standards provided in the kit. The experiment was repeated three times. Recombinant protein expression, purification, and preparation of polyclonal antisera The His-tagged recombinant ArcA protein was constructed as previously described ( ). Briefly, primers were used to amplify the full-length ArcA-encoded open reading frames. The PCR product was ligated into the pET28a vector and transformed into E. coli BL21 (DE3). Transformed bacteria expressing ArcA were grown to OD 600 = 0.6 at 37°C and induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 18°C for 16 h in LB liquid medium. Recombinant proteins labeled with His were purified by nickel-nitrilotriacetic acid affinity chromatography. Eight 6-week-old CD1 female mice, provided by the animal center of Zunyi Medical University (Zunyi, China), were housed in a specific pathogen-free environment and used to prepare anti-ArcA serum for western blot analysis. Mice were subcutaneously inoculated with 30 µg of recombinant ArcA on days 0, 14, and 28, along with the same volume of aluminum adjuvant. Anti-ArcA antiserum was prepared from blood collected 2 weeks after the last immunization. Western blot All strains were grown to stationary phase (OD 600 = 1.2) in an LB liquid medium. To prepare whole bacterial proteins, the culture was subjected to centrifugation, resulting in the separation of a supernatant layer. The precipitate was resuspended with 100 μL of distilled deionized water and 25 μL of 5× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) protein sampling buffer was added and boiled. The bacterial proteins were resolved by SDS-PAGE and transferred to a polyvinylidene fluoride membrane. After blocking, the membrane was incubated with a primary antibody (anti-ArcA serum at a dilution of 1:1000, prepared in our laboratory) overnight at 4°C. The membrane was incubated with a 1:5000 dilution of horseradish peroxidase-conjugated sheep anti-mouse IgG as a secondary antibody for 2 h after three washes with 1× Tris-buffered saline containing 0.1% Tween-20 (TBST). Finally, the membrane was washed three times with TBST, and color developed after the addition of chemiluminescent reagents (Epizyme Biomedical Technology Co. Ltd, Shanghai, China). The experiment was repeated three times. Bacterial growth curves Growth curves were generated as described previously ( ) with certain modifications. Briefly, the bacteria were cultured in LB liquid medium at 37°C until the stationary phase (OD 600 = 1.2). Aliquots of the culture were inoculated (1:500 v/v) into M9 liquid medium containing 0.4% glucose and incubated at 37°C with shaking at 200 rpm. Measurements were taken hourly for absorbance at 600 nm. The experiment was repeated three times. Suckling mouse colonization assay Six-day-old CD1 suckling mice which were randomized into the experimental and control groups (n = 8 per group). All animal experiments were approved by the Ethics Committee of Zunyi Medical University (No. ZMU21-2301-069). All strains were grown at 37°C to stationary phase (OD 600 = 1.2), and the bacteria were collected by centrifugation at 1,200 × g for 5 min. The bacterial concentration was adjusted to 1 × 10 7 CFU/mL with PBS. Each suckling mouse in the experimental group was gavaged with 50 μL of the bacterial suspension. The same volume of 1× PBS was used in the negative control. In the 18th hour following gavage, mice were euthanized. The small intestinal tissue was subsequently dissected, weighed, and homogenized ( ). After 100-fold dilution of this preparation, 100 μL aliquots were added to 0.5 mg/L gentamicin agar plates, and the colonies were counted after 18 h of incubation at 37°C. The final results are presented as the logarithm CFU/g. Statistical analyses Data are expressed as mean ± standard deviation. Non-paired two-tailed t-tests were used to analyze differences between two groups, and a one-way analysis of variance was used to analyze differences between multiple groups. SPSS version 29.0 (IBM Corp., Armonk, NY, USA) was used for the analyses. P <0.05 indicated statistical significance.
Non-O1/non-O139 V. cholerae HN375 from the China Center for Type Culture Collection (CCTCCAB209168) was used as the wild-type (WT) strain ( ). Cloning was carried out using Escherichia coli DH5 and DH5-λpir, and conjugation were carried out using WM3064. Each strain was grown on Luria-Bertani (LB) medium at 37°C until the stationary phase was achieved unless otherwise indicated. The culture medium was modified by adding 0.1% arabinose and 100 g/mL ampicillin depending on the situation. Details of all plasmids and strains used are presented in .
From the WT HN375 strain, deletion mutants were constructed with pWM91, a suicide plasmid ( ). A list of the primers used can be found in . To construct complementary mutants, the entire arcA encoding region by cloning into the pBAD24 plasmid vector, which was then transformed into ΔdegSΔarcA by electroporation to obtain ΔdegSΔarcA:arcA . A similar method was used to construct ΔdegS -overexpressed icdh strains ( ΔdegS+icdh ). As previously described, we used the pBAD24- arcA plasmid as a template and constructed a point mutation in D54E of arcA using site-directed mutagenesis ( ). The pBAD24- arcA D54E plasmid vector was then transformed into ΔdegSΔarcA by electroporation to obtain ΔdegSΔarcA:arcA D54E . The complement and overexpression strains were grown in LB liquid medium with 0.1% arabinose for gene expression induction.
The WT and ΔdegS strains were added into sterile LB liquid medium and shaken at 220 rpm and incubated at 37°C until the logarithmic growth phase (optical density at 600 nm [OD 600 ] = 0.6). Both cultures were centrifuged at 10,000 g during 10 min at 4°C. After collection, the pellets were washed twice with 50 mM PBS, and used for untargeted metabolomic analysis. This analysis was performed by Biotech-Pack Scientific Co., Ltd. (Beijing, China). The Analysis Base File (ABF) converter software was used to convert the liquid chromatography-mass spectrometry (LC-MS) raw data into the ABF format ( ). The ABF format file was imported into MS-DIAL 4.10 software for preprocessing ( ), including peak extraction, noise removal, inverse convolution, and alignment. The three-dimensional (3D) data matrix was exported in the CSV format (raw data matrix). Finally, the extracted peak information was searched against the MassBank, Respect, and Global Natural Product Social Molecular Networking (GNPS), for a full library comparison.
All strains were grown to the stationary phase (OD 600 = 1.2) in an LB liquid medium. Bacterial cultures were collected via centrifugation at 8000 rpm for 5 min. Total RNA was extracted with TRIzol reagent and reverse-transcribed into cDNA. qRT-PCR was performed using the TB Green Premix Ex TaqII (TaKaRa Bio, Shiga, Japan) ( ). The 2 -ΔΔct method was used to calculate mRNA levels relative to each other ( ). For each qRT-PCR, two independent experiments were performed, each with three technical replicates.
To examine the NADH levels of bacteria grown in LB liquid medium or M9 liquid medium (containing 0.4% glucose) to stationary phase (OD 600 = 1.2), the concentration of bacteria was adjusted to approximately 1 × 10 8 CFU/mL. The bacteria were ultrasonically lysed. NADH levels were measured using the Amplite™ Colorimetric NADH Assay Kit (AAT Bioquest, Pleasanton, CA, USA). Briefly, equal volumes of bacterial suspension and Amplite™ Colorimetric NADH Assay Kit working solution were mixed and dispensed in wells of a clear 96-well plate, followed by incubation at 26°C for 15 min to 2 h. Read the absorbance at 460 nm using an enzyme marker and construct a standard curve using the kit’s NADH standard ( ). The experiment was repeated three times.
ATP levels were determined using an ATP Assay Kit (Beyotime, Shanghai, China) following the manufacturer’s instructions. Briefly, the bacteria were cultured to stationary phase in LB liquid medium or M9 liquid medium (containing 0.4% glucose) and the bacterial concentration was adjusted to 1 × 10 8 CFU/mL. The bacteria were ultrasonically lysed. Equal volumes of the bacterial suspension and the working solution in the ATP reagent were mixed in a black opaque 96-well plate and incubated at 26°C for 5 min ( ). Luminescence was detected using a multifunctional enzyme labeler (Thermo Fisher Scientific Inc, Waltham, MA, USA). A standard curve was plotted using the standards provided in the kit. The experiment was repeated three times.
The His-tagged recombinant ArcA protein was constructed as previously described ( ). Briefly, primers were used to amplify the full-length ArcA-encoded open reading frames. The PCR product was ligated into the pET28a vector and transformed into E. coli BL21 (DE3). Transformed bacteria expressing ArcA were grown to OD 600 = 0.6 at 37°C and induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 18°C for 16 h in LB liquid medium. Recombinant proteins labeled with His were purified by nickel-nitrilotriacetic acid affinity chromatography. Eight 6-week-old CD1 female mice, provided by the animal center of Zunyi Medical University (Zunyi, China), were housed in a specific pathogen-free environment and used to prepare anti-ArcA serum for western blot analysis. Mice were subcutaneously inoculated with 30 µg of recombinant ArcA on days 0, 14, and 28, along with the same volume of aluminum adjuvant. Anti-ArcA antiserum was prepared from blood collected 2 weeks after the last immunization.
All strains were grown to stationary phase (OD 600 = 1.2) in an LB liquid medium. To prepare whole bacterial proteins, the culture was subjected to centrifugation, resulting in the separation of a supernatant layer. The precipitate was resuspended with 100 μL of distilled deionized water and 25 μL of 5× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) protein sampling buffer was added and boiled. The bacterial proteins were resolved by SDS-PAGE and transferred to a polyvinylidene fluoride membrane. After blocking, the membrane was incubated with a primary antibody (anti-ArcA serum at a dilution of 1:1000, prepared in our laboratory) overnight at 4°C. The membrane was incubated with a 1:5000 dilution of horseradish peroxidase-conjugated sheep anti-mouse IgG as a secondary antibody for 2 h after three washes with 1× Tris-buffered saline containing 0.1% Tween-20 (TBST). Finally, the membrane was washed three times with TBST, and color developed after the addition of chemiluminescent reagents (Epizyme Biomedical Technology Co. Ltd, Shanghai, China). The experiment was repeated three times.
Growth curves were generated as described previously ( ) with certain modifications. Briefly, the bacteria were cultured in LB liquid medium at 37°C until the stationary phase (OD 600 = 1.2). Aliquots of the culture were inoculated (1:500 v/v) into M9 liquid medium containing 0.4% glucose and incubated at 37°C with shaking at 200 rpm. Measurements were taken hourly for absorbance at 600 nm. The experiment was repeated three times.
Six-day-old CD1 suckling mice which were randomized into the experimental and control groups (n = 8 per group). All animal experiments were approved by the Ethics Committee of Zunyi Medical University (No. ZMU21-2301-069). All strains were grown at 37°C to stationary phase (OD 600 = 1.2), and the bacteria were collected by centrifugation at 1,200 × g for 5 min. The bacterial concentration was adjusted to 1 × 10 7 CFU/mL with PBS. Each suckling mouse in the experimental group was gavaged with 50 μL of the bacterial suspension. The same volume of 1× PBS was used in the negative control. In the 18th hour following gavage, mice were euthanized. The small intestinal tissue was subsequently dissected, weighed, and homogenized ( ). After 100-fold dilution of this preparation, 100 μL aliquots were added to 0.5 mg/L gentamicin agar plates, and the colonies were counted after 18 h of incubation at 37°C. The final results are presented as the logarithm CFU/g.
Data are expressed as mean ± standard deviation. Non-paired two-tailed t-tests were used to analyze differences between two groups, and a one-way analysis of variance was used to analyze differences between multiple groups. SPSS version 29.0 (IBM Corp., Armonk, NY, USA) was used for the analyses. P <0.05 indicated statistical significance.
Non-targeted metabolomic analysis of the degS mutant Our previous RNA-seq data suggested that the knockout of degS results in the downregulation of genes related to the TCA cycle of the aerobic respiration pathway ( ). The finding implies that DegS may affect aerobic respiration in V. cholerae . To further test this hypothesis, we conducted untargeted metabolomics on degS knockout mutants ( ΔdegS ). The analysis identified a total of 109 metabolites ( ; ). Significantly ( P <0.05) differentially expressed metabolites were screened according to fold changes >2 or <0.5. A combination of multidimensional and unidimensional analyses identified 19 significantly differentially expressed metabolites ( ). One-dimensional statistical analyses were performed using multiplicity and t-tests. The resulting data were plotted as volcano plots ( ). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of enriched genes mainly revealed genes involved in purine and glutathione metabolism ( ). Glutathione metabolism is an important component of aerobic respiration and provides important redox buffers ( ; ). Purine metabolism provides purine nucleotides that are essential for ATP production from aerobic respiration ( ). These findings indicate that DegS has an impact on aerobic respiration in V. cholerae . DegS positively affects NADH and ATP levels in V. cholerae Given the transcriptome and metabolome results, we hypothesized that DegS may affect the production of the aerobic respiration pathway products NADH and ATP. To test this hypothesis, we determined the levels of NADH in WT and ΔdegS strains. The NADH levels of the WT strain were approximately twice as high as those of the ΔdegS mutants, whereas the NADH levels of the complemented strain ΔdegS::degS were able to restore NADH levels close to those of the WT strain ( ). The pBAD24 null plasmid was unable to recover the NADH levels of ΔdegS . Subsequently, we investigated ATP levels in the above strains, which is the final energy product of the aerobic respiration pathway. The ATP level of the WT strain was approximately thrice that of ΔdegS , whereas the ATP levels of ΔdegS::degS were partially restored, with no restoration using the pBAD24 empty plasmid ( ). The above data suggest that DegS influences positively NADH and ATP levels in V. cholerae. DegS positively affects NADH and ATP levels in V. cholerae independent of σ E DegS regulates stress response and motility through σ E and DegS deficiency significantly reduces σ E activity ( ). To investigate whether DegS affects V. cholerae NADH and ATP levels via σ E , we first examined the transcript levels of rpoE . The qRT-PCR results showed that the rpoE gene transcript level in the WT strain was approximately four times higher than that of ΔdegS , while the transcript level of the rpoE gene in ΔdegS::degS was almost the same as that of the WT strain ( ). Next, we performed experiments for the detection of NADH and ATP levels using the rpoE deletion mutant ( ΔrpoE ) and corresponding complemented strain ( ΔrpoE::rpoE ). The NADH level of the ΔrpoE mutant was not statistically different from the WT and ΔrpoE::rpoE strains ( ). The ATP levels of the rpoE mutation did not differ from that of the WT and ΔrpoE::rpoE strains ( ). We used qRT-PCR to screen for changes in the expression of some aerobic respiratory genes in different strains, and the expression levels of genes encoding type I glyceraldehyde-3-phosphate dehydrogenase (GAP), isocitrate dehydrogenase (ICDH), and phosphoenolpyruvate carboxykinase (PckA) were significantly reduced in the ΔdegS mutant as compared with the WT strain ( ). However, the expression of the above genes did not change after rpoE knockout. These results indicate that DegS positively affects NADH and ATP levels in V. cholerae independent of σ E . Effect of DegS on NADH and ATP levels in V. cholerae involves ArcA In S. typhimurium , ArcA may negatively regulate ATP and NADH levels by inhibiting gene transcription levels of the pyruvate dehydrogenase complex (PDH) in the TCA cycle ( ). Using qRT-PCR, we observed that the transcript level of arcA in ΔdegS was approximately six times higher than that of the WT strain ( ), suggesting that DegS is a negative regulator of ArcA. Therefore, we hypothesized that DegS influences the NADH and ATP levels in V. cholerae through ArcA. To assess the hypothesis, we constructed ΔdegSΔarcA and ΔdegSΔarcA::arcA and measured the levels of NADH and ATP. Both levels in ΔdegSΔarcA could be partially restored compared to ΔdegS ( ). Next, we detected the expression level of ArcA protein in each strain. ArcA protein expression was almost the same in WT strains, ΔdegS , and ΔdegS::degS ( ).ArcA is a response factor in a two-component system that can activate downstream genes in a phosphorylated form. Meanwhile, in E. coli , ArcA is an important inhibitor, and its phosphorylated form directly inhibits the expression of some genes in the TCA cycle, such as citrate synthase (GltA) and malate dehydrogenase (MDH) ( ). Therefore, we speculated whether its phosphorylation modifications are involved in this regulatory process. Next, we constructed a point mutation model ( ΔdegSΔarcA::arcA D54E ) to mimic dephosphorylation ( ) to explore whether ArcA phosphorylation is associated with DegS affecting NADH and ATP levels in V. cholerae . Both NADH and ATP levels were decreased in the ΔdegSΔarcA::arcA D54E strain compared to the ΔdegSΔarcA strain. The ΔdegSΔarcA::arcA D54E strain had a smaller decrease in NADH and ATP levels than the ΔdegSΔarcA::arcA strain ( ). These results suggest that DegS affects NADH and ATP levels, which are partially dependent on ArcA phosphorylation. Effect of DegS on NADH and ATP levels in V. cholerae involved in expressing ICDH ICDH is a key rate-limiting enzyme of the TCA cycle; the knockdown of ICDH leads to a decrease in bacterial NADH and ATP levels ( ). Our qRT-PCR results revealed that the transcription level of icdh in ΔdegS strains was approximately five times lower than that of WT strains and that the transcriptional level of icdh was recovered in part in the ΔdegSΔarcA strain ( ). These findings suggest that DegS may control the transcription of icdh through the ArcA pathway. To determine whether ICDH is involved in regulating V. cholerae NADH and ATP levels in DegS, we overexpressed ICDH based on the ΔdegS strain and measured NADH and ATP levels. Both levels were partially restored in the ΔdegS+icdh strain, but not to the level of the WT strain ( ). Collectively, these results suggest that DegS is required for high levels of ATP and NADH because it indirectly increases ICDH expression. DegS affects the growth of V. cholerae through the ArcA-ICDH pathway Many enzymes and metabolites associated with bacterial energy metabolism have direct regulatory roles in bacterial growth ( ; ; ; ). Our experiments showed that DegS affects ATP and NADH levels in V. cholerae through the ArcA-ICDH signaling pathway. To confirm whether DegS affects growth in V. cholerae through this pathway, we conducted growth curve experiments in the M9 medium. The growth rate of the ΔdegS strain was lower than that of the WT strain during the logarithmic growth phase ( ). Compared to the ΔdegS strain, the ΔdegSΔarcA strain grew faster during the logarithmic growth period. The ΔdegS + icdh strain had a faster growth rate than the ΔdegS strain during the logarithmic growth phase ( ). Concurrently, the trends of NADH and ATP levels of the ΔdegSΔarcA strain and the ΔdegS + icdh strain in M9 medium corresponded to the trends of their growth rates in the logarithmic growth phase ( ). These evidences demonstrate that DegS affects the growth of V. cholerae through the ArcA-ICDH pathway. DegS affects V. cholerae intestinal colonization Inhibition of NADH and ATP production in bacteria affects their colonization ( ; ). To examine whether the regulation of V. cholerae NADH and ATP levels mediated by DegS is critical for bacterial colonization, we used a suckling mouse model of intestinal colonization. The in vivo results showed that compared with the WT strain, the colonization capacity of the ΔdegS strain was significantly reduced, while the colonization capacity of the ΔdegS::degS strain was similar to that of the WT strain, and the colonization ability of the ΔdegS+icdh strain was partially restored ( ). However, the colonization ability of ΔdegSΔarcA was comparable to that of ΔdegS , and the colonization ability of ΔdegSΔarcA::arcA was stronger than that of ΔdegSΔarcA strain.
degS mutant Our previous RNA-seq data suggested that the knockout of degS results in the downregulation of genes related to the TCA cycle of the aerobic respiration pathway ( ). The finding implies that DegS may affect aerobic respiration in V. cholerae . To further test this hypothesis, we conducted untargeted metabolomics on degS knockout mutants ( ΔdegS ). The analysis identified a total of 109 metabolites ( ; ). Significantly ( P <0.05) differentially expressed metabolites were screened according to fold changes >2 or <0.5. A combination of multidimensional and unidimensional analyses identified 19 significantly differentially expressed metabolites ( ). One-dimensional statistical analyses were performed using multiplicity and t-tests. The resulting data were plotted as volcano plots ( ). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of enriched genes mainly revealed genes involved in purine and glutathione metabolism ( ). Glutathione metabolism is an important component of aerobic respiration and provides important redox buffers ( ; ). Purine metabolism provides purine nucleotides that are essential for ATP production from aerobic respiration ( ). These findings indicate that DegS has an impact on aerobic respiration in V. cholerae .
V. cholerae Given the transcriptome and metabolome results, we hypothesized that DegS may affect the production of the aerobic respiration pathway products NADH and ATP. To test this hypothesis, we determined the levels of NADH in WT and ΔdegS strains. The NADH levels of the WT strain were approximately twice as high as those of the ΔdegS mutants, whereas the NADH levels of the complemented strain ΔdegS::degS were able to restore NADH levels close to those of the WT strain ( ). The pBAD24 null plasmid was unable to recover the NADH levels of ΔdegS . Subsequently, we investigated ATP levels in the above strains, which is the final energy product of the aerobic respiration pathway. The ATP level of the WT strain was approximately thrice that of ΔdegS , whereas the ATP levels of ΔdegS::degS were partially restored, with no restoration using the pBAD24 empty plasmid ( ). The above data suggest that DegS influences positively NADH and ATP levels in V. cholerae.
V. cholerae independent of σ E DegS regulates stress response and motility through σ E and DegS deficiency significantly reduces σ E activity ( ). To investigate whether DegS affects V. cholerae NADH and ATP levels via σ E , we first examined the transcript levels of rpoE . The qRT-PCR results showed that the rpoE gene transcript level in the WT strain was approximately four times higher than that of ΔdegS , while the transcript level of the rpoE gene in ΔdegS::degS was almost the same as that of the WT strain ( ). Next, we performed experiments for the detection of NADH and ATP levels using the rpoE deletion mutant ( ΔrpoE ) and corresponding complemented strain ( ΔrpoE::rpoE ). The NADH level of the ΔrpoE mutant was not statistically different from the WT and ΔrpoE::rpoE strains ( ). The ATP levels of the rpoE mutation did not differ from that of the WT and ΔrpoE::rpoE strains ( ). We used qRT-PCR to screen for changes in the expression of some aerobic respiratory genes in different strains, and the expression levels of genes encoding type I glyceraldehyde-3-phosphate dehydrogenase (GAP), isocitrate dehydrogenase (ICDH), and phosphoenolpyruvate carboxykinase (PckA) were significantly reduced in the ΔdegS mutant as compared with the WT strain ( ). However, the expression of the above genes did not change after rpoE knockout. These results indicate that DegS positively affects NADH and ATP levels in V. cholerae independent of σ E .
V. cholerae involves ArcA In S. typhimurium , ArcA may negatively regulate ATP and NADH levels by inhibiting gene transcription levels of the pyruvate dehydrogenase complex (PDH) in the TCA cycle ( ). Using qRT-PCR, we observed that the transcript level of arcA in ΔdegS was approximately six times higher than that of the WT strain ( ), suggesting that DegS is a negative regulator of ArcA. Therefore, we hypothesized that DegS influences the NADH and ATP levels in V. cholerae through ArcA. To assess the hypothesis, we constructed ΔdegSΔarcA and ΔdegSΔarcA::arcA and measured the levels of NADH and ATP. Both levels in ΔdegSΔarcA could be partially restored compared to ΔdegS ( ). Next, we detected the expression level of ArcA protein in each strain. ArcA protein expression was almost the same in WT strains, ΔdegS , and ΔdegS::degS ( ).ArcA is a response factor in a two-component system that can activate downstream genes in a phosphorylated form. Meanwhile, in E. coli , ArcA is an important inhibitor, and its phosphorylated form directly inhibits the expression of some genes in the TCA cycle, such as citrate synthase (GltA) and malate dehydrogenase (MDH) ( ). Therefore, we speculated whether its phosphorylation modifications are involved in this regulatory process. Next, we constructed a point mutation model ( ΔdegSΔarcA::arcA D54E ) to mimic dephosphorylation ( ) to explore whether ArcA phosphorylation is associated with DegS affecting NADH and ATP levels in V. cholerae . Both NADH and ATP levels were decreased in the ΔdegSΔarcA::arcA D54E strain compared to the ΔdegSΔarcA strain. The ΔdegSΔarcA::arcA D54E strain had a smaller decrease in NADH and ATP levels than the ΔdegSΔarcA::arcA strain ( ). These results suggest that DegS affects NADH and ATP levels, which are partially dependent on ArcA phosphorylation.
V. cholerae involved in expressing ICDH ICDH is a key rate-limiting enzyme of the TCA cycle; the knockdown of ICDH leads to a decrease in bacterial NADH and ATP levels ( ). Our qRT-PCR results revealed that the transcription level of icdh in ΔdegS strains was approximately five times lower than that of WT strains and that the transcriptional level of icdh was recovered in part in the ΔdegSΔarcA strain ( ). These findings suggest that DegS may control the transcription of icdh through the ArcA pathway. To determine whether ICDH is involved in regulating V. cholerae NADH and ATP levels in DegS, we overexpressed ICDH based on the ΔdegS strain and measured NADH and ATP levels. Both levels were partially restored in the ΔdegS+icdh strain, but not to the level of the WT strain ( ). Collectively, these results suggest that DegS is required for high levels of ATP and NADH because it indirectly increases ICDH expression.
V. cholerae through the ArcA-ICDH pathway Many enzymes and metabolites associated with bacterial energy metabolism have direct regulatory roles in bacterial growth ( ; ; ; ). Our experiments showed that DegS affects ATP and NADH levels in V. cholerae through the ArcA-ICDH signaling pathway. To confirm whether DegS affects growth in V. cholerae through this pathway, we conducted growth curve experiments in the M9 medium. The growth rate of the ΔdegS strain was lower than that of the WT strain during the logarithmic growth phase ( ). Compared to the ΔdegS strain, the ΔdegSΔarcA strain grew faster during the logarithmic growth period. The ΔdegS + icdh strain had a faster growth rate than the ΔdegS strain during the logarithmic growth phase ( ). Concurrently, the trends of NADH and ATP levels of the ΔdegSΔarcA strain and the ΔdegS + icdh strain in M9 medium corresponded to the trends of their growth rates in the logarithmic growth phase ( ). These evidences demonstrate that DegS affects the growth of V. cholerae through the ArcA-ICDH pathway.
V. cholerae intestinal colonization Inhibition of NADH and ATP production in bacteria affects their colonization ( ; ). To examine whether the regulation of V. cholerae NADH and ATP levels mediated by DegS is critical for bacterial colonization, we used a suckling mouse model of intestinal colonization. The in vivo results showed that compared with the WT strain, the colonization capacity of the ΔdegS strain was significantly reduced, while the colonization capacity of the ΔdegS::degS strain was similar to that of the WT strain, and the colonization ability of the ΔdegS+icdh strain was partially restored ( ). However, the colonization ability of ΔdegSΔarcA was comparable to that of ΔdegS , and the colonization ability of ΔdegSΔarcA::arcA was stronger than that of ΔdegSΔarcA strain.
Aerobic respiration is a major driver of V. cholerae proliferation during infection. V. cholerae require energy from aerobic respiration for subsequent proliferation and infection ( ). Here, we observed that DegS protease plays a vital role in NADH and ATP levels, growth, and colonization of V. cholerae . We propose a model whereby DegS positively regulates ATP and NADH levels to promote the growth of V. cholerae , which is in part dependent on the ArcA-ICDH pathway. In addition, there may be other factors (X) involved in the effects of DegS on V. cholerae NADH and ATP levels, and growth ( ). The DegS serine protease is located within the bacterial periplasm. The protein is thought to be involved in initiating the σ E stress response pathway, where active DegS catalyzes the cleavage of RseA, releasing active σ E , which activates σ E -regulated gene expression ( ; ). Although σ E is involved in a variety of biological processes, such as stress response, biofilm formation, and motility ( ), its relevance to aerobic respiration has remained unclear. Our study reveals that the levels of NADH and ATP, which are key products of aerobic respiration, decreased upon deletion of degS ( ). However, deletion of rpoE had little effect on NADH and ATP levels ( ). In addition, qRT-PCR results revealed no statistically significant changes in any of the relevant aerobic respiration genes in the ΔrpoE strain ( ). Given these results, we speculate that the effect of DegS on V. cholerae NADH and ATP levels is independent of σ E . This suggests that DegS may have a different pathway than the previous dependence on σ E . Further investigating the mechanism by which DegS affects NADH and ATP levels in V. cholerae , we observed using RNA-seq that deletion of degS mainly inhibits the TCA cycle, carbon metabolism, and pyruvate metabolism ( ). Meanwhile, qRT-PCR results showed that aerobic respiratory-related genes were altered in the ΔdegS mutant ( ). Among them, the expression of the gap gene, which is a key enzyme involved in glycolysis, was significantly reduced. Expression of the pckA gene, which is involved in gluconeogenesis, was also reduced. Notably, the expression of the icdh gene, a key gene in the TCA cycle, was significantly reduced. Since the TCA cycle is a major biochemical hub in most heterotrophic organisms, it is essential for aerobic respiration ( ; ). Therefore, we chose icdh to further investigate the mechanism by which DegS affects NADH and ATP levels in V. cholerae . ArcA acts as a response factor in a two-component system to directly or indirectly inhibit the TCA cycle, thereby reshuffling bacterial metabolic pathways and optimizing energy conversion ( ; ; ). ArcA as a global transcription factor responds to NADH and ATP ( ). In addition, the ΔarcA strain of Salmonella enterica displays higher levels of NADH ( ). In this study, we observed that an increase in transcript levels of arcA after knockout of degS ( ) and knockout of the arcA gene partially restored the low levels of NADH and ATP levels in the ΔdegS strain ( ). At the protein level, western blot experiments revealed no difference in ArcA protein expression in the ΔdegS strain ( ). Therefore, we speculate that the post-translational modification of ArcA may be involved in the regulation of NADH and ATP by DegS. ArcA can be activated as a transcription factor via phosphorylation to regulate the expression of downstream genes ( ). Therefore, suspecting that ArcA may play a role in phosphorylation, we constructed a model of dephosphorylation by point mutation ( ΔdegSΔarcA::arcA D54E ). Both NADH and ATP levels were lower significantly in the point mutant strain compared to the ΔdegSΔarcA strain, but not as much as in the ΔdegSΔarcA::arcA strain ( ). Thus, we speculated that DegS affects NADH and ATP levels is partially dependent on ArcA phosphorylation. This phenomenon is similar to the EnvZ/OmpR two-component system in Klebsiella pneumoniae , where the ΔompR mutant completely loses mucoviscosity compared to the wild-type strain, while the unphosphorylated ompR D55A mutant reduces mucoviscosity only to a lesser extent, suggesting that phosphorylation only partially affects its phenotype ( ). Brown et al. show that the conserved metabolic regulator ArcA responds to host-mediated cell envelope damage ( ). Meanwhile, DegS is a serine protease that mediates the cell envelope stress response, so we hypothesized that they might be linked through the cell envelope stress response pathway. ICDH is one of the vital rate-limiting enzymes in the TCA cycle ( ; ) and its transcription is dependent on ArcA ( ). In addition, knockout of icdh in the TCA cycle results in changes in the central metabolism of E. coli , such as a decrease in intracellular NADH and ATP levels and a decrease in the rate of glucose consumption ( ). In the current study, qRT-PCR showed that DegS positively regulated icdh , and ArcA negatively regulated the icdh gene ( ). Overexpression of ICDH partially restored NADH and ATP levels in the ΔdegS strain ( ). Based on these results, we suggest that DegS affects NADH and ATP levels in V. cholerae via ArcA, in relation to ICDH. Bacteria require energy to grow, and a decrease in the energy supply can inhibit their growth ( ; ). ATP and NADH are important components of the energy supply and are essential for bacterial growth. Previous studies demonstrated that ArcA affects bacterial NADH levels and growth ( ; ). In addition, deletion of E. coli icdh leads to alterations in NADH and ATP levels, thereby affecting specific growth ( ). Here, we observed that the trends in NADH and ATP levels of the ΔdegSΔarcA and ΔdegS+icdh strains in the M9 medium ( ) were consistent with the same growth rate trends in the logarithmic growth phase ( ). Therefore, we propose that DegS regulates V. cholerae NADH and ATP levels through the ArcA-ICDH pathway, thereby affecting V. cholerae growth. Given that the ΔdegSΔarcA strain and the ΔdegS+icdh strain only partially restored the growth rate of ΔdegS in the logarithmic growth phase, NADH and ATP levels were not fully restored in the ΔdegS+icdh strain, we speculate that DegS regulation of V. cholerae growth and the levels of ATP and NADH may involve other factors. Energy is an important driver of bacterial colonization ( ). We observed a highly significantly reduced colonization ability in the ΔdegS mutant, consistent with our previous study ( ). However, the colonization ability of ΔdegSΔarcA was not restored to a certain extent. This result may be due to the fact that arcA is required for V. cholerae biofilm formation ( ), which is important for intestinal colonization ( ). Compared to the ΔdegSΔarcA strain, the ΔdegS+icdh strain are directly overexpressing the icdh gene and do not involve the knockout of the arcA gene, so their colonization ability can be partially restored.
In summary, we demonstrate that deletion of degS leads to a decrease in NADH and ATP levels in V. cholerae and is independent of σ E , thus inhibiting V. cholerae growth. Furthermore, our findings indicated that DegS may be a prospective mechanism for regulating ICDH expression through ArcA. These findings enhance the knowledge of the biological functions of DegS and offer new perspectives on the regulation of NADH and ATP levels in V. cholerae . The role of DegS in ICDH regulation through ArcA independent of σ E needs to be further investigated.
|
Associations Between Sociodemographic Characteristics, eHealth Literacy, and Health-Promoting Lifestyle Among University Students in Taipei: Cross-Sectional Validation Study of the Chinese Version of the eHealth Literacy Scale | 8c689cff-4599-49b8-a597-6fbb4075a0aa | 11294764 | Health Literacy[mh] | Background In recent years, the accessibility and convenience of the internet have increased. This has allowed the public to use it more frequently for communication, education, work, or recreation. A Taiwan Network Information Center (TWNIC) survey revealed that the percentage of individuals with internet access has remained above 80% since 2015 and reached 84.3% in 2022 . Notably, generation Z (aged ≤25 years) had an internet access rate of 100%, and 20.39% of the general population consistently maintained an active online status . Nevertheless, there exists an opportunity for enhancing public digital literacy . Recent studies have highlighted that 81% of adults in the United States possess the ability to search the internet, with 72% using online sources for health-related information . The rapid and discreet nature of the internet considerably increases the public’s inclination to use the internet for accessing health-related information . However, online health information may contain complex, inaccurate, or even misleading content, resulting in low comprehensibility and reliability . Individuals may inadvertently jeopardize their well-being if they lack the ability to comprehend and critically evaluate online health information . Consequently, the concept of eHealth literacy has gradually attracted attention. Health literacy refers to the ability of an individual to engage with health information. The World Health Organization defines health literacy as “the cognitive and social skills which determine the motivation and ability of individuals to gain access to, understand and use information in ways which promote and maintain good health” . This concept can be further divided into 3 levels, namely basic/functional, communicative/interactive, and critical . Studies have indicated that individuals with high health literacy tend to effectively comprehend medical information and frequently engage in health-promoting behaviors, thereby cultivating a healthier lifestyle . eHealth literacy refers to the aptitude for sourcing, comprehending, assessing, and applying health information from the internet to address health problems. Scholars have used the lily model to delineate the 6 fundamental competencies in eHealth literacy . In addition to the aforementioned health literacy, these 6 competencies in eHealth literacy extend to traditional, information, scientific, media, and computer literacies . eHealth literacy has been positively correlated with health literacy . Studies have highlighted that eHealth literacy may potentially affect the intention and behavior of an individual to use online health information . Individuals with high eHealth literacy tend to actively search and review health information online, leveraging it to enhance their health behaviors and self-manage their health care needs . The eHealth Literacy Scale (eHEALS) was the first lily model–based tool developed for measuring eHealth literacy . Exploratory factor analysis (EFA) confirms that eHEALS consists of a single factor with 8 questions . The scale is designed to be user friendly, demonstrating strong validity, reliability, and stability . In systematic reviews, eHEALS is the most commonly used tool for eHealth literacy evaluation beyond its initial publication . Both the original and translated eHEALS versions are widely used across different countries and populations . However, limitations exist. Studies have highlighted that eHEALS only assesses 1 dimension, rendering it difficult to effectively evaluate diverse eHealth literacy aspects . Another study noted that the rapid spread of social media and mobile devices in recent years could potentially render eHEALS inadequate in completely capturing the contemporary eHealth literacy of individuals . Nonetheless, although EFA is extremely useful for reducing many questions to a manageable amount, only confirmatory factor analysis (CFA) of a multiple-factor model can rigorously evaluate the one-dimensionality of the scale . Recent studies using CFA have found that eHEALS is not a unidimensional concept, and the fit of the 2-factor model is better than that of the 1-factor model but still not good enough . Therefore, researchers have further recommended using a 3-factor model of the original eHEALS as it has a better fit and can effectively measure an individual’s current eHealth information skills and comfort . An attempt was made to divide the Chinese eHEALS into 4 factors for discussion; however, some factors only cover 1 question, and the method for dividing the factors and the model fit have not been determined . In summary, as eHEALS has only 8 questions, it is more suitable to divide it into 2 or 3 factors. However, further study is required to determine the fit, validity, and reliability of the Chinese eHEALS with 2-factor or 3-factor models. Researchers have emphasized that eHealth literacy is not a static trait but evolves in response to changes in individual circumstances, societal dynamics, and environmental factors . Several studies have identified variations in eHealth literacy across different sexes, educational backgrounds, income levels, health status, degree of health concern, and frequency of health-related discussions . eHealth literacy is considered to be positively correlated with many health behaviors adopted by an individual. Recent studies have highlighted that individuals with high eHealth literacy have positive social relationships, a balanced diet, and safe sex practice . Researchers have found that people with high eHealth literacy exercise and eat breakfast regularly . Studies have also proved that high eHealth literacy predicts balanced eating, regular exercise, and good sleep behaviors . Additionally, cross-sectional and longitudinal studies have indicated that people with high eHealth literacy can successfully cultivate a health-promoting lifestyle that includes health responsibility, exercise, nutrition, self-actualization, stress management, and interpersonal support . In Taiwan, adults aged 18-29 years, often called digital natives, have grown up with the internet. The TWNIC survey revealed that less than 1% of this demographic has never used the internet . Notably, permanent online engagement is particularly pronounced in this population, inherently amplifying opportunities for internet-based information retrieval and usage . Entering university is an important stage when adolescents transition into adulthood. Studies have shown that when presented with substantial internet-based health information encompassing both accurate and misleading content, university students may encounter challenges in accessing dependable sources and using effective evaluation methods, underscoring the need for continued strengthening of their eHealth literacy . Moreover, although in good health, university students tend to exhibit risky health behaviors . However, limited Taiwanese studies have explored the correlation between eHEALS and the health-promoting lifestyle profile (HPLP) among university students, highlighting an urgent need to ascertain and address their eHealth literacy educational requirement. Objective This study aimed to evaluate the fit, validity, and reliability of Chinese eHEALS 2- and 3-factor models. Moreover, the relationship between eHealth literacy and the HPLP among university students was explored. Specifically, this study sought to uncover sociodemographic factors capable of confounding a health-promoting lifestyle among university students and to ascertain the predictive effects of eHealth literacy on adopting a health-promoting lifestyle by excluding the influence of sociodemographic confounders. Finally, the study proposed health education advice that aligns with the current trends, addressing the specific requirements of individuals with lower eHealth literacy who need prompt intervention.
In recent years, the accessibility and convenience of the internet have increased. This has allowed the public to use it more frequently for communication, education, work, or recreation. A Taiwan Network Information Center (TWNIC) survey revealed that the percentage of individuals with internet access has remained above 80% since 2015 and reached 84.3% in 2022 . Notably, generation Z (aged ≤25 years) had an internet access rate of 100%, and 20.39% of the general population consistently maintained an active online status . Nevertheless, there exists an opportunity for enhancing public digital literacy . Recent studies have highlighted that 81% of adults in the United States possess the ability to search the internet, with 72% using online sources for health-related information . The rapid and discreet nature of the internet considerably increases the public’s inclination to use the internet for accessing health-related information . However, online health information may contain complex, inaccurate, or even misleading content, resulting in low comprehensibility and reliability . Individuals may inadvertently jeopardize their well-being if they lack the ability to comprehend and critically evaluate online health information . Consequently, the concept of eHealth literacy has gradually attracted attention. Health literacy refers to the ability of an individual to engage with health information. The World Health Organization defines health literacy as “the cognitive and social skills which determine the motivation and ability of individuals to gain access to, understand and use information in ways which promote and maintain good health” . This concept can be further divided into 3 levels, namely basic/functional, communicative/interactive, and critical . Studies have indicated that individuals with high health literacy tend to effectively comprehend medical information and frequently engage in health-promoting behaviors, thereby cultivating a healthier lifestyle . eHealth literacy refers to the aptitude for sourcing, comprehending, assessing, and applying health information from the internet to address health problems. Scholars have used the lily model to delineate the 6 fundamental competencies in eHealth literacy . In addition to the aforementioned health literacy, these 6 competencies in eHealth literacy extend to traditional, information, scientific, media, and computer literacies . eHealth literacy has been positively correlated with health literacy . Studies have highlighted that eHealth literacy may potentially affect the intention and behavior of an individual to use online health information . Individuals with high eHealth literacy tend to actively search and review health information online, leveraging it to enhance their health behaviors and self-manage their health care needs . The eHealth Literacy Scale (eHEALS) was the first lily model–based tool developed for measuring eHealth literacy . Exploratory factor analysis (EFA) confirms that eHEALS consists of a single factor with 8 questions . The scale is designed to be user friendly, demonstrating strong validity, reliability, and stability . In systematic reviews, eHEALS is the most commonly used tool for eHealth literacy evaluation beyond its initial publication . Both the original and translated eHEALS versions are widely used across different countries and populations . However, limitations exist. Studies have highlighted that eHEALS only assesses 1 dimension, rendering it difficult to effectively evaluate diverse eHealth literacy aspects . Another study noted that the rapid spread of social media and mobile devices in recent years could potentially render eHEALS inadequate in completely capturing the contemporary eHealth literacy of individuals . Nonetheless, although EFA is extremely useful for reducing many questions to a manageable amount, only confirmatory factor analysis (CFA) of a multiple-factor model can rigorously evaluate the one-dimensionality of the scale . Recent studies using CFA have found that eHEALS is not a unidimensional concept, and the fit of the 2-factor model is better than that of the 1-factor model but still not good enough . Therefore, researchers have further recommended using a 3-factor model of the original eHEALS as it has a better fit and can effectively measure an individual’s current eHealth information skills and comfort . An attempt was made to divide the Chinese eHEALS into 4 factors for discussion; however, some factors only cover 1 question, and the method for dividing the factors and the model fit have not been determined . In summary, as eHEALS has only 8 questions, it is more suitable to divide it into 2 or 3 factors. However, further study is required to determine the fit, validity, and reliability of the Chinese eHEALS with 2-factor or 3-factor models. Researchers have emphasized that eHealth literacy is not a static trait but evolves in response to changes in individual circumstances, societal dynamics, and environmental factors . Several studies have identified variations in eHealth literacy across different sexes, educational backgrounds, income levels, health status, degree of health concern, and frequency of health-related discussions . eHealth literacy is considered to be positively correlated with many health behaviors adopted by an individual. Recent studies have highlighted that individuals with high eHealth literacy have positive social relationships, a balanced diet, and safe sex practice . Researchers have found that people with high eHealth literacy exercise and eat breakfast regularly . Studies have also proved that high eHealth literacy predicts balanced eating, regular exercise, and good sleep behaviors . Additionally, cross-sectional and longitudinal studies have indicated that people with high eHealth literacy can successfully cultivate a health-promoting lifestyle that includes health responsibility, exercise, nutrition, self-actualization, stress management, and interpersonal support . In Taiwan, adults aged 18-29 years, often called digital natives, have grown up with the internet. The TWNIC survey revealed that less than 1% of this demographic has never used the internet . Notably, permanent online engagement is particularly pronounced in this population, inherently amplifying opportunities for internet-based information retrieval and usage . Entering university is an important stage when adolescents transition into adulthood. Studies have shown that when presented with substantial internet-based health information encompassing both accurate and misleading content, university students may encounter challenges in accessing dependable sources and using effective evaluation methods, underscoring the need for continued strengthening of their eHealth literacy . Moreover, although in good health, university students tend to exhibit risky health behaviors . However, limited Taiwanese studies have explored the correlation between eHEALS and the health-promoting lifestyle profile (HPLP) among university students, highlighting an urgent need to ascertain and address their eHealth literacy educational requirement.
This study aimed to evaluate the fit, validity, and reliability of Chinese eHEALS 2- and 3-factor models. Moreover, the relationship between eHealth literacy and the HPLP among university students was explored. Specifically, this study sought to uncover sociodemographic factors capable of confounding a health-promoting lifestyle among university students and to ascertain the predictive effects of eHealth literacy on adopting a health-promoting lifestyle by excluding the influence of sociodemographic confounders. Finally, the study proposed health education advice that aligns with the current trends, addressing the specific requirements of individuals with lower eHealth literacy who need prompt intervention.
Study Design and Participants This cross-sectional quantitative study was conducted among university students in Taipei, the capital of Taiwan. Two rounds of testing (pretest and formal) were performed, and 2-stage sampling was used in both rounds and the subjects were not repeated. The pretest was conducted to determine the reliability of the Chinese versions of eHEALS and the HPLP in the study population and to conduct EFA to extract the 2- and 3-factor models of the Chinese eHEALS. Next, the formal test was conducted to conduct CFA to further determine the fit, validity, and reliability of the Chinese eHEALS multiple-factor models and to perform inferential statistics on the predictive effects of eHEALS on the HPLP. In the first stage of the pretest, stratification was conducted based on the types of universities in Taipei. The Taiwan Ministry of Education has classified 24 universities in Taipei based on ownership and educational goals into 6 (25%) public general universities, 5 (21%) public vocational colleges, 8 (33%) private general universities, and 5 (21%) private vocational colleges . One school was randomly selected from each stratum for testing. Subsequently, in stage 2, convenience sampling was conducted in the 4 schools. Responses from 205 subjects aged 18-22 years were collected from September to October 2020, resulting in 201 valid questionnaires being completed and returned, with an effective recovery rate of 98%. In stage 1 of the formal test, the same method was used to divide universities in Taipei into 4 strata, and 9 (38%) of 24 schools were randomly selected from the 24 universities based on proportional stratification, including 2 (22%) public general universities, 2 (22%) public vocational colleges, 3 (33%) private general universities, and 2 (22%) private vocational colleges. Before stage 2, this study required participation from at least 384 respondents. This need was calculated using the following formula for determining the sample size : where χ 2 , P , and d are known in the reference and represent the value of the chi-square for 1 degree of freedom at the desired confidence level (3.841), the population proportion (assumed to be .5, as this would provide the maximum sample size), and the degree of accuracy expressed as a proportion (0.05), respectively . N represents the population size. The total number of university students in Taipei in 2021 (N=250,939) according to the Taiwan Ministry of Education , was substituted into in the following formula: Considering the possibility of 5%-7% of the questionnaires being invalid, the number of respondents in the formal test was expected to increase to 410. Additionally, the number of respondents for each school was calculated based on the ratio of the total students in each school to those in Taipei. Subsequently, stage 2 was performed from March to April 2021, with 39-50 university students randomly selected based on their student ID numbers obtained from each school. Invitations to participate in this study were sent using the students’ school email addresses. Students who considered themselves physically and mentally stable and who accepted the invitation were included in this study. They completed the questionnaires at their respective schools at the agreed time. If the initially selected students declined to participate, respondents were drawn again as substitutes. Ultimately, 411 formal questionnaires were collected, resulting in 406 valid completed and returned questionnaires, for an effective recovery rate of 98.8%. Ethical Considerations This study was reviewed and approved by the Behavioral and Social Science Research Ethics Committee of National Taiwan University (approval number 202004ES028). The ethical principles of the Declaration of Helsinki were adhered to during the entire study. Interviewers provided participants with questionnaires and explained the study’s objectives, procedures, benefits, and potential risks. The self-administered questionnaires were anonymously completed by participants in both rounds of testing after providing signed informed consent. Participants received a small gift—a pen worth New Taiwan dollar (NT $) 38 (US $1.2)—upon questionnaire completion as a token of appreciation. Measurements The structured questionnaire with closed-ended items encompassed sociodemographic characteristics, eHealth literacy, and health-promoting lifestyle. Sociodemographic variables considered as potential confounders in this study included sex, institution ownership, institution orientation, living status, parental education level, religious affiliation, monthly disposable amount, daily reading time, daily screen time (on mobile devices and computers), primary information channel, and perceived health status. eHealth literacy and health-promoting lifestyle were the study’s main predictor and outcome. The Chinese eHEALS, a translation of the original eHEALS by Norman and Skinner conducted by Cheng et al , was used to assess eHealth literacy . As the translation process of the Chinese eHEALS was not mentioned by Cheng et al , 2 language teachers from the Center for General Education of the China University of Technology were invited to inspect and confirm the translation accuracy and fluency of the scale. Six experts on health science and health education reviewed the scale and found that the content validity was good (mean item content validity index=1.00, SD 0.02). This scale was initially constructed with a single factor and comprised 8 questions. A 4-point Likert scale was used for scoring in this study, ranging from 1 (strongly disagree) to 4 (strongly agree). A higher mean score reflected better self-perceived online health information skills and comfort. In binary logistic regression, mean scores for eHEALS of 1.00-3.16 and 3.17-4.00 denoted relatively low and high eHealth literacy, respectively, to facilitate subsequent explanation and application. The scale exhibited good reliability with a Cronbach α of .94. Recent studies have pointed out that the German eHEALS 2-factor model and the English eHEALS 3-factor model have a better fit than the original 1-factor model and are more meaningful . Therefore, EFA was performed on pretest data to extract 2- and 3-factor structures from the Chinese eHEALS. The 3-factor model included search (3 questions), usage (2 questions), and evaluation (3 questions), with factor loadings ranging from 0.777-0.829, 0.753-0.833, and 0.717-0.840, respectively. The cumulative explained variance was 85.9%. The 2-factor model included search/usage (5 questions) and evaluation (3 questions), with factor loadings ranging from 0.773-0.826 and 0.758-0.853, respectively. The cumulative explained variance was 79.6%. Subsequently, CFA was performed on the formal test data to determine the fit, validity, and reliability of the 1-, 2-, and 3-factor models. The HPLP, initially developed by Walker et al , was translated into Chinese by Huang and Chiou and further adapted for simplification by Wei and Lu . The simplified version of the Chinese HPLP was used in this study to assess a health-promoting lifestyle . The Chinese HPLP has undergone language revision and content validity review , is considered to be faithful to the original scale, and does not have excessive additional content . The scale included 6 subscales, namely self-actualization, health responsibility, exercise, nutrition, interpersonal support, and stress management, each containing 4 items. A 5-point Likert scale was used for scoring, ranging from 1 (never) to 5 (always). A higher mean score indicated a more favorable health-promoting lifestyle. In binary logistic regression, mean scores above or equal to the middle (3/5) denoted relatively positive responses , indicating active adherence to a health-promoting lifestyle to facilitate subsequent explanation and application. The Cronbach α of the pretest data for the scale was .94, demonstrating its robust reliability. Statistical Analysis CFA was conducted using SPSS AMOS 28.0 (IBM Corporation). Recent studies have shown that a model exhibits a good fit when χ 2 / df <3; the comparative fit index (CFI), Tucker-Lewis index (TLI), and relative fit index (RFI)>0.95; the goodness-of-fit index (GFI), normed fit index (NFI), and incremental fit index (IFI)>0.9; and the root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR)<0.08 . In addition, the scale is considered to have convergent validity if the standardization factor loadings of various questions are >0.7 and the average variance extracted (AVE) of various factors are >0.5 . The correlation coefficient of 2 factors is lower than the square root of the AVE of various factors and is considered to have discriminant validity . If the Cronbach α and composite reliability (CR) of the various factors are >0.7, this indicates that the reliability is good . In this study, SPSS 23.0 was used for other inferential statistics. An independent sample 2-tailed t test or 1-way ANOVA combined with Scheffé post hoc analysis was conducted to present the relationship between sociodemographic characteristics and the HPLP. Multiple linear regression was performed to examine the sociodemographic variables that may affect the HPLP. The total HPLP score was used as the dependent variable, while the 12 sociodemographic variables were transformed into 16 dummy variables to serve as independent variables. Sex (reference=female), institution ownership (reference=public), institution orientation (reference=vocational colleges), living status (reference=alone), father’s education level (reference=high school or lower), mother’s education level (reference=high school or lower), religious affiliation (reference=with), and primary information channel (reference=self-searching) were each transformed into 1 dummy variable, while monthly disposable amount (reference=NT $15,001 [US $462] or more), daily reading time (reference=less than 1 hour), daily screen time (reference=6 hours or more), and perceived health status (reference=good) were each transformed into 2 dummy variables. Stepwise regression analysis used an inclusion criterion of P <.05 and an exclusion criterion of P >.10. Tolerance>0.1 and variance inflation factor (VIF)<10 were deemed free from collinearity between independent variables. Subsequently, Pearson product-moment correlation analysis was conducted to depict the correlation between eHEALS and the HPLP. Binary logistic regression was performed to determine the predictive effects of relatively low (1.00-3.16) and relatively high (3.17-4.00) eHEALS scores on both positive (3-5) and negative (1.00-2.99) HPLPs, while accounting for relevant sociodemographic factors associated with the HPLP (ie, sex, institution orientation, daily reading time, daily screen time, primary information channel, and perceived health status). A nonsignificant Hosmer-Lemeshow test indicated a good fit for the logistic regression model. P <.05 was considered statistically significant.
This cross-sectional quantitative study was conducted among university students in Taipei, the capital of Taiwan. Two rounds of testing (pretest and formal) were performed, and 2-stage sampling was used in both rounds and the subjects were not repeated. The pretest was conducted to determine the reliability of the Chinese versions of eHEALS and the HPLP in the study population and to conduct EFA to extract the 2- and 3-factor models of the Chinese eHEALS. Next, the formal test was conducted to conduct CFA to further determine the fit, validity, and reliability of the Chinese eHEALS multiple-factor models and to perform inferential statistics on the predictive effects of eHEALS on the HPLP. In the first stage of the pretest, stratification was conducted based on the types of universities in Taipei. The Taiwan Ministry of Education has classified 24 universities in Taipei based on ownership and educational goals into 6 (25%) public general universities, 5 (21%) public vocational colleges, 8 (33%) private general universities, and 5 (21%) private vocational colleges . One school was randomly selected from each stratum for testing. Subsequently, in stage 2, convenience sampling was conducted in the 4 schools. Responses from 205 subjects aged 18-22 years were collected from September to October 2020, resulting in 201 valid questionnaires being completed and returned, with an effective recovery rate of 98%. In stage 1 of the formal test, the same method was used to divide universities in Taipei into 4 strata, and 9 (38%) of 24 schools were randomly selected from the 24 universities based on proportional stratification, including 2 (22%) public general universities, 2 (22%) public vocational colleges, 3 (33%) private general universities, and 2 (22%) private vocational colleges. Before stage 2, this study required participation from at least 384 respondents. This need was calculated using the following formula for determining the sample size : where χ 2 , P , and d are known in the reference and represent the value of the chi-square for 1 degree of freedom at the desired confidence level (3.841), the population proportion (assumed to be .5, as this would provide the maximum sample size), and the degree of accuracy expressed as a proportion (0.05), respectively . N represents the population size. The total number of university students in Taipei in 2021 (N=250,939) according to the Taiwan Ministry of Education , was substituted into in the following formula: Considering the possibility of 5%-7% of the questionnaires being invalid, the number of respondents in the formal test was expected to increase to 410. Additionally, the number of respondents for each school was calculated based on the ratio of the total students in each school to those in Taipei. Subsequently, stage 2 was performed from March to April 2021, with 39-50 university students randomly selected based on their student ID numbers obtained from each school. Invitations to participate in this study were sent using the students’ school email addresses. Students who considered themselves physically and mentally stable and who accepted the invitation were included in this study. They completed the questionnaires at their respective schools at the agreed time. If the initially selected students declined to participate, respondents were drawn again as substitutes. Ultimately, 411 formal questionnaires were collected, resulting in 406 valid completed and returned questionnaires, for an effective recovery rate of 98.8%.
This study was reviewed and approved by the Behavioral and Social Science Research Ethics Committee of National Taiwan University (approval number 202004ES028). The ethical principles of the Declaration of Helsinki were adhered to during the entire study. Interviewers provided participants with questionnaires and explained the study’s objectives, procedures, benefits, and potential risks. The self-administered questionnaires were anonymously completed by participants in both rounds of testing after providing signed informed consent. Participants received a small gift—a pen worth New Taiwan dollar (NT $) 38 (US $1.2)—upon questionnaire completion as a token of appreciation.
The structured questionnaire with closed-ended items encompassed sociodemographic characteristics, eHealth literacy, and health-promoting lifestyle. Sociodemographic variables considered as potential confounders in this study included sex, institution ownership, institution orientation, living status, parental education level, religious affiliation, monthly disposable amount, daily reading time, daily screen time (on mobile devices and computers), primary information channel, and perceived health status. eHealth literacy and health-promoting lifestyle were the study’s main predictor and outcome. The Chinese eHEALS, a translation of the original eHEALS by Norman and Skinner conducted by Cheng et al , was used to assess eHealth literacy . As the translation process of the Chinese eHEALS was not mentioned by Cheng et al , 2 language teachers from the Center for General Education of the China University of Technology were invited to inspect and confirm the translation accuracy and fluency of the scale. Six experts on health science and health education reviewed the scale and found that the content validity was good (mean item content validity index=1.00, SD 0.02). This scale was initially constructed with a single factor and comprised 8 questions. A 4-point Likert scale was used for scoring in this study, ranging from 1 (strongly disagree) to 4 (strongly agree). A higher mean score reflected better self-perceived online health information skills and comfort. In binary logistic regression, mean scores for eHEALS of 1.00-3.16 and 3.17-4.00 denoted relatively low and high eHealth literacy, respectively, to facilitate subsequent explanation and application. The scale exhibited good reliability with a Cronbach α of .94. Recent studies have pointed out that the German eHEALS 2-factor model and the English eHEALS 3-factor model have a better fit than the original 1-factor model and are more meaningful . Therefore, EFA was performed on pretest data to extract 2- and 3-factor structures from the Chinese eHEALS. The 3-factor model included search (3 questions), usage (2 questions), and evaluation (3 questions), with factor loadings ranging from 0.777-0.829, 0.753-0.833, and 0.717-0.840, respectively. The cumulative explained variance was 85.9%. The 2-factor model included search/usage (5 questions) and evaluation (3 questions), with factor loadings ranging from 0.773-0.826 and 0.758-0.853, respectively. The cumulative explained variance was 79.6%. Subsequently, CFA was performed on the formal test data to determine the fit, validity, and reliability of the 1-, 2-, and 3-factor models. The HPLP, initially developed by Walker et al , was translated into Chinese by Huang and Chiou and further adapted for simplification by Wei and Lu . The simplified version of the Chinese HPLP was used in this study to assess a health-promoting lifestyle . The Chinese HPLP has undergone language revision and content validity review , is considered to be faithful to the original scale, and does not have excessive additional content . The scale included 6 subscales, namely self-actualization, health responsibility, exercise, nutrition, interpersonal support, and stress management, each containing 4 items. A 5-point Likert scale was used for scoring, ranging from 1 (never) to 5 (always). A higher mean score indicated a more favorable health-promoting lifestyle. In binary logistic regression, mean scores above or equal to the middle (3/5) denoted relatively positive responses , indicating active adherence to a health-promoting lifestyle to facilitate subsequent explanation and application. The Cronbach α of the pretest data for the scale was .94, demonstrating its robust reliability.
CFA was conducted using SPSS AMOS 28.0 (IBM Corporation). Recent studies have shown that a model exhibits a good fit when χ 2 / df <3; the comparative fit index (CFI), Tucker-Lewis index (TLI), and relative fit index (RFI)>0.95; the goodness-of-fit index (GFI), normed fit index (NFI), and incremental fit index (IFI)>0.9; and the root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR)<0.08 . In addition, the scale is considered to have convergent validity if the standardization factor loadings of various questions are >0.7 and the average variance extracted (AVE) of various factors are >0.5 . The correlation coefficient of 2 factors is lower than the square root of the AVE of various factors and is considered to have discriminant validity . If the Cronbach α and composite reliability (CR) of the various factors are >0.7, this indicates that the reliability is good . In this study, SPSS 23.0 was used for other inferential statistics. An independent sample 2-tailed t test or 1-way ANOVA combined with Scheffé post hoc analysis was conducted to present the relationship between sociodemographic characteristics and the HPLP. Multiple linear regression was performed to examine the sociodemographic variables that may affect the HPLP. The total HPLP score was used as the dependent variable, while the 12 sociodemographic variables were transformed into 16 dummy variables to serve as independent variables. Sex (reference=female), institution ownership (reference=public), institution orientation (reference=vocational colleges), living status (reference=alone), father’s education level (reference=high school or lower), mother’s education level (reference=high school or lower), religious affiliation (reference=with), and primary information channel (reference=self-searching) were each transformed into 1 dummy variable, while monthly disposable amount (reference=NT $15,001 [US $462] or more), daily reading time (reference=less than 1 hour), daily screen time (reference=6 hours or more), and perceived health status (reference=good) were each transformed into 2 dummy variables. Stepwise regression analysis used an inclusion criterion of P <.05 and an exclusion criterion of P >.10. Tolerance>0.1 and variance inflation factor (VIF)<10 were deemed free from collinearity between independent variables. Subsequently, Pearson product-moment correlation analysis was conducted to depict the correlation between eHEALS and the HPLP. Binary logistic regression was performed to determine the predictive effects of relatively low (1.00-3.16) and relatively high (3.17-4.00) eHEALS scores on both positive (3-5) and negative (1.00-2.99) HPLPs, while accounting for relevant sociodemographic factors associated with the HPLP (ie, sex, institution orientation, daily reading time, daily screen time, primary information channel, and perceived health status). A nonsignificant Hosmer-Lemeshow test indicated a good fit for the logistic regression model. P <.05 was considered statistically significant.
Confirmatory Factor Analysis of eHEALS In this study, CFA was performed to determine the fit of the Chinese eHEALS. The 1-, 2-, and 3-factor models proposed in the study were evaluated for fit . Results showed that the fit of the 3-factor model was significantly better than that of the other models, and the diverse indicators satisfied the recommended fit indicators ( χ 2 / df =2.574, CFI=0.991, TLI=0.984, RFI=0.975, GFI=0.975, NFI=0.985, IFI=0.991, RMSEA=0.062, SRMR=0.018). Furthermore, the fit of the 3-factor model used for empirical data in this study outperformed the 3-factor model proposed by Sudbury-Riley et al . shows the factor loadings, AVE, Cronbach α, and CR of various subscales in the eHEALS 3-factor model, which all met the ideal criteria for convergent validity and reliability. In the search subscale of eHEALS, the questions had factor loadings=0.886-0.950, AVE=0.827, Cronbach α=.93, and CR=0.935. In the usage subscale, the questions had factor loadings=0.857-0.893, AVE=0.766, Cronbach α=.87, and CR=0.867. In the evaluation subscale, the questions had factor loadings=0.785-0.900, AVE=0.710, Cronbach α=.88, and CR=0.880. shows the correlation coefficients between various eHEALS factors. The correlation coefficients of 2 factors were lower than the square roots of the AVE of various factors, which suggests that the eHEALS 3-factor model has good discriminant validity. Sociodemographic Characteristics shows the sociodemographic variables in this study. In total, 406 students were enrolled in this study. Overall, 252 (62.1%) of the 406 participants were female, 224 (55.2%) lived with family members or friends, and 269 (66.3%) did not have specific religious beliefs. Regarding institution ownership and educational goals, the sample ratio was close to the distribution ratio of various universities in Taipei. More than half of the students were enrolled in private universities (n=226, 55.7%) than in public universities. Furthermore, the ratio of students attending general universities (n=227, 55.9%) was higher than that of students attending vocational colleges. Regarding the parental education level, 237 (58.4%) of the participants had fathers with a university degree or higher, while 252 (62.1%) had mothers with a university education level or higher. Most participants had a monthly disposable amount of NT $10,000 (US $308) or less (n=182, 44.8%). Additionally, a significant proportion of the participants spent less than 1 hour reading per day (n=198, 48.8%), while the majority spent 6 hours or more on mobile devices and computers daily (n=225, 55.4%). The participants indicated that their primary information-acquiring channel was self-searching (n=361, 88.9%), with only a minority (n=45, 11.1%) relying on asking others. Notably, 253 (62.3%) of the participants reported having a good perceived health status. Current eHEALS and the HPLP In eHEALS, the mean total score was 3.17 (SD 0.48). The score of the usage subscale was the highest (mean 3.25, SD 0.50), followed by the search subscale (mean 3.20, SD 0.52) and the evaluation subscale (mean 3.08, SD 0.56). In the HPLP, the mean total score was 3.55 (SD 0.62). The score of the interpersonal support subscale was the highest (mean 3.87, SD 0.70), followed by the self-actualization subscale (mean 3.85, SD 0.74), the stress management subscale (mean 3.74, SD 0.74), the nutrition subscale (mean 3.41, SD 0.79), and the health responsibility subscale (mean 3.26, SD 0.87), with the exercise subscale (mean 3.18, SD 0.90) being the lowest. Relationship Between Sociodemographic Variables and the HPLP As shown in , the total HPLP score showed significant differences between sexes ( t 404 =2.346, P =.02), institution orientation ( t 404 =2.564, P =.01), daily reading time ( F 2,403 =17.618, P <.001), daily screen time ( F 2,403 =7.148, P <.001), primary information channel ( t 404 =3.892, P <.001), and perceived health status ( F 2,403 =24.366, P <.001). Specifically, the HPLP score was higher for male participants (mean 3.65, SD 0.71) than female ones (mean 3.49, SD 0.55). Participants attending general university (mean 3.62, SD 0.59) had a higher HPLP score than those attending vocational college (mean 3.46, SD 0.65). Regarding daily reading time, participants who read for 1.0-2.9 (mean 3.67, SD 0.58) and ≥3 hours (mean 3.87, SD 0.49) had higher HPLP scores than those who read for <1 hour (mean 3.38, SD 0.64). Regarding daily screen time, participants who spent <3 hours (mean 3.70, SD 0.61) had higher HPLP scores than those who spent ≥6 hours (mean 3.45, SD 0.60). Additionally, the HPLP score was higher for participants who acquired information from others (mean 3.89, SD 0.59) than those who acquired information by themselves (mean 3.51, SD 0.62). Participants with a good perceived health status (mean 3.71, SD 0.62) had higher HPLP scores than those with an average (mean 3.32, SD 0.55) or a poor (mean 3.09, SD 0.50) perceived health status. Stepwise multiple linear regression was performed to analyze sociodemographic variables that affected the HPLP of participants . Results showed that sex, institution orientation, daily reading time, primary information channel, and perceived health status are confounders of the overall HPLP. In particular, male participants, participants attending general university, those who read for ≥1 hour daily, those who acquired information from others, and those with a good perceived health status had a better HPLP. Collinearity was absent between the independent variables (tolerance=0.825-0.969, VIF=1.032-1.212), and the factors explained 19.8% of the variance (adjusted R 2 =0.198, F 7,398 =15.290, P <.001). Relationship Between eHEALS and the HPLP Pearson product-moment correlation was performed to analyze the correlation between eHEALS and the HPLP . The overall eHEALS showed a significantly moderate positive correlation with the overall HPLP among participants ( r =0.512, P <.001). Furthermore, different eHEALS dimensions showed a significantly low-to-moderate positive correlation with the various HPLP dimensions ( r =0.291-0.522, P <.001). Binary logistic regression was performed to analyze the predictive effects of the overall eHEALS and its various dimensions on the overall HPLP among participants . After adjusting for sociodemographic variables, compared with participants with relatively high overall eHEALS scores, those with relatively low eHEALS scores had 3.37 times the risk of a negative HPLP (adjusted odds ratio [OR]=3.37, 95% CI 1.49-7.61). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =2.128, P =.98), could explain 14.7%-24.4% of the variance (Cox-Snell R 2 =0.147, Nagelkerke R 2 =0.244), and had an accurate classification rate of 83.3%. Compared with participants with relatively high eHEALS search subscale scores, those with relatively low search abilities had 3.38 times the risk of a negative overall HPLP (adjusted OR=3.38, 95% CI 1.54-7.42). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =3.052, P =.93), could explain 14.8%-24.7% of the variance (Cox-Snell R 2 =0.148, Nagelkerke R 2 =0.247), and had an accurate classification rate of 83.3%. Compared with participants with relatively high eHEALS usage subscale scores, those with relatively low usage abilities had 2.25 times the risk of a negative HPLP (adjusted OR=2.25, 95% CI 1.11-4.59). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =10.538, P =.23), could explain 13.7%-22.8% of the variance (Cox-Snell R 2 =0.137, Nagelkerke R 2 =0.228), and had an accurate classification rate of 82.8%. Moreover, compared with participants with relatively high eHEALS evaluation subscale scores, those with relatively low evaluation abilities had 3.20 times the risk of a negative HPLP (adjusted OR=3.20, 95% CI 1.35-7.59). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =2.916, P =.94), could explain 14.3%-23.8% of the variance (Cox-Snell R 2 =0.143, Nagelkerke R 2 =0.238), and had an accurate classification rate of 83.7%. Further analysis of the prediction results of eHEALS on various dimensions of the HPLP was conducted . Results showed that compared with participants with relatively high overall eHEALS scores, those with relatively low eHEALS scores had 2.74 times the risk of negative health responsibility (adjusted OR=2.74, 95% CI 1.55-4.84), 2.41 times the risk of negative exercise (adjusted OR=2.41, 95% CI 1.43-4.07), and 1.86 times the risk of negative nutrition (adjusted OR=1.86, 95% CI 1.07-3.22). Compared with participants with relatively high eHEALS subscales scores, those with relatively low search, usage, and evaluation abilities, respectively, had 2.66 (adjusted OR=2.66, 95% CI 1.52-4.62), 2.00 (adjusted OR=2.00, 95% CI 1.18-3.37), and 3.01 (adjusted OR=3.01, 95% CI 1.63-5.55) times the risk of negative health responsibility; 2.02 (adjusted OR=2.02, 95% CI 1.22-3.35), 2.12 (adjusted OR=2.12, 95% CI 1.29-3.50), and 2.71 (adjusted OR=2.71, 95% CI 1.54-4.76) times the risk of negative exercise; and 2.08 (adjusted OR=2.08, 95% CI 1.12-3.86), 1.83 (adjusted OR=1.83, 95% CI 1.08-3.11), and 2.08 (adjusted OR=2.08, 95% CI 1.07-4.06) times the risk of negative nutrition. In addition, compared with participants with relatively high eHEALS evaluation subscale scores, those with relatively low evaluation abilities had 2.06 times the risk of negative stress management (adjusted OR=2.06, 95% CI 1.01-4.22).
In this study, CFA was performed to determine the fit of the Chinese eHEALS. The 1-, 2-, and 3-factor models proposed in the study were evaluated for fit . Results showed that the fit of the 3-factor model was significantly better than that of the other models, and the diverse indicators satisfied the recommended fit indicators ( χ 2 / df =2.574, CFI=0.991, TLI=0.984, RFI=0.975, GFI=0.975, NFI=0.985, IFI=0.991, RMSEA=0.062, SRMR=0.018). Furthermore, the fit of the 3-factor model used for empirical data in this study outperformed the 3-factor model proposed by Sudbury-Riley et al . shows the factor loadings, AVE, Cronbach α, and CR of various subscales in the eHEALS 3-factor model, which all met the ideal criteria for convergent validity and reliability. In the search subscale of eHEALS, the questions had factor loadings=0.886-0.950, AVE=0.827, Cronbach α=.93, and CR=0.935. In the usage subscale, the questions had factor loadings=0.857-0.893, AVE=0.766, Cronbach α=.87, and CR=0.867. In the evaluation subscale, the questions had factor loadings=0.785-0.900, AVE=0.710, Cronbach α=.88, and CR=0.880. shows the correlation coefficients between various eHEALS factors. The correlation coefficients of 2 factors were lower than the square roots of the AVE of various factors, which suggests that the eHEALS 3-factor model has good discriminant validity.
shows the sociodemographic variables in this study. In total, 406 students were enrolled in this study. Overall, 252 (62.1%) of the 406 participants were female, 224 (55.2%) lived with family members or friends, and 269 (66.3%) did not have specific religious beliefs. Regarding institution ownership and educational goals, the sample ratio was close to the distribution ratio of various universities in Taipei. More than half of the students were enrolled in private universities (n=226, 55.7%) than in public universities. Furthermore, the ratio of students attending general universities (n=227, 55.9%) was higher than that of students attending vocational colleges. Regarding the parental education level, 237 (58.4%) of the participants had fathers with a university degree or higher, while 252 (62.1%) had mothers with a university education level or higher. Most participants had a monthly disposable amount of NT $10,000 (US $308) or less (n=182, 44.8%). Additionally, a significant proportion of the participants spent less than 1 hour reading per day (n=198, 48.8%), while the majority spent 6 hours or more on mobile devices and computers daily (n=225, 55.4%). The participants indicated that their primary information-acquiring channel was self-searching (n=361, 88.9%), with only a minority (n=45, 11.1%) relying on asking others. Notably, 253 (62.3%) of the participants reported having a good perceived health status.
In eHEALS, the mean total score was 3.17 (SD 0.48). The score of the usage subscale was the highest (mean 3.25, SD 0.50), followed by the search subscale (mean 3.20, SD 0.52) and the evaluation subscale (mean 3.08, SD 0.56). In the HPLP, the mean total score was 3.55 (SD 0.62). The score of the interpersonal support subscale was the highest (mean 3.87, SD 0.70), followed by the self-actualization subscale (mean 3.85, SD 0.74), the stress management subscale (mean 3.74, SD 0.74), the nutrition subscale (mean 3.41, SD 0.79), and the health responsibility subscale (mean 3.26, SD 0.87), with the exercise subscale (mean 3.18, SD 0.90) being the lowest.
As shown in , the total HPLP score showed significant differences between sexes ( t 404 =2.346, P =.02), institution orientation ( t 404 =2.564, P =.01), daily reading time ( F 2,403 =17.618, P <.001), daily screen time ( F 2,403 =7.148, P <.001), primary information channel ( t 404 =3.892, P <.001), and perceived health status ( F 2,403 =24.366, P <.001). Specifically, the HPLP score was higher for male participants (mean 3.65, SD 0.71) than female ones (mean 3.49, SD 0.55). Participants attending general university (mean 3.62, SD 0.59) had a higher HPLP score than those attending vocational college (mean 3.46, SD 0.65). Regarding daily reading time, participants who read for 1.0-2.9 (mean 3.67, SD 0.58) and ≥3 hours (mean 3.87, SD 0.49) had higher HPLP scores than those who read for <1 hour (mean 3.38, SD 0.64). Regarding daily screen time, participants who spent <3 hours (mean 3.70, SD 0.61) had higher HPLP scores than those who spent ≥6 hours (mean 3.45, SD 0.60). Additionally, the HPLP score was higher for participants who acquired information from others (mean 3.89, SD 0.59) than those who acquired information by themselves (mean 3.51, SD 0.62). Participants with a good perceived health status (mean 3.71, SD 0.62) had higher HPLP scores than those with an average (mean 3.32, SD 0.55) or a poor (mean 3.09, SD 0.50) perceived health status. Stepwise multiple linear regression was performed to analyze sociodemographic variables that affected the HPLP of participants . Results showed that sex, institution orientation, daily reading time, primary information channel, and perceived health status are confounders of the overall HPLP. In particular, male participants, participants attending general university, those who read for ≥1 hour daily, those who acquired information from others, and those with a good perceived health status had a better HPLP. Collinearity was absent between the independent variables (tolerance=0.825-0.969, VIF=1.032-1.212), and the factors explained 19.8% of the variance (adjusted R 2 =0.198, F 7,398 =15.290, P <.001).
Pearson product-moment correlation was performed to analyze the correlation between eHEALS and the HPLP . The overall eHEALS showed a significantly moderate positive correlation with the overall HPLP among participants ( r =0.512, P <.001). Furthermore, different eHEALS dimensions showed a significantly low-to-moderate positive correlation with the various HPLP dimensions ( r =0.291-0.522, P <.001). Binary logistic regression was performed to analyze the predictive effects of the overall eHEALS and its various dimensions on the overall HPLP among participants . After adjusting for sociodemographic variables, compared with participants with relatively high overall eHEALS scores, those with relatively low eHEALS scores had 3.37 times the risk of a negative HPLP (adjusted odds ratio [OR]=3.37, 95% CI 1.49-7.61). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =2.128, P =.98), could explain 14.7%-24.4% of the variance (Cox-Snell R 2 =0.147, Nagelkerke R 2 =0.244), and had an accurate classification rate of 83.3%. Compared with participants with relatively high eHEALS search subscale scores, those with relatively low search abilities had 3.38 times the risk of a negative overall HPLP (adjusted OR=3.38, 95% CI 1.54-7.42). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =3.052, P =.93), could explain 14.8%-24.7% of the variance (Cox-Snell R 2 =0.148, Nagelkerke R 2 =0.247), and had an accurate classification rate of 83.3%. Compared with participants with relatively high eHEALS usage subscale scores, those with relatively low usage abilities had 2.25 times the risk of a negative HPLP (adjusted OR=2.25, 95% CI 1.11-4.59). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =10.538, P =.23), could explain 13.7%-22.8% of the variance (Cox-Snell R 2 =0.137, Nagelkerke R 2 =0.228), and had an accurate classification rate of 82.8%. Moreover, compared with participants with relatively high eHEALS evaluation subscale scores, those with relatively low evaluation abilities had 3.20 times the risk of a negative HPLP (adjusted OR=3.20, 95% CI 1.35-7.59). The model exhibited a good fit (Hosmer-Lemeshow χ 2 8 =2.916, P =.94), could explain 14.3%-23.8% of the variance (Cox-Snell R 2 =0.143, Nagelkerke R 2 =0.238), and had an accurate classification rate of 83.7%. Further analysis of the prediction results of eHEALS on various dimensions of the HPLP was conducted . Results showed that compared with participants with relatively high overall eHEALS scores, those with relatively low eHEALS scores had 2.74 times the risk of negative health responsibility (adjusted OR=2.74, 95% CI 1.55-4.84), 2.41 times the risk of negative exercise (adjusted OR=2.41, 95% CI 1.43-4.07), and 1.86 times the risk of negative nutrition (adjusted OR=1.86, 95% CI 1.07-3.22). Compared with participants with relatively high eHEALS subscales scores, those with relatively low search, usage, and evaluation abilities, respectively, had 2.66 (adjusted OR=2.66, 95% CI 1.52-4.62), 2.00 (adjusted OR=2.00, 95% CI 1.18-3.37), and 3.01 (adjusted OR=3.01, 95% CI 1.63-5.55) times the risk of negative health responsibility; 2.02 (adjusted OR=2.02, 95% CI 1.22-3.35), 2.12 (adjusted OR=2.12, 95% CI 1.29-3.50), and 2.71 (adjusted OR=2.71, 95% CI 1.54-4.76) times the risk of negative exercise; and 2.08 (adjusted OR=2.08, 95% CI 1.12-3.86), 1.83 (adjusted OR=1.83, 95% CI 1.08-3.11), and 2.08 (adjusted OR=2.08, 95% CI 1.07-4.06) times the risk of negative nutrition. In addition, compared with participants with relatively high eHEALS evaluation subscale scores, those with relatively low evaluation abilities had 2.06 times the risk of negative stress management (adjusted OR=2.06, 95% CI 1.01-4.22).
Principal Findings Comparison of the Chinese eHEALS 3-Factor Model With Previous Studies Norman and Skinner developed eHEALS and highlighted that men’s eHEALS scores are significantly higher than those of women, which could be used as an a priori hypothesis. Similar results were obtained by using the Chinese eHEALS in this study; in other words, significant differences were observed in eHEALS scores between sexes ( t 404 =2.708, P =.007), with males having higher scores (mean 3.25, SD 0.51) than females (mean 3.12, SD 0.46). This shows that the Chinese eHEALS has known-groups validity. Moreover, in this study, the original 8 eHEALS questions were classified into 3 factors, namely search (questions 1-3), usage (questions 4 and 5), and evaluation (questions 6-8). Compared with the initial 1-factor model , CFA showed that the 3-factor model exhibits a better fit and good validity and reliability. The findings were similar to those of a recent study on the Chinese eHEALS multifactorial model ; however, this study showed more robust evidence of fit, validity, and reliability. In contrast to Sudbury-Riley et al , who used a 3-factor eHEALS model and defined question 3 as “I know how to find helpful health resources and information on the internet” and questions 4 and 5 as the ability to acquire and use health resources and information, this study defined questions 1-3 as the ability to search for health resources on the internet and questions 4 and 5 as the ability to use online health information. Results revealed that differences in the delineation of questions lead to variations in the model fit. Notably, empirical data showed that the fit of the 3-factor model in this study is superior to that of Sudbury-Riley et al’s model. This can be attributed to 2 potential explanations. First, EFA was performed in the pretest to delineate the 3 factors, which differed from Sudbury-Riley et al’s method, who carefully reviewed and partitioned the factors based on social cognitive and self-efficacy theories. Second, minor differences in participants’ perceptions of the translated scale may have contributed to these disparities . In the English eHEALS, questions 3-5 start with “I know how to,” which may have caused participants to perceive them as belonging to the same factor . In the Chinese eHEALS, participants tended to consider questions 1-3 as search factors due to words such as “what,” “where,” and “find,” while the word “use” in questions 4 and 5 led participants to classify it as a usage factor. Nonetheless, the 3-factor model used in this study complies with the foundational theories of the eHEALS lily model (ie, social cognitive theory and self-efficacy theory) . This model may be more suitable for regions where the Chinese eHEALS is used in eHealth literacy studies. eHealth Literacy Level In this study, the overall eHEALS score of the university students was moderate or higher, and the search and usage dimensions had higher scores. In contrast, the evaluation dimension had a lower score. This reveals that students perceive themselves to have good search and usage capabilities of eHealth information; however, they possess low confidence in evaluating such information and using it for decision-making. In recent studies, the mean scores for eHEALS questions 6-8 were lower than those for questions 1-3 and questions 4 and 5 , similar to scores obtained in this study. A Taiwanese study used a self-formulated scale to evaluate the eHealth literacy of university students and divided the questions into functional, interactive, and critical literacies . Interactive literacy encompasses the ability to select, comprehend, and use online health information, which was similar to the search and usage dimensions in this study. Critical literacy refers to the ability to analyze, criticize, and respond to online health information, which was similar to the evaluation dimension in this study. The score for critical literacy was visibly lower than that for interactive literacy in the previous study , which was similar to this study. Researchers found that although most university students mentioned that they can understand the general idea of online health information, they have a vague understanding of the jargon, foreign languages, and data . In addition, some university students lack confidence in the quality of online health information and express difficulty in determining the quality of such information . Therefore, in the contemporary landscape characterized by the unlimited accumulation and dissemination of internet-based health information of uncertain veracity, imparting fundamental health knowledge to Taiwanese university students is imperative. This includes fostering a sense of caution toward eHealth information among students and equipping them with the ability to critically assess and validate uncertainties. Association Between Sociodemographic Variables and the HPLP The overall HPLP of university students in this study was moderate or higher, wherein interpersonal support and self-actualization scores were the highest, while nutrition, exercise, and health responsibility scores were the lowest, similar to those of the most recent studies . Among sociodemographic variables, stepwise multiple linear regression showed that female students, students attending vocational colleges, those with a daily reading time of <1 hour, those who acquired information by themselves, and those with an average or a poor perceived health status were confounders of a poor overall HPLP. This was consistent with the significant differences in the overall HPLP in these sociodemographic variables. Recent studies have found that sex affects the HPLP and health behaviors, such as exercise and sleep . The frequency of discussions of health problems with others has been highlighted to positively affect the dietary behavior of university students . Individuals with a good perceived health status or great concern for health have a better HPLP and show several health behaviors, such as eating, exercise, and sleep . In addition, this study found that a daily reading time of ≥1 hour is a confounder of a good HPLP among university students. This may be because information in books, newspapers, and magazines usually undergoes review and proofreading, and reading more accurate and reliable hardcopy information may lead to a tendency to adopt a positive lifestyle profile. Some studies have highlighted that reading hardcopy materials can promote better comprehension results than reading from screens . However, the increased screen time on digital media today has greatly decreased the reading time in print. In this study, only 208 (51.2%) of 406 university students read for ≥1 hour per day, but 362 (89.2%) spent ≥3 hours on mobile devices or computers daily. Recent studies have shown that newspapers and magazines are the media that Taiwanese university students spend the least time on, far below the time spent on mobile devices and computers . Furthermore, mobile devices and computers have diverse online functions. The TWNIC survey found that the most commonly used internet functions among generation Z are real-time messaging, social networks, free videos, online news, online games, and ecommerce; however, online learning is not their priority . In the ANOVA in this study, the HPLP score of participants with a daily screen time of ≥6 hours was significantly low. However, multiple linear regression excluded the daily screen time from the HPLP confounders. It is believed that frequent usage of mobile devices and computers by university students consumes the time spent on reading. Therefore, the effect of screen time in the multiple linear regression may be explained by the reading time factor. In the binary logistic regression in this study, daily screen time was still considered a confounder of HPLP scores. Association of eHEALS With the HPLP After adjusting for sociodemographic factors that may affect the HPLP, this study revealed that eHEALS consistently and significantly affects the HPLP of university students. Compared with students with relatively high overall eHEALS scores, those with relatively low eHEALS scores had a higher probability of a negative overall HPLP, similar to the results of studies in other regions . Other researchers have used their own created scales to measure eHealth literacy and proved that it predicts multiple HPLP dimensions in university students . Many studies have found that eHealth literacy has positive effects on exercise, diet, and sleep behaviors , or even safe sex practice and COVID-19 prevention among university students. This study found that among the various HPLP dimensions, compared with a relatively high overall eHEALS score, a relatively low eHEALS score is associated with negative health responsibility, exercise, and nutrition. University students in this study had low HPLP health responsibility, exercise, and nutrition scores—dimensions that require improvements. At the same time, many recent studies have found that these health behaviors are poor in university students . Additionally, among the 3 eHEALS dimensions in this study, compared with participants with relatively high search, usage, and evaluation literacies, those with relatively low scores had a higher probability of a negative overall HPLP and its health responsibility, exercise, and nutrition dimensions, similar to the overall eHEALS results. Regarding evaluation literacy, this study found that in addition to predicting the negative health responsibility, exercise, and nutrition dimensions of the HPLP, a relatively low eHEALS evaluation score can also reflect poor stress management. This result is similar to that of a recent study indicating that critical eHealth literacy can predict more HPLP dimensions . Limitations and Strengths This study has certain limitations, which can provide a reference for future studies. First, this study included only university students without major diseases from the capital of Taiwan, and the results can only be generalized to the eHealth literacy and health-promoting lifestyle of this population. It is recommended that future studies extend to other regions in Taiwan or university students with other health statuses. Second, some variables may be related to eHealth literacy and a health-promoting lifestyle, such as majors and health risk behaviors, and it is recommended that future studies expand to include these variables. In addition, although participants were advised that the entire process was anonymized, the self-administered questionnaire may have caused their answers to be exposed to memory recall errors, environmental effects, and social desirability bias. Lastly, a cross-sectional study design was used in this study, and the causal relationship between eHealth literacy and a health-promoting lifestyle, as well as changes in these 2 factors with time, could not be confirmed. Hence, further repeated-measures or longitudinal studies are required for clarification. Nonetheless, this study confirmed the feasibility of using the Chinese version of the eHEALS 3-factor model to examine eHealth literacy and highlighted that eHealth literacy affects and predicts the HPLP in university students. In the contemporary world where internet use is widespread and portable mobile devices are rapidly advancing, using the internet, mobile phones, tablets, or computers as aids in daily life has become an unstoppable trend. If university students can cultivate the online learning habit early on and establish the concept of consulting to acquire information and reading in print, actively nurturing their skills to search, use, and access internet-based health information, it will undoubtedly positively impact their health-promoting behavior and lifestyle. Conclusion This study is the first to validate the Chinese eHEALS 3-factor model, encompassing search, usage, and evaluation dimensions. Notably, eHEALS is the first eHealth literacy measurement tool to be developed and is the most widely used. This 3-factor model results in more definite eHEALS content and undoubtedly increases the practicality and applicability of the scale to satisfy the eHealth literacy evaluation needs of health promoting–related studies, particularly in Chinese-speaking regions. Higher education represents the most significant and final opportunity for behavioral development and learning in young people. Behavioral health during this period impacts lifetime health outcomes. This study found that alongside specific sociodemographic characteristics, the overall eHEALS and its dimensions are independent predictors of the HPLP. Compared to university students with relatively high overall eHEALS and various dimension scores, those with relatively low scores had a negative overall HPLP and HPLP health responsibility, exercise, and nutrition. University students with relatively low eHEALS evaluation scores compared to those with relatively high evaluation scores also had negative stress management. These findings can be used to screen university students who require HPLP improvement so that health education suitable for their needs can be provided. In addition, there is room for improving overall eHEALS scores among university students, with particular attention to improving evaluation literacy. It is recommended that the centers for general education, digital learning, and health of the universities and colleges in Taipei, as well as targeting populations with relatively low eHealth literacy (eHEALS score<3.17), be integrated to provide appropriate health education and programs. Courses should be conducted to educate students on identifying objective, credible, and understandable online health information platforms, while cultivating vigilance and critical judgment in evaluating eHealth information. Additionally, fostering a supportive and user-friendly online health information environment is essential. It is recommended that universities and colleges further establish good campus eHealth literacy learning and support channels. For example, good health information online platforms could be recommended on school websites, and in-person or virtual health information consultation could be applied within schools. These measures would collectively contribute to improving university students’ eHealth literacy, thereby encouraging their adoption of health-promoting lifestyles.
Comparison of the Chinese eHEALS 3-Factor Model With Previous Studies Norman and Skinner developed eHEALS and highlighted that men’s eHEALS scores are significantly higher than those of women, which could be used as an a priori hypothesis. Similar results were obtained by using the Chinese eHEALS in this study; in other words, significant differences were observed in eHEALS scores between sexes ( t 404 =2.708, P =.007), with males having higher scores (mean 3.25, SD 0.51) than females (mean 3.12, SD 0.46). This shows that the Chinese eHEALS has known-groups validity. Moreover, in this study, the original 8 eHEALS questions were classified into 3 factors, namely search (questions 1-3), usage (questions 4 and 5), and evaluation (questions 6-8). Compared with the initial 1-factor model , CFA showed that the 3-factor model exhibits a better fit and good validity and reliability. The findings were similar to those of a recent study on the Chinese eHEALS multifactorial model ; however, this study showed more robust evidence of fit, validity, and reliability. In contrast to Sudbury-Riley et al , who used a 3-factor eHEALS model and defined question 3 as “I know how to find helpful health resources and information on the internet” and questions 4 and 5 as the ability to acquire and use health resources and information, this study defined questions 1-3 as the ability to search for health resources on the internet and questions 4 and 5 as the ability to use online health information. Results revealed that differences in the delineation of questions lead to variations in the model fit. Notably, empirical data showed that the fit of the 3-factor model in this study is superior to that of Sudbury-Riley et al’s model. This can be attributed to 2 potential explanations. First, EFA was performed in the pretest to delineate the 3 factors, which differed from Sudbury-Riley et al’s method, who carefully reviewed and partitioned the factors based on social cognitive and self-efficacy theories. Second, minor differences in participants’ perceptions of the translated scale may have contributed to these disparities . In the English eHEALS, questions 3-5 start with “I know how to,” which may have caused participants to perceive them as belonging to the same factor . In the Chinese eHEALS, participants tended to consider questions 1-3 as search factors due to words such as “what,” “where,” and “find,” while the word “use” in questions 4 and 5 led participants to classify it as a usage factor. Nonetheless, the 3-factor model used in this study complies with the foundational theories of the eHEALS lily model (ie, social cognitive theory and self-efficacy theory) . This model may be more suitable for regions where the Chinese eHEALS is used in eHealth literacy studies. eHealth Literacy Level In this study, the overall eHEALS score of the university students was moderate or higher, and the search and usage dimensions had higher scores. In contrast, the evaluation dimension had a lower score. This reveals that students perceive themselves to have good search and usage capabilities of eHealth information; however, they possess low confidence in evaluating such information and using it for decision-making. In recent studies, the mean scores for eHEALS questions 6-8 were lower than those for questions 1-3 and questions 4 and 5 , similar to scores obtained in this study. A Taiwanese study used a self-formulated scale to evaluate the eHealth literacy of university students and divided the questions into functional, interactive, and critical literacies . Interactive literacy encompasses the ability to select, comprehend, and use online health information, which was similar to the search and usage dimensions in this study. Critical literacy refers to the ability to analyze, criticize, and respond to online health information, which was similar to the evaluation dimension in this study. The score for critical literacy was visibly lower than that for interactive literacy in the previous study , which was similar to this study. Researchers found that although most university students mentioned that they can understand the general idea of online health information, they have a vague understanding of the jargon, foreign languages, and data . In addition, some university students lack confidence in the quality of online health information and express difficulty in determining the quality of such information . Therefore, in the contemporary landscape characterized by the unlimited accumulation and dissemination of internet-based health information of uncertain veracity, imparting fundamental health knowledge to Taiwanese university students is imperative. This includes fostering a sense of caution toward eHealth information among students and equipping them with the ability to critically assess and validate uncertainties. Association Between Sociodemographic Variables and the HPLP The overall HPLP of university students in this study was moderate or higher, wherein interpersonal support and self-actualization scores were the highest, while nutrition, exercise, and health responsibility scores were the lowest, similar to those of the most recent studies . Among sociodemographic variables, stepwise multiple linear regression showed that female students, students attending vocational colleges, those with a daily reading time of <1 hour, those who acquired information by themselves, and those with an average or a poor perceived health status were confounders of a poor overall HPLP. This was consistent with the significant differences in the overall HPLP in these sociodemographic variables. Recent studies have found that sex affects the HPLP and health behaviors, such as exercise and sleep . The frequency of discussions of health problems with others has been highlighted to positively affect the dietary behavior of university students . Individuals with a good perceived health status or great concern for health have a better HPLP and show several health behaviors, such as eating, exercise, and sleep . In addition, this study found that a daily reading time of ≥1 hour is a confounder of a good HPLP among university students. This may be because information in books, newspapers, and magazines usually undergoes review and proofreading, and reading more accurate and reliable hardcopy information may lead to a tendency to adopt a positive lifestyle profile. Some studies have highlighted that reading hardcopy materials can promote better comprehension results than reading from screens . However, the increased screen time on digital media today has greatly decreased the reading time in print. In this study, only 208 (51.2%) of 406 university students read for ≥1 hour per day, but 362 (89.2%) spent ≥3 hours on mobile devices or computers daily. Recent studies have shown that newspapers and magazines are the media that Taiwanese university students spend the least time on, far below the time spent on mobile devices and computers . Furthermore, mobile devices and computers have diverse online functions. The TWNIC survey found that the most commonly used internet functions among generation Z are real-time messaging, social networks, free videos, online news, online games, and ecommerce; however, online learning is not their priority . In the ANOVA in this study, the HPLP score of participants with a daily screen time of ≥6 hours was significantly low. However, multiple linear regression excluded the daily screen time from the HPLP confounders. It is believed that frequent usage of mobile devices and computers by university students consumes the time spent on reading. Therefore, the effect of screen time in the multiple linear regression may be explained by the reading time factor. In the binary logistic regression in this study, daily screen time was still considered a confounder of HPLP scores. Association of eHEALS With the HPLP After adjusting for sociodemographic factors that may affect the HPLP, this study revealed that eHEALS consistently and significantly affects the HPLP of university students. Compared with students with relatively high overall eHEALS scores, those with relatively low eHEALS scores had a higher probability of a negative overall HPLP, similar to the results of studies in other regions . Other researchers have used their own created scales to measure eHealth literacy and proved that it predicts multiple HPLP dimensions in university students . Many studies have found that eHealth literacy has positive effects on exercise, diet, and sleep behaviors , or even safe sex practice and COVID-19 prevention among university students. This study found that among the various HPLP dimensions, compared with a relatively high overall eHEALS score, a relatively low eHEALS score is associated with negative health responsibility, exercise, and nutrition. University students in this study had low HPLP health responsibility, exercise, and nutrition scores—dimensions that require improvements. At the same time, many recent studies have found that these health behaviors are poor in university students . Additionally, among the 3 eHEALS dimensions in this study, compared with participants with relatively high search, usage, and evaluation literacies, those with relatively low scores had a higher probability of a negative overall HPLP and its health responsibility, exercise, and nutrition dimensions, similar to the overall eHEALS results. Regarding evaluation literacy, this study found that in addition to predicting the negative health responsibility, exercise, and nutrition dimensions of the HPLP, a relatively low eHEALS evaluation score can also reflect poor stress management. This result is similar to that of a recent study indicating that critical eHealth literacy can predict more HPLP dimensions .
Norman and Skinner developed eHEALS and highlighted that men’s eHEALS scores are significantly higher than those of women, which could be used as an a priori hypothesis. Similar results were obtained by using the Chinese eHEALS in this study; in other words, significant differences were observed in eHEALS scores between sexes ( t 404 =2.708, P =.007), with males having higher scores (mean 3.25, SD 0.51) than females (mean 3.12, SD 0.46). This shows that the Chinese eHEALS has known-groups validity. Moreover, in this study, the original 8 eHEALS questions were classified into 3 factors, namely search (questions 1-3), usage (questions 4 and 5), and evaluation (questions 6-8). Compared with the initial 1-factor model , CFA showed that the 3-factor model exhibits a better fit and good validity and reliability. The findings were similar to those of a recent study on the Chinese eHEALS multifactorial model ; however, this study showed more robust evidence of fit, validity, and reliability. In contrast to Sudbury-Riley et al , who used a 3-factor eHEALS model and defined question 3 as “I know how to find helpful health resources and information on the internet” and questions 4 and 5 as the ability to acquire and use health resources and information, this study defined questions 1-3 as the ability to search for health resources on the internet and questions 4 and 5 as the ability to use online health information. Results revealed that differences in the delineation of questions lead to variations in the model fit. Notably, empirical data showed that the fit of the 3-factor model in this study is superior to that of Sudbury-Riley et al’s model. This can be attributed to 2 potential explanations. First, EFA was performed in the pretest to delineate the 3 factors, which differed from Sudbury-Riley et al’s method, who carefully reviewed and partitioned the factors based on social cognitive and self-efficacy theories. Second, minor differences in participants’ perceptions of the translated scale may have contributed to these disparities . In the English eHEALS, questions 3-5 start with “I know how to,” which may have caused participants to perceive them as belonging to the same factor . In the Chinese eHEALS, participants tended to consider questions 1-3 as search factors due to words such as “what,” “where,” and “find,” while the word “use” in questions 4 and 5 led participants to classify it as a usage factor. Nonetheless, the 3-factor model used in this study complies with the foundational theories of the eHEALS lily model (ie, social cognitive theory and self-efficacy theory) . This model may be more suitable for regions where the Chinese eHEALS is used in eHealth literacy studies.
In this study, the overall eHEALS score of the university students was moderate or higher, and the search and usage dimensions had higher scores. In contrast, the evaluation dimension had a lower score. This reveals that students perceive themselves to have good search and usage capabilities of eHealth information; however, they possess low confidence in evaluating such information and using it for decision-making. In recent studies, the mean scores for eHEALS questions 6-8 were lower than those for questions 1-3 and questions 4 and 5 , similar to scores obtained in this study. A Taiwanese study used a self-formulated scale to evaluate the eHealth literacy of university students and divided the questions into functional, interactive, and critical literacies . Interactive literacy encompasses the ability to select, comprehend, and use online health information, which was similar to the search and usage dimensions in this study. Critical literacy refers to the ability to analyze, criticize, and respond to online health information, which was similar to the evaluation dimension in this study. The score for critical literacy was visibly lower than that for interactive literacy in the previous study , which was similar to this study. Researchers found that although most university students mentioned that they can understand the general idea of online health information, they have a vague understanding of the jargon, foreign languages, and data . In addition, some university students lack confidence in the quality of online health information and express difficulty in determining the quality of such information . Therefore, in the contemporary landscape characterized by the unlimited accumulation and dissemination of internet-based health information of uncertain veracity, imparting fundamental health knowledge to Taiwanese university students is imperative. This includes fostering a sense of caution toward eHealth information among students and equipping them with the ability to critically assess and validate uncertainties.
The overall HPLP of university students in this study was moderate or higher, wherein interpersonal support and self-actualization scores were the highest, while nutrition, exercise, and health responsibility scores were the lowest, similar to those of the most recent studies . Among sociodemographic variables, stepwise multiple linear regression showed that female students, students attending vocational colleges, those with a daily reading time of <1 hour, those who acquired information by themselves, and those with an average or a poor perceived health status were confounders of a poor overall HPLP. This was consistent with the significant differences in the overall HPLP in these sociodemographic variables. Recent studies have found that sex affects the HPLP and health behaviors, such as exercise and sleep . The frequency of discussions of health problems with others has been highlighted to positively affect the dietary behavior of university students . Individuals with a good perceived health status or great concern for health have a better HPLP and show several health behaviors, such as eating, exercise, and sleep . In addition, this study found that a daily reading time of ≥1 hour is a confounder of a good HPLP among university students. This may be because information in books, newspapers, and magazines usually undergoes review and proofreading, and reading more accurate and reliable hardcopy information may lead to a tendency to adopt a positive lifestyle profile. Some studies have highlighted that reading hardcopy materials can promote better comprehension results than reading from screens . However, the increased screen time on digital media today has greatly decreased the reading time in print. In this study, only 208 (51.2%) of 406 university students read for ≥1 hour per day, but 362 (89.2%) spent ≥3 hours on mobile devices or computers daily. Recent studies have shown that newspapers and magazines are the media that Taiwanese university students spend the least time on, far below the time spent on mobile devices and computers . Furthermore, mobile devices and computers have diverse online functions. The TWNIC survey found that the most commonly used internet functions among generation Z are real-time messaging, social networks, free videos, online news, online games, and ecommerce; however, online learning is not their priority . In the ANOVA in this study, the HPLP score of participants with a daily screen time of ≥6 hours was significantly low. However, multiple linear regression excluded the daily screen time from the HPLP confounders. It is believed that frequent usage of mobile devices and computers by university students consumes the time spent on reading. Therefore, the effect of screen time in the multiple linear regression may be explained by the reading time factor. In the binary logistic regression in this study, daily screen time was still considered a confounder of HPLP scores.
After adjusting for sociodemographic factors that may affect the HPLP, this study revealed that eHEALS consistently and significantly affects the HPLP of university students. Compared with students with relatively high overall eHEALS scores, those with relatively low eHEALS scores had a higher probability of a negative overall HPLP, similar to the results of studies in other regions . Other researchers have used their own created scales to measure eHealth literacy and proved that it predicts multiple HPLP dimensions in university students . Many studies have found that eHealth literacy has positive effects on exercise, diet, and sleep behaviors , or even safe sex practice and COVID-19 prevention among university students. This study found that among the various HPLP dimensions, compared with a relatively high overall eHEALS score, a relatively low eHEALS score is associated with negative health responsibility, exercise, and nutrition. University students in this study had low HPLP health responsibility, exercise, and nutrition scores—dimensions that require improvements. At the same time, many recent studies have found that these health behaviors are poor in university students . Additionally, among the 3 eHEALS dimensions in this study, compared with participants with relatively high search, usage, and evaluation literacies, those with relatively low scores had a higher probability of a negative overall HPLP and its health responsibility, exercise, and nutrition dimensions, similar to the overall eHEALS results. Regarding evaluation literacy, this study found that in addition to predicting the negative health responsibility, exercise, and nutrition dimensions of the HPLP, a relatively low eHEALS evaluation score can also reflect poor stress management. This result is similar to that of a recent study indicating that critical eHealth literacy can predict more HPLP dimensions .
This study has certain limitations, which can provide a reference for future studies. First, this study included only university students without major diseases from the capital of Taiwan, and the results can only be generalized to the eHealth literacy and health-promoting lifestyle of this population. It is recommended that future studies extend to other regions in Taiwan or university students with other health statuses. Second, some variables may be related to eHealth literacy and a health-promoting lifestyle, such as majors and health risk behaviors, and it is recommended that future studies expand to include these variables. In addition, although participants were advised that the entire process was anonymized, the self-administered questionnaire may have caused their answers to be exposed to memory recall errors, environmental effects, and social desirability bias. Lastly, a cross-sectional study design was used in this study, and the causal relationship between eHealth literacy and a health-promoting lifestyle, as well as changes in these 2 factors with time, could not be confirmed. Hence, further repeated-measures or longitudinal studies are required for clarification. Nonetheless, this study confirmed the feasibility of using the Chinese version of the eHEALS 3-factor model to examine eHealth literacy and highlighted that eHealth literacy affects and predicts the HPLP in university students. In the contemporary world where internet use is widespread and portable mobile devices are rapidly advancing, using the internet, mobile phones, tablets, or computers as aids in daily life has become an unstoppable trend. If university students can cultivate the online learning habit early on and establish the concept of consulting to acquire information and reading in print, actively nurturing their skills to search, use, and access internet-based health information, it will undoubtedly positively impact their health-promoting behavior and lifestyle.
This study is the first to validate the Chinese eHEALS 3-factor model, encompassing search, usage, and evaluation dimensions. Notably, eHEALS is the first eHealth literacy measurement tool to be developed and is the most widely used. This 3-factor model results in more definite eHEALS content and undoubtedly increases the practicality and applicability of the scale to satisfy the eHealth literacy evaluation needs of health promoting–related studies, particularly in Chinese-speaking regions. Higher education represents the most significant and final opportunity for behavioral development and learning in young people. Behavioral health during this period impacts lifetime health outcomes. This study found that alongside specific sociodemographic characteristics, the overall eHEALS and its dimensions are independent predictors of the HPLP. Compared to university students with relatively high overall eHEALS and various dimension scores, those with relatively low scores had a negative overall HPLP and HPLP health responsibility, exercise, and nutrition. University students with relatively low eHEALS evaluation scores compared to those with relatively high evaluation scores also had negative stress management. These findings can be used to screen university students who require HPLP improvement so that health education suitable for their needs can be provided. In addition, there is room for improving overall eHEALS scores among university students, with particular attention to improving evaluation literacy. It is recommended that the centers for general education, digital learning, and health of the universities and colleges in Taipei, as well as targeting populations with relatively low eHealth literacy (eHEALS score<3.17), be integrated to provide appropriate health education and programs. Courses should be conducted to educate students on identifying objective, credible, and understandable online health information platforms, while cultivating vigilance and critical judgment in evaluating eHealth information. Additionally, fostering a supportive and user-friendly online health information environment is essential. It is recommended that universities and colleges further establish good campus eHealth literacy learning and support channels. For example, good health information online platforms could be recommended on school websites, and in-person or virtual health information consultation could be applied within schools. These measures would collectively contribute to improving university students’ eHealth literacy, thereby encouraging their adoption of health-promoting lifestyles.
|
The Indian Health Service Primary Care-Based Teleophthalmology Program for Diabetic Eye Disease Surveillance and Management | f09daf36-2850-4b58-8ae7-9982e15351ad | 7757525 | Ophthalmology[mh] | At 15.1%, the age-adjusted rate of diagnosed diabetes is higher in American Indians and Alaska Natives (AI/AN) than any other major race/ethnic group in the United States (U.S.). AI/AN also tend to have higher rates of diabetes-related complications. Diabetic retinopathy (DR) is the most common microvascular complication of diabetes and is the leading cause of moderate and severe vision loss among working age adults. Timely diagnosis and treatment of DR is effective in substantially reducing vision loss due to DR, but approximately half of patients with diabetes in the United States do not receive the recommended annual DR examination. , Many factors contribute to this low exam rate, including access to care and patient awareness. , To increase the rate that AI/AN with diabetes receive DR examinations, the Indian Health Service launched the Joslin Vision Network (IHS-JVN) Teleophthalmology Program in 2000. The program arose from a federal, multicenter grant to support research for improved evaluation and management of diabetes and DR. Work leading to the IHS-JVN began in 1999 with the development of the technical infrastructure, Reading Center, and clinical protocols modeled after the JVN at the Joslin Diabetes Center (JDC). Discrete aspects of the IHS-JVN and the JDC JVN have been described. This article gives a broad overview of the program's main components and outcomes.
DR best practices combined with telemedicine standards have been used extensively in the IHS-JVN. In turn, the IHS-JVN has participated in standards development for telemedicine, particularly the American Telemedicine Association (ATA) Telehealth Practice Recommendations for Diabetic Retinopathy. , The practice recommendations describe four validation categories, based on the Early Treatment Diabetic Retinopathy Study (ETDRS), the gold standard for DR diagnosis and treatment. The IHS-JVN is an ATA Category 3 program, meaning the program's clinicians identify the ETDRS defined clinical levels of DR and diabetic macular edema (DME; central involved and non-central involved) severity with a diagnostic accuracy that matches clinical retinal examination through dilated pupils or ETDRS photographs ( , last column ). Functionally, this means that levels of non-proliferative DR (NPDR) that are not yet considered sight-threatening can be found, referred for eye or other specialty care as needed, and medically managed appropriately before they progress to severe levels of DR. A Category 3 program also provides a management plan commensurate with the clinical recommendations that would have resulted from a traditional ophthalmology-based clinical retinal examination through dilated pupils. Category 1 programs differentiate between patients who have no DR and those who have any form of DR. Category 2 programs differentiate between patients who do or do not have potentially sight-threatening DR requiring prompt referral for eye care. The ATA validation categories do not constitute a quality continuum, but rather exist to optimize teleretinal services for a variety of settings. In the IHS, services are provided directly or indirectly through contracts or compacts with tribes or Tribal Organizations (TOs) and grants to Urban Indian Organizations (UIOs). The IHS currently includes 46 hospitals (24 are IHS operated, 22 are tribally operated) and 568 ambulatory facilities (92 are IHS operated, 476 are tribally operated) in 35 states, serving a user population (i.e., the number of patients who used IHS-funded services at least once in the most recent 3 years) of 1.6 million. Most of the ambulatory facilities are rural with limited availability of specialty providers. Given the largely rural context with limited access to specialists, the IHS-JVN program was designed to recruit patients opportunistically when they present for primary care and to minimize referrals for specialty care. The ATA validation categories do not include measures for non-DR findings, but such findings are common and may be associated with severe vision loss and/or general health risks. Thus, the JVN protocol was studied to ensure its capacity to identify non-DR findings as well (e.g., cataract, age-related maculopathy choroidal lesions, etc.). , A few telemedicine programs for DR are validated to ATA Category 3. The IHS-JVN is the largest primary care-based ATA Category 3 program in the U.S.
The IHS-JVN conforms with Government Performance and Results Act (GPRA), which provides for annual, national reporting of defined clinical outcomes considered key indicators of health care quality. The GPRA defines a qualified retinal evaluation as: (1) a dilated retinal evaluation by an optometrist or ophthalmologist; (2) seven standard fields stereoscopic photos (ETDRS) evaluated by an optometrist or ophthalmologist; or (3) any photographic method formally validated to seven standard fields (ETDRS). The IHS-JVN satisfies the latter of these options.
Clinical workflow of the IHS-JVN begins with opportunistic recruitment of patients during primary care encounters . A certified Imager acquires a standardized series of retinal photographs specific to one of two imaging technologies (described below) and conducts limited diabetes and DR education during the encounter, using the patients' retinal images as conversation points. The Imager also looks at the patients' images to determine whether urgent assessment is needed from the Reading Center for possible urgent or sight-threatening findings (i.e., STAT read). The images are saved locally and then securely transmitted to the IHS-JVN National Reading Center at the Phoenix Indian Medical Center (PIMC) for assessment according to a standardized protocol performed by certified optometrist, teleretinal Readers with ophthalmologist oversight. The Reader diagnoses DR and DME severity level by using ETDRS rules-based, validated computer-assisted decision support and recommends a management plan in a report sent to the patient's primary care provider and imported into the electronic health record (EHR). If follow-up eye or other specialty care is needed, it is the responsibility of the primary care staff to make that referral. There are currently four Readers at PIMC.
The IHS-JVN technology has a modular architecture and a suite of applications for image acquisition, database management/monitoring/reporting, picture archiving and communication system, image display/post-imaging processing, study reading/reporting with computer-assisted decision support, and quality assurance (QA). Image acquisition The program uses two configurations of commercially available technology for image acquisition. The first configuration uses a low-illumination, nonmydriatic fundus photography (NMFP) digital imaging system (Topcon NW6S; Topcon Medical Systems, Inc., Paramus, NJ) with a custom digital camera back that provides low-light exposure capability to reduce reflexive pupillary constriction from multi-image light exposures (Megavision Retinal Image Capture; Santa Barbara, CA). Three non-simultaneous stereo pair 45° (degree) and two non-simultaneous stereo pair 30° digital images of the retina of each eye are obtained. One external image of the anterior segment of each eye is also obtained. The NMFP showed substantial agreement with ETDRS controls for diagnosis of DR severity level (unweighted κ = 0.81, 95% confidence interval = 0.73–0.89). The second configuration is nonmydriatic ultrawide-field imaging (UWFI) scanning laser ophthalmoscopy (Daytona, Optos, plc, Dunfermline, United Kingdom). The UWFI protocol includes non-simultaneous stereo pair 200° images from each eye, centered on the macula. The UWFI showed perfect agreement with ETDRS photography in 84% of cases, and agreement within one level of severity in 91% of cases (unweighted κ = 0.79). The IHS-JVN introduced UWFI to the program in 2014. It has become the default imaging system for new deployments and upgrades, because it has greater clinical accuracy, requires fewer images, incorporates automatic image acquisition, and lowers the proportion of poor images from which Readers cannot make a diagnosis of DR/DME severity level. However, the NMFP retinal imager is portable and durable, so it is still used in smaller and more remotely located settings requiring a durable and/or portable imager. Image grading The IHS-JVN images are transmitted to the IHS-JVN National Reading Center via an internet-based medical informatics program called the Comprehensive Diabetes Management Program. The software automatically imports patient demographics, diabetes duration, medications, laboratory data, blood pressure, and problem list from the EHR. The Imager manually supplements this information if needed. The IHS-JVN DR grading details are shown in (Category 3 column). In addition, grading for DME severity is based on stereoscopic assessment. Grades include: (1) absent; (2) not clinically significant, characterized by retinal thickening or hard exudates at or within 3,000 μm from the fovea or thickening in the posterior pole within the arcades, but outside the threshold for Clinically Significant Macular Edema (CSME); and (3) CSME, characterized by retinal thickening at or within 500 μm of the fovea, hard exudates at or within 500 μm of the fovea with adjacent retinal thickening, or one or more disk areas of retinal thickening, any part of which is within 1,500 μm of the fovea or with center involvement. If a conclusive determination of the level of DR or DME cannot be made, the Reader reports the study as ungradable for that condition. Stereo imaging and overlapping retinal fields provide redundancy of data within a single retinal field, so images may be ungradable for one condition but sufficient to grade for another. An ungradable outcome triggers a recommendation for referral. Usually, the recommended follow-up is within 3 months; however, the timeframe may be shorter depending on the basis for ungradability, whether other pathology is found, the patient's health status, and other factors. Non-DR abnormalities evident in the images are included in the management recommendations. , Image reporting The Reader generates a report for each IHS-JVN study, which is sent to the originating site for communication to the patient via their primary care physician or any other member of the primary care team. The report includes available information on patient risk factors for progression of DR/DME, DR/DME severity, findings of nondiabetic eye disease, and the recommended management plan . The recommended plan may suggest modification of identified DR risk factors, repeat IHS-JVN imaging, or referral to an eye care provider, primary care provider, or appropriate specialty provider.
The program uses two configurations of commercially available technology for image acquisition. The first configuration uses a low-illumination, nonmydriatic fundus photography (NMFP) digital imaging system (Topcon NW6S; Topcon Medical Systems, Inc., Paramus, NJ) with a custom digital camera back that provides low-light exposure capability to reduce reflexive pupillary constriction from multi-image light exposures (Megavision Retinal Image Capture; Santa Barbara, CA). Three non-simultaneous stereo pair 45° (degree) and two non-simultaneous stereo pair 30° digital images of the retina of each eye are obtained. One external image of the anterior segment of each eye is also obtained. The NMFP showed substantial agreement with ETDRS controls for diagnosis of DR severity level (unweighted κ = 0.81, 95% confidence interval = 0.73–0.89). The second configuration is nonmydriatic ultrawide-field imaging (UWFI) scanning laser ophthalmoscopy (Daytona, Optos, plc, Dunfermline, United Kingdom). The UWFI protocol includes non-simultaneous stereo pair 200° images from each eye, centered on the macula. The UWFI showed perfect agreement with ETDRS photography in 84% of cases, and agreement within one level of severity in 91% of cases (unweighted κ = 0.79). The IHS-JVN introduced UWFI to the program in 2014. It has become the default imaging system for new deployments and upgrades, because it has greater clinical accuracy, requires fewer images, incorporates automatic image acquisition, and lowers the proportion of poor images from which Readers cannot make a diagnosis of DR/DME severity level. However, the NMFP retinal imager is portable and durable, so it is still used in smaller and more remotely located settings requiring a durable and/or portable imager.
The IHS-JVN images are transmitted to the IHS-JVN National Reading Center via an internet-based medical informatics program called the Comprehensive Diabetes Management Program. The software automatically imports patient demographics, diabetes duration, medications, laboratory data, blood pressure, and problem list from the EHR. The Imager manually supplements this information if needed. The IHS-JVN DR grading details are shown in (Category 3 column). In addition, grading for DME severity is based on stereoscopic assessment. Grades include: (1) absent; (2) not clinically significant, characterized by retinal thickening or hard exudates at or within 3,000 μm from the fovea or thickening in the posterior pole within the arcades, but outside the threshold for Clinically Significant Macular Edema (CSME); and (3) CSME, characterized by retinal thickening at or within 500 μm of the fovea, hard exudates at or within 500 μm of the fovea with adjacent retinal thickening, or one or more disk areas of retinal thickening, any part of which is within 1,500 μm of the fovea or with center involvement. If a conclusive determination of the level of DR or DME cannot be made, the Reader reports the study as ungradable for that condition. Stereo imaging and overlapping retinal fields provide redundancy of data within a single retinal field, so images may be ungradable for one condition but sufficient to grade for another. An ungradable outcome triggers a recommendation for referral. Usually, the recommended follow-up is within 3 months; however, the timeframe may be shorter depending on the basis for ungradability, whether other pathology is found, the patient's health status, and other factors. Non-DR abnormalities evident in the images are included in the management recommendations. ,
The Reader generates a report for each IHS-JVN study, which is sent to the originating site for communication to the patient via their primary care physician or any other member of the primary care team. The report includes available information on patient risk factors for progression of DR/DME, DR/DME severity, findings of nondiabetic eye disease, and the recommended management plan . The recommended plan may suggest modification of identified DR risk factors, repeat IHS-JVN imaging, or referral to an eye care provider, primary care provider, or appropriate specialty provider.
Standardized training of Imagers and Readers and structured QA protocols are used to promote ongoing fidelity with the validation studies by both existing and newly added sites. Imager training begins with home-based, self-paced pre-training on ocular anatomy, DR, and telemedicine fundamentals before reporting to PIMC for a 2- or 3-day syllabus on UWFI or NMFP technology, respectively. This training includes structured exposure to Reading Center administrative and reading activities to facilitate the Imager's ability to request an urgent reading for possible urgent or sight-threatening findings. Provisional Imager certification is conveyed based on demonstrated technical, clinical, and administrative proficiency. Full certification is conveyed after 6 months of experience and 100 IHS-JVN patient encounters demonstrating successful quality reviews. Full certification is maintained by ongoing successful quality reviews or by formal retraining and recertification. Reader training is provided to optometrist applicants with recent clinical experience. Most applicants come from IHS health care facilities. Home-based, self-paced pre-training emphasizes DR pathophysiology and foundational clinical trials. , , This is followed by 3 days of Reader training at JDC (Boston). Provisional Reader certification is conveyed on demonstrated proficiency of: (1) application of ETDRS-based interpretation of JVN digital images (NMFP and UWFI) in native stereoscopic format, with and without post-imaging processing enhancement, as compared with reference ETDRS image sets; and (2) post-imaging processing and other reading tools, computer-assisted decision support, and standardized reporting. Full Reader certification is conveyed after 6 months of successful interpretation of the images and use of the technology as determined by senior Reading Center staff. Full certification is maintained by ongoing successful quality reviews, focused training stemming from structured peer review, or by formal retraining and recertification.
Ongoing QA is a requirement of accreditation regulations and the ATA DR practice guidelines. The IHS-JVN is carefully monitored end-to-end, with the outcome used for performance improvement as described by ATA recommendations. , Imagers and Readers are assessed for volume, recency, and proficiency of administrative and clinical performance. Technical support quality is similarly assessed. The overall program is assessed through monitoring of Reading Center latency (routine and STAT reads), reading queue, hosting site productivity, and hosting site annual DR examination rate. QA reporting is used internally to guide staff and provider education, provider re-credentialing, and regulatory compliance.
Outcomes to measure success of the program include: (1) geographical adoption by IHS and Tribal programs thus far and clinical volume; (2) improved DR surveillance rate with an associated improvement in DR treatment rate; (3) cost-effectiveness compared with a conventional examination; (4) technology testing for improved clinical performance and operational efficiency; and (5) expansion of epidemiologic data on DR and DME in AI/AN. Geographical adoption and clinical volume IHS health care services are administered through 12 administrative area offices and the IHS-JVN is currently active in 11: Alaska; Albuquerque; Bemidji; Billings; Great Plains; Nashville; Navajo; Oklahoma; Phoenix; Portland; and Tucson. As of the final quarter of fiscal year 2019, the IHS-JVN program had 99 active sites, distributed across 23 states . In fiscal year 2019, the IHS-JVN imaged 29,958 unique patients. Since the program was initiated, it has conducted 226,333 studies cumulatively. In the past 3 years, during which the program cumulatively imaged 62,091 patients, we estimate that the program served about 25% of the IHS user population with diabetes. Facility participation in the IHS-JVN program is optional for IHS and tribal facilities. Facilities apply to participate and the program evaluates their eligibility, taking into consideration the facility's number of patients with diabetes, their existing annual DR examination rate (GPRA), and the clinical, technical, and business ability of the facility to support ongoing operations of the IHS-JVN. As facilities apply, the program will continue to expand to serve more patients. DR examination rate A retrospective study of the annual (1999–2003) rates of DR examinations and follow-up laser treatment was conducted at PIMC, the site of the first IHS-JVN deployment in 2000. A nearby PIMC satellite clinic without the IHS-JVN was used for comparison. Although the annual DR examination rates at the comparison clinic remained stable over the period of study, ranging from 51% to 59%, PIMC's annual DR examination rate increased from 50% in 1999 to 75% in 2003. The rate of laser treatment for the satellite clinic was not reported, but there was a parallel 51% increase in laser treatment for DR at PIMC in the same population during the same period. There has been a similar trend in the national GPRA reports. Specifically, the national rate of annual DR examinations increased from 49.0% to 61.3% during 2007–2015. This 25% increase in exam rates was due to both IHS-JVN exams and conventional eye exams. Because there was not a significant change in conventional eye examination capacity within the IHS in this period, we believe that the IHS-JVN contributed substantially to the rate change. Cost-effectiveness A modeled economic analysis found that the IHS-JVN was more cost-effective than a conventional dilated retinal examination in detecting proliferative DR (PDR) and identifying patients requiring intervention to prevent vision loss. Specifically, the analysis showed that the IHS-JVN would detect an additional 148 cases of PDR over conventional eye exams while lowering costs by $525,690. One hundred thirty-nine more patients would receive laser treatment for their PDR, at a cost savings of $195,210. Five fewer patients would develop severe vision loss while saving another $324,810. Technology testing The performance of UWFI was compared with NMFP in the IHS-JVN over a 1-year period and 25,635 patients. The UWFI resulted in more than an 80% decrease in the image ungradable rate, doubled the rate of diagnosed DR, identified a more severe rate of DR in 9% of patients, and prevented unnecessary referrals in more than 4,000 patients. DR/DME epidemiology Prior reports of DR prevalence in AI/AN were dated, and research on DME in AI/AN was lacking. Thus, the IHS-JVN analyzed retrospective data from 53,998 consecutive patients imaged nationally by the program between November 1, 2011 and October 31, 2016 to determine the prevalence of: levels of DR and DME; presence/absence of any DR; and presence/absence of sight-threatening retinopathy (STR), defined as severe NPDR, any PDR, or any DME. The analysis found any level of DR in 17.7% of patients and STR in 4.2% of patients. A subset analysis of the UWFI images ( n = 16,535) found any level of DR in 28.2% of patients and STR in 5.4%. This latter finding suggests that the wider field of view and lower ungradable rate of UWFI particularly make it easier for Readers to see mild and moderate NPDR. That analysis was reported in the most comprehensive, peer-reviewed publication describing DR and DME prevalence and severity in AI/AN to date.
IHS health care services are administered through 12 administrative area offices and the IHS-JVN is currently active in 11: Alaska; Albuquerque; Bemidji; Billings; Great Plains; Nashville; Navajo; Oklahoma; Phoenix; Portland; and Tucson. As of the final quarter of fiscal year 2019, the IHS-JVN program had 99 active sites, distributed across 23 states . In fiscal year 2019, the IHS-JVN imaged 29,958 unique patients. Since the program was initiated, it has conducted 226,333 studies cumulatively. In the past 3 years, during which the program cumulatively imaged 62,091 patients, we estimate that the program served about 25% of the IHS user population with diabetes. Facility participation in the IHS-JVN program is optional for IHS and tribal facilities. Facilities apply to participate and the program evaluates their eligibility, taking into consideration the facility's number of patients with diabetes, their existing annual DR examination rate (GPRA), and the clinical, technical, and business ability of the facility to support ongoing operations of the IHS-JVN. As facilities apply, the program will continue to expand to serve more patients.
A retrospective study of the annual (1999–2003) rates of DR examinations and follow-up laser treatment was conducted at PIMC, the site of the first IHS-JVN deployment in 2000. A nearby PIMC satellite clinic without the IHS-JVN was used for comparison. Although the annual DR examination rates at the comparison clinic remained stable over the period of study, ranging from 51% to 59%, PIMC's annual DR examination rate increased from 50% in 1999 to 75% in 2003. The rate of laser treatment for the satellite clinic was not reported, but there was a parallel 51% increase in laser treatment for DR at PIMC in the same population during the same period. There has been a similar trend in the national GPRA reports. Specifically, the national rate of annual DR examinations increased from 49.0% to 61.3% during 2007–2015. This 25% increase in exam rates was due to both IHS-JVN exams and conventional eye exams. Because there was not a significant change in conventional eye examination capacity within the IHS in this period, we believe that the IHS-JVN contributed substantially to the rate change.
A modeled economic analysis found that the IHS-JVN was more cost-effective than a conventional dilated retinal examination in detecting proliferative DR (PDR) and identifying patients requiring intervention to prevent vision loss. Specifically, the analysis showed that the IHS-JVN would detect an additional 148 cases of PDR over conventional eye exams while lowering costs by $525,690. One hundred thirty-nine more patients would receive laser treatment for their PDR, at a cost savings of $195,210. Five fewer patients would develop severe vision loss while saving another $324,810.
The performance of UWFI was compared with NMFP in the IHS-JVN over a 1-year period and 25,635 patients. The UWFI resulted in more than an 80% decrease in the image ungradable rate, doubled the rate of diagnosed DR, identified a more severe rate of DR in 9% of patients, and prevented unnecessary referrals in more than 4,000 patients.
Prior reports of DR prevalence in AI/AN were dated, and research on DME in AI/AN was lacking. Thus, the IHS-JVN analyzed retrospective data from 53,998 consecutive patients imaged nationally by the program between November 1, 2011 and October 31, 2016 to determine the prevalence of: levels of DR and DME; presence/absence of any DR; and presence/absence of sight-threatening retinopathy (STR), defined as severe NPDR, any PDR, or any DME. The analysis found any level of DR in 17.7% of patients and STR in 4.2% of patients. A subset analysis of the UWFI images ( n = 16,535) found any level of DR in 28.2% of patients and STR in 5.4%. This latter finding suggests that the wider field of view and lower ungradable rate of UWFI particularly make it easier for Readers to see mild and moderate NPDR. That analysis was reported in the most comprehensive, peer-reviewed publication describing DR and DME prevalence and severity in AI/AN to date.
The IHS-JVN is a large, nationally distributed, primary care-based, teleophthalmology program for AI/AN that is validated at ATA Validation Category 3 for the remote diagnosis and management of DR/DME severity levels and certain non-DR disease. Because Category 3 programs are complex and costly to operate, they are reserved for use in specific circumstances. The IHS circumstances are such that referred specialty care may be limited. This makes the higher DR severity referral threshold, lower referral rate, and specific management plan reported to the primary care staff particularly important in the IHS. Cost-avoidance is also critical in the IHS due to lower per capita funding of health care for AI/AN than that of the general U.S. population. , Improving the timeliness of DR examinations with the IHS-JVN decreases DR/DME complications and allows for less costly and more effective interventions. Adoption of innovation can be challenging for organizations wishing to improve processes and outcomes. Although the value of teleophthalmology for DR has been established, , the IHS has shown less rapid adoption of the IHS-JVN than anticipated. This is unexpected since the program funds all costs of the equipment needed, the Imager training, the Reading Center services, and ongoing technical support. Sites pay their Imager's salary, usually a fraction of a full-time equivalent, and provide space for the imaging encounters. Imager salary costs can potentially be offset if sites bill third party payers for the technical component of the telemedicine services. Based on the literature and anecdote, we speculate that clinical inertia, , professional resistance, the threat of workflow disruption to the clinic, and competition for staffing/space (anecdote) are likely barriers to program adoption. Looking ahead, a next step for the IHS-JVN program is to quantify its public health impact. As mentioned, the annual DR examination rate among IHS tribal facilities (as tracked by GPRA) has increased annually, in parallel with expansion of the IHS-JVN. This implies a favorable correlation between the two historical trends. But various administrative barriers have precluded retrospective statistical tests of this correlation. In addition, after 2015, changes in the data collection process of this DR GPRA element limit comparison of current and prior (pre-2015) examination rates. Prospectively examining the DR examination rate at existing and new IHS-JVN sites and counting which annual DR exams at these sites are from the program versus a conventional eye exam is needed. This would give administrators and providers guidance on how the IHS-JVN can help clinical performance, regulatory compliance, and utilization at existing sites and improve public health overall. Recently, artificial intelligence has gained attention as a way to increase access to cost-effective screening for DR, particularly in low- and middle-income countries. However, existing artificial intelligence systems for diagnosing DR are not validated at ATA Category 3, nor has the performance of these systems been evaluated in AI/AN populations. For the IHS-JVN context, artificial intelligence is an opportunity for the program to enhance Reading Center performance, rather than replace manual grading and interpretation of retinal images. Specifically, the program plans to use artificial intelligence algorithms to triage patients with no or mild DR so that patients with more severe DR can be prioritized in the Readers' queue, shortening the reading latency for these cases.
|
Postmortem Immunohistochemical Findings in Early Acute Myocardial Infarction: A Systematic Review | e62b3ca9-71e1-424f-bed1-77352873910e | 11277133 | Anatomy[mh] | One of the most frequent causes of sudden cardiac death worldwide is acute myocardial infarction . In some cases, given the short survival time after the acute coronary event, there are no obvious changes identifiable either grossly or using classical histopathology methods, which causes uncertainty regarding a positive diagnosis . Nowadays, the postmortem diagnosis of acute myocardial infarction may employ, besides the analysis of gross and microscopic slides (which may allow its detection if the person survives for more 12 h from the onset of the coronary occlusion), various molecular, genetic and biochemical markers, with higher costs and limited availability . Recently, minimally invasive or non-invasive postmortem imaging techniques have been shown to be useful to diagnose acute myocardial infarction . If death occurs within six hours from the onset of the occlusion, the methods that can be used in forensic practice are more limited, mainly due to their costs. In these cases, the diagnosis of acute myocardial infarction is mostly indirect, undertaken by detecting the narrowing of the coronary lumen without observing myocardial tissue with the morphological appearance of an infarction, which increases the risk of over-diagnosing it and the omission of other potential non-ischemic myocardial pathologies . Nowadays, a relatively cheap and widely available technique that has been shown to be useful for the diagnosis of acute myocardial infarction is immunohistochemistry, numerous markers being tested, with varying success, for a positive and differential diagnosis. Although there are useful pre-existing reviews on this topic , the particular aim of our review is to analyze and systematize the immunohistochemical markers presented in the literature concerning effectiveness in the postmortem detection of early acute myocardial infarction, without omitting necessary and useful laboratory elements in this regard, to help focus on a reliable and reproducible range of potential immunohistochemical markers. A number of 1986 studies were initially found. Following their analysis based on the inclusion criteria, we have included 15 articles in this review, as described below . The following IHC markers were discussed in the included studies as potentially effective in dating acute myocardial infarction within 6 h of onset . 2.1. Complement Factors and CD 59 Irreversible ischemic degenerative changes were observed as early as the first 20 min after total coronary obstruction, which allows for the identification of complement complexes at the myocardial level . The membrane attack complex (MAC) of the complement system is part of the C5b-9 membrane complex, an immunohistochemical marker that can be detected in renal , muscle , and myocardial necrotic biopsy samples. C5b-9 accumulates specifically at the level of necrotic myocardial cells and is associated with cell membrane damage by opening membrane pores . Its concentration depends on the diffusion rate of the complement substrate in the infarcted area, and therefore on reperfusion . Jasra et al., in a human necropsy casework study in which they compared myocardial infarctions more recent than 6 h and myocardial infarctions older than 6 h, concluded that ischemic and infarcted cardiac myofibrils show a positive C9 staining as early as the first 6 h . CD 59, also called protectin, inhibits the cytolytic activity of the complement system, and therefore myocardial loss and its plasma release occur prior to complement activation . From an immunohistochemical standpoint, Väkevä et al. in their studies found that CD59 expression exists at the level of the sarcolemal membrane of the normal myocardium, whereas in the infarcted region after 1–14 days, it is low or absent, accompanied by the concomitant deposition of MAC in CD59-negative areas. Small CD59 vesicles were found at the normal myocardium-infarcted myocardial boundary, suggesting its removal by a possible excretory mechanism . 2.2. Myoglobin (MB) Myoglobin is a morpho-functionally comparable protein to hemoglobin; the difference between them lies in the number of polypeptide chains and the number of oxygen-binding sites. Myoglobin has a greater affinity for oxygen compared to hemoglobin and is very effective in extracting oxygen from the blood. It is found in skeletal muscle fiber sarcoplasm and cardiac myocyte cytoplasm . Also, in much lower concentrations, it can be found in smooth muscle, endothelial, and tumor cells, such as in breast, lung, and colon cancer, etc. , which makes this protein nonspecific. There are studies evaluating the morphological changes and myoglobin content in the normal myocardium versus the ischemic and necrotic myocardium. These studies found a loss of myoglobin from the necrotic myocardium . Considering the criterion for the biohumoral detection of myoglobin level, the concentration peaks 6 h from the moment of coronary obstruction; however, for the microscopic criterion, this time interval is lower, allowing an evaluation of necrotic myocardial tissue . Xiaohong et al. found subendocardial cellular myoglobin depletion in mice 30 min after coronary ligation, which expanded transmurally within 3 h . Amin et al., in their necropsy casework study in which they also compared the immunohistochemical expression of MB in cases with definite signs of ischemia, cases with probable signs of ischemia, and cases without signs of ischemia but with other causes of myocardial damage, found variable myoglobin depletion in all cases with definite signs of ischemia and in all cases with other causes of myocardial damage . 2.3. Fibrinogen Fibrinogen is a glycoprotein complex that is converted by thrombi to fibrin, and further to a fibrin-based blood clot during the coagulation process. Its increase is known to be a risk factor for AMI and stroke , and it was one of the earliest immunohistochemical markers evaluated for the diagnosis of AMI. Raza-Ahmad showed that, in the first 12 h postmortem, a positive reaction for fibrinogen was obtained not only on fibers demonstrating coagulative necrosis, contraction band necrosis, or a wavy pattern, but also in fibers suspected of acute ischemia in the absence of visible changes in H&E . Brinkman et al. performed a case–control study using four groups: AMI macroscopically present, occlusive coronary thrombosis without infarction, coronary atherosclerosis without AMI/occlusion, and a control group. The second group (representing coronary occlusion without gross signs of AMI) is representative of supra-acute MI. In this group, the authors found a patchy and moderate reaction, which was much weaker than the grossly visible AMI group . Xiaohong, in a study using an animal model for early myocardial ischemia, found that, after 30 min, fibrinogen staining was identified as light brown dots in subendocardial myocardial cells. One hour later, the immunostaining was identifiable in the middle myocardium, and after another two hours, in the subepicardial cells. After another three hours, the entire myocardial layer was positive . Sabatasso et al., also on an animal model, found the earliest positive reaction for fibrinogen within an hour of the coronary occlusion, only in the ischemic area, while the non-affected myocardium showed a negative reaction . The intensification of the positive reaction of myocardial fibrinogen together with myoglobin depletion denotes possible irreversible myocardial lesions detectable from the first 30 min of the moment of coronary blood flow arrest . However, the reliability of fibrinogen is considered to be low given the potential false-positive results from contaminated samples . 2.4. Desmin Desmin is a protein specific to muscle tissue; it is structural to the heart muscle but also skeletal and smooth muscles. It plays an essential role in maintaining the structural and mechanical integrity of the contractile structure. For example, in rat studies, a deficiency in desmin synthesis was found to cause dilated cardiomyopathy, smooth muscle defects, and skeletal myopathy . In myocardial ischemia, the depletion of desmin from the cytoplasm of the heart cell begins 30 min after the onset of ischemia, and within 90–120 min, depletion is complete . 2.5. Tumor Necrosis Factor Alpha (TNF-α), P-38, and JNK (Jun N Terminal Kinase) TNF-α is a cytokine secreted by macrophages that can be detected mainly in vascular smooth muscle cells. Increased values have been reported in heart failure , but also in infarcted myocardial tissue, with TNF-α having the role of stimulating the proliferation and expression of fibronectin at the level of fibroblasts in the injured myocardium . P-38-activated kinase (MPAK) is interconnected with cardiac activity; its isoforms have a role in the growth and differentiation of cardiomyocytes . It is also closely related to oxidative stress, along with nuclear factor kappa B (NF-k-B), which plays a role in the production of TNF-α and JNK (stress-signaling kinases) . 2.6. Transforming Growth Factor β1 (TGF-β1) TGF-β1 is a profibrotic cytokine found in increased amounts in acute myocardial infarction. It also precedes increased collagen expression . 2.7. Cardiac Troponins cTnT, a marker of myocardial lesions, is a component of the troponin complex of the muscle cell. Serum levels of cTnT increase within 4–6 h of the onset of acute myocardial infarction and peak at about 24 h . Fishbein et. al, in an experimental immunohistochemical study in animals, found an uneven loss of cTnT alongside cTnI in the periphery of the necrotic myocardium, with the loss of cTnT being more pronounced compared to that of cTnI. In some cases, histopathological changes are detectable as early as the first 30 min after coronary occlusion . In the study by Amin et al. on human subjects, the results were similar . Diaz et al., in their necropsy casework study, found in cardiac deaths a more diffuse and pronounced immunohistochemical expression of cTnC compared to that of cTnT within the first hour after the onset of acute myocardial infarction . 2.8. H-FABP (Heart Fatty Acid Binding Protein) H-FABP (heart fatty acid binding protein) is a cardiac protein that plays an essential role in the metabolism of cardiomyocyte fatty acids . At the time of ischemic myocardial injury, H-FABP is released into the blood and eliminated renally. Biohumorally, it can be detected from about an hour from the onset of myocardial infarction , and immunohistopathologically it is considered objectifiable in the first 4 h after blocking the coronary flow. 2.9. Dityrosine Dityrosine is a stable product synthesized by the myeloperoxidase–hydrogen peroxidase system of neutrophils and macrophages. This is a marker of oxidative protein stress . In the study on human necropsy cases conducted by Mayer et al., it was found that dityrosine, although not specific to infarcts, can also be a marker of myocardial infarction with a short survival time, being detectable immunohistochemically after the first 4 h of survival after the onset of infarction . 2.10. Fibronectin Fibronectin is an extracellular matrix protein assembled into viscoelastic fibrils that can bind growth factors and cytokines. These fibrils play an essential role in injury healing processes, possessing unique mechanical properties that allow them to modify detected mechanotransduction signals transmitted by cells . In the experimental study conducted by Casscells et al. to observe the immunohistochemical aspects of fibronectin in acute myocardial infarction, it was found that fibronectin was irregularly located in the cytoplasm and in the interstitial space of myocytes of the area irrigated by the ligated coronary infarction . Saleki et al. performed a semiquantitative analysis of fibronectin immunostaining in three study groups—confirmed AMI, suspected AMI, and negative controls—and employed a four-stage grading for each case (negative to 3+). From 15 cases of confirmed AMI, a strong immunostaining pattern was identified in 10, both in the cytoplasm and nuclei of the affected fibers, in 4—a moderate pattern, and in 1—a weak staining. In the negative controls, 13 had a negative reaction, 4 had a weak reaction, and 1 a moderate reaction . This suggests that the sensibility of a strong immunostaining pattern is high for AMI, but moderate and weak patterns may be encountered in other pathologies as well, possibly caused by terminal ischemia associated with an increased agonal period. Hu et al. tried to evaluate the specificity of positive fibronectin immunostaining for the detection of AMI by quantifying the total positive area using the Interactive Building Analysis System. They found that, in AMI, the positive staining area was around 100 times higher compared to normal controls, hemorrhagic shock, cardiac contusion, and organophosphate poisoning, around 75 times higher compared to mechanical asphyxia and electrocution, and only around 2.5 times higher compared to myocarditis. Their study of fibronectin in necropsy cases additionally found that it is not affected by postmortem autolysis nor by prior formalin fixation, which also allows for the retrospective analysis of cases of forensic interest to detect early myocardial infarction postmortem . 2.11. CD 15 CD15, or Lewis x antigen, is a cell-surface glycan, considered a useful marker in identifying cells of the granulocyte series . Given the hypothesis that from the first hours from the onset of acute myocardial infarction, leukocytes are gradually attracted to the infarcted myocardial area, some studies have tried to estimate the age of acute myocardial infarction based on the number of CD15-positive marked cells . The number of CD-15 positive cells is one of the reliable immunohistochemical markers in dating myocardial infarction, as a linear relationship with time has been found regarding the accumulation of polymorphonuclear leukocytes at this level . For example, in the study conducted by Mortensen et al., it was estimated that several myocardial-positive CD15 cells between 20 and 30 raised the suspicion of an acute myocardial infarction, and values above 30 were considered relevant for an acute myocardial infarction occurring for 5–6 h . 2.12. CD 18 CD 18 is a cell-surface adhesion molecule that shows increased expression on the cell membrane in the infarcted area. It is associated with neutrophil activation. CD18 leukocyte integrins are adhesion receptors released in the acute phase and easily measured. They are mainly involved in neutrophil extravasation . Hill et al., in their experimental rabbit heart study, found an increase in CD18 expression as early as the first 20 min after infarction and correlated it with the extent of myocardial necrosis, observing, however, an interindividual variation that did not allow for percentage/numerical appreciations . 2.13. ICAM 1 ICAM 1 is an endothelial receptor and ligand for leukocyte integrin CD 18. It allows their firm adhesion to the vascular endothelium . There are studies that have demonstrated the increase in soluble ICAM 1 in the context of acute myocardial infarction , but also its overregulation in human cardiomyocytes after infarction. However, this is not detectable in infarcts more recent than one day . 2.14. ILs (Interleukins) ILs are important inflammatory mediators of acute myocardial infarction, with a role in exacerbating (e.g., IL6) or attenuating (e.g., IL10, IL 15) the acute inflammatory response. They are released from various cells into the bloodstream and interstitium, where they bind to interleukin receptors on cell surfaces, causing cell activation. Some studies have also analyzed the prognosis of acute myocardial infarction according to the level of serum interleukins . 2.15. Monocyte Chemoattractant Protein-1 (MCP 1) MCP 1 is a chemokine that induces the recruitment and activation of monocytes, T cells, and NK cells . It is produced by many cells in response to injury factors or exposure to other cytokines, and is also involved in acute myocardial infarction . The antiapoptotic and chemotactic effects of MCP 1 may be mediated by different signaling pathways in the infarcted myocardial area . In the study conducted by Turillazzi et al., a strongly expressed response of MCP-1 was found in the early phase of acute myocardial infarction in the first 0–4 h . 2.16. Tryptase Tryptase is a protease released by mast cells. The increase in mast cell tryptase plays a central role in the immediate inflammatory and allergic reactions initiated by IgE . It induces fibroblast proliferation, stimulates fibroblast chemotaxis, and regulates the increased production of collagen type I . The same study mentioned above by Turillazzi et al. found a moderate immunohistochemical reaction for tryptase at myocardial infarction sites 0–6 h old, along with CD15, IL-1β, IL-6, TNF-α, IL-8, CD18, ICAM-1, IL-15, and MCP-1 . 2.17. Adiponectin Adiponectin is an adipokine secreted by adipocytes, with anti-inflammatory, antifibrotic, antioxidant, and cardioprotective effects . Experimental studies on animals have proved that in ischemic myocardial lesions, adiponectin accumulates in the heart tissue, where it passes from the vascular compartment. It has a longer half-life in heart tissue than in plasma . 2.18. Macrophage Migration Inhibitory Factor (MIF) MIF is an immunoregulatory cytokine considered a potential biomarker in numerous diseases that also have an inflammatory component . It can be produced by monocytes, macrophages, endocrine, epithelial, and endothelial cells, is stored cytoplasmically, and is rapidly released to stimuli such as microbial products, proliferative signals, and hypoxia . In the case of acute myocardial infarction, MIF increased in myocytes within a few hours, preceding macrophage infiltration in the infarcted area . Irreversible ischemic degenerative changes were observed as early as the first 20 min after total coronary obstruction, which allows for the identification of complement complexes at the myocardial level . The membrane attack complex (MAC) of the complement system is part of the C5b-9 membrane complex, an immunohistochemical marker that can be detected in renal , muscle , and myocardial necrotic biopsy samples. C5b-9 accumulates specifically at the level of necrotic myocardial cells and is associated with cell membrane damage by opening membrane pores . Its concentration depends on the diffusion rate of the complement substrate in the infarcted area, and therefore on reperfusion . Jasra et al., in a human necropsy casework study in which they compared myocardial infarctions more recent than 6 h and myocardial infarctions older than 6 h, concluded that ischemic and infarcted cardiac myofibrils show a positive C9 staining as early as the first 6 h . CD 59, also called protectin, inhibits the cytolytic activity of the complement system, and therefore myocardial loss and its plasma release occur prior to complement activation . From an immunohistochemical standpoint, Väkevä et al. in their studies found that CD59 expression exists at the level of the sarcolemal membrane of the normal myocardium, whereas in the infarcted region after 1–14 days, it is low or absent, accompanied by the concomitant deposition of MAC in CD59-negative areas. Small CD59 vesicles were found at the normal myocardium-infarcted myocardial boundary, suggesting its removal by a possible excretory mechanism . Myoglobin is a morpho-functionally comparable protein to hemoglobin; the difference between them lies in the number of polypeptide chains and the number of oxygen-binding sites. Myoglobin has a greater affinity for oxygen compared to hemoglobin and is very effective in extracting oxygen from the blood. It is found in skeletal muscle fiber sarcoplasm and cardiac myocyte cytoplasm . Also, in much lower concentrations, it can be found in smooth muscle, endothelial, and tumor cells, such as in breast, lung, and colon cancer, etc. , which makes this protein nonspecific. There are studies evaluating the morphological changes and myoglobin content in the normal myocardium versus the ischemic and necrotic myocardium. These studies found a loss of myoglobin from the necrotic myocardium . Considering the criterion for the biohumoral detection of myoglobin level, the concentration peaks 6 h from the moment of coronary obstruction; however, for the microscopic criterion, this time interval is lower, allowing an evaluation of necrotic myocardial tissue . Xiaohong et al. found subendocardial cellular myoglobin depletion in mice 30 min after coronary ligation, which expanded transmurally within 3 h . Amin et al., in their necropsy casework study in which they also compared the immunohistochemical expression of MB in cases with definite signs of ischemia, cases with probable signs of ischemia, and cases without signs of ischemia but with other causes of myocardial damage, found variable myoglobin depletion in all cases with definite signs of ischemia and in all cases with other causes of myocardial damage . Fibrinogen is a glycoprotein complex that is converted by thrombi to fibrin, and further to a fibrin-based blood clot during the coagulation process. Its increase is known to be a risk factor for AMI and stroke , and it was one of the earliest immunohistochemical markers evaluated for the diagnosis of AMI. Raza-Ahmad showed that, in the first 12 h postmortem, a positive reaction for fibrinogen was obtained not only on fibers demonstrating coagulative necrosis, contraction band necrosis, or a wavy pattern, but also in fibers suspected of acute ischemia in the absence of visible changes in H&E . Brinkman et al. performed a case–control study using four groups: AMI macroscopically present, occlusive coronary thrombosis without infarction, coronary atherosclerosis without AMI/occlusion, and a control group. The second group (representing coronary occlusion without gross signs of AMI) is representative of supra-acute MI. In this group, the authors found a patchy and moderate reaction, which was much weaker than the grossly visible AMI group . Xiaohong, in a study using an animal model for early myocardial ischemia, found that, after 30 min, fibrinogen staining was identified as light brown dots in subendocardial myocardial cells. One hour later, the immunostaining was identifiable in the middle myocardium, and after another two hours, in the subepicardial cells. After another three hours, the entire myocardial layer was positive . Sabatasso et al., also on an animal model, found the earliest positive reaction for fibrinogen within an hour of the coronary occlusion, only in the ischemic area, while the non-affected myocardium showed a negative reaction . The intensification of the positive reaction of myocardial fibrinogen together with myoglobin depletion denotes possible irreversible myocardial lesions detectable from the first 30 min of the moment of coronary blood flow arrest . However, the reliability of fibrinogen is considered to be low given the potential false-positive results from contaminated samples . Desmin is a protein specific to muscle tissue; it is structural to the heart muscle but also skeletal and smooth muscles. It plays an essential role in maintaining the structural and mechanical integrity of the contractile structure. For example, in rat studies, a deficiency in desmin synthesis was found to cause dilated cardiomyopathy, smooth muscle defects, and skeletal myopathy . In myocardial ischemia, the depletion of desmin from the cytoplasm of the heart cell begins 30 min after the onset of ischemia, and within 90–120 min, depletion is complete . TNF-α is a cytokine secreted by macrophages that can be detected mainly in vascular smooth muscle cells. Increased values have been reported in heart failure , but also in infarcted myocardial tissue, with TNF-α having the role of stimulating the proliferation and expression of fibronectin at the level of fibroblasts in the injured myocardium . P-38-activated kinase (MPAK) is interconnected with cardiac activity; its isoforms have a role in the growth and differentiation of cardiomyocytes . It is also closely related to oxidative stress, along with nuclear factor kappa B (NF-k-B), which plays a role in the production of TNF-α and JNK (stress-signaling kinases) . TGF-β1 is a profibrotic cytokine found in increased amounts in acute myocardial infarction. It also precedes increased collagen expression . cTnT, a marker of myocardial lesions, is a component of the troponin complex of the muscle cell. Serum levels of cTnT increase within 4–6 h of the onset of acute myocardial infarction and peak at about 24 h . Fishbein et. al, in an experimental immunohistochemical study in animals, found an uneven loss of cTnT alongside cTnI in the periphery of the necrotic myocardium, with the loss of cTnT being more pronounced compared to that of cTnI. In some cases, histopathological changes are detectable as early as the first 30 min after coronary occlusion . In the study by Amin et al. on human subjects, the results were similar . Diaz et al., in their necropsy casework study, found in cardiac deaths a more diffuse and pronounced immunohistochemical expression of cTnC compared to that of cTnT within the first hour after the onset of acute myocardial infarction . H-FABP (heart fatty acid binding protein) is a cardiac protein that plays an essential role in the metabolism of cardiomyocyte fatty acids . At the time of ischemic myocardial injury, H-FABP is released into the blood and eliminated renally. Biohumorally, it can be detected from about an hour from the onset of myocardial infarction , and immunohistopathologically it is considered objectifiable in the first 4 h after blocking the coronary flow. Dityrosine is a stable product synthesized by the myeloperoxidase–hydrogen peroxidase system of neutrophils and macrophages. This is a marker of oxidative protein stress . In the study on human necropsy cases conducted by Mayer et al., it was found that dityrosine, although not specific to infarcts, can also be a marker of myocardial infarction with a short survival time, being detectable immunohistochemically after the first 4 h of survival after the onset of infarction . Fibronectin is an extracellular matrix protein assembled into viscoelastic fibrils that can bind growth factors and cytokines. These fibrils play an essential role in injury healing processes, possessing unique mechanical properties that allow them to modify detected mechanotransduction signals transmitted by cells . In the experimental study conducted by Casscells et al. to observe the immunohistochemical aspects of fibronectin in acute myocardial infarction, it was found that fibronectin was irregularly located in the cytoplasm and in the interstitial space of myocytes of the area irrigated by the ligated coronary infarction . Saleki et al. performed a semiquantitative analysis of fibronectin immunostaining in three study groups—confirmed AMI, suspected AMI, and negative controls—and employed a four-stage grading for each case (negative to 3+). From 15 cases of confirmed AMI, a strong immunostaining pattern was identified in 10, both in the cytoplasm and nuclei of the affected fibers, in 4—a moderate pattern, and in 1—a weak staining. In the negative controls, 13 had a negative reaction, 4 had a weak reaction, and 1 a moderate reaction . This suggests that the sensibility of a strong immunostaining pattern is high for AMI, but moderate and weak patterns may be encountered in other pathologies as well, possibly caused by terminal ischemia associated with an increased agonal period. Hu et al. tried to evaluate the specificity of positive fibronectin immunostaining for the detection of AMI by quantifying the total positive area using the Interactive Building Analysis System. They found that, in AMI, the positive staining area was around 100 times higher compared to normal controls, hemorrhagic shock, cardiac contusion, and organophosphate poisoning, around 75 times higher compared to mechanical asphyxia and electrocution, and only around 2.5 times higher compared to myocarditis. Their study of fibronectin in necropsy cases additionally found that it is not affected by postmortem autolysis nor by prior formalin fixation, which also allows for the retrospective analysis of cases of forensic interest to detect early myocardial infarction postmortem . CD15, or Lewis x antigen, is a cell-surface glycan, considered a useful marker in identifying cells of the granulocyte series . Given the hypothesis that from the first hours from the onset of acute myocardial infarction, leukocytes are gradually attracted to the infarcted myocardial area, some studies have tried to estimate the age of acute myocardial infarction based on the number of CD15-positive marked cells . The number of CD-15 positive cells is one of the reliable immunohistochemical markers in dating myocardial infarction, as a linear relationship with time has been found regarding the accumulation of polymorphonuclear leukocytes at this level . For example, in the study conducted by Mortensen et al., it was estimated that several myocardial-positive CD15 cells between 20 and 30 raised the suspicion of an acute myocardial infarction, and values above 30 were considered relevant for an acute myocardial infarction occurring for 5–6 h . CD 18 is a cell-surface adhesion molecule that shows increased expression on the cell membrane in the infarcted area. It is associated with neutrophil activation. CD18 leukocyte integrins are adhesion receptors released in the acute phase and easily measured. They are mainly involved in neutrophil extravasation . Hill et al., in their experimental rabbit heart study, found an increase in CD18 expression as early as the first 20 min after infarction and correlated it with the extent of myocardial necrosis, observing, however, an interindividual variation that did not allow for percentage/numerical appreciations . ICAM 1 is an endothelial receptor and ligand for leukocyte integrin CD 18. It allows their firm adhesion to the vascular endothelium . There are studies that have demonstrated the increase in soluble ICAM 1 in the context of acute myocardial infarction , but also its overregulation in human cardiomyocytes after infarction. However, this is not detectable in infarcts more recent than one day . ILs are important inflammatory mediators of acute myocardial infarction, with a role in exacerbating (e.g., IL6) or attenuating (e.g., IL10, IL 15) the acute inflammatory response. They are released from various cells into the bloodstream and interstitium, where they bind to interleukin receptors on cell surfaces, causing cell activation. Some studies have also analyzed the prognosis of acute myocardial infarction according to the level of serum interleukins . MCP 1 is a chemokine that induces the recruitment and activation of monocytes, T cells, and NK cells . It is produced by many cells in response to injury factors or exposure to other cytokines, and is also involved in acute myocardial infarction . The antiapoptotic and chemotactic effects of MCP 1 may be mediated by different signaling pathways in the infarcted myocardial area . In the study conducted by Turillazzi et al., a strongly expressed response of MCP-1 was found in the early phase of acute myocardial infarction in the first 0–4 h . Tryptase is a protease released by mast cells. The increase in mast cell tryptase plays a central role in the immediate inflammatory and allergic reactions initiated by IgE . It induces fibroblast proliferation, stimulates fibroblast chemotaxis, and regulates the increased production of collagen type I . The same study mentioned above by Turillazzi et al. found a moderate immunohistochemical reaction for tryptase at myocardial infarction sites 0–6 h old, along with CD15, IL-1β, IL-6, TNF-α, IL-8, CD18, ICAM-1, IL-15, and MCP-1 . Adiponectin is an adipokine secreted by adipocytes, with anti-inflammatory, antifibrotic, antioxidant, and cardioprotective effects . Experimental studies on animals have proved that in ischemic myocardial lesions, adiponectin accumulates in the heart tissue, where it passes from the vascular compartment. It has a longer half-life in heart tissue than in plasma . MIF is an immunoregulatory cytokine considered a potential biomarker in numerous diseases that also have an inflammatory component . It can be produced by monocytes, macrophages, endocrine, epithelial, and endothelial cells, is stored cytoplasmically, and is rapidly released to stimuli such as microbial products, proliferative signals, and hypoxia . In the case of acute myocardial infarction, MIF increased in myocytes within a few hours, preceding macrophage infiltration in the infarcted area . Our review has exposed the main immunohistochemical markers considered over time for diagnosing myocardial infarction more recent than 6 h. They derive from knowledge of the pathophysiological mechanism, and appropriate laboratory elements are required for their determination, as well as complex and expensive forensic analysis . As far as laboratory methods are concerned, there is a need to simultaneously analyze as many markers of recent acute myocardial infarction as possible on the same tissue sample . However, particular aspects of each case must also be considered, such as cardiopulmonary resuscitation, catecholamine injection, drug use (ecstasy, cocaine), agonal artifacts, postmortem interval, autolysis, and pre-existing ischemic events, which can influence immunohistochemical results . 3.1. Agonal Myocardial Immunohistochemical Changes Myocardial contraction bands can be observed in multiple circumstances of death, as a result of artefactual causes such as poor sampling techniques (e.g., a low-temperature fixator). In experimental studies on ligated pig hearts, bands were found with abundance in the first 20–30 min . In the study conducted by Morita et al. on necropsy fragments of human heart, in order to differentiate pathological contraction bands from artefactual ones by immunohistochemical analysis, CCC9-positive reactions (complement C9 fraction) and SIRT 1-negative reactions were found in cases of myocarditis and myocardial ischemia, and in the rest of the cases, who benefited from cardiopulmonary resuscitation but had various causes of death, a positive reaction was found only in the case of SIRT 1, concluding that the positive CCC9 marker allows for the differentiation between myocardial changes in acute myocardial infarction and myocardial changes occurring after attempts at cardiopulmonary resuscitation or other terminal conditions . Depending on the immunohistochemical reaction obtained, they proposed the following classification of contraction bands . In addition, CCC9 is detectable from the early stage of acute myocardial infarction. Classical hematoxylin–eosin staining does not allow the detection of changes in this regard. Therefore, a myocardial-positive CCC9 immunohistochemical reaction is a reliable element supporting early acute myocardial infarction , while also allowing the differential diagnosis of agonal changes . Sabatosso et al., in their comparative immunohistochemical study of cardiac changes in myocardial ischemia, acute myocardial infarction, and hanging, using necropsy casework, found that myoglobin and TnT showed no expressions, supporting the statistically significant differences between the three groups. This, for example purposes, makes the specificity of immunohistochemical markers in humans inconclusive. Moreover, in the same study, a greater positive expression of Cx34 and Jun B markers was observed in cases of hanging than in cases of acute myocardial infarction or myocardial ischemia, demonstrating their non-specificity in humans, unlike experimental animal studies . 3.2. Myocardial Immunohistochemical Changes Following Autolysis and Putrefaction In general, immunohistochemical results can also be influenced by biological changes during a prolonged postmortem interval with subsequent autolysis and putrefactive phenomena, which, against the background of acidosis, determines the activation of intracellular enzymes with the degradation of protein structures . In this regard, the study by Thomsen and Held demonstrated that the marker C5b-9 can be detected in the myocardium at a postmortem interval of up to 11 days . Ortmann et al. also obtained a positive reaction for C5b-9 at postmortem intervals of about 8 weeks . Fibronectin and myoglobin could be detected immunohistologically at approximately 3 days postmortem at the myocardial level . However, these studies were not case-controlled and were performed on a small number of subjects, which may make it difficult to differentiate between false-positive results, against the background of protein degradation by autolysis and putrefaction, and positive results in a real pathological context following acute myocardial infarction . Hu et al., in their study in which they analyzed the expression of desmin, actin, and myoglobin on heart fragments, stored at 4C and analyzed at different postmortem intervals, found that the postmortem interval had a significant influence on the experience of the markers studied, with the markers being resistant to autolysis only for an interval of 2 days postmortem . 3.3. Perimortem Catecholamines High levels of catecholamines (either exogenous or endogenous, e.g., in pheochromocytoma or brainstem lesions affecting the nucleus tractus solitari ) have myotoxic action. In experimental animal studies, immunohistochemical myocytic apoptotic phenomena were observed 3–6 h after catecholamine injection, and necrotic phenomena were observed at about 18 h. In the experimental study on mice conducted by Goldspink et al., these phenomena were detected as heterogeneous myocardial, and were more pronounced in the left ventricular subendocardial with the predominance of necrosis phenomena and positive caspase 3 and myosin reactions . Another experimental study on rats by Lu et al. noted that an overdose of norepinephrine (NE) can cause severe cardiopulmonary dysfunction, with cardiac immunohistochemical positivity within the first 6 h of TnT and Cx43 at the biventricular and septal levels . 3.4. Gender and Myocardial Ischemic Preconditions A person’s gender and pre-existing ischemic conditions impact the severity and course of acute myocardial infarction. Some studies have concluded that the severity of apoptosis and myocardial cell necrosis is more pronounced in men . Even experimental studies in mice have detected smaller areas of acute myocardial infarction in females than in males . Pre-existing ischemic conditions refer to transient myocardial ischemic episodes in the subject’s pathological history and have been found to have a protective role for subsequent cardiac ischemic injury, resulting in infarction in a smaller area and a better preservation of left ventricular function . In the experimental case–control study conducted by Scholl et al. in mice, in which they analyzed the expression of dityrosine, TnT, TnI, and Connexin 43 (Cx43) according to the sex of the subjects and pre-existing myocardial ischemic conditions, it was concluded that only pre-existing ischemic conditions had a significant impact on the expression of troponins and Cx43, with sex having no discriminatory influence on the results obtained . 3.5. Overall Considerations Overall, the results of this review find their usefulness in forensic practice, systematizing the main immunohistochemical markers to be analyzed in the case of suspicion of an early acute myocardial infarction. After the first 6 h, these markers can be detectable for certain time intervals (see —Included studies) with the coexistence of the usual morphological changes in acute myocardial infarction, detectable by classical optical microscopy, which can help better characterize acute myocardial infarction. Many markers show significant changes dependent on the time interval, as has been extensively evaluated elsewhere in extensive, comparative analyses. The studies existing so far in the analyzed databases are mainly experimental and performed on non-human subjects, but have an important contribution to forensic practice in the instance of the suspicion of early acute myocardial infarction, undetected by classical microscopy techniques. In future, the simultaneous analysis of a panel of markers will be necessary to increase diagnostic accuracy, requiring appropriate laboratory infrastructure and reagents . Also, extensive comparative studies are needed to analyze the specificity of the presented immunohistochemical markers. The limitations of this review are represented by the absence of the specificity of these markers for early acute myocardial infarction, being rather markers of acute myocyte necrosis in general, regardless of etiology, and the small number of studies aimed at the immunohistochemical analysis of acute myocardial infarction more recent than 6 h. Another limit is represented by interspecies variability in marker analysis, with most markers being analyzed on non-human subjects. In experimental studies, human tissue fragments may show limited specificity. Consequently, this requires the survey of a larger number of tissue samples, the simultaneous analysis of several immunohistochemical markers, and the careful analysis of survey data regarding perimortem events, resuscitation, and postmortem intervals. Myocardial contraction bands can be observed in multiple circumstances of death, as a result of artefactual causes such as poor sampling techniques (e.g., a low-temperature fixator). In experimental studies on ligated pig hearts, bands were found with abundance in the first 20–30 min . In the study conducted by Morita et al. on necropsy fragments of human heart, in order to differentiate pathological contraction bands from artefactual ones by immunohistochemical analysis, CCC9-positive reactions (complement C9 fraction) and SIRT 1-negative reactions were found in cases of myocarditis and myocardial ischemia, and in the rest of the cases, who benefited from cardiopulmonary resuscitation but had various causes of death, a positive reaction was found only in the case of SIRT 1, concluding that the positive CCC9 marker allows for the differentiation between myocardial changes in acute myocardial infarction and myocardial changes occurring after attempts at cardiopulmonary resuscitation or other terminal conditions . Depending on the immunohistochemical reaction obtained, they proposed the following classification of contraction bands . In addition, CCC9 is detectable from the early stage of acute myocardial infarction. Classical hematoxylin–eosin staining does not allow the detection of changes in this regard. Therefore, a myocardial-positive CCC9 immunohistochemical reaction is a reliable element supporting early acute myocardial infarction , while also allowing the differential diagnosis of agonal changes . Sabatosso et al., in their comparative immunohistochemical study of cardiac changes in myocardial ischemia, acute myocardial infarction, and hanging, using necropsy casework, found that myoglobin and TnT showed no expressions, supporting the statistically significant differences between the three groups. This, for example purposes, makes the specificity of immunohistochemical markers in humans inconclusive. Moreover, in the same study, a greater positive expression of Cx34 and Jun B markers was observed in cases of hanging than in cases of acute myocardial infarction or myocardial ischemia, demonstrating their non-specificity in humans, unlike experimental animal studies . In general, immunohistochemical results can also be influenced by biological changes during a prolonged postmortem interval with subsequent autolysis and putrefactive phenomena, which, against the background of acidosis, determines the activation of intracellular enzymes with the degradation of protein structures . In this regard, the study by Thomsen and Held demonstrated that the marker C5b-9 can be detected in the myocardium at a postmortem interval of up to 11 days . Ortmann et al. also obtained a positive reaction for C5b-9 at postmortem intervals of about 8 weeks . Fibronectin and myoglobin could be detected immunohistologically at approximately 3 days postmortem at the myocardial level . However, these studies were not case-controlled and were performed on a small number of subjects, which may make it difficult to differentiate between false-positive results, against the background of protein degradation by autolysis and putrefaction, and positive results in a real pathological context following acute myocardial infarction . Hu et al., in their study in which they analyzed the expression of desmin, actin, and myoglobin on heart fragments, stored at 4C and analyzed at different postmortem intervals, found that the postmortem interval had a significant influence on the experience of the markers studied, with the markers being resistant to autolysis only for an interval of 2 days postmortem . High levels of catecholamines (either exogenous or endogenous, e.g., in pheochromocytoma or brainstem lesions affecting the nucleus tractus solitari ) have myotoxic action. In experimental animal studies, immunohistochemical myocytic apoptotic phenomena were observed 3–6 h after catecholamine injection, and necrotic phenomena were observed at about 18 h. In the experimental study on mice conducted by Goldspink et al., these phenomena were detected as heterogeneous myocardial, and were more pronounced in the left ventricular subendocardial with the predominance of necrosis phenomena and positive caspase 3 and myosin reactions . Another experimental study on rats by Lu et al. noted that an overdose of norepinephrine (NE) can cause severe cardiopulmonary dysfunction, with cardiac immunohistochemical positivity within the first 6 h of TnT and Cx43 at the biventricular and septal levels . A person’s gender and pre-existing ischemic conditions impact the severity and course of acute myocardial infarction. Some studies have concluded that the severity of apoptosis and myocardial cell necrosis is more pronounced in men . Even experimental studies in mice have detected smaller areas of acute myocardial infarction in females than in males . Pre-existing ischemic conditions refer to transient myocardial ischemic episodes in the subject’s pathological history and have been found to have a protective role for subsequent cardiac ischemic injury, resulting in infarction in a smaller area and a better preservation of left ventricular function . In the experimental case–control study conducted by Scholl et al. in mice, in which they analyzed the expression of dityrosine, TnT, TnI, and Connexin 43 (Cx43) according to the sex of the subjects and pre-existing myocardial ischemic conditions, it was concluded that only pre-existing ischemic conditions had a significant impact on the expression of troponins and Cx43, with sex having no discriminatory influence on the results obtained . Overall, the results of this review find their usefulness in forensic practice, systematizing the main immunohistochemical markers to be analyzed in the case of suspicion of an early acute myocardial infarction. After the first 6 h, these markers can be detectable for certain time intervals (see —Included studies) with the coexistence of the usual morphological changes in acute myocardial infarction, detectable by classical optical microscopy, which can help better characterize acute myocardial infarction. Many markers show significant changes dependent on the time interval, as has been extensively evaluated elsewhere in extensive, comparative analyses. The studies existing so far in the analyzed databases are mainly experimental and performed on non-human subjects, but have an important contribution to forensic practice in the instance of the suspicion of early acute myocardial infarction, undetected by classical microscopy techniques. In future, the simultaneous analysis of a panel of markers will be necessary to increase diagnostic accuracy, requiring appropriate laboratory infrastructure and reagents . Also, extensive comparative studies are needed to analyze the specificity of the presented immunohistochemical markers. The limitations of this review are represented by the absence of the specificity of these markers for early acute myocardial infarction, being rather markers of acute myocyte necrosis in general, regardless of etiology, and the small number of studies aimed at the immunohistochemical analysis of acute myocardial infarction more recent than 6 h. Another limit is represented by interspecies variability in marker analysis, with most markers being analyzed on non-human subjects. In experimental studies, human tissue fragments may show limited specificity. Consequently, this requires the survey of a larger number of tissue samples, the simultaneous analysis of several immunohistochemical markers, and the careful analysis of survey data regarding perimortem events, resuscitation, and postmortem intervals. We conducted a study in adherence to the PRISMA guidelines for reporting systematic literature reviews. In this regard, we searched the Web of Science and PubMed databases from inception to 2023, using the keywords “myocardial infarction” and “immunohistochemistry”. The inclusion criteria were observational studies that analyzed the effectiveness of immunohistochemical markers in the postmortem diagnosis of early acute myocardial infarction (the first 6 h after the occurrence) and exposed the laboratory techniques and materials used. We excluded reviews and meta-analyses. We also analyzed the reference lists of the studies included from the 2 databases to add potential supplementary studies to this literature review. The main analyzed elements in each study were as follows: the number of subjects, type of subjects (humans/non-humans), age of the subjects, documented pathologies of the subjects, symptom–death time interval, laboratory equipment and laboratory techniques used (see —Laboratory methodology in the included studies), and immunohistochemical marker(s) analyzed. The review is registered in the Open Science Framework: https://doi.org/10.17605/OSF.IO/2GWUT (accessed on 8 July 2024). Sudden cardiac death still poses a challenge to necropsy practice. Although the pathophysiological mechanisms of acute myocardial infarction have been carefully analyzed and unraveled through studies conducted on both human and animal subjects, we have a reduced range of feasible immunohistochemical markers useful in diagnosing early acute myocardial infarction in human subjects. Extensive studies on human necrotic tissue fragments are required with the simultaneous analysis of several immunohistochemical markers, including survey data for perimortem events, resuscitation, and postmortem intervals in the context of a uniform laboratory methodology. |
Exploring the Potential of Optical Genome Mapping in the Diagnosis and Prognosis of Soft Tissue and Bone Tumors | de505a0c-22d0-414e-ac37-b14a0380caee | 11942867 | Neoplasms[mh] | Sarcomas are a rare type of malignant neoplasm arising from mesenchymal tissues, affecting both soft tissue and bone [ , , , , ]. The existing literature reports a misdiagnosis rate of 20–30% , which significantly impacts treatment quality and patient prognosis. The difficulty in diagnosing these tumors originates from their low incidence and heterogeneity, as they comprise over 100 subtypes with overlapping histological characteristics, making their study and classification particularly challenging. From a cytogenetic point of view, they can be divided into two large groups: those with simple gene alterations and those with complex karyotypes with multiple structural variations (SVs) and copy number variations (CNVs) without defining alterations [ , , ]. However, this simplification in classification does not help to predict the clinical behavior, response to treatment, or prognosis of these tumors. This suggests that better characterization of cytogenetic alterations could have a major impact on patient stratification and prognosis. Concerning sarcoma diagnosis, a biopsy is required for histological, phenotypic, and molecular analysis, which may be performed using various techniques, including percutaneous core needle biopsy. The histological study includes the morphological analysis of the tissue, which allows the determination of the tumor grade based on the study of the parameters described by the ‘Fédération Nationale des Centres de Lutte Contre le Cancer’ (FNCLCC) . In addition, immunohistochemical studies help to determine tumor differentiation but not its grade or aggressiveness . Molecular studies are increasingly becoming the standard of care in sarcoma diagnosis. However, they are currently performed only when a specific histological diagnosis suggests a simple genetic alteration, such as a characteristic translocation or amplification, particularly in cases of diagnostic uncertainty, unusual clinicopathological presentation, or potential prognostic relevance . Fluorescence In Situ Hybridization (FISH) and Next-Generation Sequencing (NGS) panels are widely used molecular techniques with high sensitivity and specificity, but both are inherently targeted. FISH relies on specific probes for predefined genes, restricting its use to known alterations and limiting the ability to analyze all possible genetic changes in the tumor . Similarly, NGS panels, despite being fast and cost-effective tools for detecting point mutations, gene fusions, and small genetic abnormalities, require prior knowledge of fusion breakpoints [ , , ]. Additionally, conventional karyotyping and array comparative genomic hybridization (aCGH) lack diagnostic value . Optical genome mapping (OGM) is a high-resolution cytogenetic technique that uses ultra-high molecular weight DNA (UHMW DNA) (>150 Kbp) to detect SVs (500 bp-1 Mbp), CNVs (>5 Mbp), and complex rearrangements with a Variant Allele Frequency (VAF) higher than 5–10% and with 100–20,000 times more accuracy than traditional karyotyping . UHMW DNA molecules are enzymatically labeled by an enzyme that recognizes a six bp sequence (CTTAAG) present 14 to 17 times per 100 Kbp and then pass through nanochannels where images are taken. The images obtained are converted into digital molecules and assembled bioinformatically to compare the label pattern of the molecules with a reference genome . The overall result is a high-resolution genome-wide analysis that, in a single assay, equals or exceeds the diagnostic scope of multiple combined techniques currently used in clinical cytogenetics. To date, the use of OGM has been explored mainly in the area of oncohematological diseases, demonstrating that it helps in more accurate diagnosis, better risk stratification, and, thus, more precise treatment [ , , , ]. Recently, OGM has started to be used as a diagnostic tool for solid tumors . However, only one study has been described where OGM is performed in soft tissue and bone sarcomas, and it was able to detect alterations in 91% of the assessable samples which carried SVs or CNVs . Therefore, the application of OGM could improve the diagnostic classification of sarcomas, their prognosis, and the use or development of targeted therapies, as it could contribute to a better understanding of the molecular mechanisms of these neoplasms. This study aims to compare the findings from the analysis of 53 patients diagnosed with soft tissue or bone sarcomas using OGM, focusing on the expected alterations for each sarcoma subtype and the feasibility of using this technique in routine diagnosis.
2.1. General Data The age of the patients included in the analysis ranged from 18 to 93 years, with a mean of 59 years. Of the 53 samples included in this study, 33 were from male patients (62.3%) and 20 from female patients (37.7%) ( ). provides information regarding tumor groups and sarcoma subtypes included in this study. Regarding the histological nature of the 53 initial tumors, 21 were adipocytic (39.6%), 15 were non-adipocytic (28.3%), 6 were bone tumors (11.3%), 7 were of uncertain differentiation (13.2%), and 4 were small round cell sarcomas (7.6%). 2.2. Optical Genome Mapping Analysis 2.2.1. Performance of the Technique Of the 53 samples, 33 (62.3%) were evaluable for OGM analysis, with the technique’s effectiveness varying based on the type of starting tissue. High evaluability rates were observed in non-adipocytic samples (11/15, 73.3%), sarcomas of uncertain differentiation (5/7, 71.4%), and small round cell sarcomas (SRCSs) (3/4, 75.0%). In contrast, lower evaluability rates were found in adipose tissue samples (12/21, 57.1%) and bone samples (2/6, 33.3%). Additionally, non-adipocytic tumors with a high extracellular matrix and low cellularity, such as myxofibrosarcomas, also showed reduced evaluability (7/11, 63.6%) ( ). However, the underlying causes of these limitations varied by tissue type: in adipocytic and myxofibrosarcoma samples, the challenge was primarily related to the difficulty of extracting UHMW DNA, whereas, in bone samples, the main issue was obtaining an assessable OGM analysis. From now on, only the results of samples considered evaluable by OGM analysis according to the criteria set out in the section on materials and methods are reported. 2.2.2. Quality Parameters Regarding the quality parameters, 24 of the 33 (72.7%) samples showed good quality with respect to the established values of total DNA collected (>1500 Gbp), map rate (>70%), and coverage (>300×). The average DNA collected was 1435 Gbp (standard deviation: 518.0), average map rate was 72% (standard deviation: 14.2), and average coverage was 328× (standard deviation: 130.8). In addition, nine samples obtained values below the recommended thresholds for some parameters but were still evaluable. In addition, case 28, corresponding to a dedifferentiated liposarcoma (DDLPS), was loaded in duplicate to observe the reproducibility between runs of the OGM technique. The results show a high consistency for the OGM technique ( ). 2.2.3. Detection of Genetic Aberrations The analysis of the 33 cases by OGM resulted in an average of 408.4 alterations per sample (range: 22–2470) ( ). Expected alterations were identified in 97% (32/33) of the analyzed sarcomas: 11 with fusion genes, 10 with MDM2 amplifications, and 11 with non-specific SVs and CNVs. The RVA pipeline detected alterations in 31/33 (94%) tumors ( ). These were determined based on routine diagnostic techniques. Both morphology and immunohistochemistry provide indications that suggest the presence of specific alterations. Additionally, all the fusions and CNV described in this study have primarily been investigated using FISH and a custom RNA-based NGS panel (Agilent, Santa Clara, CA, USA). De Novo Analysis, which identifies SVs > 500 bp, detected additional alterations in one tumor, corresponding to a solitary fibrous tumor with a NAB2::STAT6 fusion ( ). These genes are contiguous on chromosome 12 (12q13) but have opposite transcription directions, requiring a small inversion of a few kilobases to form the fusion gene . The undetected case in OGM analysis was due to low percentage of neoplastic cells described in tenosynovial giant cell tumors ( ). shows the CircosPlots, all fusion genes, and the annotations of the alterations found in each case analyzed by OGM according to the standards of the International System for Human Cytogenomic Nomenclature 2024 (ISCN) . 2.2.4. Diagnosis Refinement OGM analysis allowed a diagnosis refinement in 3/33 patients (9.1%). Patient #20, initially diagnosed with a low-grade myofibroblastic lesion, was reclassified as nodular fasciitis. OGM analysis detected the t(6;17)(p21.31;p13.2) translocation, confirmed by FISH, resulting in the SRSF3::USP6 fusion gene, which is present in both aneurysmal bone cysts and nodular fasciitis ( a) . Patient #30 was initially diagnosed with a small round cell sarcoma with Ewing-like morphology. However, OGM and FISH analysis did not detect any fusion typical of Ewing sarcomas. Nevertheless, OGM detected multiple translocations, among them the t(5;22)(q31.2;q12.1) translocation resulting in the MN1::CXXC5 fusion gene. This translocation is typical of astroblastomas; cases of MN1::BEND2 (also typical of astroblastomas) and MN1::TAF3 fusions have begun to be described in soft tissue sarcomas . This tumor has been classified as a small round cell sarcoma, opening the door for future research to more precisely determine its classification. ( b). Patient #41, previously classified as a high-grade myxofibrosarcoma, was reclassified as a myxoinflammatory fibroblastic sarcoma, due to amplification of the VGLL3 , monosomy of chromosome 13, and the t(1;10)(p22;q24) translocation, which results in the OGA::TGFBR3 fusion gene, observed by OGM ( c) . 2.2.5. CNV Aberrations In addition to the defining alterations detected by OGM, amplifications in oncogenes and deletions in tumor suppressor genes of great importance in soft tissue and bone sarcomas were also detected ( and ). As expected, MDM2 was the most amplified oncogene (13 cases, 39.4%), followed by CDK4 , GLI2, and HMGA2 (11 cases each, 33.3%), with co-amplifications in 5 cases (15%). NTRK1 and RUNX2 were also amplified in 10 cases (30.3% each). Regarding tumor suppressor genes, the most deleted genes were CDKN2A (12 cases, 36.36%), CDKN2B (11 cases, 33.3%), TP53 (11 cases, 33.3%), and RB1 (10 cases, 30.3%). 2.2.6. Complex Cases Chromoanagenesis was detected in 17/33 patients (51.5%): 6 (35.3%) with chromoplexia, 10 (58.8%) with chromoplexia and chromothripsis, and 1 (5.9%) with chromothripsis. Furthermore, all dedifferentiated liposarcomas (DDLPSs), well-differentiated liposarcomas (WDLPSs), and atypical lipomatous tumors (ALTs) present chromothripsis in the long arm of chromosome 12, where amplified CDK4 , HMGA2, and MDM2 genes are localized.
The age of the patients included in the analysis ranged from 18 to 93 years, with a mean of 59 years. Of the 53 samples included in this study, 33 were from male patients (62.3%) and 20 from female patients (37.7%) ( ). provides information regarding tumor groups and sarcoma subtypes included in this study. Regarding the histological nature of the 53 initial tumors, 21 were adipocytic (39.6%), 15 were non-adipocytic (28.3%), 6 were bone tumors (11.3%), 7 were of uncertain differentiation (13.2%), and 4 were small round cell sarcomas (7.6%).
2.2.1. Performance of the Technique Of the 53 samples, 33 (62.3%) were evaluable for OGM analysis, with the technique’s effectiveness varying based on the type of starting tissue. High evaluability rates were observed in non-adipocytic samples (11/15, 73.3%), sarcomas of uncertain differentiation (5/7, 71.4%), and small round cell sarcomas (SRCSs) (3/4, 75.0%). In contrast, lower evaluability rates were found in adipose tissue samples (12/21, 57.1%) and bone samples (2/6, 33.3%). Additionally, non-adipocytic tumors with a high extracellular matrix and low cellularity, such as myxofibrosarcomas, also showed reduced evaluability (7/11, 63.6%) ( ). However, the underlying causes of these limitations varied by tissue type: in adipocytic and myxofibrosarcoma samples, the challenge was primarily related to the difficulty of extracting UHMW DNA, whereas, in bone samples, the main issue was obtaining an assessable OGM analysis. From now on, only the results of samples considered evaluable by OGM analysis according to the criteria set out in the section on materials and methods are reported. 2.2.2. Quality Parameters Regarding the quality parameters, 24 of the 33 (72.7%) samples showed good quality with respect to the established values of total DNA collected (>1500 Gbp), map rate (>70%), and coverage (>300×). The average DNA collected was 1435 Gbp (standard deviation: 518.0), average map rate was 72% (standard deviation: 14.2), and average coverage was 328× (standard deviation: 130.8). In addition, nine samples obtained values below the recommended thresholds for some parameters but were still evaluable. In addition, case 28, corresponding to a dedifferentiated liposarcoma (DDLPS), was loaded in duplicate to observe the reproducibility between runs of the OGM technique. The results show a high consistency for the OGM technique ( ). 2.2.3. Detection of Genetic Aberrations The analysis of the 33 cases by OGM resulted in an average of 408.4 alterations per sample (range: 22–2470) ( ). Expected alterations were identified in 97% (32/33) of the analyzed sarcomas: 11 with fusion genes, 10 with MDM2 amplifications, and 11 with non-specific SVs and CNVs. The RVA pipeline detected alterations in 31/33 (94%) tumors ( ). These were determined based on routine diagnostic techniques. Both morphology and immunohistochemistry provide indications that suggest the presence of specific alterations. Additionally, all the fusions and CNV described in this study have primarily been investigated using FISH and a custom RNA-based NGS panel (Agilent, Santa Clara, CA, USA). De Novo Analysis, which identifies SVs > 500 bp, detected additional alterations in one tumor, corresponding to a solitary fibrous tumor with a NAB2::STAT6 fusion ( ). These genes are contiguous on chromosome 12 (12q13) but have opposite transcription directions, requiring a small inversion of a few kilobases to form the fusion gene . The undetected case in OGM analysis was due to low percentage of neoplastic cells described in tenosynovial giant cell tumors ( ). shows the CircosPlots, all fusion genes, and the annotations of the alterations found in each case analyzed by OGM according to the standards of the International System for Human Cytogenomic Nomenclature 2024 (ISCN) . 2.2.4. Diagnosis Refinement OGM analysis allowed a diagnosis refinement in 3/33 patients (9.1%). Patient #20, initially diagnosed with a low-grade myofibroblastic lesion, was reclassified as nodular fasciitis. OGM analysis detected the t(6;17)(p21.31;p13.2) translocation, confirmed by FISH, resulting in the SRSF3::USP6 fusion gene, which is present in both aneurysmal bone cysts and nodular fasciitis ( a) . Patient #30 was initially diagnosed with a small round cell sarcoma with Ewing-like morphology. However, OGM and FISH analysis did not detect any fusion typical of Ewing sarcomas. Nevertheless, OGM detected multiple translocations, among them the t(5;22)(q31.2;q12.1) translocation resulting in the MN1::CXXC5 fusion gene. This translocation is typical of astroblastomas; cases of MN1::BEND2 (also typical of astroblastomas) and MN1::TAF3 fusions have begun to be described in soft tissue sarcomas . This tumor has been classified as a small round cell sarcoma, opening the door for future research to more precisely determine its classification. ( b). Patient #41, previously classified as a high-grade myxofibrosarcoma, was reclassified as a myxoinflammatory fibroblastic sarcoma, due to amplification of the VGLL3 , monosomy of chromosome 13, and the t(1;10)(p22;q24) translocation, which results in the OGA::TGFBR3 fusion gene, observed by OGM ( c) . 2.2.5. CNV Aberrations In addition to the defining alterations detected by OGM, amplifications in oncogenes and deletions in tumor suppressor genes of great importance in soft tissue and bone sarcomas were also detected ( and ). As expected, MDM2 was the most amplified oncogene (13 cases, 39.4%), followed by CDK4 , GLI2, and HMGA2 (11 cases each, 33.3%), with co-amplifications in 5 cases (15%). NTRK1 and RUNX2 were also amplified in 10 cases (30.3% each). Regarding tumor suppressor genes, the most deleted genes were CDKN2A (12 cases, 36.36%), CDKN2B (11 cases, 33.3%), TP53 (11 cases, 33.3%), and RB1 (10 cases, 30.3%). 2.2.6. Complex Cases Chromoanagenesis was detected in 17/33 patients (51.5%): 6 (35.3%) with chromoplexia, 10 (58.8%) with chromoplexia and chromothripsis, and 1 (5.9%) with chromothripsis. Furthermore, all dedifferentiated liposarcomas (DDLPSs), well-differentiated liposarcomas (WDLPSs), and atypical lipomatous tumors (ALTs) present chromothripsis in the long arm of chromosome 12, where amplified CDK4 , HMGA2, and MDM2 genes are localized.
Of the 53 samples, 33 (62.3%) were evaluable for OGM analysis, with the technique’s effectiveness varying based on the type of starting tissue. High evaluability rates were observed in non-adipocytic samples (11/15, 73.3%), sarcomas of uncertain differentiation (5/7, 71.4%), and small round cell sarcomas (SRCSs) (3/4, 75.0%). In contrast, lower evaluability rates were found in adipose tissue samples (12/21, 57.1%) and bone samples (2/6, 33.3%). Additionally, non-adipocytic tumors with a high extracellular matrix and low cellularity, such as myxofibrosarcomas, also showed reduced evaluability (7/11, 63.6%) ( ). However, the underlying causes of these limitations varied by tissue type: in adipocytic and myxofibrosarcoma samples, the challenge was primarily related to the difficulty of extracting UHMW DNA, whereas, in bone samples, the main issue was obtaining an assessable OGM analysis. From now on, only the results of samples considered evaluable by OGM analysis according to the criteria set out in the section on materials and methods are reported.
Regarding the quality parameters, 24 of the 33 (72.7%) samples showed good quality with respect to the established values of total DNA collected (>1500 Gbp), map rate (>70%), and coverage (>300×). The average DNA collected was 1435 Gbp (standard deviation: 518.0), average map rate was 72% (standard deviation: 14.2), and average coverage was 328× (standard deviation: 130.8). In addition, nine samples obtained values below the recommended thresholds for some parameters but were still evaluable. In addition, case 28, corresponding to a dedifferentiated liposarcoma (DDLPS), was loaded in duplicate to observe the reproducibility between runs of the OGM technique. The results show a high consistency for the OGM technique ( ).
The analysis of the 33 cases by OGM resulted in an average of 408.4 alterations per sample (range: 22–2470) ( ). Expected alterations were identified in 97% (32/33) of the analyzed sarcomas: 11 with fusion genes, 10 with MDM2 amplifications, and 11 with non-specific SVs and CNVs. The RVA pipeline detected alterations in 31/33 (94%) tumors ( ). These were determined based on routine diagnostic techniques. Both morphology and immunohistochemistry provide indications that suggest the presence of specific alterations. Additionally, all the fusions and CNV described in this study have primarily been investigated using FISH and a custom RNA-based NGS panel (Agilent, Santa Clara, CA, USA). De Novo Analysis, which identifies SVs > 500 bp, detected additional alterations in one tumor, corresponding to a solitary fibrous tumor with a NAB2::STAT6 fusion ( ). These genes are contiguous on chromosome 12 (12q13) but have opposite transcription directions, requiring a small inversion of a few kilobases to form the fusion gene . The undetected case in OGM analysis was due to low percentage of neoplastic cells described in tenosynovial giant cell tumors ( ). shows the CircosPlots, all fusion genes, and the annotations of the alterations found in each case analyzed by OGM according to the standards of the International System for Human Cytogenomic Nomenclature 2024 (ISCN) .
OGM analysis allowed a diagnosis refinement in 3/33 patients (9.1%). Patient #20, initially diagnosed with a low-grade myofibroblastic lesion, was reclassified as nodular fasciitis. OGM analysis detected the t(6;17)(p21.31;p13.2) translocation, confirmed by FISH, resulting in the SRSF3::USP6 fusion gene, which is present in both aneurysmal bone cysts and nodular fasciitis ( a) . Patient #30 was initially diagnosed with a small round cell sarcoma with Ewing-like morphology. However, OGM and FISH analysis did not detect any fusion typical of Ewing sarcomas. Nevertheless, OGM detected multiple translocations, among them the t(5;22)(q31.2;q12.1) translocation resulting in the MN1::CXXC5 fusion gene. This translocation is typical of astroblastomas; cases of MN1::BEND2 (also typical of astroblastomas) and MN1::TAF3 fusions have begun to be described in soft tissue sarcomas . This tumor has been classified as a small round cell sarcoma, opening the door for future research to more precisely determine its classification. ( b). Patient #41, previously classified as a high-grade myxofibrosarcoma, was reclassified as a myxoinflammatory fibroblastic sarcoma, due to amplification of the VGLL3 , monosomy of chromosome 13, and the t(1;10)(p22;q24) translocation, which results in the OGA::TGFBR3 fusion gene, observed by OGM ( c) .
In addition to the defining alterations detected by OGM, amplifications in oncogenes and deletions in tumor suppressor genes of great importance in soft tissue and bone sarcomas were also detected ( and ). As expected, MDM2 was the most amplified oncogene (13 cases, 39.4%), followed by CDK4 , GLI2, and HMGA2 (11 cases each, 33.3%), with co-amplifications in 5 cases (15%). NTRK1 and RUNX2 were also amplified in 10 cases (30.3% each). Regarding tumor suppressor genes, the most deleted genes were CDKN2A (12 cases, 36.36%), CDKN2B (11 cases, 33.3%), TP53 (11 cases, 33.3%), and RB1 (10 cases, 30.3%).
Chromoanagenesis was detected in 17/33 patients (51.5%): 6 (35.3%) with chromoplexia, 10 (58.8%) with chromoplexia and chromothripsis, and 1 (5.9%) with chromothripsis. Furthermore, all dedifferentiated liposarcomas (DDLPSs), well-differentiated liposarcomas (WDLPSs), and atypical lipomatous tumors (ALTs) present chromothripsis in the long arm of chromosome 12, where amplified CDK4 , HMGA2, and MDM2 genes are localized.
This study explores the use of OGM as an advanced tool to diagnose bone and soft tissue sarcomas, seeking to overcome the limitations of current techniques and adding information to predict behavior. The principal limitation of this technique was the UHMW DNA extraction in adipocytic samples and myxofibrosarcomas, with a successful rate of around 65%. Challenges in extracting DNA in myxofibrosarcoma arise due to the low cellularity and myxoid matrix, while, in adipocytic tumors, it is due to the low nucleus-to-cytoplasm ratio and the presence of large lipid-rich cytoplasm in WDLPS, ALT, and lipomas. Consequently, using small tissue samples as the starting material makes UHMW DNA extraction particularly difficult. This issue can be mitigated by prioritizing tissue from surgical resections in lipomatous and myxofibrosarcoma tumors whenever available. Regarding DDLPSs, these tumors are composed of two morphologically distinct regions: WDLPS-like or ALT-like regions and dedifferentiated denser regions with a higher cell count where cytogenetic alterations responsible for the tumor’s increased aggressiveness are found. For optimal UHMW DNA extraction and tumor analysis, core biopsies must be performed in the latter region. On the other hand, patients with malignant tenosynovial giant cell tumors, defined by COL6A3::CSF1 fusion and low percentage neoplastic cells, are better diagnosed by target PCR instead of OGM . OGM identified defining alterations in 95% of tumors and, additionally, non-defining SVs and CNVs in 36%. Furthermore, it is important to highlight that the RVA pipeline was enough to find the expected alterations in 94% of tumors. Furthermore, in the OGM analysis workflow, we recommend first using the RVA pipeline, as it was enough to find the expected alterations in 94% of tumors. If the expected alterations are not identified, a De Novo Analysis is recommended, as it can detect smaller alterations. OGM enhances the understanding of characteristic genomic alterations in soft tissue and bone sarcomas. In this context, our study highlights the utility of OGM, revealing amplifications in oncogenes— MDM2 (39%), CDK4 (33%), HMGA2 (33%), and GLI2 (33%)—as well as deletions in tumor suppressor genes— CDKN2A (36%), CDKN2B (33%), TP53 (33%), and RB1 (30%). Regarding oncogenes, amplification of MDM2 and CDK4 is commonly observed in WDLPS, DDLPS, ALT, dedifferentiated parosteal osteosarcoma, and osteosarcoma NOS . Previous studies have demonstrated that higher amplification levels of these genes are associated with poorer prognosis in DDLPS patients [ , , , ]. Additionally, for patients with DDLPS, targeted therapies inhibiting MDM2 , such as milademetan and brigimadlin, are currently in phase 3 clinical trials (NCT04979442 and NCT06058793), while CDK4/6 inhibitors, such as palbociclib, are already used in clinical practice . However, in osteosarcomas, the clinical significance of these amplifications remains poorly understood. Otherwise, amplification of HMGA2 has been reported in adipocytic tumors associated with a favorable prognosis, as it may contribute to tumor differentiation . In fact, it can act as a prognostic marker in DDLPS, where the MDM2/HMGA2 amplification ratio correlates with prognosis—lower ratio, better prognosis . In addition, GLI2 amplifications have been associated with osteosarcoma development and metastasis, making it a potential target for therapy in these tumors . However, its role in WDLPS, DDLPS, and ALT is still unknown and could be interesting. Referring to tumor suppressor genes, loss of CDKN2A/B has been reported to be associated with poor prognosis in soft tissue sarcomas . More specifically, it has been described in Ewing sarcomas, osteosarcomas, and myxofibrosarcomas [ , , ], being a potentially actionable gene for targeted therapies . In addition, deletions, SVs, and LOH regions of TP53 and RB1 have also been identified in several sarcoma subtypes, particularly in leiomyosarcomas, myxofibrosarcomas, and undifferentiated pleomorphic sarcomas, where they play a significant role in tumor oncogenesis and progression . Given the high percentage of patients with TP53 losses (33%) detected by OGM, an extensive study is recommended to discard a possible Li–Fraumeni syndrome in young patients. However, despite their relevance, the impact of these tumor suppressor genes on patient prognosis and clinical management remains insufficiently explored. Moreover, this study demonstrates that OGM provides a deeper understanding of the genomic complexity of soft tissue and bone sarcomas, detecting complex karyotypes with chromoanagenesis in 51.5% of patients. Regarding chromoplexy, rearrangement loops in fusion genes responsible for Ewing sarcomas have been reported as common in tumors with more aggressive behavior ; however, no cases have been detected in our cohort. Referring to chromothripsis, it is associated with more aggressive tumor behavior and poor prognosis in cancer patients [ , , , , ]. In fact, Mandahl et al. described chromothripsis-driven amplifications in the 12p regions, as well as in the 5p and 20q regions, in WDLPS and DDLPS, being more frequent in the latter, suggesting that it contributes to tumor aggressiveness. This aligns with our observations, as all patients with these tumors in our study presented chromothripsis in chromosome 12p and other chromosomes. Overall, further study of these complex events could lead to improved patient prognosis and even possible targeted therapies. In addition, regarding osteosarcomas, a novel mechanism known as Loss-Translocation-Amplification (LTA) chromothripsis was recently described by Espejo Valle-Inclán et al. . This process involves the loss of TP53 , followed by translocation and amplification of various oncogenes, and appears to drive increased intratumoral heterogeneity and tumor clonal evolution . Moreover, LOH regions have been identified as prognostic biomarkers for survival in osteosarcoma patients . Although neither of these phenomena was observed in the two high-grade osteosarcoma cases in our study, both can be reliably detected using the OGM technique, enabling future research in this field. In terms of time, OGM can generally be completed within 5–7 days, depending on sample quality and processing conditions, whereas NGS typically takes 7–14 days, especially when whole-exome or whole-genome sequencing is performed. Although targeted NGS panels provide faster results for known mutations, OGM offers a comprehensive, untargeted approach that enables the detection of a wider range of structural variations. This can be particularly valuable in rare cases or tumors with complex karyotypes, where traditional sequencing methods may fail to identify crucial alterations. Despite its advantages, the OGM technique does come with certain limitations. One significant challenge is that formalin-fixed paraffin-embedded (FFPE) tissues often result in DNA that is unsuitable for this method. To facilitate the routine use of OGM in clinical laboratories, workflows would need to be adjusted to include the processing of fresh or appropriately frozen tissue. This necessitates that samples be frozen using liquid nitrogen and stored at −80 °C, which can present logistical challenges in many medical centers. Nevertheless, we strongly recommend the freezing of tumor biopsies for OGM studies, as this approach could substantially enhance prognostic accuracy and improve patient management. Moreover, an intrinsic limitation of OGM is its inability to detect point mutations, important in some sarcomas such as gastrointestinal stromal tumors (GISTs) or desmoid tumors . To improve diagnosis, we propose the use of OGM together with NGS gene panels, achieving a more complete approach and better characterization of the patient’s tumor at the time of diagnosis.
4.1. Sample Processing and Selection We selected biobanked samples from 53 adult patients, aged between 18 and 93 years old, diagnosed with soft tissue or bone sarcoma between 2022 and 2024 at our institution, a reference center, Hospital Universitari i Politècnic La Fe; which samples were preserved at the Biobanco La Fe. The samples were sent fresh to the pathology lab and obtained by two different methods: core needle diagnostic biopsies percutaneously performed and surgical resections. Both types of samples were sent in an interval of no more than 30 min to prevent cold ischemia and DNA degradation. Hematoxylin and eosin (HE)-stained slides were prepared and reviewed by an expert pathologist to ensure the presence of at least 10% tumor cells in the tissue. The samples were weighed, frozen by immersion in liquid nitrogen for 3 min and stored in aliquots of 2 mL eppendorfs at −80 °C at Biobanco La Fe. This study was approved by the Clinical Research Ethics Committee of the Hospital Universitari i Politècnic La Fe (No. 2023-984-1), and all patients signed an informed consent form in accordance with the recommendations of the Declaration of Human Rights, the Helsinki Conference, and institutional regulations. 4.2. Ultra-High Molecular Weight DNA Extraction and Labeling For the OGM analysis, very long DNA molecules (>150 Kbp) are required. For this purpose, the Bionano Prep SP Tissue and Tumor DNA Isolation Extraction Kit was used according to the manufacturer’s instructions (Bionano Genomics, San Diego, CA, USA) , which allows the extraction of DNA molecules with a low degree of fragmentation due to the use of paramagnetic nanodiscs. After elution, UHMW DNA was homogenized and quantified using the Qubit Broad Range dsDNA Assay Kit, to ensure DNA concentration was between 50 and 150 ng/µL, as recommended by Bionano Genomics . Subsequently, UHMW DNA labeling was performed using the Bionano Prep Direct Label and Stain Kit (Bionano Genomics) , which employs an enzymatic fluorescent labeling method targeting the CTTAAG motif with DLE-1 enzyme. This method does not introduce nicks in the gDNA, allowing the generation of long, contiguous genome maps (20–100 Mbp). Final quantification was performed using the Qubit High Sensitivity dsDNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) to ensure the DNA concentration was between 4 and 12 ng/µL, as recommended by Bionano Genomics. 4.3. Sample Loading on Chip, Reading by Saphyr, and Data Analysis Samples labeled with concentrations between 4 and 12 ng/µL were loaded onto the Saphyr 3.3 chip (Bionano Genomics) and inserted into the Saphyr ® (Bionano Genomics) for imaging. UHMW DNA molecules were moved via electrophoresis, linearized, and imaged by a high-resolution camera. The images were converted into digital representations and compared to the fluorescently labeled human reference genome GRCh38 . Results were obtained with quality parameters such as DNA collected (>150 Kbp), number of analyzed UHMW DNA strands (>150 Kbp and >20 Kbp), label density, mapping rate, genomic coverage, and positive (PLV) and negative (NLV) label variances. Aiming for 1500 Gbp of DNA, 300× coverage, and >70% mapping rate , samples were considered evaluable if the map rate was 40–50% with >150× coverage, or 50–70% with >100× coverage, based on our own experience analyzing solid tumors. Data analysis was conducted using Bionano Access ® software (version 1.8.2), which consists of two bioinformatics analysis pipelines that were applied to all samples. First, Rare Variant Analysis (RVA), detects low VAF alterations, including SVs and CNVs larger than 5 Mbp by aligning DNA molecules >150 Kbp to the human reference genome and generating consensus maps. Secondly, samples were reanalyzed with the De Novo Analysis pipeline, designed to identify smaller SVs (<500 bp), differentiate homozygous from heterozygous alterations, and detect loss of heterozygosity (LOH) regions. This pipeline aligns molecules >200 Kbp and generates refined consensus maps for each allele . Finally, a BED file containing 167 sarcoma-related genes was used for prioritization, focusing on SVs > 500 bp and CNVs > 500 bp. 4.4. Highly Complex Genomic Cases Analysis High genomic complexity cases involve complex karyotypes with multiple inter- and intrachromosomal rearrangements, known as chromoanagenesis. First described in 2012 by Holland and Cleveland, chromoanagenesis refers to a catastrophic event involving multiple complex rearrangements in one or more chromosomal regions, encompassing processes like chromoplexy and chromothripsis . Chromoplexy: In the present study, chromoplexies are defined as chained multichromosomal rearrangements (≥3 chromosomes) with breakpoints less than five tags apart, unless deletion bridges were present [ , , ]. Chromothripsis: In this study, chromothripsis is determined as regions with copy number variations coexisting with more than seven chromosomal rearrangements in 50 Mb; furthermore, the rearrangements must be interspersed and uniformed in the affected region [ , , , ]. All chromoanagenesis events were manually inspected for validation.
We selected biobanked samples from 53 adult patients, aged between 18 and 93 years old, diagnosed with soft tissue or bone sarcoma between 2022 and 2024 at our institution, a reference center, Hospital Universitari i Politècnic La Fe; which samples were preserved at the Biobanco La Fe. The samples were sent fresh to the pathology lab and obtained by two different methods: core needle diagnostic biopsies percutaneously performed and surgical resections. Both types of samples were sent in an interval of no more than 30 min to prevent cold ischemia and DNA degradation. Hematoxylin and eosin (HE)-stained slides were prepared and reviewed by an expert pathologist to ensure the presence of at least 10% tumor cells in the tissue. The samples were weighed, frozen by immersion in liquid nitrogen for 3 min and stored in aliquots of 2 mL eppendorfs at −80 °C at Biobanco La Fe. This study was approved by the Clinical Research Ethics Committee of the Hospital Universitari i Politècnic La Fe (No. 2023-984-1), and all patients signed an informed consent form in accordance with the recommendations of the Declaration of Human Rights, the Helsinki Conference, and institutional regulations.
For the OGM analysis, very long DNA molecules (>150 Kbp) are required. For this purpose, the Bionano Prep SP Tissue and Tumor DNA Isolation Extraction Kit was used according to the manufacturer’s instructions (Bionano Genomics, San Diego, CA, USA) , which allows the extraction of DNA molecules with a low degree of fragmentation due to the use of paramagnetic nanodiscs. After elution, UHMW DNA was homogenized and quantified using the Qubit Broad Range dsDNA Assay Kit, to ensure DNA concentration was between 50 and 150 ng/µL, as recommended by Bionano Genomics . Subsequently, UHMW DNA labeling was performed using the Bionano Prep Direct Label and Stain Kit (Bionano Genomics) , which employs an enzymatic fluorescent labeling method targeting the CTTAAG motif with DLE-1 enzyme. This method does not introduce nicks in the gDNA, allowing the generation of long, contiguous genome maps (20–100 Mbp). Final quantification was performed using the Qubit High Sensitivity dsDNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) to ensure the DNA concentration was between 4 and 12 ng/µL, as recommended by Bionano Genomics.
Samples labeled with concentrations between 4 and 12 ng/µL were loaded onto the Saphyr 3.3 chip (Bionano Genomics) and inserted into the Saphyr ® (Bionano Genomics) for imaging. UHMW DNA molecules were moved via electrophoresis, linearized, and imaged by a high-resolution camera. The images were converted into digital representations and compared to the fluorescently labeled human reference genome GRCh38 . Results were obtained with quality parameters such as DNA collected (>150 Kbp), number of analyzed UHMW DNA strands (>150 Kbp and >20 Kbp), label density, mapping rate, genomic coverage, and positive (PLV) and negative (NLV) label variances. Aiming for 1500 Gbp of DNA, 300× coverage, and >70% mapping rate , samples were considered evaluable if the map rate was 40–50% with >150× coverage, or 50–70% with >100× coverage, based on our own experience analyzing solid tumors. Data analysis was conducted using Bionano Access ® software (version 1.8.2), which consists of two bioinformatics analysis pipelines that were applied to all samples. First, Rare Variant Analysis (RVA), detects low VAF alterations, including SVs and CNVs larger than 5 Mbp by aligning DNA molecules >150 Kbp to the human reference genome and generating consensus maps. Secondly, samples were reanalyzed with the De Novo Analysis pipeline, designed to identify smaller SVs (<500 bp), differentiate homozygous from heterozygous alterations, and detect loss of heterozygosity (LOH) regions. This pipeline aligns molecules >200 Kbp and generates refined consensus maps for each allele . Finally, a BED file containing 167 sarcoma-related genes was used for prioritization, focusing on SVs > 500 bp and CNVs > 500 bp.
High genomic complexity cases involve complex karyotypes with multiple inter- and intrachromosomal rearrangements, known as chromoanagenesis. First described in 2012 by Holland and Cleveland, chromoanagenesis refers to a catastrophic event involving multiple complex rearrangements in one or more chromosomal regions, encompassing processes like chromoplexy and chromothripsis . Chromoplexy: In the present study, chromoplexies are defined as chained multichromosomal rearrangements (≥3 chromosomes) with breakpoints less than five tags apart, unless deletion bridges were present [ , , ]. Chromothripsis: In this study, chromothripsis is determined as regions with copy number variations coexisting with more than seven chromosomal rearrangements in 50 Mb; furthermore, the rearrangements must be interspersed and uniformed in the affected region [ , , , ]. All chromoanagenesis events were manually inspected for validation.
Briefly, this study confirms the high utility of OGM as a diagnostic tool for soft tissue and bone sarcomas, identifying fusion genes, chromothripsis, chromoplexies, SVs, and CNVs, many of them affecting oncogenes and tumor suppressor genes. These findings open the door for future research focused on better understanding the mechanisms of sarcomas and developing targeted treatments.
|
Rhizosphere competent inoculants modulate the apple root–associated microbiome and plant phytoalexins | f438a8b0-0efb-4408-a38a-527786770e3d | 11129989 | Microbiology[mh] | Apple replant disease (ARD) is a phenomenon occurring in tree nurseries and apple orchards worldwide. Replanting apples on the same soil as previous apple cultures leads to severe disease symptoms, resulting in reduced plant growth and yield losses (Mazzola and Manici ; Winkelmann et al. ; Somera and Mazzola ). Despite decades of research trying to elucidate the etiology of ARD, the disease is still not fully understood. ARD is hypothesized to be caused by a dysbiosis of microorganisms in the soil that was previously grown with apple plants (Winkelmann et al. ). Typical disease symptoms are root blackening, reduced root branching, and root infections with plant pathogens (Caruso et al. ; Grunewaldt-Stöcker et al. , ). The symptoms are associated with a strong plant stress response to ARD-affected soil in which plant defense molecules like phenolic compounds, especially phytoalexins, are accumulated in the roots and differentially exuded to the soil (Henfrey et al. ; Weiß et al. ; Busnena et al. ; Reim et al. ). Several studies revealed that the rhizosphere microbial community composition of apple plants grown in ARD-affected soils is distinct from that found in plants grown in soil from the same site but with no history of apple cultivation (Sun et al. ; Franke-Whittle et al. ; Yim et al. ; Tilston et al. ; Balbín-Suárez et al. , ). For the root endosphere, species belonging to the genera Streptomyces , Ilyonectria , Thelonectria , Rhizoctonia , or Pythium were often reported in higher densities in ARD-affected roots and therefore considered to contribute to the ARD causing complex (Manici et al. , ; Popp et al. ; Mahnkopp-Dirks et al. ). However, the abundance, diversity, and even presence or absence of taxa related to ARD vary highly between regions and orchards (Mazzola et al. ; Tewoldemedhin et al. ; Manici et al. ). The predominant measure to counteract ARD remains chemical soil fumigation (Mai and Abawi ; Willett et al. ; Yim et al. ). However, due to their toxicity and environmental harm, the application of these chemicals was prohibited in most European countries (Ruzo ; Porter et al. ; Prashar and Shah ). In Germany, the use of the pesticide Basamid® granular, mainly used for chemical mitigation of ARD, will be prohibited from June 2024 (Federal Office of Consumer Protection and Food Safety ). Thus, there is an urgent need to develop sustainable alternative treatment options to counteract ARD. Over the past years, alternatives to soil fumigation like spatial reorganization of planting in the orchards, biofumigation, or breeding of tolerant rootstocks have been evaluated (Leinfelder and Merwin ; Rumberger et al. ; Yim et al. , ). Surprisingly, reports on the effects of microbial inoculants on the rhizosphere microbiome and the response of the plant in ARD-affected soils are rare (Somera and Mazzola ). Only a few studies investigated the effect of beneficial bacteria, mainly members of the genus Bacillus , with regard to ARD, focusing on plant growth but typically not considering effects on the microbiome (Utkhede et al. ; Karlidag et al. ; Duan et al. , , ). Currently, reports on effects of bacterial inoculants on the microbial community composition and microbiome modulation in apple are missing. Also, the molecular plant response to inoculation has not been investigated before. Isolates belonging to Bacillus spp. have several traits that make them promising candidates for biostimulation. The Gram-positive bacterium is a spore-former allowing survival under extreme conditions (Santoyo et al. ; Shafi et al. ). Bacillus strains are easy to ferment and harvest as spores at a large-scale, enabling their production as a biocontrol product (Fira et al. ). Bacillus velezensis FZB42 (formerly classified as Bacillus amyloliquefaciens ) is considered “the Gram-positive model strain for plant growth promotion and biocontrol” (Fan et al. ). This strain and its potential for plant growth promotion was studied for many years, and the genome sequence is available (Chen et al. ). Also, Pseudomonas commonly occur in the rhizosphere and characteristics like the ability to colonize and proliferate in the rhizosphere, the effective use of root exudates, and the potential to suppress soil-borne pathogens make also isolates belonging to Pseudomonas spp. promising candidates for biostimulation (Weller et al. ; Preston ; Weller ; Santoyo et al. ). The strain Pseudomonas sp. RU47 (previously Pseudomonas jessenii RU47) was isolated from a disease suppressive soil (Adesina et al. ) and the analyses of its genome revealed several genes coding for plant beneficial traits (Kuzmanović et al. ). Previous studies using FZB42 or RU47 mainly focused on a variety of annual plants, including tomato, cucumber, lettuce, cotton, or tobacco (Grosch et al. ; Yao et al. ; Gül et al. ; Adesina et al. ; Wang et al. ; Chowdhury et al. ; Windisch et al. ; Schreiter et al. ; Eltlbany et al. ). PGPR traits of FZB42, like the secretion of secondary metabolites and production of hydrolytic enzymes, the stimulation of induced systemic resistance and positive effects on the microbiome, have been reviewed extensively (Chowdhury et al. ; Fan et al. ; Amaresan et al. ). The potential of RU47 to establish in the rhizosphere and to enhance plant growth or suppress plant pathogens was shown for potato, tomato, or lettuce (Adesina et al. ; Schreiter et al. ; Eltlbany et al. ). The strain’s capability to solubilize phosphate and to produce indole-3-acetic acid, siderophores, HCN, and protease was confirmed by in vitro testing (Adesina et al. ; Kuzmanović et al. ). Successful plant growth promotion or biostimulation require the establishment of the inoculants in the rhizosphere of the targeted host plant. Recently, Behr et al. showed in field trials that RU47 was capable to survive in the rhizosphere of winter rye over one winter period. Berg et al. stated that the effective colonization of inoculants in situ is one of the essential steps for successful interaction with the plant and the native microbiome. Although both strains, FZB42 and RU47, are well characterized and were successfully used as inoculants on several cultivated plants, to our knowledge, their interplay with apple plants and their native microbiome has not been investigated before. In this study, our objective was to elucidate the colonization potential of strains FZB42 and RU47 in root-affected soil (RA) and on the rhizoplane (RP) of apple plants. Additionally, we sought to understand their impact on the microbiome within the RA, RP, and the root endosphere (RE). We hypothesized that the plant response to the inoculation depends on the inoculants’ rhizosphere competence and the degree of microbiome modulation. Therefore, we determined the inoculants’ colony forming units (CFU) counts 3, 16, or 28 days post inoculation (dpi) of young apple plants grown under greenhouse conditions in soil from the same site affected by ARD or not (grass soils). We analyzed the microbiome in the different microhabitats RA, RP, and RE by a DNA-based meta-barcoding approach. Plant response to the inoculants was analyzed by investigating root phytoalexin content and morphology.
Soils Topsoils (0–20 cm) were collected from a site in Ellerhoop, Germany (53° 42′ 51.71″ N, 9° 46′ 12.16″ E), in May 2020. Soil from this site was classified as Endostagnic Luvisol (FAO and ITPS ). Details on soil texture and abiotic soil properties were reported previously (Mahnkopp et al. ). The site was established in 2009 with ARD plots ( n = 4) by replanting apple rootstocks ‘Bittenfelder Sämling’ every other year (ARD soil) and grass plots ( n = 4) with no apple cultivation history (Mahnkopp et al. ). Soils from both variants were taken as pooled samples from all plots and homogenized by sieving through a 2-mm mesh. For the greenhouse experiment, both soils were mixed with 50% (v/v) sterilized sand and fertilized with 2 g L -1 Osmocote exact 3–4 M (16% N + 9% P 2 O 5 + 12% K 2 O + 2% MgO, ICL Deutschland, Nordhorn, Germany). These mixtures will further be referred to as ARD and grass soils. Bacterial inoculant strains The rifampicin-resistant strain Bacillus velezensis FZB42 (DSMZ, Braunschweig, Germany, No.: DSM23117) was provided as a ready-to-use spore suspension of 6.7 × 10 9 CFU mL −1 and derived from the commercial product Rhizovital (ABiTEP GmbH, Berlin, Germany). The cultivation was performed on Reasoners’ 2 agar (R2A, Merck Millipore, Burlington, MA, USA) supplemented with rifampicin (75 µg mL −1 ) and cycloheximide (100 µg mL −1 ) (thereafter called medium “MB”). Strain Pseudomonas sp. RU47 (DSMZ, No.: DSM117411) was obtained from our laboratory strain collection and was resistant to rifampicin, tetracycline, chloramphenicol, and ampicillin. Initial cultivation of the strain was carried out on King’s B agar (KB, Carl Roth, Karlsruhe, Germany) supplemented with rifampicin (75 µg mL −1 ), ampicillin (100 µg mL −1 ), chloramphenicol (30 µg mL −1 ), tetracycline (10 µg mL −1 ), and cycloheximide (100 µg mL −1 ) (thereafter called medium “MP”). Agar plates for both strains were incubated at 28 °C for 48 h until single colonies were observed. Overnight cultures of RU47 were grown in Luria Bertani broth (LB, Carl Roth, Karlsruhe, Germany) supplemented with the aforementioned antibiotics at 28 °C and 150 rpm on a shaker. Plant material and greenhouse experiment Plant material of the ARD-susceptible rootstock genotype M26 was propagated in vitro as described earlier (Rohr et al. ). In brief, shoot cultures were multiplied on MS (Murashige and Skoog ) medium containing 3% (w/v) sucrose, 4.4 µM 6-benzylaminopurine, and 0.5 µM indole-3-butyric acid (IBA). For rooting, single shoots were transferred to rooting medium (1/2 MS with 3% (w/v) sucrose and 4.92 µM IBA) for 3 weeks, before the plants were acclimatized in a commercial peat substrate (Steckmedium, Klasmann-Deilmann GmbH, Geeste, Germany). Forty-three-day-old plants (counting from transfer of the plants into the substrate) were used for the greenhouse experiment. At set-up, the substrate was carefully removed around the roots. Both bacterial strains, FZB42 and RU47, were inoculated by root-dipping followed by drenching around the stem after planting in the soils. Root dipping and drenching with sterile tap water without inoculum served as control (treatment C). For root dipping, inoculation suspensions were prepared. For FZB42, the provided spore suspension was mixed thoroughly and diluted in sterile tap water to 1 × 10 7 spores mL −1 (treatment B). For RU47, overnight cultures were grown and pelleted at 4000 g for 20 min at 4 °C. The cell pellet was washed in 50 mL sterile 0.85% NaCl twice, and the centrifugation was repeated. Finally, the cells were diluted to 1 × 10 7 CFU mL −1 in sterile tap water (treatment P). The roots of plants to be treated were placed in 250 mL of inoculation suspension for 30 min. After root dipping, each plant was transferred into a pot with 400 mL soil (1.2 g mL −1 ) and subsequently subjected to drenching with 10 mL sterile H 2 O (C), 10 mL 1 × 10 8 spores mL −1 of FZB42 or 10 mL 1 × 10 8 CFU mL −1 of RU47. Pots were placed in trays laid out with fleece mats facilitating steady watering from below. Plants were irrigated every other day by evenly wetting the fleece mats. Cultivation occurred from June 06, 2020, until July 07, 2020, in a greenhouse chamber at a mean temperature of 19.9 ± 1.8/18.4 ± 0.9 °C (day/night) with a 16-h photoperiod. The average relative humidity during the experiment was 71.2 ± 10.9%. If the photosynthetic active radiation (PAR) was below 182 µmol m 2 s −1 during the photoperiod, additional light was supplied by high-pressure sodium lamps (MASTER SON-T PIA Plus, Phillips Lightning, Eindhoven, Netherlands). The two soils and three treatments led to six differently treated variants: M26 grown in ARD or grass soil and treated with C, B, or P (Fig. ). Sample collection and processing Destructive samplings of RA, RP, and roots for the extraction of phytoalexins (PA) were conducted with four replicates per variant 3, 16, and 28 dpi (Fig. ). Twenty-eight dpi, RE and root morphology were analyzed using six and seven replicates per variant, respectively. Plants were carefully taken out of the pots. The soil remaining in the pots (RA) was mixed well, and 1 g was resuspended in 1:10 (w/v) 0.85% NaCl and vortexed for 1 min. Root systems were separated from the shoot using sterile scalpels and split in half. One half of the root system was rinsed gently under water, dried on a paper towel, immediately frozen in liquid nitrogen, and stored at − 80 °C until PA extraction. From the second half, loosely adhering soil was gently removed from the roots using toothbrushes to detach the rhizosphere which was not analyzed further. To obtain RP, the brushed roots were cut in 3–4 cm pieces and RP was detached by vortexing in 1:10 (w/v) 0.85% NaCl for 1 min. RA and RP suspensions were used for selective plating within 2 h after obtaining the solution. For subsequent molecular analysis, the remaining RA and RP solutions were centrifuged at 4000 g for 20 min at 4 °C and the pellets were stored at − 20 °C until microbial community DNA was extracted. Twenty-eight dpi, root systems were split into three parts: (I) for PA extraction, (II) for the harvest of RP and subsequent storage of roots in 50% alcohol solution (i.e. diluted Rotisol®) until analysis of root morphology (four split root systems and three additional entire root systems), and (III) for the analysis of RE (four split root systems and two additional root systems). RE samples were rinsed under tap water and stored in 50 mL reaction tubes in the dark (100% humidity, 4 °C). The next day, roots were surface-disinfected as described in Mahnkopp-Dirks et al. . The surface-disinfected roots were cut into approximately 1 cm pieces and dried for a short period on sterile filter paper under a laminar flow. Approximately 60–100 mg were transferred to a 2 mL reaction tube, frozen in liquid nitrogen and stored at − 80 °C until DNA extraction. Phytoalexin extraction and quantification by GC–MS Aliquots of the root samples described above were lyophilized and homogenized to a fine powder (29 Hz, 1 min; Mixer Mill MM400, Retsch, Haan, Germany). The protocols used for PA extraction as well as detection and quantification by GC–MS after silylation are well-established techniques, which were described previously (Weiß et al. ; Balbín-Suárez et al. ; Busnena et al. ). Briefly, the root powder was extracted with 1 mL methanol containing 25 µg 4-hydroxybiphenyl (internal standard for relative quantification) by vigorous vortexing (2700 rpm, 20 min, Vortex Genie2, Scientific Industries, Bohemia, NY, USA). The extracts obtained were centrifuged (13,439 g , 10 min), and the supernatants were air-stream dried. The residues were redissolved in 1 mL dichloromethane:chloroform (1:1, v/v), centrifuged (13,439 g , 10 min), and the supernatants were air-stream dried. The residues were redissolved in 200 µL ethyl acetate and centrifuged (13,439 g , 10 min). The supernatants were transferred to GC–MS vials with glass inlet, and the ethyl acetate was air-stream dried. Residuals were resuspended in 50 µL N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA; ABCR, Karlsruhe, Germany) and silyated for 30 min at 60 °C. Silyated samples were analyzed on a GC–MS at 70 °C for 3 min, 70–310 °C in 24 min [10 °C min −1 ], 310 °C for 5 min with a helium flow of 1 mL min −1 , an injection volume of 1 µL and a split ratio of 1:10. Relative quantification of the individual compounds was achieved using the added internal standard 4-hydroxybiphenyl (response factor 1), which allowed a relative quantitative comparison of the levels of phytoalexin content in all samples. A set of co-injected hydrocarbons (even-numbered C14 to C32) was used to calculate the retention indices by linear extrapolation, as described in the literature (Busnena et al. ). Analysis of root morphology For root morphological analyses, roots sampled were scanned at 720 dots per inch with 35 μm resolution using a flatbed scanner (EPSON perfection V700). Root traits were analyzed using the software WinRhizo 2019 (Regent Instruments, Canada). Root length was measured in 15 diameter classes divided in 100 µm intervals ranging from < 100 µm to > 1.4 mm. For statistical analysis, the four subsamples, and the three additional samples of entire root systems were considered. To check the quality of the measured root data, the root length of all samples included in the analysis was plotted against the respective root surface without any grouping (Fig. ). All samples fitted to a linear relation indicating that the subsampling of root fractions ( n = 4) was successful as the root-to-surface ratio was in line with that of entire root systems ( n = 3). Thus, all 42 samples were included in the analysis of root morphology. Detection of inoculants To check the colonization of the inoculated strains, serial dilutions of the detached cells from RA and RP were prepared 3, 16, and 28 dpi. Dilutions were plated and incubated for 48 h on media MB and MP for FZB42 and RU47, respectively. For the cultivation-independent quantification of total bacteria in RA and RP, a quantitative real-time PCR (qPCR) of the 16S rRNA gene fragment was performed according to Suzuki et al. using RA and RP DNA extracts. The abundance of RU47 was quantified using newly developed primers ( aombb-F and aombb-R ) and a TaqMan probe ( aombb -P) targeting an autotransporter outer membrane beta-barrel domain-containing protein encoding gene ( aombb ) (Eltlbany ). Details on primer and probe design and qPCR conditions are described in supplemental File S1 and supplemental Table . The relative abundance of RU47 was calculated based on the absolute abundances of aombb and 16S rRNA gene copies. All reactions were performed on a CFX Connect real-time PCR cycler (Bio-Rad Laboratories Inc., Hercules, CA, USA). DNA extraction and amplicon sequencing Microbial community DNA of 0.5 g RA and total RP (~ 0.1 g) samples was extracted using the FastPrep-24 bead-beating system and the FastDNA Spin Kit for Soil (MP Biomedicals, Eschwege, Germany) following the manufacturers’ instructions. DNA extracts were purified using the Geneclean Spin Kit (MP Biomedicals, Eschwege, Germany). For samples taken 28 dpi, the V3–V4 regions of bacterial and archaeal 16S rRNA gene and the fungal ITS2 region were amplified from RA and RP DNA using primer pairs 341F/806R (Sundberg et al. ) and gITS7/ITS4 (White et al. ; Ihrmark et al. ). Samples were shipped to the sequencing service provider Novogene Co. (Cambridge, UK) where PCR amplification, library preparation, and sequencing were performed using Illumina MiSeq v2 PE250 according to the companies’ standard procedures. For RE, frozen samples were homogenized using sterile metal beads (5 mm diameter) and a mixer mill MM400 (RETSCH GmbH, Haan, Germany) at 27 s −1 for 30 s. This step was repeated if the samples were not completely homogenized. DNA extraction was performed using the Invisorb Spin Plant Mini Kit (Invitek Molecular, Berlin, Germany), following the manufacturers’ instructions with the following modifications. (I) Samples were centrifuged at 11,000 g for 7 min before transferring them to the pre-filter to prevent filter clogging. (II) The elution step was performed in two steps with 50 µL elution buffer each instead of once with 100 µL. The 16S rRNA gene fragment of the variable V3–V4 region was amplified using primers 335F/769R (Dorn-In et al. ), which were used to exclude amplification of 16S rRNA genes derived from chloroplasts and mitochondria. The amplified DNA was sequenced on an Illumina MiSeq sequencer as described previously (Mahnkopp-Dirks et al. ) with the following modifications: PCR was done using NEBNext high-fidelity polymerase (New England Biolabs, Ipswich, USA) in a total volume of 25 µl (15 ng DNA template, 12.5 µl polymerase, 5 pmol of each primer) for 5 min at 98 °C; 30 cycles of 10 s at 98 °C, 30 s at 60 °C, 30 s at 72 °C; 5 min at 72 °C. Bioinformatic analysis Datasets from 16S rRNA gene and ITS amplicon sequencing were handled independently. For 16S rRNA gene amplicons from RA and RP, paired-end reads were assigned to samples according to their unique barcodes and truncated by cutting off the barcode and primer sequences. Paired-ends were merged using FLASH (version 1.2.7) resulting in the overlapping splicing sequences, called raw tags. Quality filtering of the raw tags was performed using Qiime2 (Bolyen et al. ) to obtain high-quality clean tags. To detect chimera sequences, clean tags were compared with the Gold reference database (Edgar et al. ) using the UCHIME algorithm. Chimera sequences were removed, finally resulting in effective tags. For sequences derived from RE samples, the pipeline was modified as follows: Sequences were analyzed on the Galaxy web platform version 1.20 (Galaxy ). FASTQ files were trimmed with a minimum read length of 50 using Cutadapt (Martin ). Quality control was performed via FastQC. For subsequent data analysis, the DADA2 pipeline (Callahan et al. ) was used with the following trimming and filtering parameters: 20 bp were removed n-terminally, and reads were truncated at position 280 (forward) and 220 (reverse), with an expected error of 4 (forward) and 6 (reverse), respectively. Amplicon sequence variants (ASVs) were annotated using SILVA database version 132 (Quast et al. ). The primers for the 16S rRNA gene were designed to bind to areas inside the bacterial domain but partial binding to the archaeal domain is also possible. We will refer to this subset of the microbiota as the bacterial community. For the analysis of ITS amplicons derived from RA and RP, PCR primers were trimmed of raw sequence reads. Read pairs in which any of the primers were not detected were discarded using cutadapt (version 2.3) (Martin ). Error-correction was performed for the trimmed sequence-reads, sequences were merged and ASVs were annotated using DADA2 (version 1.10.0) (Callahan et al. ) within Qiime2 (Bolyen et al. ). ASV annotation was done using UNITE database version 8.3 (Abarenkov et al. ). For all datasets, reads were excluded if classified as mitochondria, chloroplast, or plant tissue (mainly Malus ) or if the phylum was missing. ASVs occurring in PCR no template controls were excluded as potential contaminants. For RA and RP, ASVs that were found uniquely in samples of treatments B or P and matched the genome sequence of FZB42 or RU47, respectively, were removed to depict the modulation of the microbiome without potential bias of ASVs affiliated to the inoculants. This was not done for RE, enabling to affirm if internal plant tissue was colonized by FBZ42 or RU47, a comparison between the sequences of the inoculants and the endophytes at ASV level was carried out in Bioedit (Hall ). The sequences were aligned using ClustalW Multiple alignment (Thompson et al. ) with 1000 bootstraps. Statistical analysis and data visualization The analyses were performed using R version 4.2.1 (R Core Team ) in RStudio version 2022.02.1. (R Studio Team ). Pearson’s product-moment correlation was calculated to establish a correlation between CFU counts of RU47 and its relative abundance measured by qPCR. Analysis of the microbial community composition in RA, RP, and RE was performed on rarefied data following as recently recommended by Schloss . Data were rarefied by randomly subsampling to the lowest number of reads for RA and RP samples: 40,118 for 16S rRNA gene and 62,855 for ITS datasets and 18,788 for 16S rRNA gene sequencing in RE using phyloseq (Love et al. ). Average α-diversity indices for species richness, diversity (Shannon), and evenness (Pielou) were calculated using phyloseq (Love et al. ) and forcats (Wickham ) packages. Data were checked for normal distribution by the Shapiro-Wilks test and considered normally distributed at an adjusted p > 0.05 (Benjamini-Hochberg correction). For normally distributed data, the significance of differences in α-diversities was tested by two-factorial analysis of variance (ANOVA) followed by Tukey’s HSD test. For non-normally distributed data, the Kruskal-Wallis test followed by Dunn’s test was applied using vegan (Oksanen et al. ) and agricolae (De Mendiburu and Yaseen ). Effects of microhabitat (RA and RP), soil, and treatment on the microbial communities were tested by permutational analysis of variance (PERMANOVA) based on Bray-Curtis dissimilarity using 10,000 permutations based on square root-transformed count data using vegan (Oksanen et al. ). Ordination of microbial community compositions was obtained by non-metric multidimensional scaling (NMDS) and Bray-Curtis dissimilarity based on square root-transformed count data using vegan (Oksanen et al. ). A negative binomial Wald test (Love et al. ) was used for differential abundance on rarefied reads to identify species with significant differences across RA, RP, and RE among the 20 most abundant ASVs in each microhabitat using DESeq2 v1.18.1 inside phyloseq (Love et al. ). Figures were generated using ggplot2 (Wickham ), ggpubr (Kassambara ), pheatmap (Kolde ), and RColorBrewer (Neuwirth ) packages. For normally distributed parametric data of two variables, statistical differences were tested using paired t -test.
Topsoils (0–20 cm) were collected from a site in Ellerhoop, Germany (53° 42′ 51.71″ N, 9° 46′ 12.16″ E), in May 2020. Soil from this site was classified as Endostagnic Luvisol (FAO and ITPS ). Details on soil texture and abiotic soil properties were reported previously (Mahnkopp et al. ). The site was established in 2009 with ARD plots ( n = 4) by replanting apple rootstocks ‘Bittenfelder Sämling’ every other year (ARD soil) and grass plots ( n = 4) with no apple cultivation history (Mahnkopp et al. ). Soils from both variants were taken as pooled samples from all plots and homogenized by sieving through a 2-mm mesh. For the greenhouse experiment, both soils were mixed with 50% (v/v) sterilized sand and fertilized with 2 g L -1 Osmocote exact 3–4 M (16% N + 9% P 2 O 5 + 12% K 2 O + 2% MgO, ICL Deutschland, Nordhorn, Germany). These mixtures will further be referred to as ARD and grass soils.
The rifampicin-resistant strain Bacillus velezensis FZB42 (DSMZ, Braunschweig, Germany, No.: DSM23117) was provided as a ready-to-use spore suspension of 6.7 × 10 9 CFU mL −1 and derived from the commercial product Rhizovital (ABiTEP GmbH, Berlin, Germany). The cultivation was performed on Reasoners’ 2 agar (R2A, Merck Millipore, Burlington, MA, USA) supplemented with rifampicin (75 µg mL −1 ) and cycloheximide (100 µg mL −1 ) (thereafter called medium “MB”). Strain Pseudomonas sp. RU47 (DSMZ, No.: DSM117411) was obtained from our laboratory strain collection and was resistant to rifampicin, tetracycline, chloramphenicol, and ampicillin. Initial cultivation of the strain was carried out on King’s B agar (KB, Carl Roth, Karlsruhe, Germany) supplemented with rifampicin (75 µg mL −1 ), ampicillin (100 µg mL −1 ), chloramphenicol (30 µg mL −1 ), tetracycline (10 µg mL −1 ), and cycloheximide (100 µg mL −1 ) (thereafter called medium “MP”). Agar plates for both strains were incubated at 28 °C for 48 h until single colonies were observed. Overnight cultures of RU47 were grown in Luria Bertani broth (LB, Carl Roth, Karlsruhe, Germany) supplemented with the aforementioned antibiotics at 28 °C and 150 rpm on a shaker.
Plant material of the ARD-susceptible rootstock genotype M26 was propagated in vitro as described earlier (Rohr et al. ). In brief, shoot cultures were multiplied on MS (Murashige and Skoog ) medium containing 3% (w/v) sucrose, 4.4 µM 6-benzylaminopurine, and 0.5 µM indole-3-butyric acid (IBA). For rooting, single shoots were transferred to rooting medium (1/2 MS with 3% (w/v) sucrose and 4.92 µM IBA) for 3 weeks, before the plants were acclimatized in a commercial peat substrate (Steckmedium, Klasmann-Deilmann GmbH, Geeste, Germany). Forty-three-day-old plants (counting from transfer of the plants into the substrate) were used for the greenhouse experiment. At set-up, the substrate was carefully removed around the roots. Both bacterial strains, FZB42 and RU47, were inoculated by root-dipping followed by drenching around the stem after planting in the soils. Root dipping and drenching with sterile tap water without inoculum served as control (treatment C). For root dipping, inoculation suspensions were prepared. For FZB42, the provided spore suspension was mixed thoroughly and diluted in sterile tap water to 1 × 10 7 spores mL −1 (treatment B). For RU47, overnight cultures were grown and pelleted at 4000 g for 20 min at 4 °C. The cell pellet was washed in 50 mL sterile 0.85% NaCl twice, and the centrifugation was repeated. Finally, the cells were diluted to 1 × 10 7 CFU mL −1 in sterile tap water (treatment P). The roots of plants to be treated were placed in 250 mL of inoculation suspension for 30 min. After root dipping, each plant was transferred into a pot with 400 mL soil (1.2 g mL −1 ) and subsequently subjected to drenching with 10 mL sterile H 2 O (C), 10 mL 1 × 10 8 spores mL −1 of FZB42 or 10 mL 1 × 10 8 CFU mL −1 of RU47. Pots were placed in trays laid out with fleece mats facilitating steady watering from below. Plants were irrigated every other day by evenly wetting the fleece mats. Cultivation occurred from June 06, 2020, until July 07, 2020, in a greenhouse chamber at a mean temperature of 19.9 ± 1.8/18.4 ± 0.9 °C (day/night) with a 16-h photoperiod. The average relative humidity during the experiment was 71.2 ± 10.9%. If the photosynthetic active radiation (PAR) was below 182 µmol m 2 s −1 during the photoperiod, additional light was supplied by high-pressure sodium lamps (MASTER SON-T PIA Plus, Phillips Lightning, Eindhoven, Netherlands). The two soils and three treatments led to six differently treated variants: M26 grown in ARD or grass soil and treated with C, B, or P (Fig. ).
Destructive samplings of RA, RP, and roots for the extraction of phytoalexins (PA) were conducted with four replicates per variant 3, 16, and 28 dpi (Fig. ). Twenty-eight dpi, RE and root morphology were analyzed using six and seven replicates per variant, respectively. Plants were carefully taken out of the pots. The soil remaining in the pots (RA) was mixed well, and 1 g was resuspended in 1:10 (w/v) 0.85% NaCl and vortexed for 1 min. Root systems were separated from the shoot using sterile scalpels and split in half. One half of the root system was rinsed gently under water, dried on a paper towel, immediately frozen in liquid nitrogen, and stored at − 80 °C until PA extraction. From the second half, loosely adhering soil was gently removed from the roots using toothbrushes to detach the rhizosphere which was not analyzed further. To obtain RP, the brushed roots were cut in 3–4 cm pieces and RP was detached by vortexing in 1:10 (w/v) 0.85% NaCl for 1 min. RA and RP suspensions were used for selective plating within 2 h after obtaining the solution. For subsequent molecular analysis, the remaining RA and RP solutions were centrifuged at 4000 g for 20 min at 4 °C and the pellets were stored at − 20 °C until microbial community DNA was extracted. Twenty-eight dpi, root systems were split into three parts: (I) for PA extraction, (II) for the harvest of RP and subsequent storage of roots in 50% alcohol solution (i.e. diluted Rotisol®) until analysis of root morphology (four split root systems and three additional entire root systems), and (III) for the analysis of RE (four split root systems and two additional root systems). RE samples were rinsed under tap water and stored in 50 mL reaction tubes in the dark (100% humidity, 4 °C). The next day, roots were surface-disinfected as described in Mahnkopp-Dirks et al. . The surface-disinfected roots were cut into approximately 1 cm pieces and dried for a short period on sterile filter paper under a laminar flow. Approximately 60–100 mg were transferred to a 2 mL reaction tube, frozen in liquid nitrogen and stored at − 80 °C until DNA extraction.
Aliquots of the root samples described above were lyophilized and homogenized to a fine powder (29 Hz, 1 min; Mixer Mill MM400, Retsch, Haan, Germany). The protocols used for PA extraction as well as detection and quantification by GC–MS after silylation are well-established techniques, which were described previously (Weiß et al. ; Balbín-Suárez et al. ; Busnena et al. ). Briefly, the root powder was extracted with 1 mL methanol containing 25 µg 4-hydroxybiphenyl (internal standard for relative quantification) by vigorous vortexing (2700 rpm, 20 min, Vortex Genie2, Scientific Industries, Bohemia, NY, USA). The extracts obtained were centrifuged (13,439 g , 10 min), and the supernatants were air-stream dried. The residues were redissolved in 1 mL dichloromethane:chloroform (1:1, v/v), centrifuged (13,439 g , 10 min), and the supernatants were air-stream dried. The residues were redissolved in 200 µL ethyl acetate and centrifuged (13,439 g , 10 min). The supernatants were transferred to GC–MS vials with glass inlet, and the ethyl acetate was air-stream dried. Residuals were resuspended in 50 µL N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA; ABCR, Karlsruhe, Germany) and silyated for 30 min at 60 °C. Silyated samples were analyzed on a GC–MS at 70 °C for 3 min, 70–310 °C in 24 min [10 °C min −1 ], 310 °C for 5 min with a helium flow of 1 mL min −1 , an injection volume of 1 µL and a split ratio of 1:10. Relative quantification of the individual compounds was achieved using the added internal standard 4-hydroxybiphenyl (response factor 1), which allowed a relative quantitative comparison of the levels of phytoalexin content in all samples. A set of co-injected hydrocarbons (even-numbered C14 to C32) was used to calculate the retention indices by linear extrapolation, as described in the literature (Busnena et al. ).
For root morphological analyses, roots sampled were scanned at 720 dots per inch with 35 μm resolution using a flatbed scanner (EPSON perfection V700). Root traits were analyzed using the software WinRhizo 2019 (Regent Instruments, Canada). Root length was measured in 15 diameter classes divided in 100 µm intervals ranging from < 100 µm to > 1.4 mm. For statistical analysis, the four subsamples, and the three additional samples of entire root systems were considered. To check the quality of the measured root data, the root length of all samples included in the analysis was plotted against the respective root surface without any grouping (Fig. ). All samples fitted to a linear relation indicating that the subsampling of root fractions ( n = 4) was successful as the root-to-surface ratio was in line with that of entire root systems ( n = 3). Thus, all 42 samples were included in the analysis of root morphology.
To check the colonization of the inoculated strains, serial dilutions of the detached cells from RA and RP were prepared 3, 16, and 28 dpi. Dilutions were plated and incubated for 48 h on media MB and MP for FZB42 and RU47, respectively. For the cultivation-independent quantification of total bacteria in RA and RP, a quantitative real-time PCR (qPCR) of the 16S rRNA gene fragment was performed according to Suzuki et al. using RA and RP DNA extracts. The abundance of RU47 was quantified using newly developed primers ( aombb-F and aombb-R ) and a TaqMan probe ( aombb -P) targeting an autotransporter outer membrane beta-barrel domain-containing protein encoding gene ( aombb ) (Eltlbany ). Details on primer and probe design and qPCR conditions are described in supplemental File S1 and supplemental Table . The relative abundance of RU47 was calculated based on the absolute abundances of aombb and 16S rRNA gene copies. All reactions were performed on a CFX Connect real-time PCR cycler (Bio-Rad Laboratories Inc., Hercules, CA, USA).
Microbial community DNA of 0.5 g RA and total RP (~ 0.1 g) samples was extracted using the FastPrep-24 bead-beating system and the FastDNA Spin Kit for Soil (MP Biomedicals, Eschwege, Germany) following the manufacturers’ instructions. DNA extracts were purified using the Geneclean Spin Kit (MP Biomedicals, Eschwege, Germany). For samples taken 28 dpi, the V3–V4 regions of bacterial and archaeal 16S rRNA gene and the fungal ITS2 region were amplified from RA and RP DNA using primer pairs 341F/806R (Sundberg et al. ) and gITS7/ITS4 (White et al. ; Ihrmark et al. ). Samples were shipped to the sequencing service provider Novogene Co. (Cambridge, UK) where PCR amplification, library preparation, and sequencing were performed using Illumina MiSeq v2 PE250 according to the companies’ standard procedures. For RE, frozen samples were homogenized using sterile metal beads (5 mm diameter) and a mixer mill MM400 (RETSCH GmbH, Haan, Germany) at 27 s −1 for 30 s. This step was repeated if the samples were not completely homogenized. DNA extraction was performed using the Invisorb Spin Plant Mini Kit (Invitek Molecular, Berlin, Germany), following the manufacturers’ instructions with the following modifications. (I) Samples were centrifuged at 11,000 g for 7 min before transferring them to the pre-filter to prevent filter clogging. (II) The elution step was performed in two steps with 50 µL elution buffer each instead of once with 100 µL. The 16S rRNA gene fragment of the variable V3–V4 region was amplified using primers 335F/769R (Dorn-In et al. ), which were used to exclude amplification of 16S rRNA genes derived from chloroplasts and mitochondria. The amplified DNA was sequenced on an Illumina MiSeq sequencer as described previously (Mahnkopp-Dirks et al. ) with the following modifications: PCR was done using NEBNext high-fidelity polymerase (New England Biolabs, Ipswich, USA) in a total volume of 25 µl (15 ng DNA template, 12.5 µl polymerase, 5 pmol of each primer) for 5 min at 98 °C; 30 cycles of 10 s at 98 °C, 30 s at 60 °C, 30 s at 72 °C; 5 min at 72 °C.
Datasets from 16S rRNA gene and ITS amplicon sequencing were handled independently. For 16S rRNA gene amplicons from RA and RP, paired-end reads were assigned to samples according to their unique barcodes and truncated by cutting off the barcode and primer sequences. Paired-ends were merged using FLASH (version 1.2.7) resulting in the overlapping splicing sequences, called raw tags. Quality filtering of the raw tags was performed using Qiime2 (Bolyen et al. ) to obtain high-quality clean tags. To detect chimera sequences, clean tags were compared with the Gold reference database (Edgar et al. ) using the UCHIME algorithm. Chimera sequences were removed, finally resulting in effective tags. For sequences derived from RE samples, the pipeline was modified as follows: Sequences were analyzed on the Galaxy web platform version 1.20 (Galaxy ). FASTQ files were trimmed with a minimum read length of 50 using Cutadapt (Martin ). Quality control was performed via FastQC. For subsequent data analysis, the DADA2 pipeline (Callahan et al. ) was used with the following trimming and filtering parameters: 20 bp were removed n-terminally, and reads were truncated at position 280 (forward) and 220 (reverse), with an expected error of 4 (forward) and 6 (reverse), respectively. Amplicon sequence variants (ASVs) were annotated using SILVA database version 132 (Quast et al. ). The primers for the 16S rRNA gene were designed to bind to areas inside the bacterial domain but partial binding to the archaeal domain is also possible. We will refer to this subset of the microbiota as the bacterial community. For the analysis of ITS amplicons derived from RA and RP, PCR primers were trimmed of raw sequence reads. Read pairs in which any of the primers were not detected were discarded using cutadapt (version 2.3) (Martin ). Error-correction was performed for the trimmed sequence-reads, sequences were merged and ASVs were annotated using DADA2 (version 1.10.0) (Callahan et al. ) within Qiime2 (Bolyen et al. ). ASV annotation was done using UNITE database version 8.3 (Abarenkov et al. ). For all datasets, reads were excluded if classified as mitochondria, chloroplast, or plant tissue (mainly Malus ) or if the phylum was missing. ASVs occurring in PCR no template controls were excluded as potential contaminants. For RA and RP, ASVs that were found uniquely in samples of treatments B or P and matched the genome sequence of FZB42 or RU47, respectively, were removed to depict the modulation of the microbiome without potential bias of ASVs affiliated to the inoculants. This was not done for RE, enabling to affirm if internal plant tissue was colonized by FBZ42 or RU47, a comparison between the sequences of the inoculants and the endophytes at ASV level was carried out in Bioedit (Hall ). The sequences were aligned using ClustalW Multiple alignment (Thompson et al. ) with 1000 bootstraps.
The analyses were performed using R version 4.2.1 (R Core Team ) in RStudio version 2022.02.1. (R Studio Team ). Pearson’s product-moment correlation was calculated to establish a correlation between CFU counts of RU47 and its relative abundance measured by qPCR. Analysis of the microbial community composition in RA, RP, and RE was performed on rarefied data following as recently recommended by Schloss . Data were rarefied by randomly subsampling to the lowest number of reads for RA and RP samples: 40,118 for 16S rRNA gene and 62,855 for ITS datasets and 18,788 for 16S rRNA gene sequencing in RE using phyloseq (Love et al. ). Average α-diversity indices for species richness, diversity (Shannon), and evenness (Pielou) were calculated using phyloseq (Love et al. ) and forcats (Wickham ) packages. Data were checked for normal distribution by the Shapiro-Wilks test and considered normally distributed at an adjusted p > 0.05 (Benjamini-Hochberg correction). For normally distributed data, the significance of differences in α-diversities was tested by two-factorial analysis of variance (ANOVA) followed by Tukey’s HSD test. For non-normally distributed data, the Kruskal-Wallis test followed by Dunn’s test was applied using vegan (Oksanen et al. ) and agricolae (De Mendiburu and Yaseen ). Effects of microhabitat (RA and RP), soil, and treatment on the microbial communities were tested by permutational analysis of variance (PERMANOVA) based on Bray-Curtis dissimilarity using 10,000 permutations based on square root-transformed count data using vegan (Oksanen et al. ). Ordination of microbial community compositions was obtained by non-metric multidimensional scaling (NMDS) and Bray-Curtis dissimilarity based on square root-transformed count data using vegan (Oksanen et al. ). A negative binomial Wald test (Love et al. ) was used for differential abundance on rarefied reads to identify species with significant differences across RA, RP, and RE among the 20 most abundant ASVs in each microhabitat using DESeq2 v1.18.1 inside phyloseq (Love et al. ). Figures were generated using ggplot2 (Wickham ), ggpubr (Kassambara ), pheatmap (Kolde ), and RColorBrewer (Neuwirth ) packages. For normally distributed parametric data of two variables, statistical differences were tested using paired t -test.
Bacterial inoculants successfully colonized root-affected soil and rhizoplane The CFU counts determined on selective media for RA and RP 3, 16, and 28 dpi revealed that both strains established well in ARD and grass soils, indicating the successful colonization of the inoculants (ARD vs. grass: Fig. ). Only for RU47 in RA, significantly lower CFU numbers were recorded in ARD soil 16 and 28 dpi. The CFU counts of FZB42 were stable over 4 weeks in RA ARD and grass soil, and only 28 dpi, the CFU counts in ARD soil were significantly lower compared to grass. In contrast, CFU counts of RU47 significantly decreased in RA over time with lower CFU counts 16 and 28 dpi in ARD soil compared to grass. Also, in RP, CFU counts of FZB42 remained stable over time at around 10 6 CFU g −1 root fresh mass (RFW) for both, ARD and grass soils. Generally, CFU counts of RU47 were higher in RP than in RA but decreased significantly over time. Both inoculant strains successfully colonized RA and RP of apple plants M26 grown in both, ARD or grass soils for at least 4 weeks. Bacterial 16S rRNA gene abundance ranged from 5 × 10 7 copies g −1 SFW to 2 × 10 8 copies g −1 SFW in RA and 8 × 10 8 copies g −1 RFW to 2 × 10 9 copies g −1 RFW in RP. In RA and RP, absolute abundances of 16S rRNA gene copies g −1 of SFW or RFM were not significantly different between treatments, soils, and time points. For the cultivation-independent detection of RU47, a qPCR was used to quantify aombb in RA and RP DNA extracts from 3, 16, and 28 dpi. In no sample of treatments C and B, amplification of specific aombb fragments was detected, confirming primer specificity. The qPCR data indicated a high competence of RU47 to colonize RP (Fig. ). The strain established less well in RA. Already 3 dpi its relative abundance was at LOG(− 3.7) copies aombb/16S rRNA in ARD and LOG(− 3.6) copies aombb/16S rRNA in grass soil and further decreased significantly over time. At the same time points, no significant differences between grass and ARD soil were observed in RA. In RP, the abundance of RU47 was about 2 orders of magnitude higher 3 dpi compared to RA and the relative abundance of RU47 significantly decreased over time. There was a high positive correlation ( r = 0.901, p < 0.001) between the results of the selective plating method using Log 10 -transformed CFU counts of RU47 (Fig. ) and the copy numbers determined by qPCR of the aombb gene (Fig. ). Microbial diversity in root-affected soil, rhizoplane, and root endosphere Rarefaction curves (Fig. ) showed that all samples reached a plateau with a sample size of > 20.000 reads per sample for 16S rRNA gene and ITS amplicon sequencing of RA and RP and with > 5000 reads per sample for 16S rRNA gene amplicon sequencing for RE. In all samples, the microbial diversity was covered sufficiently by the size of the sequence library. Calculating the α-diversity indices Shannon, richness and evenness for each microhabitat individually did not reveal differences for the bacterial or fungal diversity regarding the different treatments or soils (Fig. ) except that significantly lower richness and Shannon indices were observed in the RP of apple plants grown in ARD soil compared to grass soil (C treatment). In the B and P treatments, α-diversity was restored. We analyzed the β-diversity for B and P in comparison to C, individually for each inoculant. The bacterial β-diversity was mainly shaped by the microhabitat (RA vs. RP), explaining 23% and 24% of the variance for B and P (Fig. A; Table ). Also, the soil (ARD vs. grass) had a significant influence, explaining 6% of the variance for both B and P, while inoculation had no significant influence on bacterial β-diversity. Similar to the bacterial community, the fungal community composition was mainly shaped by microhabitat (RA vs. RP) followed by soil (ARD vs. grass) (Fig. A; Table ). Both factors significantly influenced the fungal community composition irrespective of inoculation. However, a significant effect of the inoculum on fungal β-diversity, explaining 10% and 4% of the variance was observed for treatment B and P, respectively (Fig. A; Table ). In RE, no evident clustering of samples separating treatments B and P from C was observed using NMDS-plotting (Fig. B), but the RE of plants grown in grass or ARD soils were clearly distinguished. This is supported by the results of PERMANOVA, showing highly significant differences between ARD and grass soils (Table ). Effect of soil and inoculation on dominant taxa in root-affected soil, rhizoplane, and root endosphere The most pronounced differences in the relative abundance of the dominant bacterial taxa were observed between grass and ARD soils, irrespective of the different treatments in RA and RP (Fig. ; Table ). In RA, Bacillus (ASV39) and Bradyrhizobium (ASV7) were higher in relative abundance in ARD_C compared to Grass_C, while Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium (ASV3) was significantly lower in ARD_C than in Grass_C. In RA of ARD soil, the relative abundance of Gaiellales (ASV24) and Gaiella (ASV33) was significantly higher in ARD_C compared to ARD_B and ARD_P. Two ASVs affiliated to Bacillus (ASV34 and ASV39) were significantly lower in relative abundance in ARD_C and ARD_B compared to ARD_P. Contrary, the relative abundance of Streptomyces (ASV1) was significantly higher in ARD_C and ARD_B compared to ARD_P. In RP, bacterial ASVs of Streptomyces (ASV1), Novosphingobium , or taxa belonging to the families Enterobacteriaceae and Sphingomonadaceae were overall highly abundant in both soils and all treatments (Fig. ; Table ). The relative abundances of Burkholderia-Caballeronia-Paraburkholderia (ASV13) and Enterobacteriaceae (ASV2) were significantly higher in RP of ARD_C compared to Grass_C. In contrast, the relative abundance of Sphingobium (ASV26) and Phenylobacterium (ASV6) was significantly lower in ARD_C than in Grass_C. Four of the 20 most abundant bacterial ASVs in RP affiliated to Novosphingobium (ASV14, ASV16, ASV17, and ASV23) were differently abundant either in ARD or grass soil. A taxon from the family Enterobacteriaceae (ASV2) was higher in relative abundance in ARD_C (11.8%) compared to ARD_B (5.9%) and ARD_P (4.2%), with ARD_P being significantly lower. The same trend was observed for the relative abundance of Enterobacteriaceae (ASV10) which was highest in ARD_C (3.69%), followed by ARD_B (1.82%) or ARD_P (1.32%). Interestingly, in both, RA and RP, the different treatments significantly affected the relative abundance of dominant ASVs exclusively in ARD soil while no significant differences between treatments were observed in grass soils. As described for bacteria, the most pronounced differences among the most abundant fungal ASVs were observed between ARD and grass soil in both microhabitats, RA and RP (Fig. ; Table ). In RA, pairwise comparisons of each treatment revealed that Mortierella (ASV15) was significantly higher only in ARD_C compared to Grass_C, but not after treatments B or P. Overall, a clear trend that indicated lower relative abundances of ASVs identified as Mortierella (ASV5, ASV6, ASV7, and ASV8) in ARD_C compared to Grass_C was observed. In RA from grass, only the relative abundance of Cladosporium (ASV5) was significantly lower after treatment B and P compared to treatment C. The present dataset revealed that in RP, Thelonectria (ASV1, ASV2, ASV3, and ASV4) and Ilyonectria (ASV40) were dominant in all treatments with ASV1, ASV4, and ASV40 being significantly higher in relative abundance in RP from ARD soil compared to grass for all treatments except ASV4 in treatment P. In the RP of grass soil, the relative abundance of Cladosporium (ASV5) and Moesziomyces (ASV28) was significantly higher in treatment C than in treatments B or P. In RE, differences in relative abundances among the most abundant ASVs were observed mainly between ARD and grass soil (Fig. ; Table ). The relative abundance of Pseudomonas (ASV8) was significantly higher in ARD_C compared to Grass_C as well as compared to ARD_B and ARD_P. The relative abundance of Burkholderia-Caballeronia-Paraburkholderia (ASV18) was significantly higher in ARD_C (12.6%) than in Grass_C (2.3%). The relative abundance of Streptomyces (ASV29) was significantly increased in RE of ARD soils compared to grass in all treatments. In RE of all ARD soils, Bosea (ASV9) was low in relative abundance in all treatments (< 0.1%). It was mainly present in grass soil and significantly lower in relative abundance in Grass_C (2.4%) compared to Grass_B (4.1%) and Grass_P (2.5%). Delftia (ASV4) was almost exclusively present in the RE of grass soil. It accounted for 10.4%, 12.7%, and 10.0% of the relative abundance in RE of Grass_C, Grass_B and Grass_P, but only for 0.01%, 0.16%, and 0% in ARD_C, ARD_B and ARD_P, respectively. In RE, the relative abundance of two bacterial ASVs was significantly different due to inoculation with B or P in RE from either ARD or grass soil. In comparison, in RA and RP a higher number of the dominant bacterial ASVs was significantly different due to inoculation, exclusively in ARD but not in grass soil. The comparison of ASVs affiliated with the genus Bacillus with the genome sequence of FZB42 revealed that the sequence of ASV1040 overlapped 100% with the sequence of inoculant FZB42. Based on amplicon sequencing analysis, ASV1040 was annotated as Bacillus velezensis and occurred exclusively in RE samples of ARD_B and Grass_B with a relative abundance between 0.052 and 0.228%. For RU47, ASV50 overlapped 100% with at least one of the multiple 16S rRNA gene copies of the RU47 genome. It occurred almost exclusively in samples of treatment P, with one exception where it occurred in relatively low abundance in one sample of Grass_C. Based on the results of amplicon sequencing, ASV50 was annotated as Pseudomonas baetica and occurred with a relative abundance between 0.064 and 8.61%. The data indicate that the inoculants FZB42 and RU47 were detected in the RE. Altogether, inoculation with FZB42 or RU47 did not lead to a significantly increased relative abundance of the genera Bacillus or Pseudomonas in RE of treatments B or P. Phytoalexins In general, the total PA content in roots of M26 increased over 4 weeks, resulting in significantly higher PA contents 28 dpi compared to 3 dpi (Fig. ). At all time points, the content of PAs in the roots was the lowest in Grass_C. Three dpi, the total PA content in roots of ARD_C (191.7 µg g −1 root dry weight (RDW)) was significantly higher than that in roots of Grass_C (31.6 µg g −1 RDW; p < 0.05). In contrast, no significant differences existed in total PA content between ARD_B and Grass_B, ARD_P and Grass_P. When comparing the treatments 3 dpi, the total PA content in ARD_C was significantly higher than those of ARD_B and ARD_P. In contrast, no significant effect of the inoculation treatments was observed in grass soil. Sixteen dpi, PA contents in roots increased in all variants with no significant differences between treatments and soils. Twenty-eight dpi, PA contents in roots of ARD_C (1022.7 µg g −1 RDW) and ARD_P (1224.4 µg g −1 RDW) were significantly higher than those of Grass_C (218.8 µg g −1 RDW) and Grass_P (768.4 µg g −1 RDW). Interestingly, the total PA contents in Grass_B (733.0 µg g −1 RDW; p < 0.05) and Grass_P (768.1 µg g −1 RDW; p < 0.05) were significantly higher than that of Grass_C (218.8 µg g −1 RDW). Peak PA concentrations were measured in ARD soils at the latest time point in ARD_B (1301.3 µg g −1 RDW), ARD_P (1224.4 µg g −1 RDW). No significant differences were detected among the treatments in ARD soil with PA contents in ARD_C (1022.7 µg g −1 RDW) being slightly lower than in ARD_B and ARD_P. These results indicate that inoculation with B or P to apple plants grown in grass soil significantly induced the production of PAs in the roots at the late time point (28 dpi). The inoculation with B or P to apple plants grown in ARD soil did not significantly, but slightly increase PA production. The dominant PA was 2-hydroxy-4-methoxydibenzofuran in all variants (Fig. ). Noraucuparin and 3-hydroxy-5-methoxybiphenyl ranked second and third. Over time, the concentrations of hydroxyeriobofuran isomer 2 and noreriobofuran in roots increased. Three dpi, the PAs at the end of their biosynthetic pathway, like noreriobofuan and eriobofuran, were absent in all variants. Still, 16 dpi and 28 dpi, they appeared in all variants except for Grass_C with increased contents 28 dpi compared to 16 dpi. Root morphology Root length was not significantly different between treatments or when comparing root systems grown in ARD or grass soil (Fig. ). However, some trends were observed: longest root systems were present in Grass_C. The length of inoculated root systems was reduced to 69% in Grass_B and 58% in Grass_P compared to Grass_C. This reduction was not observed for ARD soils: root systems grown in ARD soils were smaller and relative root length was 40–50% lower in ARD soils compared to Grass_C with no effect of inoculation (Fig. ). A comparison of root diameters measured in 100 µm steps indicated that growth in ARD soil led to the formation of significantly less fine roots and higher portions of thicker roots compared to Grass_C, which was observed for all treatments (Fig. ). Inoculation with both B and P resulted in significant shifts in root morphology towards less fine roots and higher portions of thicker roots in inoculated root systems, exclusively in grass, but not in ARD soil.
The CFU counts determined on selective media for RA and RP 3, 16, and 28 dpi revealed that both strains established well in ARD and grass soils, indicating the successful colonization of the inoculants (ARD vs. grass: Fig. ). Only for RU47 in RA, significantly lower CFU numbers were recorded in ARD soil 16 and 28 dpi. The CFU counts of FZB42 were stable over 4 weeks in RA ARD and grass soil, and only 28 dpi, the CFU counts in ARD soil were significantly lower compared to grass. In contrast, CFU counts of RU47 significantly decreased in RA over time with lower CFU counts 16 and 28 dpi in ARD soil compared to grass. Also, in RP, CFU counts of FZB42 remained stable over time at around 10 6 CFU g −1 root fresh mass (RFW) for both, ARD and grass soils. Generally, CFU counts of RU47 were higher in RP than in RA but decreased significantly over time. Both inoculant strains successfully colonized RA and RP of apple plants M26 grown in both, ARD or grass soils for at least 4 weeks. Bacterial 16S rRNA gene abundance ranged from 5 × 10 7 copies g −1 SFW to 2 × 10 8 copies g −1 SFW in RA and 8 × 10 8 copies g −1 RFW to 2 × 10 9 copies g −1 RFW in RP. In RA and RP, absolute abundances of 16S rRNA gene copies g −1 of SFW or RFM were not significantly different between treatments, soils, and time points. For the cultivation-independent detection of RU47, a qPCR was used to quantify aombb in RA and RP DNA extracts from 3, 16, and 28 dpi. In no sample of treatments C and B, amplification of specific aombb fragments was detected, confirming primer specificity. The qPCR data indicated a high competence of RU47 to colonize RP (Fig. ). The strain established less well in RA. Already 3 dpi its relative abundance was at LOG(− 3.7) copies aombb/16S rRNA in ARD and LOG(− 3.6) copies aombb/16S rRNA in grass soil and further decreased significantly over time. At the same time points, no significant differences between grass and ARD soil were observed in RA. In RP, the abundance of RU47 was about 2 orders of magnitude higher 3 dpi compared to RA and the relative abundance of RU47 significantly decreased over time. There was a high positive correlation ( r = 0.901, p < 0.001) between the results of the selective plating method using Log 10 -transformed CFU counts of RU47 (Fig. ) and the copy numbers determined by qPCR of the aombb gene (Fig. ).
Rarefaction curves (Fig. ) showed that all samples reached a plateau with a sample size of > 20.000 reads per sample for 16S rRNA gene and ITS amplicon sequencing of RA and RP and with > 5000 reads per sample for 16S rRNA gene amplicon sequencing for RE. In all samples, the microbial diversity was covered sufficiently by the size of the sequence library. Calculating the α-diversity indices Shannon, richness and evenness for each microhabitat individually did not reveal differences for the bacterial or fungal diversity regarding the different treatments or soils (Fig. ) except that significantly lower richness and Shannon indices were observed in the RP of apple plants grown in ARD soil compared to grass soil (C treatment). In the B and P treatments, α-diversity was restored. We analyzed the β-diversity for B and P in comparison to C, individually for each inoculant. The bacterial β-diversity was mainly shaped by the microhabitat (RA vs. RP), explaining 23% and 24% of the variance for B and P (Fig. A; Table ). Also, the soil (ARD vs. grass) had a significant influence, explaining 6% of the variance for both B and P, while inoculation had no significant influence on bacterial β-diversity. Similar to the bacterial community, the fungal community composition was mainly shaped by microhabitat (RA vs. RP) followed by soil (ARD vs. grass) (Fig. A; Table ). Both factors significantly influenced the fungal community composition irrespective of inoculation. However, a significant effect of the inoculum on fungal β-diversity, explaining 10% and 4% of the variance was observed for treatment B and P, respectively (Fig. A; Table ). In RE, no evident clustering of samples separating treatments B and P from C was observed using NMDS-plotting (Fig. B), but the RE of plants grown in grass or ARD soils were clearly distinguished. This is supported by the results of PERMANOVA, showing highly significant differences between ARD and grass soils (Table ).
The most pronounced differences in the relative abundance of the dominant bacterial taxa were observed between grass and ARD soils, irrespective of the different treatments in RA and RP (Fig. ; Table ). In RA, Bacillus (ASV39) and Bradyrhizobium (ASV7) were higher in relative abundance in ARD_C compared to Grass_C, while Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium (ASV3) was significantly lower in ARD_C than in Grass_C. In RA of ARD soil, the relative abundance of Gaiellales (ASV24) and Gaiella (ASV33) was significantly higher in ARD_C compared to ARD_B and ARD_P. Two ASVs affiliated to Bacillus (ASV34 and ASV39) were significantly lower in relative abundance in ARD_C and ARD_B compared to ARD_P. Contrary, the relative abundance of Streptomyces (ASV1) was significantly higher in ARD_C and ARD_B compared to ARD_P. In RP, bacterial ASVs of Streptomyces (ASV1), Novosphingobium , or taxa belonging to the families Enterobacteriaceae and Sphingomonadaceae were overall highly abundant in both soils and all treatments (Fig. ; Table ). The relative abundances of Burkholderia-Caballeronia-Paraburkholderia (ASV13) and Enterobacteriaceae (ASV2) were significantly higher in RP of ARD_C compared to Grass_C. In contrast, the relative abundance of Sphingobium (ASV26) and Phenylobacterium (ASV6) was significantly lower in ARD_C than in Grass_C. Four of the 20 most abundant bacterial ASVs in RP affiliated to Novosphingobium (ASV14, ASV16, ASV17, and ASV23) were differently abundant either in ARD or grass soil. A taxon from the family Enterobacteriaceae (ASV2) was higher in relative abundance in ARD_C (11.8%) compared to ARD_B (5.9%) and ARD_P (4.2%), with ARD_P being significantly lower. The same trend was observed for the relative abundance of Enterobacteriaceae (ASV10) which was highest in ARD_C (3.69%), followed by ARD_B (1.82%) or ARD_P (1.32%). Interestingly, in both, RA and RP, the different treatments significantly affected the relative abundance of dominant ASVs exclusively in ARD soil while no significant differences between treatments were observed in grass soils. As described for bacteria, the most pronounced differences among the most abundant fungal ASVs were observed between ARD and grass soil in both microhabitats, RA and RP (Fig. ; Table ). In RA, pairwise comparisons of each treatment revealed that Mortierella (ASV15) was significantly higher only in ARD_C compared to Grass_C, but not after treatments B or P. Overall, a clear trend that indicated lower relative abundances of ASVs identified as Mortierella (ASV5, ASV6, ASV7, and ASV8) in ARD_C compared to Grass_C was observed. In RA from grass, only the relative abundance of Cladosporium (ASV5) was significantly lower after treatment B and P compared to treatment C. The present dataset revealed that in RP, Thelonectria (ASV1, ASV2, ASV3, and ASV4) and Ilyonectria (ASV40) were dominant in all treatments with ASV1, ASV4, and ASV40 being significantly higher in relative abundance in RP from ARD soil compared to grass for all treatments except ASV4 in treatment P. In the RP of grass soil, the relative abundance of Cladosporium (ASV5) and Moesziomyces (ASV28) was significantly higher in treatment C than in treatments B or P. In RE, differences in relative abundances among the most abundant ASVs were observed mainly between ARD and grass soil (Fig. ; Table ). The relative abundance of Pseudomonas (ASV8) was significantly higher in ARD_C compared to Grass_C as well as compared to ARD_B and ARD_P. The relative abundance of Burkholderia-Caballeronia-Paraburkholderia (ASV18) was significantly higher in ARD_C (12.6%) than in Grass_C (2.3%). The relative abundance of Streptomyces (ASV29) was significantly increased in RE of ARD soils compared to grass in all treatments. In RE of all ARD soils, Bosea (ASV9) was low in relative abundance in all treatments (< 0.1%). It was mainly present in grass soil and significantly lower in relative abundance in Grass_C (2.4%) compared to Grass_B (4.1%) and Grass_P (2.5%). Delftia (ASV4) was almost exclusively present in the RE of grass soil. It accounted for 10.4%, 12.7%, and 10.0% of the relative abundance in RE of Grass_C, Grass_B and Grass_P, but only for 0.01%, 0.16%, and 0% in ARD_C, ARD_B and ARD_P, respectively. In RE, the relative abundance of two bacterial ASVs was significantly different due to inoculation with B or P in RE from either ARD or grass soil. In comparison, in RA and RP a higher number of the dominant bacterial ASVs was significantly different due to inoculation, exclusively in ARD but not in grass soil. The comparison of ASVs affiliated with the genus Bacillus with the genome sequence of FZB42 revealed that the sequence of ASV1040 overlapped 100% with the sequence of inoculant FZB42. Based on amplicon sequencing analysis, ASV1040 was annotated as Bacillus velezensis and occurred exclusively in RE samples of ARD_B and Grass_B with a relative abundance between 0.052 and 0.228%. For RU47, ASV50 overlapped 100% with at least one of the multiple 16S rRNA gene copies of the RU47 genome. It occurred almost exclusively in samples of treatment P, with one exception where it occurred in relatively low abundance in one sample of Grass_C. Based on the results of amplicon sequencing, ASV50 was annotated as Pseudomonas baetica and occurred with a relative abundance between 0.064 and 8.61%. The data indicate that the inoculants FZB42 and RU47 were detected in the RE. Altogether, inoculation with FZB42 or RU47 did not lead to a significantly increased relative abundance of the genera Bacillus or Pseudomonas in RE of treatments B or P.
In general, the total PA content in roots of M26 increased over 4 weeks, resulting in significantly higher PA contents 28 dpi compared to 3 dpi (Fig. ). At all time points, the content of PAs in the roots was the lowest in Grass_C. Three dpi, the total PA content in roots of ARD_C (191.7 µg g −1 root dry weight (RDW)) was significantly higher than that in roots of Grass_C (31.6 µg g −1 RDW; p < 0.05). In contrast, no significant differences existed in total PA content between ARD_B and Grass_B, ARD_P and Grass_P. When comparing the treatments 3 dpi, the total PA content in ARD_C was significantly higher than those of ARD_B and ARD_P. In contrast, no significant effect of the inoculation treatments was observed in grass soil. Sixteen dpi, PA contents in roots increased in all variants with no significant differences between treatments and soils. Twenty-eight dpi, PA contents in roots of ARD_C (1022.7 µg g −1 RDW) and ARD_P (1224.4 µg g −1 RDW) were significantly higher than those of Grass_C (218.8 µg g −1 RDW) and Grass_P (768.4 µg g −1 RDW). Interestingly, the total PA contents in Grass_B (733.0 µg g −1 RDW; p < 0.05) and Grass_P (768.1 µg g −1 RDW; p < 0.05) were significantly higher than that of Grass_C (218.8 µg g −1 RDW). Peak PA concentrations were measured in ARD soils at the latest time point in ARD_B (1301.3 µg g −1 RDW), ARD_P (1224.4 µg g −1 RDW). No significant differences were detected among the treatments in ARD soil with PA contents in ARD_C (1022.7 µg g −1 RDW) being slightly lower than in ARD_B and ARD_P. These results indicate that inoculation with B or P to apple plants grown in grass soil significantly induced the production of PAs in the roots at the late time point (28 dpi). The inoculation with B or P to apple plants grown in ARD soil did not significantly, but slightly increase PA production. The dominant PA was 2-hydroxy-4-methoxydibenzofuran in all variants (Fig. ). Noraucuparin and 3-hydroxy-5-methoxybiphenyl ranked second and third. Over time, the concentrations of hydroxyeriobofuran isomer 2 and noreriobofuran in roots increased. Three dpi, the PAs at the end of their biosynthetic pathway, like noreriobofuan and eriobofuran, were absent in all variants. Still, 16 dpi and 28 dpi, they appeared in all variants except for Grass_C with increased contents 28 dpi compared to 16 dpi.
Root length was not significantly different between treatments or when comparing root systems grown in ARD or grass soil (Fig. ). However, some trends were observed: longest root systems were present in Grass_C. The length of inoculated root systems was reduced to 69% in Grass_B and 58% in Grass_P compared to Grass_C. This reduction was not observed for ARD soils: root systems grown in ARD soils were smaller and relative root length was 40–50% lower in ARD soils compared to Grass_C with no effect of inoculation (Fig. ). A comparison of root diameters measured in 100 µm steps indicated that growth in ARD soil led to the formation of significantly less fine roots and higher portions of thicker roots compared to Grass_C, which was observed for all treatments (Fig. ). Inoculation with both B and P resulted in significant shifts in root morphology towards less fine roots and higher portions of thicker roots in inoculated root systems, exclusively in grass, but not in ARD soil.
The current study demonstrated the competence of Pseudomonas sp. RU47 and Bacillus velezensis FZB42 to colonize the rhizoplane and root-affected soil of apple plants grown in ARD or grass soil. This phenomenon was observed through cultivation-dependent methods for FZB42 and RU47 and cultivation-independent methods for RU47. Twenty-eight dpi, amplicon sequencing data indicated that the inoculants colonized also the root endosphere. Inoculation resulted in a soil- and inoculant strain-dependent plant response. Plants responded strongly to the inoculants by increasing phytoalexins in roots mainly in grass soil. Soil, rhizoplane, and root endosphere microbiome modulation by the inoculant strains investigated by amplicon sequence analysis showed that the microhabitat (RA vs. RP) and the soil (ARD vs. Grass) shaped the microbiome more strongly than the inoculants. Application of bacterial inoculants in apple cultivation To our knowledge, this study is the first one to describe an inoculation procedure for the two bacterial strains on apple plants, which can be easily transferred from greenhouse to field scale. Root dipping can be easily implemented in apple tree planting in tree nurseries and orchards. Usually, roots of the apple plants are stored in a water bath the night before planting which can be supplemented with the bacterial inoculants, causing no additional work for farmers. For repeated inoculation during the growing period, the inoculum suspension could be applied by drenching around the trees with little effort. Successful drenching was demonstrated by Utkhede et al. with Bacillus subtilis EBW4 upon planting apple trees in an orchard in British Columbia, which led to significantly increased fruit yield, reduced disease severity and improved trunk radial growth compared to untreated apple plants. Recently, two other bacterial strains of the genus Bacillus were isolated from healthy apple roots from ARD-affected soils in China (Duan et al. , ). Both strains were investigated for their biological control ability with a focus on the potential to reduce the growth and germination of different Fusarium species that are assumed to add to the ARD disease complex in Chinese soils (Duan et al. , ). Both studies mainly focused on in vitro experiments and did not assess the inoculants rhizosphere competence or potential to modulate the microbiome. Several studies claimed that Fusarium was one of the primary causal pathogens of ARD in China (Wang et al. ; Jiang et al. ). However, there are no studies so far providing clear evidence for the contribution of Fusarium to ARD as discussed by Somera and Mazzola . Our study revealed two ASVs assigned to Fusarium to occur in significantly higher relative abundance in ARD than grass soil. Regarding the relative abundances however, these differences were small (~ 0.1%). Moreover, the increased relative abundance of a certain taxon, here Fusarium , does not provide evidence for the contribution of this taxon to ARD, as the data are based on amplicon sequencing only. Furthermore, previous microscopic studies employed on the same ARD soil did not find evidence for an important role of Fusarium in ARD soils from three sites in Germany (Grunewaldt-Stöcker et al. ). While the studies mentioned above investigated inoculation effects on plant growth, fruit yield or reduction of potential pathogens, our study is the first to examine the modulation of the microbial community composition of apples across microhabitats, phytoalexin content and root morphology triggered by two bacterial inoculants. Colonization competence and quantification techniques of the inoculant strains The present study provides first evidence that both, FZB42 and RU47, are rhizosphere competent on apple rootstock M26 for at least 4 weeks. The use of rifampicin-resistant mutants allowed a specific cultivation-dependent quantification of the inoculants. Compared to amplicon sequencing of the 16S rRNA gene, CFU counting is highly sensitive and provides actual numbers of CFU instead of only relative abundances. The CFU counting method is fast, cost-effective, and easily applicable in every wet lab. In addition to selective plating, a TaqMan-based qPCR-system was developed to quantify RU47. In contrast to the spore-former FZB42, the cultivability of RU47 can be affected by environmental stress and therefore a DNA-based quantification for this strain was also used. We propose that the newly developed qPCR system allows specific detection of the inoculant strain in rhizoplane and soil of RU47-inoculated apple plants. In this study, the results of CFU counting and quantification of the aombb gene of RU47 were highly positively correlated, indicating that the newly developed qPCR assay provides reliable and accurate results for the cultivation-independent quantification of RU47. Combining selective plating and qPCR complements the benefits of both methods. While CFU counting is more sensitive at a lower detection limit than qPCR, the latter allows the detection of cells in a non-cultivable state. Additionally, amplicon sequencing data indicated that both, RU47 and FZB42, colonized the root endosphere. Generally, roots are considered the most common entry point for the colonization of the root interior of plants, including apple (Frank et al. ). Previously, Chen et al. were able to detect the inoculants Burkholderia sp. and Pseudomonas thivervalensis in the root interior of rape 60 dpi. Lacava et al. showed that citrus tree roots ( Citrus sinensis ) inoculated with Klebsiella pneumoniae Kp342 colonized the root interior. The present study revealed two ASVs exclusively in samples of treatments B (ASV1024) and P (ASV50) which were identical with genome sequences of the inoculant strains FZB42 and RU47, respectively. This indicates the potential colonization of the root endosphere by the inoculant strains. However, due to the short sequences obtained by amplicon sequencing, the approach does not allow an identification of the inoculants on species or strain level leaving some uncertainty regarding their identification. Nonetheless, considering the ability of FZB42 and RU47 to colonize soil and rhizoplane, we conclude that the inoculant strains are likely to successfully colonize the root endosphere. Differences in microbial community composition in ARD and grass soil The microbial community composition of ARD and grass soil from the same experimental site used in this study was described previously for different microhabitats (Balbín-Suárez et al. , ; Mahnkopp-Dirks et al. ). As described before (Balbín-Suárez et al. ), the composition of bacteria was mainly driven by the microhabitat and significantly differed between rhizoplane and root-affected soil. These results support the expectation that the microhabitat marks the main differences in microbial community composition (Reinhold-Hurek et al. ; van der Heijden and Schlaeppi ). Balbín-Suárez et al. identified several taxa to be higher in abundance in either ARD or grass soil and identified taxa that uniquely responded to the respective soil. In ARD soil or rhizoplane, they identified Burkholderia , Variovorax , Streptomyces , and Nectria in increased relative abundance. The present study showed ASVs affiliated with Burkholderia-Caballeronia-Paraburkholderia , Thelonectria , and Ilyonectria to be of significantly higher relative abundance in the rhizoplane of plants grown in ARD soils. The potential contribution of Nectriaceae to the ARD disease complex has recently been studied intensively (Popp et al. , ) and is further supported by the notably high occurrence of four ASVs (fungal ASV1-ASV4 in rhizoplane) belonging to the genus Thelonectria and ASV40 belonging to the genus Ilyonectria in the present dataset. Our study found ASVs affiliated to the order Enterobacteriaceae highly abundant in soil and rhizoplane of apple plants grown in ARD soil. Enterobacteriaceae were not observed before in rhizoplane and soil originating from the same site (Balbín-Suárez et al. , ), but one ASV identified as Enterobacteriaceae was found in the root endosphere of apple grown in ARD soil from the same site that negatively correlated with shoot growth and shoot fresh mass (Mahnkopp-Dirks et al. ). In the rhizosphere and the root endosphere of apple trees affected by bitter rot and leaf spot disease in Brazil, Dos Passos et al. found cultivable Enterobacter to be the most abundant genus. Members of the family Enterobacteriaceae can indirectly influence a plant’s defense response by supporting plant pathogens and have the potential to decompose plant tissue (Berg et al. ). In the treatments with the bioinoculants a lower relative abundances of Enterobacteriaceae than non-inoculated controls which could be an indicator for successful modulation of bacteria that accumulate in ARD-affected soils. Certain ASVs of Rhizobium and Streptomyces were previously reported to occur in increased relative abundances in apple roots grown in ARD soil (Mahnkopp-Dirks et al. ). In the present study, Streptomyces was one of the most abundant taxa in the root endosphere, with ASV29 being significantly higher in relative abundance in roots from ARD than in grass soil. Our study revealed a strikingly lower relative abundance of Delftia of almost 0% in ARD soils, which was not observed before. Endophytic strains of the genus Delftia are commonly associated with plant beneficial traits (Han et al. ; Woźniak et al. , ; Da Silveira et al. ; Chen et al. ; Islam et al. ). We assume that a reduction of endophytic Delftia in ARD-affected apple roots might contribute to the plant growth depression. Similarly, also the relative abundance of Bosea was remarkably low with almost 0% in ARD soils, but was found in relative abundances of 2.5% or higher in grass soils. The genus Bosea belongs to the novel family Bosaceae that recently emanated from the family Bradyrhizobiaceae (Hördt et al. ) . Most members of Bosaceae were isolated from legume nodules pointing towards a close association between Bosea and legumes. Also, the co-existence of endophytic Bosea spatocytisi sp. nov. with rhizobia in root nodules was observed (Pulido-Suárez et al. ). The composite mix of plants used as grass cover in our study site included Trifolium , which might explain why Bosea was observed exclusively in samples from grass, but not ARD soils. In rhizoplane of grass soil, e.g., Sphingobium and Novosphingobium occurred in significantly higher relative abundance than in rhizoplane of ARD soil. Several members of the genera Sphingobium and Novosphingobium are well-known environmental strains involved in bioremediation and biodegradation (Yang et al. ; Boss et al. ). Also, various isolates of the family Sphingomonadaceae were identified as PGPR with the potential to produce phytobeneficial compounds or to increase root and shoot length when applied as inoculant (Yang et al. ; Krishnan et al. ). The significant higher relative abundance of Sphingobium and Novosphingobium in grass soil could have promoted the root growth observed in grass soils. To validate this hypothesis, isolation, in vitro testing, and inoculation experiments using Sphingomonadaceae isolates would be needed. Microbiome modulation of the microbial community composition Over the past years, biocontrol, inoculation of beneficial microbes, and measures to increase microbial diversity and activity have gained increasing attention as eco-friendly management options in agriculture (Berg et al. ; Tosi et al. ; Müller and Behrendt ). Recently, microbiome modulation was introduced as an effective mode of action for microbial inoculants (Berg et al. ). Six different types of microbiome modulation were formulated including the restoration of a dysbiosis, the targeted shift towards potentially beneficial taxa, or the depletion of potential pathogens (Berg et al. ). These authors summarized that the degree of microbiome modulation highly depends on the sampling time and mode of inoculant application and that shifts are usually only evident shortly after inoculation. This might explain why in our study no or only little effects of the inoculants on the bacterial ß-diversity were observed 28 dpi. However, fungal ß-diversity was significantly affected by both inoculants likely due to the antifungal activities of both inoculant strains. For instance, the potential to suppress the fungal plant pathogen Rhizoctonia solani was previously demonstrated for both, FZB42 and RU47 (Chowdhury et al. ; Schreiter et al. ). The significant shifts in fungal β-diversity as well as changes in relative abundances of the dominant fungal taxa indicate that FZB42 and RU47 modulated the fungal microbiome in both, ARD and grass soil. These results give perspective for future experiments in which the potential of the inoculants to suppress ARD-associated microorganisms like members of Ilyonectria , Thelonectria , or Pythium should be evaluated not only by amplicon sequencing-based approaches, but by in vitro studies following Koch’s postulates and molecular quantification tools such as qPCR assays targeting taxa that potentially contribute to ARD. Interestingly, regarding the dominant bacteria, only in root-affected soil and rhizoplane of ARD soil, many taxa that were detected changed significantly in relative abundance due to the inoculation while no significant changes in relative abundance were observed for any of the 20 most abundant taxa in both microhabitats from grass soil. We assume that due to the imbalanced microbiome of the ARD-affected soil and rhizoplane, the microbiome was less stable and could be modulated more easily. Plant stress response and plant growth are affected by inoculants The production of phytoalexins is a common plant defense strategy to combat pathogen invasion. Apple and other rosaceous plants, in particular, form biphenyls and dibenzofurans to inhibit microbial growth and cell propagation in a local environment around the plant (Chizzali and Beerhues ; Busnena et al. ). Balbín-Suárez et al. showed that biphenyl and dibenzofuran phytoalexins predominantly and significantly accumulate in diseased roots of apple grown in ARD soils. This result was confirmed in the present study as significantly higher PA contents were observed in the roots of apple plants grown in ARD soil compared to grass soil. However, our results showed that phytoalexins were also induced by the inoculation of FZB42 or RU47 in the roots of apple plants especially in grass soil. Hence, the overall production of biphenyl and dibenzofuran phytoalexins seems not a specific response of apple roots to ARD stress but indicating an unbalanced microbiome. Phytoalexin production upon exposure to microbes is a general defense response of plants, which is also true for apple (Busnena et al. ). In , Weiß et al. reported for the first time that apple roots form biphenyl and dibenzofuran phytoalexins upon microbial stress, here ARD soil. However, production of these phytoalexins after application of microbial inoculants, such as the two bacteria used in this study, has not been reported before. In the few years from 2017 up to now, to our knowledge, there was no report about the formation of the phytoalexins in apple roots in response to inoculants. We assume that the inoculation of high cell or spore numbers disturbed the previous equilibrium of the native soil microbial communities in grass soil, which might have led to the induction of the apple roots’ stress response. The increase in phytoalexin content in the FZB42 and RU47 treatments was weaker in the roots of plants grown in dysbiotic ARD soil. Further detailed research investigating the response of apple plants to microbial inoculants applied in different cell or spore densities is needed to answer this question. Analyzing root length in the different soils and treatments did not reveal significant differences. Most likely, this can be explained by high plant-to-plant variations, impeding the detection of significant differences. However, we observed some trends that might indicate a link to phytoalexin contents. Twenty-eight dpi, roots that were inoculated had significantly higher phytoalexin contents compared to the respective non-inoculated controls in grass soil. We assume that the high phytoalexin content in roots of inoculated plants might have affected root growth. However, as the results regarding the root length were not significant, these observations need to be considered with care. Moreover, inoculation led to significantly higher proportions of thicker roots compared to the control only in grass soil which might have been caused by changes in the microbial community composition and activity. High ethylene concentrations are known to increase cortex width and hence root diameters (Gebauer et al. ). Here, it can only be speculated whether the differences in root diameter classes observed in the present experiment between grass soil and ARD soil are related to the documented differences in microbiome composition and their potential functional differences in hormone production. In summary, we explored the potential of two microbial inoculants, B. velezensis FZB42 and Pseudomonas sp. RU47, on apple, to mitigate apple replant disease. We used inoculation techniques that enabled the inoculants to colonize different root-associated microhabitats of apple, which at the same time, can be easily implemented in current horticultural practices. As a first step, we showed that both inoculant strains had the potential to establish across microhabitats, to modulate the microbiome and to induce shifts in the relative abundance of dominant bacterial and fungal ASVs. The effects of microhabitat and replanting history of the soils on the bacterial community were stronger than the inoculation effect. To our surprise, the inoculation effect of FZB42 on the fungal community composition was stronger than the effect of replanting history likely due to the strong antifungal capacity of FZB42, potentially enabling the suppression of ARD-associated fungi. The inoculants decreased the relative abundance of ARD-related Enterobacteriaceae . Unexpectedly, inoculation increased phytoalexin content in roots of apple plants grown in grass soil. Whether increased phytoalexin contents can be linked to changes in root morphology that were observed needs additional investigation. To further unravel the interplay between inoculants and plants, and in particular their potential to reduce ARD, further field experiments and additional methods such as the measurement of volatile organic compounds or functional microbiome analysis, are needed. It will also be essential to investigate long-term effects of the inoculants in field trials.
To our knowledge, this study is the first one to describe an inoculation procedure for the two bacterial strains on apple plants, which can be easily transferred from greenhouse to field scale. Root dipping can be easily implemented in apple tree planting in tree nurseries and orchards. Usually, roots of the apple plants are stored in a water bath the night before planting which can be supplemented with the bacterial inoculants, causing no additional work for farmers. For repeated inoculation during the growing period, the inoculum suspension could be applied by drenching around the trees with little effort. Successful drenching was demonstrated by Utkhede et al. with Bacillus subtilis EBW4 upon planting apple trees in an orchard in British Columbia, which led to significantly increased fruit yield, reduced disease severity and improved trunk radial growth compared to untreated apple plants. Recently, two other bacterial strains of the genus Bacillus were isolated from healthy apple roots from ARD-affected soils in China (Duan et al. , ). Both strains were investigated for their biological control ability with a focus on the potential to reduce the growth and germination of different Fusarium species that are assumed to add to the ARD disease complex in Chinese soils (Duan et al. , ). Both studies mainly focused on in vitro experiments and did not assess the inoculants rhizosphere competence or potential to modulate the microbiome. Several studies claimed that Fusarium was one of the primary causal pathogens of ARD in China (Wang et al. ; Jiang et al. ). However, there are no studies so far providing clear evidence for the contribution of Fusarium to ARD as discussed by Somera and Mazzola . Our study revealed two ASVs assigned to Fusarium to occur in significantly higher relative abundance in ARD than grass soil. Regarding the relative abundances however, these differences were small (~ 0.1%). Moreover, the increased relative abundance of a certain taxon, here Fusarium , does not provide evidence for the contribution of this taxon to ARD, as the data are based on amplicon sequencing only. Furthermore, previous microscopic studies employed on the same ARD soil did not find evidence for an important role of Fusarium in ARD soils from three sites in Germany (Grunewaldt-Stöcker et al. ). While the studies mentioned above investigated inoculation effects on plant growth, fruit yield or reduction of potential pathogens, our study is the first to examine the modulation of the microbial community composition of apples across microhabitats, phytoalexin content and root morphology triggered by two bacterial inoculants.
The present study provides first evidence that both, FZB42 and RU47, are rhizosphere competent on apple rootstock M26 for at least 4 weeks. The use of rifampicin-resistant mutants allowed a specific cultivation-dependent quantification of the inoculants. Compared to amplicon sequencing of the 16S rRNA gene, CFU counting is highly sensitive and provides actual numbers of CFU instead of only relative abundances. The CFU counting method is fast, cost-effective, and easily applicable in every wet lab. In addition to selective plating, a TaqMan-based qPCR-system was developed to quantify RU47. In contrast to the spore-former FZB42, the cultivability of RU47 can be affected by environmental stress and therefore a DNA-based quantification for this strain was also used. We propose that the newly developed qPCR system allows specific detection of the inoculant strain in rhizoplane and soil of RU47-inoculated apple plants. In this study, the results of CFU counting and quantification of the aombb gene of RU47 were highly positively correlated, indicating that the newly developed qPCR assay provides reliable and accurate results for the cultivation-independent quantification of RU47. Combining selective plating and qPCR complements the benefits of both methods. While CFU counting is more sensitive at a lower detection limit than qPCR, the latter allows the detection of cells in a non-cultivable state. Additionally, amplicon sequencing data indicated that both, RU47 and FZB42, colonized the root endosphere. Generally, roots are considered the most common entry point for the colonization of the root interior of plants, including apple (Frank et al. ). Previously, Chen et al. were able to detect the inoculants Burkholderia sp. and Pseudomonas thivervalensis in the root interior of rape 60 dpi. Lacava et al. showed that citrus tree roots ( Citrus sinensis ) inoculated with Klebsiella pneumoniae Kp342 colonized the root interior. The present study revealed two ASVs exclusively in samples of treatments B (ASV1024) and P (ASV50) which were identical with genome sequences of the inoculant strains FZB42 and RU47, respectively. This indicates the potential colonization of the root endosphere by the inoculant strains. However, due to the short sequences obtained by amplicon sequencing, the approach does not allow an identification of the inoculants on species or strain level leaving some uncertainty regarding their identification. Nonetheless, considering the ability of FZB42 and RU47 to colonize soil and rhizoplane, we conclude that the inoculant strains are likely to successfully colonize the root endosphere.
The microbial community composition of ARD and grass soil from the same experimental site used in this study was described previously for different microhabitats (Balbín-Suárez et al. , ; Mahnkopp-Dirks et al. ). As described before (Balbín-Suárez et al. ), the composition of bacteria was mainly driven by the microhabitat and significantly differed between rhizoplane and root-affected soil. These results support the expectation that the microhabitat marks the main differences in microbial community composition (Reinhold-Hurek et al. ; van der Heijden and Schlaeppi ). Balbín-Suárez et al. identified several taxa to be higher in abundance in either ARD or grass soil and identified taxa that uniquely responded to the respective soil. In ARD soil or rhizoplane, they identified Burkholderia , Variovorax , Streptomyces , and Nectria in increased relative abundance. The present study showed ASVs affiliated with Burkholderia-Caballeronia-Paraburkholderia , Thelonectria , and Ilyonectria to be of significantly higher relative abundance in the rhizoplane of plants grown in ARD soils. The potential contribution of Nectriaceae to the ARD disease complex has recently been studied intensively (Popp et al. , ) and is further supported by the notably high occurrence of four ASVs (fungal ASV1-ASV4 in rhizoplane) belonging to the genus Thelonectria and ASV40 belonging to the genus Ilyonectria in the present dataset. Our study found ASVs affiliated to the order Enterobacteriaceae highly abundant in soil and rhizoplane of apple plants grown in ARD soil. Enterobacteriaceae were not observed before in rhizoplane and soil originating from the same site (Balbín-Suárez et al. , ), but one ASV identified as Enterobacteriaceae was found in the root endosphere of apple grown in ARD soil from the same site that negatively correlated with shoot growth and shoot fresh mass (Mahnkopp-Dirks et al. ). In the rhizosphere and the root endosphere of apple trees affected by bitter rot and leaf spot disease in Brazil, Dos Passos et al. found cultivable Enterobacter to be the most abundant genus. Members of the family Enterobacteriaceae can indirectly influence a plant’s defense response by supporting plant pathogens and have the potential to decompose plant tissue (Berg et al. ). In the treatments with the bioinoculants a lower relative abundances of Enterobacteriaceae than non-inoculated controls which could be an indicator for successful modulation of bacteria that accumulate in ARD-affected soils. Certain ASVs of Rhizobium and Streptomyces were previously reported to occur in increased relative abundances in apple roots grown in ARD soil (Mahnkopp-Dirks et al. ). In the present study, Streptomyces was one of the most abundant taxa in the root endosphere, with ASV29 being significantly higher in relative abundance in roots from ARD than in grass soil. Our study revealed a strikingly lower relative abundance of Delftia of almost 0% in ARD soils, which was not observed before. Endophytic strains of the genus Delftia are commonly associated with plant beneficial traits (Han et al. ; Woźniak et al. , ; Da Silveira et al. ; Chen et al. ; Islam et al. ). We assume that a reduction of endophytic Delftia in ARD-affected apple roots might contribute to the plant growth depression. Similarly, also the relative abundance of Bosea was remarkably low with almost 0% in ARD soils, but was found in relative abundances of 2.5% or higher in grass soils. The genus Bosea belongs to the novel family Bosaceae that recently emanated from the family Bradyrhizobiaceae (Hördt et al. ) . Most members of Bosaceae were isolated from legume nodules pointing towards a close association between Bosea and legumes. Also, the co-existence of endophytic Bosea spatocytisi sp. nov. with rhizobia in root nodules was observed (Pulido-Suárez et al. ). The composite mix of plants used as grass cover in our study site included Trifolium , which might explain why Bosea was observed exclusively in samples from grass, but not ARD soils. In rhizoplane of grass soil, e.g., Sphingobium and Novosphingobium occurred in significantly higher relative abundance than in rhizoplane of ARD soil. Several members of the genera Sphingobium and Novosphingobium are well-known environmental strains involved in bioremediation and biodegradation (Yang et al. ; Boss et al. ). Also, various isolates of the family Sphingomonadaceae were identified as PGPR with the potential to produce phytobeneficial compounds or to increase root and shoot length when applied as inoculant (Yang et al. ; Krishnan et al. ). The significant higher relative abundance of Sphingobium and Novosphingobium in grass soil could have promoted the root growth observed in grass soils. To validate this hypothesis, isolation, in vitro testing, and inoculation experiments using Sphingomonadaceae isolates would be needed.
Over the past years, biocontrol, inoculation of beneficial microbes, and measures to increase microbial diversity and activity have gained increasing attention as eco-friendly management options in agriculture (Berg et al. ; Tosi et al. ; Müller and Behrendt ). Recently, microbiome modulation was introduced as an effective mode of action for microbial inoculants (Berg et al. ). Six different types of microbiome modulation were formulated including the restoration of a dysbiosis, the targeted shift towards potentially beneficial taxa, or the depletion of potential pathogens (Berg et al. ). These authors summarized that the degree of microbiome modulation highly depends on the sampling time and mode of inoculant application and that shifts are usually only evident shortly after inoculation. This might explain why in our study no or only little effects of the inoculants on the bacterial ß-diversity were observed 28 dpi. However, fungal ß-diversity was significantly affected by both inoculants likely due to the antifungal activities of both inoculant strains. For instance, the potential to suppress the fungal plant pathogen Rhizoctonia solani was previously demonstrated for both, FZB42 and RU47 (Chowdhury et al. ; Schreiter et al. ). The significant shifts in fungal β-diversity as well as changes in relative abundances of the dominant fungal taxa indicate that FZB42 and RU47 modulated the fungal microbiome in both, ARD and grass soil. These results give perspective for future experiments in which the potential of the inoculants to suppress ARD-associated microorganisms like members of Ilyonectria , Thelonectria , or Pythium should be evaluated not only by amplicon sequencing-based approaches, but by in vitro studies following Koch’s postulates and molecular quantification tools such as qPCR assays targeting taxa that potentially contribute to ARD. Interestingly, regarding the dominant bacteria, only in root-affected soil and rhizoplane of ARD soil, many taxa that were detected changed significantly in relative abundance due to the inoculation while no significant changes in relative abundance were observed for any of the 20 most abundant taxa in both microhabitats from grass soil. We assume that due to the imbalanced microbiome of the ARD-affected soil and rhizoplane, the microbiome was less stable and could be modulated more easily.
The production of phytoalexins is a common plant defense strategy to combat pathogen invasion. Apple and other rosaceous plants, in particular, form biphenyls and dibenzofurans to inhibit microbial growth and cell propagation in a local environment around the plant (Chizzali and Beerhues ; Busnena et al. ). Balbín-Suárez et al. showed that biphenyl and dibenzofuran phytoalexins predominantly and significantly accumulate in diseased roots of apple grown in ARD soils. This result was confirmed in the present study as significantly higher PA contents were observed in the roots of apple plants grown in ARD soil compared to grass soil. However, our results showed that phytoalexins were also induced by the inoculation of FZB42 or RU47 in the roots of apple plants especially in grass soil. Hence, the overall production of biphenyl and dibenzofuran phytoalexins seems not a specific response of apple roots to ARD stress but indicating an unbalanced microbiome. Phytoalexin production upon exposure to microbes is a general defense response of plants, which is also true for apple (Busnena et al. ). In , Weiß et al. reported for the first time that apple roots form biphenyl and dibenzofuran phytoalexins upon microbial stress, here ARD soil. However, production of these phytoalexins after application of microbial inoculants, such as the two bacteria used in this study, has not been reported before. In the few years from 2017 up to now, to our knowledge, there was no report about the formation of the phytoalexins in apple roots in response to inoculants. We assume that the inoculation of high cell or spore numbers disturbed the previous equilibrium of the native soil microbial communities in grass soil, which might have led to the induction of the apple roots’ stress response. The increase in phytoalexin content in the FZB42 and RU47 treatments was weaker in the roots of plants grown in dysbiotic ARD soil. Further detailed research investigating the response of apple plants to microbial inoculants applied in different cell or spore densities is needed to answer this question. Analyzing root length in the different soils and treatments did not reveal significant differences. Most likely, this can be explained by high plant-to-plant variations, impeding the detection of significant differences. However, we observed some trends that might indicate a link to phytoalexin contents. Twenty-eight dpi, roots that were inoculated had significantly higher phytoalexin contents compared to the respective non-inoculated controls in grass soil. We assume that the high phytoalexin content in roots of inoculated plants might have affected root growth. However, as the results regarding the root length were not significant, these observations need to be considered with care. Moreover, inoculation led to significantly higher proportions of thicker roots compared to the control only in grass soil which might have been caused by changes in the microbial community composition and activity. High ethylene concentrations are known to increase cortex width and hence root diameters (Gebauer et al. ). Here, it can only be speculated whether the differences in root diameter classes observed in the present experiment between grass soil and ARD soil are related to the documented differences in microbiome composition and their potential functional differences in hormone production. In summary, we explored the potential of two microbial inoculants, B. velezensis FZB42 and Pseudomonas sp. RU47, on apple, to mitigate apple replant disease. We used inoculation techniques that enabled the inoculants to colonize different root-associated microhabitats of apple, which at the same time, can be easily implemented in current horticultural practices. As a first step, we showed that both inoculant strains had the potential to establish across microhabitats, to modulate the microbiome and to induce shifts in the relative abundance of dominant bacterial and fungal ASVs. The effects of microhabitat and replanting history of the soils on the bacterial community were stronger than the inoculation effect. To our surprise, the inoculation effect of FZB42 on the fungal community composition was stronger than the effect of replanting history likely due to the strong antifungal capacity of FZB42, potentially enabling the suppression of ARD-associated fungi. The inoculants decreased the relative abundance of ARD-related Enterobacteriaceae . Unexpectedly, inoculation increased phytoalexin content in roots of apple plants grown in grass soil. Whether increased phytoalexin contents can be linked to changes in root morphology that were observed needs additional investigation. To further unravel the interplay between inoculants and plants, and in particular their potential to reduce ARD, further field experiments and additional methods such as the measurement of volatile organic compounds or functional microbiome analysis, are needed. It will also be essential to investigate long-term effects of the inoculants in field trials.
Below is the link to the electronic supplementary material. Supplementary file1 (PDF 1.08 MB)
|
Erector spinae plane block spread patterns and its analgesic effects after computed tomography-guided hepatic tumour ablation: a randomized double-blind trial | 00dd297a-f773-459c-8509-b576e9889285 | 11924262 | Musculoskeletal System[mh] | The erector spinae plane block (ESPB) is a relatively new regional anaesthesia technique that has gained popularity in clinical practice due to its technical simplicity, analgesic efficacy, and low incidence of complications . However, there is controversy surrounding the drug spread pattern of ESPB, which is crucial for optimizing block performance and ensuring the safe and effective delivery of analgesia. Numerous cadaveric studies have focused on the issue of drug spread in ESPB. However, relying solely on these cadaveric studies is inadequate for establishing the actual drug spread pattern of this novel technique, as the drug spread in the fascial plane of living subjects can significantly differ from that observed in cadaveric models . Computed tomography (CT)-guided percutaneous radiofrequency ablation (RFA) is a commonly used treatment option for hepatic tumours in patients who are not suitable for surgery. However, patients undergoing this procedure frequently experience post-procedural pain due to inflammatory responses and necrosis of the ablated hepatic tissues . ESPB has been reported to reduce postoperative pain after laparoscopic cholecystectomy and hepatic surgery . Furthermore, ESPB has recently been reported in association with percutaneous ablation techniques like RFA or microwaves for hepatic tumour treatment . Although literature supports the analgesic efficacy of ESPB for painful hepatobiliary procedures, its spread pattern has not been thoroughly investigated. Since patient-reported sensory loss following the block may not always accurately reflect the actual spread of ESPB , concurrently assessing both its analgesic effects and spread pattern could provide valuable clinical insights. Given that CT imaging is routinely performed during CT-guided RFA, this procedure offers a unique opportunity to investigate both the spread pattern of local anaesthetics following ESPB in living subjects and its analgesic efficacy during RFA, without exposing participants to additional radiation. Therefore, we hypothesized that ESPB would result in extensive drug spread in living subjects and provide effective analgesia for RFA in hepatic tumours. Consequently, we designed this double-blinded randomized controlled trial to evaluate its efficacy and distribution.
Study design, ethics and trial registration This study was designed to investigate the spread of a 30 mL solution consisting of contrast with local anaesthetics or normal saline following a right ESPB block of T10 level, and the effects of post-procedural analgesia in patients who received computed tomography-guided radiofrequency ablation of hepatic tumours. This double-blinded, prospective, randomized, sham-controlled single-institution trial received approval from the Research Ethics Committee of National Taiwan University Hospital (Approval No. 202101007RINA) and written informed consent was obtained from all subjects participating in the trial. The trial was registered prior to patient enrolment at clinicaltrials.gov (NCT04837742; Principal investigator: Ming-Shiang Wu; Date of first submitted: 6 April 2021; First posted: 8 April 2021). Participants The study recruited patients aged from 20 years to 85 years who were undergoing CT-guided hepatic tumour RFA with a tumour size larger than 2 cm between 14 April 2021 and 18 January 2023. Patients with the following conditions were excluded: a history of allergic reactions to local anaesthetics or contrast agents, abnormal kidney function, defined as an estimated glomerular filtration rate <30 mL min −1 1.73 m −2 , and coagulopathy. We conducted the study in accordance with the Declaration of Helsinki and its later amendments or comparable ethical standards, adhering to the applicable CONSORT guidelines. Randomization and blinding Before the trial began, stratified randomization was performed by an independent statistical expert using a block size of 30 and 1:1 allocation. All patients provided written informed consent on the day before the RFA procedure to an investigator who was unaware of the randomization results. They were then equally and randomly assigned to either the ESPB group (patients who received the ESPB injection with local anaesthetics) or the sham group (patients who received the ESPB injection with an equivalent volume of normal saline). The masked drug was provided by an independent pharmacy, ensuring that the allocation was concealed from the investigators and clinicians. Therefore, the blinding was maintained for patients, clinical care providers, and the outcome investigators. Interventions and the rationales of the inclusion of sham group This study was designed to include the sham group to differentiate between the physiological effects of ESPB, placebo-related effects and systemic analgesic effects of the fascial plane block for improving the quality of blinding and reducing bias . Anaesthesia and ESPB Each patient underwent standard intraoperative monitoring using a Philips IntelliVue MP70 monitor (Philips Medical Systems, Suresnes, France). After being placed in the left lateral decubitus position, the patients received 50 μg of intravenous fentanyl. The right-sided T10 spine level was identified by counting up from the L5/S1 junction, which was determined by an indentation of the reflective surface using the low-frequency curved probe. This level was chosen to potentially cover the insertion site, as well as the greater and lesser splanchnic nerves . Before the procedure, pre-scanning was performed using both transverse and sagittal views to identify bony contours, including spinous processes, laminae, transverse processes, costotransverse junction, ribs, and erector spinae muscles at the right-sided T10 spine level. Upon recognizing the lateral edge of T10 transverse process, a 23-gauge needle (“NIPRO” KATERAN needle, 70 mm in length) was inserted in a caudal-to-cranial direction until it came into contact with the tip of the T10 transverse process. Next, the experimental drug was administered, which consisted of a mixture of 10 mL contrast media (Iohexol- Omnipaque TM , GE Healthcare, Chicago, IL, USA) with 20 mL 0.5% levobupivacaine for the ESPB group or a mixture of 10 mL contrast media (Iohexol- Omnipaque TM , GE Healthcare, Chicago, IL, USA) with 20 mL normal saline for the sham group, respectively. The distribution of the local anaesthetics was confirmed by observing the plane between the transverse process of the vertebra and the erector spinae muscle. After the completion of the ESPB, patients were sedated with a target-controlled propofol infusion (Schnider model) to maintain an effect site concentration of 2.5 to 3.5 μg/mL. During the RFA procedure, the attending anaesthesiologist, who was unaware of the group allocation, could administer opioids (fentanyl or remifentanil at the anaesthesiologist’s discretion) for the management of intraoperative analgesia. Percutaneous RFA was performed under CT guidance in all patients by the same radiologist, using a single radiofrequency ablation electrode with a 200-W generator . Subsequently, a low-dose CT scan with a minimalized scan range was obtained after adjusting each electrode. Before the RFA procedure, a whole spine CT scan was performed to evaluate the spread of ESPB in the following regions: erector spinae muscle plane, paravertebral space, intercostal space, and epidural space. All CT images were reviewed and analysed by the same independent pain specialist who had experience in interpreting spine CT images. Additionally, the thickness of the back muscle was measured to investigate potential correlations between the muscle thickness and the extent of ESPB spreads. Post-procedural pain assessment and management Post-procedural pain intensity was assessed by investigators who were independent of the clinical care team. They utilized a 100-mm visual analogue scale (VAS) to evaluate the highest pain intensity experienced by patients during either deep breaths or cough at 1 h and 24 h after the RFA procedure. During each assessment, both the investigator and the participant were blinded to the previous VAS questionnaire results to maintain objectivity. The results of the VAS were used for outcome analysis purposes only and were not utilized for clinical care decisions. To manage post-procedural pain, boluses of intravenous morphine (2 mg) were administered upon the patient’s request, aiming to maintain pain score < 4 (using 0–10 numeric rating pain score by the caring staff) in the post-anaesthetic care unit. For rescue analgesia, intravenous morphine (2–4 mg every 3 h) was provided upon patient request in the general ward. Additionally, oral acetaminophen was provided to each patient in the general ward every 6 h for 24 h following the RFA. Furthermore, the postoperative quality of recovery at 24 h was evaluated using the Quality of Recovery-15 (QoR-15) questionnaire . Outcomes The primary outcomes of thIS study were to determine the drug spread pattern and the highest pain intensity, assessed using the 100-mm Visual Analogue Scale (VAS), at 24 h after the RFA procedure. Secondary outcome included the highest pain intensity at 1 hr, intraoperative fentanyl, postoperative analgesic usage and quality of recovery (QoR-15 score). Sample size calculation Based on pilot data from patients undergoing RFA of hepatic tumours, the mean (SD) VAS score at 24 h after RFA was approximately 45 (30) mm. To detect a difference in mean VAS score of 35 mm (approximately a 80% reduction in pain intensity) with 80% power and a two-sided type I error of 0.05, a sample size of 26 patients (13 patients in each group) was calculated. To account for potential attrition, a total of 30 patients were enrolled in the study. This sample size was also considered adequate for evaluating the drug spread pattern, as it exceeded the number used in most previous studies investigating ESPB spread . Statistical analysis The normality of the distribution was assessed using the Shapiro–Wilk test, and visual inspections were conducted using histograms. Continuous variables are presented as mean (standard deviation) or median (interquartile range), depending on the distribution. For dichotomous data, Fisher’s exact test or the chi-square test was used for analysis. Student’s t -test was applied to normally distributed continuous data, while the Mann–Whitney U test was used for nonparametric ordinal data. To investigate correlations between the numbers of vertebral spread levels and the back muscle thickness at the injection level, as well as patient characteristics like body mass index, Pearson’s correlation test was utilized. Statistical analyses were conducted using the PASS Sample Size Software (NCSS, LLC, Kaysville, Utah, USA) and MedCalc Statistical Software version 19.3.1 (MedCalc Software Ltd., Ostend, Belgium).
This study was designed to investigate the spread of a 30 mL solution consisting of contrast with local anaesthetics or normal saline following a right ESPB block of T10 level, and the effects of post-procedural analgesia in patients who received computed tomography-guided radiofrequency ablation of hepatic tumours. This double-blinded, prospective, randomized, sham-controlled single-institution trial received approval from the Research Ethics Committee of National Taiwan University Hospital (Approval No. 202101007RINA) and written informed consent was obtained from all subjects participating in the trial. The trial was registered prior to patient enrolment at clinicaltrials.gov (NCT04837742; Principal investigator: Ming-Shiang Wu; Date of first submitted: 6 April 2021; First posted: 8 April 2021).
The study recruited patients aged from 20 years to 85 years who were undergoing CT-guided hepatic tumour RFA with a tumour size larger than 2 cm between 14 April 2021 and 18 January 2023. Patients with the following conditions were excluded: a history of allergic reactions to local anaesthetics or contrast agents, abnormal kidney function, defined as an estimated glomerular filtration rate <30 mL min −1 1.73 m −2 , and coagulopathy. We conducted the study in accordance with the Declaration of Helsinki and its later amendments or comparable ethical standards, adhering to the applicable CONSORT guidelines.
Before the trial began, stratified randomization was performed by an independent statistical expert using a block size of 30 and 1:1 allocation. All patients provided written informed consent on the day before the RFA procedure to an investigator who was unaware of the randomization results. They were then equally and randomly assigned to either the ESPB group (patients who received the ESPB injection with local anaesthetics) or the sham group (patients who received the ESPB injection with an equivalent volume of normal saline). The masked drug was provided by an independent pharmacy, ensuring that the allocation was concealed from the investigators and clinicians. Therefore, the blinding was maintained for patients, clinical care providers, and the outcome investigators.
This study was designed to include the sham group to differentiate between the physiological effects of ESPB, placebo-related effects and systemic analgesic effects of the fascial plane block for improving the quality of blinding and reducing bias .
Each patient underwent standard intraoperative monitoring using a Philips IntelliVue MP70 monitor (Philips Medical Systems, Suresnes, France). After being placed in the left lateral decubitus position, the patients received 50 μg of intravenous fentanyl. The right-sided T10 spine level was identified by counting up from the L5/S1 junction, which was determined by an indentation of the reflective surface using the low-frequency curved probe. This level was chosen to potentially cover the insertion site, as well as the greater and lesser splanchnic nerves . Before the procedure, pre-scanning was performed using both transverse and sagittal views to identify bony contours, including spinous processes, laminae, transverse processes, costotransverse junction, ribs, and erector spinae muscles at the right-sided T10 spine level. Upon recognizing the lateral edge of T10 transverse process, a 23-gauge needle (“NIPRO” KATERAN needle, 70 mm in length) was inserted in a caudal-to-cranial direction until it came into contact with the tip of the T10 transverse process. Next, the experimental drug was administered, which consisted of a mixture of 10 mL contrast media (Iohexol- Omnipaque TM , GE Healthcare, Chicago, IL, USA) with 20 mL 0.5% levobupivacaine for the ESPB group or a mixture of 10 mL contrast media (Iohexol- Omnipaque TM , GE Healthcare, Chicago, IL, USA) with 20 mL normal saline for the sham group, respectively. The distribution of the local anaesthetics was confirmed by observing the plane between the transverse process of the vertebra and the erector spinae muscle. After the completion of the ESPB, patients were sedated with a target-controlled propofol infusion (Schnider model) to maintain an effect site concentration of 2.5 to 3.5 μg/mL. During the RFA procedure, the attending anaesthesiologist, who was unaware of the group allocation, could administer opioids (fentanyl or remifentanil at the anaesthesiologist’s discretion) for the management of intraoperative analgesia. Percutaneous RFA was performed under CT guidance in all patients by the same radiologist, using a single radiofrequency ablation electrode with a 200-W generator . Subsequently, a low-dose CT scan with a minimalized scan range was obtained after adjusting each electrode. Before the RFA procedure, a whole spine CT scan was performed to evaluate the spread of ESPB in the following regions: erector spinae muscle plane, paravertebral space, intercostal space, and epidural space. All CT images were reviewed and analysed by the same independent pain specialist who had experience in interpreting spine CT images. Additionally, the thickness of the back muscle was measured to investigate potential correlations between the muscle thickness and the extent of ESPB spreads.
Post-procedural pain intensity was assessed by investigators who were independent of the clinical care team. They utilized a 100-mm visual analogue scale (VAS) to evaluate the highest pain intensity experienced by patients during either deep breaths or cough at 1 h and 24 h after the RFA procedure. During each assessment, both the investigator and the participant were blinded to the previous VAS questionnaire results to maintain objectivity. The results of the VAS were used for outcome analysis purposes only and were not utilized for clinical care decisions. To manage post-procedural pain, boluses of intravenous morphine (2 mg) were administered upon the patient’s request, aiming to maintain pain score < 4 (using 0–10 numeric rating pain score by the caring staff) in the post-anaesthetic care unit. For rescue analgesia, intravenous morphine (2–4 mg every 3 h) was provided upon patient request in the general ward. Additionally, oral acetaminophen was provided to each patient in the general ward every 6 h for 24 h following the RFA. Furthermore, the postoperative quality of recovery at 24 h was evaluated using the Quality of Recovery-15 (QoR-15) questionnaire .
The primary outcomes of thIS study were to determine the drug spread pattern and the highest pain intensity, assessed using the 100-mm Visual Analogue Scale (VAS), at 24 h after the RFA procedure. Secondary outcome included the highest pain intensity at 1 hr, intraoperative fentanyl, postoperative analgesic usage and quality of recovery (QoR-15 score).
Based on pilot data from patients undergoing RFA of hepatic tumours, the mean (SD) VAS score at 24 h after RFA was approximately 45 (30) mm. To detect a difference in mean VAS score of 35 mm (approximately a 80% reduction in pain intensity) with 80% power and a two-sided type I error of 0.05, a sample size of 26 patients (13 patients in each group) was calculated. To account for potential attrition, a total of 30 patients were enrolled in the study. This sample size was also considered adequate for evaluating the drug spread pattern, as it exceeded the number used in most previous studies investigating ESPB spread .
The normality of the distribution was assessed using the Shapiro–Wilk test, and visual inspections were conducted using histograms. Continuous variables are presented as mean (standard deviation) or median (interquartile range), depending on the distribution. For dichotomous data, Fisher’s exact test or the chi-square test was used for analysis. Student’s t -test was applied to normally distributed continuous data, while the Mann–Whitney U test was used for nonparametric ordinal data. To investigate correlations between the numbers of vertebral spread levels and the back muscle thickness at the injection level, as well as patient characteristics like body mass index, Pearson’s correlation test was utilized. Statistical analyses were conducted using the PASS Sample Size Software (NCSS, LLC, Kaysville, Utah, USA) and MedCalc Statistical Software version 19.3.1 (MedCalc Software Ltd., Ostend, Belgium).
Study population The original target sample size was set at 80 patients to identify potential differences in QoR-15 scores. However, enrolment was significantly impacted by the SARS-CoV-2 pandemic, and the final sample size was adjusted to focus on comparing the primary outcome, namely the difference in VAS scores, which requires a smaller sample size. Between August 2021 and May 2023, a total of 44 patients were initially assessed for inclusion in the study. After the screening process, 30 patients were ultimately included for the final analysis (see ). The baseline characteristics of the patients in the two study groups were comparable and showed no significant differences ( ). The time interval between the ESPB injection and the whole spine CT scan was 46.8 (11.8) min. However, it is worth noting that patients in the ESPB group had statistically non-significantly larger tumour size compared to the sham group (3.4 ± 1.6 cm in the ESPB group vs. 2.5 ± 1.0 cm in the sham group; p = 0.080). Primary outcome analysis Spread patterns of the erector spinae plane block and provide a summary of the profiles of drug spreads to the dorsal erector spinae muscle, intercostal space, paravertebral space, and epidural space. The cranio-caudal dorsal spread of the drug to the dorsal erector spinae muscle was observed in each patient, with a median (Q 1, Q 3 ) spread of 9 (8–11) vertebral levels ( ). Similarly, the drug spread to the intercostal space was observed in each patient, with a median (Q 1, Q 3 ) spread of 4 (3–6) vertebral levels ( ). The paravertebral spread was observed in 27 out of 30 (90%) patients, with a median (Q 1, Q 3 ) spread of 3 (2–5) vertebral levels ( ). On the other hand, epidural spread was only observed in 11 out of 30 (36.7%) patients, with a median (Q 1, Q 3 ) spread of 0 (0–2) vertebral levels ( ). Overall, there is a greater distribution of the contrast on the cranial side compared to the caudal side, starting from the needle insertion point ( ). This distribution pattern aligns with the direction of the needle insertion. The CT images depicting these patterns of spread are illustrated in . The mean (SD) thickness of the back muscle at the T10 level was 34.3 (5.6) mm. The number of vertebral levels of cranio-caudal erector spinae muscle spread was negatively correlated with the thickness of the back muscle at the T10 level ( r = −0.4; p = 0.035). However, there was no significant correlation between the number of vertebral levels of cranio-caudal erector spinae muscle spread and the number of epidural spread ( r = −0.3; p = 0.076), the number of paravertebral spread ( r = −0.4; p = 0.058), or the number of intercostal spread ( r = −0.1; p = 0.528). Furthermore, the number of vertebral levels of intercostal spread was negatively correlated with the height of the patients ( r = −0.5; p < 0.001), and the number of vertebral levels of epidural spread was negatively correlated with the weight of the patients ( r = −0.4; p = 0.030). Additionally, it was observed that female patients had a significantly higher number of intercostal spread levels compared to male patients (5.8 ± 1.0 vs. 4.3 ± 1.6 levels in female and male patients, respectively; p = 0.021). 24-hr pain intensity There were no significant differences between the two study groups in terms of the highest VAS scores at 24 h after the RFA procedure. Procedural data and secondary outcomes provides a summary of the procedural data and postoperative analgesic profiles of the two study groups. The procedural time was not significantly different between the ESPB group and the sham group [74.9 (34.4) min vs. 71.1 (29.5) min, respectively; p = 0.756]. However, patients in the sham group received a significantly lower median (Q1, Q3) fentanyl equivalent dose during the RFA procedure compared to the ESPB group [100 (100–215) μg vs. 78 (50–100) μg, respectively; p = 0.041]. Regarding postoperative analgesic profiles, there were no significant differences between the two study groups in terms of the highest VAS scores at 1 h and 24 h after the RFA procedure, the 24-h morphine dose, and the proportion of patients who requested no morphine ( ). Furthermore, the 24-h QoR-15 scores, which indicate postoperative quality of recovery, were comparable between the ESPB group and the sham group.
The original target sample size was set at 80 patients to identify potential differences in QoR-15 scores. However, enrolment was significantly impacted by the SARS-CoV-2 pandemic, and the final sample size was adjusted to focus on comparing the primary outcome, namely the difference in VAS scores, which requires a smaller sample size. Between August 2021 and May 2023, a total of 44 patients were initially assessed for inclusion in the study. After the screening process, 30 patients were ultimately included for the final analysis (see ). The baseline characteristics of the patients in the two study groups were comparable and showed no significant differences ( ). The time interval between the ESPB injection and the whole spine CT scan was 46.8 (11.8) min. However, it is worth noting that patients in the ESPB group had statistically non-significantly larger tumour size compared to the sham group (3.4 ± 1.6 cm in the ESPB group vs. 2.5 ± 1.0 cm in the sham group; p = 0.080).
Spread patterns of the erector spinae plane block and provide a summary of the profiles of drug spreads to the dorsal erector spinae muscle, intercostal space, paravertebral space, and epidural space. The cranio-caudal dorsal spread of the drug to the dorsal erector spinae muscle was observed in each patient, with a median (Q 1, Q 3 ) spread of 9 (8–11) vertebral levels ( ). Similarly, the drug spread to the intercostal space was observed in each patient, with a median (Q 1, Q 3 ) spread of 4 (3–6) vertebral levels ( ). The paravertebral spread was observed in 27 out of 30 (90%) patients, with a median (Q 1, Q 3 ) spread of 3 (2–5) vertebral levels ( ). On the other hand, epidural spread was only observed in 11 out of 30 (36.7%) patients, with a median (Q 1, Q 3 ) spread of 0 (0–2) vertebral levels ( ). Overall, there is a greater distribution of the contrast on the cranial side compared to the caudal side, starting from the needle insertion point ( ). This distribution pattern aligns with the direction of the needle insertion. The CT images depicting these patterns of spread are illustrated in . The mean (SD) thickness of the back muscle at the T10 level was 34.3 (5.6) mm. The number of vertebral levels of cranio-caudal erector spinae muscle spread was negatively correlated with the thickness of the back muscle at the T10 level ( r = −0.4; p = 0.035). However, there was no significant correlation between the number of vertebral levels of cranio-caudal erector spinae muscle spread and the number of epidural spread ( r = −0.3; p = 0.076), the number of paravertebral spread ( r = −0.4; p = 0.058), or the number of intercostal spread ( r = −0.1; p = 0.528). Furthermore, the number of vertebral levels of intercostal spread was negatively correlated with the height of the patients ( r = −0.5; p < 0.001), and the number of vertebral levels of epidural spread was negatively correlated with the weight of the patients ( r = −0.4; p = 0.030). Additionally, it was observed that female patients had a significantly higher number of intercostal spread levels compared to male patients (5.8 ± 1.0 vs. 4.3 ± 1.6 levels in female and male patients, respectively; p = 0.021). 24-hr pain intensity There were no significant differences between the two study groups in terms of the highest VAS scores at 24 h after the RFA procedure. Procedural data and secondary outcomes provides a summary of the procedural data and postoperative analgesic profiles of the two study groups. The procedural time was not significantly different between the ESPB group and the sham group [74.9 (34.4) min vs. 71.1 (29.5) min, respectively; p = 0.756]. However, patients in the sham group received a significantly lower median (Q1, Q3) fentanyl equivalent dose during the RFA procedure compared to the ESPB group [100 (100–215) μg vs. 78 (50–100) μg, respectively; p = 0.041]. Regarding postoperative analgesic profiles, there were no significant differences between the two study groups in terms of the highest VAS scores at 1 h and 24 h after the RFA procedure, the 24-h morphine dose, and the proportion of patients who requested no morphine ( ). Furthermore, the 24-h QoR-15 scores, which indicate postoperative quality of recovery, were comparable between the ESPB group and the sham group.
and provide a summary of the profiles of drug spreads to the dorsal erector spinae muscle, intercostal space, paravertebral space, and epidural space. The cranio-caudal dorsal spread of the drug to the dorsal erector spinae muscle was observed in each patient, with a median (Q 1, Q 3 ) spread of 9 (8–11) vertebral levels ( ). Similarly, the drug spread to the intercostal space was observed in each patient, with a median (Q 1, Q 3 ) spread of 4 (3–6) vertebral levels ( ). The paravertebral spread was observed in 27 out of 30 (90%) patients, with a median (Q 1, Q 3 ) spread of 3 (2–5) vertebral levels ( ). On the other hand, epidural spread was only observed in 11 out of 30 (36.7%) patients, with a median (Q 1, Q 3 ) spread of 0 (0–2) vertebral levels ( ). Overall, there is a greater distribution of the contrast on the cranial side compared to the caudal side, starting from the needle insertion point ( ). This distribution pattern aligns with the direction of the needle insertion. The CT images depicting these patterns of spread are illustrated in . The mean (SD) thickness of the back muscle at the T10 level was 34.3 (5.6) mm. The number of vertebral levels of cranio-caudal erector spinae muscle spread was negatively correlated with the thickness of the back muscle at the T10 level ( r = −0.4; p = 0.035). However, there was no significant correlation between the number of vertebral levels of cranio-caudal erector spinae muscle spread and the number of epidural spread ( r = −0.3; p = 0.076), the number of paravertebral spread ( r = −0.4; p = 0.058), or the number of intercostal spread ( r = −0.1; p = 0.528). Furthermore, the number of vertebral levels of intercostal spread was negatively correlated with the height of the patients ( r = −0.5; p < 0.001), and the number of vertebral levels of epidural spread was negatively correlated with the weight of the patients ( r = −0.4; p = 0.030). Additionally, it was observed that female patients had a significantly higher number of intercostal spread levels compared to male patients (5.8 ± 1.0 vs. 4.3 ± 1.6 levels in female and male patients, respectively; p = 0.021).
There were no significant differences between the two study groups in terms of the highest VAS scores at 24 h after the RFA procedure.
provides a summary of the procedural data and postoperative analgesic profiles of the two study groups. The procedural time was not significantly different between the ESPB group and the sham group [74.9 (34.4) min vs. 71.1 (29.5) min, respectively; p = 0.756]. However, patients in the sham group received a significantly lower median (Q1, Q3) fentanyl equivalent dose during the RFA procedure compared to the ESPB group [100 (100–215) μg vs. 78 (50–100) μg, respectively; p = 0.041]. Regarding postoperative analgesic profiles, there were no significant differences between the two study groups in terms of the highest VAS scores at 1 h and 24 h after the RFA procedure, the 24-h morphine dose, and the proportion of patients who requested no morphine ( ). Furthermore, the 24-h QoR-15 scores, which indicate postoperative quality of recovery, were comparable between the ESPB group and the sham group.
In this study, we demonstrated the drug spread patterns in a relatively large series of living subjects and revealed that a high proportion of participants exhibited patterns of both dorsal and anterior spreads. However, we observed that the unilateral ESPB was ineffective in alleviating the pain intensity after RFA of the hepatic tumour. An increasing number of studies have investigated the spread patterns of ESPB in larger living cohorts [ , , ]. For instance, Abdellav et al. assessed the injectate spread in 60 patients who received a T4 ESPB using CT imaging and reported a limited extent of both paravertebral and epidural spread . The differences in ESPB injectate distribution within paravertebral space between Abdellav et al.’s study and this study may be attributed to variations in the injection site (T4 vs. T10) and the time interval between injection and CT imaging (15 min vs. 47 min). In contrast, Sørenstua et al. recently evaluated the spread pattern of ESPB in 10 healthy volunteers using 30 mL of diluted ropivacaine at the T7 level . Their findings revealed paravertebral spread in 9 out of 10 participants, with a median spread of 4 levels; intercostal spread in all participants, with a median spread of 5.5 levels; and epidural spread in 4 out of 10 participants, with a median spread of 0 levels. These results align with the proportions of ESPB spread reported in this study. Additionally, Shan et al. assessed injectate spread in 84 patients who received a T7 ESPB under CT imaging and observed a similar extent of paravertebral and epidural spread as reported in our study . By integrating the findings from Sørenstua et al. , Shan et al. and our present study, we concluded that the drug spread patterns may be similar at the mid (T7) and low thoracic levels (T10) in living subjects. These findings of living subjects could be valuable in planning the execution of ESPB in clinical practices. Compared to the aforementioned studies conducted in living subjects, this study differs in several aspects. Firstly, the CT-guided RFA model employed in this research minimized radiation exposure to the participants by utilizing low-dose CT scans. Secondly, in the three aforementioned studies conducted in living subjects, ESPB was performed in the prone position, whereas in this study, it was performed in the lateral position. Thirdly, additional analyses were conducted to examine the relationships between gender, muscle thickness, and injectate spread, which may provide insights into the differences between studies in living subjects and cadaveric studies. For instance, the anterior spreads of ESPB, including paravertebral, intercostal, and epidural spread, were not extensively reported in cadaveric studies. Among the 16 cadaveric studies on thoracic ESPB, only 11 found evidence of paravertebral dye penetration . In addition, we observed wide cranio-caudal spread patterns, with a median spread across nine vertebral levels in this study. This spread pattern was found to be generally broader than those reported in cadaveric studies, which revealed a cranio-caudal spread of 3–6 vertebral levels . Furthermore, we noted that the drug spread was more pronounced toward the cranial end compared to the caudal end, aligning with the direction of needle insertion. Live subjects’ fascial planes and compartments are significantly influenced by dynamic forces, as muscles and fasciae slide over each other, potentially contributing to a wider transmission of the ESPB drug injection’s pressure. Interestingly, we also observed a significantly negative correlation between the back muscle thickness and the extent of ESPB dorsal spread. This finding emphasizes the potential differences in drug spread patterns between cadaveric models and living subjects. It is worth noting that the observed median numbers of levels of intercostal spread and paravertebral spread in this study were lower than those reported in case series studies of living subjects who received ESPB injections at the same T10 level. For example, Schwartzmann et al. revealed a paravertebral spread of seven levels and an intercostal spread of six levels in a female patient after the T10 ESPB injection . The same research group also reported injectate spreads of 9 [5–12] and 3 [2–6] levels to the intercostal space and neural foramina (similar to paravertebral spreads) in a series of six female patients after the T10 ESPB injection . In contrast, our present study found that female patients may exhibit a broader intercostal spread following a T10 ESPB, while the patients in the above two studies were all female. It is essential to note that our study enrolled patients with hepatocellular cancer, and over 70% of them were male . Since male subjects generally have higher muscle tone and back muscle thickness compared to female subjects , this factor might have an impact on the extent of ESPB spread patterns. This study reported findings of analgesic effects that are discordant with two recent studies on ESPB administration for ablation of hepatic tumours. Mostafa et al. reported that ESPB alleviated post-procedural pain of radiofrequency ablation of hepatic tumours , while Gergin et al. also reported similar pain relief after microwave ablation of hepatic tumours . These conflicting findings may be attributed to several reasons. Firstly, the procedures in our study included multiple tumour ablations, resulting in significantly longer average procedure times (approximately 70 min) compared to Mostafa et al.’s study (approximately 25 min) and Gergin et al.’s study (approximately 10 min). Secondly, while we applied a sham block in participants of this study, this was not performed in the other two studies. Thirdly, variations in hepatic tumour size may significantly influence both opioid requirements during RFA and post-procedural pain intensity. For instance, in the study by Gergin et al. hepatic tumours in the ESPB group were smaller than those in the control group, which may have contributed to differences in analgesic outcomes . By contrast, patients in the ESPB group had statistically non-significantly larger tumour size than those in the sham group (approximately 1 cm larger) in this study. As a result, patients in the ESPB group in the study by Gergin et al. reported lower pain scores in the post-anaesthetic care unit, whereas patients in the ESPB group in this study required a higher, though not statistically significant, fentanyl dose during RFA. Fourthly, the patients in this study underwent complex RFA for hepatic tumours, with a more prolonged procedure time than Mostafa et al.’s and Gergin et al.’s studies , which could have resulted in a higher intensity of post-RFA visceral pain due to more hepatic tissue necrosis. The unsatisfactory analgesic of unilateral ESPB to post-RFA pain may be because of several reasons. First, pain signals from the liver are primarily transmitted via visceral afferents, which travel alongside sympathetic nerves and enter the spinal dorsal horn at the T7–T12 levels. These nerve fibres predominantly pass through the prevertebral ganglia including celiac and splanchnic ganglion, and relay signals to the central nervous system via the splanchnic nerve . Consequently, effective analgesia targeting these nerves requires deeper injections, such as a splanchnic nerve block or epidural analgesia, to achieve adequate pain relief. Since the spread of local anaesthetics in ESPB primarily affects somatic nerves, as demonstrated in this study, its impact on visceral nerves is relatively limited. Consistently, we observed a low incidence of epidural spread, which may explain why ESPB failed to alleviate post-RFA visceral pain. Second, Sørenstua et al. recently reported discrepancies between what could have been expected from the images of ESPB injectate spread and the test result of cutaneous sensation blockade . This study was compatible to the above literature regarding the discrepancy between injectate spread and the unsatisfactory analgesic effect. Several alternative techniques, beyond splanchnic nerve block and epidural analgesia, may improve post-RFA pain management. For instance, bilateral ESPB has been shown to provide superior analgesic effects, , and increasing the local anaesthetic concentration in ESPB may further enhance its efficacy . This study has several limitations. Firstly, we did not conduct the sensory test due to the sham injection performed in this study. Performing the sensory test could have violated the blinding of randomization, given the use of a sham block. Moreover, previous reports indicate that sensory loss may not accurately reflect the actual spread of local anaesthetics , making it unreliable to confidently assume clinical effect based on sensory testing results . Secondly, we observed that female patients had more intercostal spread than male patients. However, it is essential to note that this study was conducted during the SARS-CoV-2 pandemic, making patient enrolment challenging. Since the majority of patients with hepatocellular carcinoma were male , the sample size of our study included an insufficient number of female patients, and therefore, the study may have been underpowered to confirm any potential influence of gender on drug spread patterns. Thirdly, the CT image was performed approximately 47 min after the ESPB, whereas late diffusion of the injectate may occur around 60 min after ESPB injection, as previously reported .
In conclusion, this study provided valuable insights into the drug spread patterns of right T10 ESPB, with a significant number of participants demonstrating both dorsal and anterior spreads. Additionally, we observed that the thickness of the back muscle and the sex of the participants may have an impact on the extent of ESPB spread. However, it is important to note that unilateral ESPB was found to be ineffective in alleviating the pain intensity after RFA of the hepatic tumour.
Supplementary materials.doc
|
Validity of algorithms for identifying five chronic conditions in MedicineInsight, an Australian national general practice database | 648b86d0-1517-4f7a-8dfc-2effe1d43878 | 8178900 | Family Medicine[mh] | Electronic health records (EHRs) are used in primary health care settings to keep patient-level records of clinical information including diagnoses, reasons for encounters, prescriptions, observations, test results and referrals . The development of tools to extract the data contained in these EHRs has allowed for the establishment of primary health care EHR databases which have proven to be a valuable resource for health research and public health surveillance. Widely used examples from across the world include the Clinical Practice Research Datalink (CPRD) and The Health Improvement Network (THIN) database in the United Kingdom (UK), and the Canadian Primary Care Sentinel Surveillance Network (CPCSSN) . Primary health care EHR data have been used to improve our understanding of the epidemiology of diseases and the use, costs and outcomes of health care practices, as well as for disease surveillance and quality improvement in primary health care [ , , ]. In Australia, the majority of general practitioners use EHRs to manage their patient care, including writing prescriptions, ordering pathology tests and filing correspondence. A variety of EHR clinical information systems are in use, all with different data structures and terminologies. This lack of interoperability means that EHR data is not routinely shared between practices, although efforts to change this are underway with the introduction of the national cloud-based My Health Record . There has been limited use of Australian primary health care EHR data for research and surveillance, because of data access barriers . These obstacles have been overcome by the establishment of centralised repositories, which are now facilitating timely access to EHR data from Australian general practices . MedicineInsight, which has national coverage, is one of the largest and most widely used of these Australian databases. Details of this resource are described elsewhere . Briefly, MedicineInsight was established in 2011 and contains de-identified EHRs from just over 700 of Australia’s 8147 general practices . MedicineInsight focuses on practices using Best Practice [BP] ™ or Medical Director [MD] ™ , the most widely used clinical information systems in Australia (over 80 % coverage) and the most similar in structure, noting that they were designed by the same individual . A whole-of-practice data collection, containing all available EHRs in the practice’s clinical information system is conducted when a practice joins MedicineInsight. Extracted data include patient demographics and clinical data entered directly into fields within the EHR by healthcare professionals. Free text fields potentially containing identifying information, such as progress notes and correspondence, are not included in the extraction. Incremental data are extracted regularly, resulting in an updated longitudinal database in which patients within each practice can be tracked over time. Data from practices using BP and MD software are merged into a single consistent data structure, and monthly builds of the database are generated and made available for use. As is the case for many of these primary health care EHR databases, MedicineInsight contains diagnostic algorithms that use information from various EHR fields to identify whether patients have specific health conditions. Such algorithms are required because there is no single field that provides definitive information on the health conditions experienced by each patient. The MedicineInsight algorithms have been developed by NPS MedicineWise, the custodian of MedicineInsight, to create efficiencies for users of the data and promote consistency between studies. Knowledge of the extent to which these algorithms accurately identify patients’ disease status is key to understanding the potential biases that may arise in analyses using these algorithms. This is essential for the appropriate interpretation of results of analyses of MedicineInsight data. Indeed, validation studies of algorithms used to identify patients with health conditions in routinely collected data have been recognized as a priority for health services research . Although the MedicineInsight algorithms for many conditions have been demonstrated to yield prevalence estimates that are similar to those produced by other reputable data sources [ – ], there has been no formal assessment of their validity. The findings from the numerous validation studies of diagnostic algorithms in primary health care EHR data in other developed countries cannot be assumed to generalise to Australian data, due to between-country differences in the operation and funding of the health care system and differences in the variables available in different databases . The purpose of this study was to examine the validity of MedicineInsight algorithms for five common chronic conditions in general practice: anxiety, asthma, depression, osteoporosis and type 2 diabetes.
We compared each patient’s disease status according to the diagnostic algorithms in the MedicineInsight database to their status determined through review of the original EHRs held in the participating practices. Study population This study was based on patients attending four general practices participating in MedicineInsight. To be eligible, practices had to meet the following criteria: i) data related to activity in October 2019 were successfully extracted; ii) at least 250 patients aged 40 years and older with an encounter in October 2019; iii) located within 40 km of the Sydney or Melbourne central business districts, to ensure ease of access for EHR reviewers (Sydney and Melbourne are the capital cities of Australia’s two most populous states); and. iv) participated in at least one MedicineInsight quality improvement activity in the period November 2018 to October 2019, to ensure interest in engaging with the MedicineInsight program. We categorised the 50 practices meeting these criteria according to the EHR software used (BP or MD) and the city in which the practice is located (Sydney or Melbourne). We randomly selected one practice from each of these four categories (BP Sydney, MD Sydney, BP Melbourne and MD Melbourne); additional practices were selected until one from each category agreed to participate. We stratified our random selection by the EHR software used so that we could examine whether the software contributed to any differences in the validity of the MedicineInsight algorithms. We stratified by city to evenly distribute the data collection between EHR reviewers based in the two cities. Five practices were issued with invitations to participate before four confirmed participation by providing written informed consent. Using MedicineInsight data, we selected patients who were aged 40 years and older and attended the participating practices in October 2019. This age restriction increased the prevalence of the evaluated conditions, thereby optimising statistical power. We randomly selected 250 of these patients per practice. We aimed to collect data for as many of these patients as possible within the five days of data collection planned at each practice. MedicineInsight diagnostic algorithms MedicineInsight personnel have developed coding algorithms that identify patients with specific health conditions. These algorithms identify conditions using information from three EHR fields: diagnosis, reason for visit and reason for prescription . These fields either contain coded terms that the user selects from a drop-down list in the EHR software, or free text. ‘Pyefinch’ coding is available in BP, while ‘Docle’ coding is available in MD. The algorithms identify patients as having the specific health condition if a coded term or text string from the pre-defined list has ever been recorded for that patient in any one of the three fields. The pre-defined list of coded terms and text strings is compiled by trained clinical coders, and is based on available Pyefinch and Docle codes, as well as commonly accepted clinical definitions and abbreviations. For records identified by a free text string alone, the context in which it is recorded is reviewed by clinical coders at the time of developing the algorithm and periodically thereafter, and irrelevant instances removed. A detailed description of the MedicineInsight algorithms for anxiety, asthma, depression, osteoporosis and type 2 diabetes is included in Additional File . For the purposes of this study, the diagnostic algorithms were applied to MedicineInsight data up to 31 October 2019. To ensure that the results of EHR reviews could not influence the classification of patients on the diagnostic algorithms, values on the diagnostic algorithms were extracted from the MedicineInsight database prior to the conduct of EHR reviews. These data extracts were provided to an analyst who did not have access to any additional MedicineInsight data. EHR reviews Information obtained from the original EHRs held in the participating practices was used as the reference standard against which accuracy was benchmarked. Three EHR reviewers visited the participating practices between January and March 2020 and accessed the original EHRs for the randomly selected patients. All EHR reviewers were health professionals registered with the Australian Health Practitioner Regulation Agency, and thus accredited for the keeping of medical records and adherence to confidentiality and privacy principles. Anonymised identifiers for these patients (extracted from the MedicineInsight data) were reassociated with patient names using the third-party data extraction tools installed on computers at each practice. EHR reviewers completed reviews for as many of the 250 selected patients as possible within the time available in the practice, which ranged from three to eight days. To minimise the inconvenience to practices, we planned only five days of data collection in each practice. In one practice, EHR reviews were particularly time consuming due to the size of the records, so an extra three days of data collection were completed. In two of the practices, it was necessary to close data collection early due to COVID-19, with three days of data collection completed in one, and four days in the other. EHR reviewers worked through the randomly ordered list of selected patients from the beginning, without skipping any. Guided by a standardised electronic data capture form, the EHR reviewers searched for evidence of the specific health conditions in the following fields: diagnosis, reason for visit, reason for prescription, correspondence and progress notes . If a diagnosis of the condition (recording of symptoms was not sufficient) was recorded in any of these fields, or if it was documented that the patient was undergoing treatment that is highly specific to the specific condition (e.g. asthma care plan), the patient was considered to have the condition. The term ‘anxiety’ was the exception; it can be used to represent symptoms, but it is often used to indicate anxiety disorder. If it was not clear from the context whether the term ‘anxiety’ was meant to represent symptoms or a diagnosis, it was assumed to be a diagnosis. For osteoporosis, the investigations/results fields were also searched for a diagnosis recorded on bone mineral density test results. The investigations/results fields were also searched for type 2 diabetes. If a diagnosis was recorded or results of fasting blood glucose tests, oral glucose tolerance tests or glycated haemoglobin tests were consistent with the Royal Australian College of General Practitioners’ diagnostic criteria for type 2 diabetes , the patient was considered to have type 2 diabetes. EHR reviewers were blinded to the patient’s disease status on the MedicineInsight algorithms. EHR reviewers were instructed to ignore any evidence documented after 31 October 2019, as the algorithms were applied to MedicineInsight data up to this date. The EHR data were collected and managed using REDCap electronic data capture tools hosted at The University of Melbourne. Analysis For each health condition, the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the MedicineInsight algorithms were calculated. These measures of accuracy are defined in Table . As the data are clustered within practices, variance was adjusted to account for correlation between observations within clusters, and confidence intervals adjusted accordingly. Analyses were conducted using R, version 3.6.2 .
This study was based on patients attending four general practices participating in MedicineInsight. To be eligible, practices had to meet the following criteria: i) data related to activity in October 2019 were successfully extracted; ii) at least 250 patients aged 40 years and older with an encounter in October 2019; iii) located within 40 km of the Sydney or Melbourne central business districts, to ensure ease of access for EHR reviewers (Sydney and Melbourne are the capital cities of Australia’s two most populous states); and. iv) participated in at least one MedicineInsight quality improvement activity in the period November 2018 to October 2019, to ensure interest in engaging with the MedicineInsight program. We categorised the 50 practices meeting these criteria according to the EHR software used (BP or MD) and the city in which the practice is located (Sydney or Melbourne). We randomly selected one practice from each of these four categories (BP Sydney, MD Sydney, BP Melbourne and MD Melbourne); additional practices were selected until one from each category agreed to participate. We stratified our random selection by the EHR software used so that we could examine whether the software contributed to any differences in the validity of the MedicineInsight algorithms. We stratified by city to evenly distribute the data collection between EHR reviewers based in the two cities. Five practices were issued with invitations to participate before four confirmed participation by providing written informed consent. Using MedicineInsight data, we selected patients who were aged 40 years and older and attended the participating practices in October 2019. This age restriction increased the prevalence of the evaluated conditions, thereby optimising statistical power. We randomly selected 250 of these patients per practice. We aimed to collect data for as many of these patients as possible within the five days of data collection planned at each practice.
MedicineInsight personnel have developed coding algorithms that identify patients with specific health conditions. These algorithms identify conditions using information from three EHR fields: diagnosis, reason for visit and reason for prescription . These fields either contain coded terms that the user selects from a drop-down list in the EHR software, or free text. ‘Pyefinch’ coding is available in BP, while ‘Docle’ coding is available in MD. The algorithms identify patients as having the specific health condition if a coded term or text string from the pre-defined list has ever been recorded for that patient in any one of the three fields. The pre-defined list of coded terms and text strings is compiled by trained clinical coders, and is based on available Pyefinch and Docle codes, as well as commonly accepted clinical definitions and abbreviations. For records identified by a free text string alone, the context in which it is recorded is reviewed by clinical coders at the time of developing the algorithm and periodically thereafter, and irrelevant instances removed. A detailed description of the MedicineInsight algorithms for anxiety, asthma, depression, osteoporosis and type 2 diabetes is included in Additional File . For the purposes of this study, the diagnostic algorithms were applied to MedicineInsight data up to 31 October 2019. To ensure that the results of EHR reviews could not influence the classification of patients on the diagnostic algorithms, values on the diagnostic algorithms were extracted from the MedicineInsight database prior to the conduct of EHR reviews. These data extracts were provided to an analyst who did not have access to any additional MedicineInsight data.
Information obtained from the original EHRs held in the participating practices was used as the reference standard against which accuracy was benchmarked. Three EHR reviewers visited the participating practices between January and March 2020 and accessed the original EHRs for the randomly selected patients. All EHR reviewers were health professionals registered with the Australian Health Practitioner Regulation Agency, and thus accredited for the keeping of medical records and adherence to confidentiality and privacy principles. Anonymised identifiers for these patients (extracted from the MedicineInsight data) were reassociated with patient names using the third-party data extraction tools installed on computers at each practice. EHR reviewers completed reviews for as many of the 250 selected patients as possible within the time available in the practice, which ranged from three to eight days. To minimise the inconvenience to practices, we planned only five days of data collection in each practice. In one practice, EHR reviews were particularly time consuming due to the size of the records, so an extra three days of data collection were completed. In two of the practices, it was necessary to close data collection early due to COVID-19, with three days of data collection completed in one, and four days in the other. EHR reviewers worked through the randomly ordered list of selected patients from the beginning, without skipping any. Guided by a standardised electronic data capture form, the EHR reviewers searched for evidence of the specific health conditions in the following fields: diagnosis, reason for visit, reason for prescription, correspondence and progress notes . If a diagnosis of the condition (recording of symptoms was not sufficient) was recorded in any of these fields, or if it was documented that the patient was undergoing treatment that is highly specific to the specific condition (e.g. asthma care plan), the patient was considered to have the condition. The term ‘anxiety’ was the exception; it can be used to represent symptoms, but it is often used to indicate anxiety disorder. If it was not clear from the context whether the term ‘anxiety’ was meant to represent symptoms or a diagnosis, it was assumed to be a diagnosis. For osteoporosis, the investigations/results fields were also searched for a diagnosis recorded on bone mineral density test results. The investigations/results fields were also searched for type 2 diabetes. If a diagnosis was recorded or results of fasting blood glucose tests, oral glucose tolerance tests or glycated haemoglobin tests were consistent with the Royal Australian College of General Practitioners’ diagnostic criteria for type 2 diabetes , the patient was considered to have type 2 diabetes. EHR reviewers were blinded to the patient’s disease status on the MedicineInsight algorithms. EHR reviewers were instructed to ignore any evidence documented after 31 October 2019, as the algorithms were applied to MedicineInsight data up to this date. The EHR data were collected and managed using REDCap electronic data capture tools hosted at The University of Melbourne.
For each health condition, the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the MedicineInsight algorithms were calculated. These measures of accuracy are defined in Table . As the data are clustered within practices, variance was adjusted to account for correlation between observations within clusters, and confidence intervals adjusted accordingly. Analyses were conducted using R, version 3.6.2 .
Within the time available for data collection, EHR reviews were conducted for 477 patient records. One of these EHR reviews was not included in the analyses because the EHR indicated that it was a test record (as opposed to belonging to a real patient), while another was excluded because the EHR review record could not be linked to a patient record in the MedicineInsight data extract due to a data entry error in the study patient identifier. This resulted in the inclusion of 475 patients in the analysis, distributed across practices as follows: BP Sydney, 3 days, n = 65 (14 %); MD Sydney, 5 days, n = 194 (41 %); BP Melbourne, 8 days, n = 110 (23 %); and MD Melbourne, 4 days, n = 106 (22 %). 40 % of the final sample were male; 61 % were aged 40 to 64 years, with the remainder 65 years or older; and 37 % had EHRs based on BP software, with the remainder in MD software. Concordance between the MedicineInsight diagnostic algorithms and EHR reviews is presented in Table . Based on EHR reviews for these 475 patients aged ≥ 40 years, 163 (34 %) patients had anxiety. The diagnostic algorithm for identifying patients with anxiety yielded excellent specificity, PPV and NPV (all 0.93 and above) and a sensitivity of 0.85. According to EHR reviews, 23 % of patients had a diagnosis of asthma recorded ever, and 11 % had osteoporosis. The diagnostic algorithms for asthma and osteoporosis both yielded excellent sensitivity, specificity, PPV and NPV (all 0.94 and above). 35 % of patients ever had a diagnosis of depression recorded, and 15 % had type 2 diabetes. The diagnostic algorithms for depression and type 2 diabetes yielded excellent specificity, PPV and NPV (all 0.94 and above), and both yielded a sensitivity of 0.89. When the calculation of these measures of accuracy was stratified according to the EHR software used (BP or MD), non-overlapping confidence intervals indicated statistically significant differences in the NPV for asthma (0.93, 95 % CI: 0.90–0.95 in BP and 1.00, 95 % CI: 0.99–1.00 in MD), the PPV for osteoporosis (1.00, 95 % CI: 0.98–1.00 in BP and 0.92, 95 % CI: 0.80–0.97 in MD), and the specificity for type 2 diabetes (0.99, 95 % CI: 0.98–0.99 in BP and 1.00, 95 % CI: 1.00–1.00 in MD). While statistically significant, these differences have no obvious clinical significance (see Table ).
This study found that all five MedicineInsight diagnostic algorithms evaluated had excellent specificity, PPV and NPV. The high specificities and PPVs indicate that these algorithms return few false positives and are therefore useful for identifying cohorts of patients who truly have the specific condition and for classifying outcomes . The asthma and osteoporosis algorithms also had excellent sensitivity, making them valuable for identifying representative cohorts of patients and for measuring the prevalence of these conditions. The algorithms for anxiety, depression and type 2 diabetes yielded sensitivities below 0.9, which indicates that some patients who have these conditions are incorrectly classified as not having these conditions. As a result, use of these algorithms will lead to undercounting of patients with these conditions and this should be borne in mind when interpreting the findings of analyses involving these algorithms. Nevertheless, this level of under ascertainment is generally considered acceptable, with many prior validation studies of primary health care EHR data interpreting sensitivities of this magnitude as evidence of good accuracy [ , , ]. Three of the evaluated MedicineInsight diagnostic algorithms have accuracy that is comparable to, or superior to, the accuracy of diagnostic algorithms in electronic primary health care databases in other parts of the world. According to a recent systematic review, other asthma algorithms have yielded sensitivities ranging from 0.74 to 0.92, specificities ranging from 0.84 to 0.98, PPVs ranging from 0.67 to 0.81 and NPVs of 0.9 and above . Depression algorithms have returned sensitivities ranging from 0.73 to 0.81, PPVs ranging from 0.79 to 0.87 and specificities and NPVs of 0.9 and above . Type 2 diabetes algorithms have yielded sensitivities ranging from 0.65 to 1.0, PPVs ranging from 0.87 to 1.0 and specificities and NPVs of 0.94 and above . To our knowledge, there have been no prior validation studies of anxiety or osteoporosis algorithms in primary health care data. Strengths and limitations A strength of this study is that EHR reviews were conducted for patients that the algorithm identified as cases as well as those the algorithm considered non-cases. Including both cases and non-cases in a study allows for the calculation of sensitivity, specificity, NPV and PPV, where all of these measures are important because each describes a different aspect of accuracy and allows the reader to consider how the algorithm will perform in a particular context . Despite this, many studies have not collected reference standard data for non-cases, instead opting to seek confirmation only for patients identified as cases by the algorithm. While this reduces the total number of patients for whom reference standard data needs to be collected, such an approach means that PPV is the only measure of accuracy that can be calculated. To attain sufficient statistical power in the current study, the sample was restricted to patients aged 40 years and older. This represents a trade-off in terms of generalisability of the PPV and NPV estimates. As estimates of PPV and NPV depend on the prevalence of the specific health condition , the PPV estimates returned in this study may be higher, and our NPV estimates may be lower, than those yielded by the diagnostic algorithms in a population with a lower prevalence of the condition. The prevalence of the five conditions in our sample was approximately twice that of the whole MedicineInsight patient sample . In addition to the age restriction, this increased prevalence is likely due to the focus on patients with a recent visit to a general practitioner, the chance of which would be higher in frequent attenders. A further threat to the generalisability of the results arises from the inclusion of only four practices in this study, potentially leading to high sampling variability, compounded by the uneven distribution of EHR reviews across these practices. As a consequence, in the estimates of concordance generated by this study, more weight has been given to those practices that contributed more EHR reviews. This uncertain generalisability should be borne in mind when applying the diagnostic algorithms for other populations within the MedicineInsight database. Recording of the diagnosis in the original EHRs was used as the reference standard against which the accuracy of the algorithms was benchmarked. The limitation of this approach is that the recording of diagnoses in the original EHR may be inaccurate or incomplete . This is a particular challenge in the Australian context, where patients are able to obtain care at multiple general practices and information is not routinely shared between practices. The extent to which diagnoses are not recorded completely may differ according to the specific condition, with fragmentation of mental health care and patient concerns about confidentiality contributing to the under-recording of mental health conditions in primary health care EHRs . Despite this, there is consensus among experts that EHR reviews are an acceptable reference standard for validation studies, with the majority of validation studies of electronic primary health care and other administrative health data using EHR reviews as the reference standard . As an alternative to EHR reviews, some prior validation studies have asked general practitioners to complete questionnaires regarding the health of their individual patients. However, this approach generally results in a low response rate and limits the number of patients for whom data can be collected . Other validation studies have used records in population-based data collections such as cancer registries, hospital admissions data and death registries as the reference standard , but this is not possible for MedicineInsight data until full-scale record linkage is enabled.
A strength of this study is that EHR reviews were conducted for patients that the algorithm identified as cases as well as those the algorithm considered non-cases. Including both cases and non-cases in a study allows for the calculation of sensitivity, specificity, NPV and PPV, where all of these measures are important because each describes a different aspect of accuracy and allows the reader to consider how the algorithm will perform in a particular context . Despite this, many studies have not collected reference standard data for non-cases, instead opting to seek confirmation only for patients identified as cases by the algorithm. While this reduces the total number of patients for whom reference standard data needs to be collected, such an approach means that PPV is the only measure of accuracy that can be calculated. To attain sufficient statistical power in the current study, the sample was restricted to patients aged 40 years and older. This represents a trade-off in terms of generalisability of the PPV and NPV estimates. As estimates of PPV and NPV depend on the prevalence of the specific health condition , the PPV estimates returned in this study may be higher, and our NPV estimates may be lower, than those yielded by the diagnostic algorithms in a population with a lower prevalence of the condition. The prevalence of the five conditions in our sample was approximately twice that of the whole MedicineInsight patient sample . In addition to the age restriction, this increased prevalence is likely due to the focus on patients with a recent visit to a general practitioner, the chance of which would be higher in frequent attenders. A further threat to the generalisability of the results arises from the inclusion of only four practices in this study, potentially leading to high sampling variability, compounded by the uneven distribution of EHR reviews across these practices. As a consequence, in the estimates of concordance generated by this study, more weight has been given to those practices that contributed more EHR reviews. This uncertain generalisability should be borne in mind when applying the diagnostic algorithms for other populations within the MedicineInsight database. Recording of the diagnosis in the original EHRs was used as the reference standard against which the accuracy of the algorithms was benchmarked. The limitation of this approach is that the recording of diagnoses in the original EHR may be inaccurate or incomplete . This is a particular challenge in the Australian context, where patients are able to obtain care at multiple general practices and information is not routinely shared between practices. The extent to which diagnoses are not recorded completely may differ according to the specific condition, with fragmentation of mental health care and patient concerns about confidentiality contributing to the under-recording of mental health conditions in primary health care EHRs . Despite this, there is consensus among experts that EHR reviews are an acceptable reference standard for validation studies, with the majority of validation studies of electronic primary health care and other administrative health data using EHR reviews as the reference standard . As an alternative to EHR reviews, some prior validation studies have asked general practitioners to complete questionnaires regarding the health of their individual patients. However, this approach generally results in a low response rate and limits the number of patients for whom data can be collected . Other validation studies have used records in population-based data collections such as cancer registries, hospital admissions data and death registries as the reference standard , but this is not possible for MedicineInsight data until full-scale record linkage is enabled.
Primary health care EHR databases are powerful resources for improving our understanding of health and healthcare practices. These databases typically provide clinical information that is richer than that available through administrative data or population surveys . However, the extent to which the findings of analyses of such data are a true reflection of patient health, and are trusted by clinicians, policymakers and researchers, depends on the accuracy of the data. This study measured the accuracy of MedicineInsight algorithms for five chronic conditions, finding that the algorithms for asthma and osteoporosis have excellent accuracy and the algorithms for anxiety, depression and type 2 diabetes have good accuracy when compared to recording of diagnoses in the original EHR. This study provides support for the use of these algorithms in the MedicineInsight data for primary health care quality improvement activities, research and health system policymaking and planning. General practices provided informed written consent to participate in this research, and a waiver of the requirement for individual patient consent was granted by the NREEC.
Additional file 1:
|
Understanding the Composition of a Comprehensive Otolaryngologist's Practice Through Medicare Reimbursements | f31aa16d-f86e-4cec-aa9a-d5804c4fc217 | 11844328 | Otolaryngology[mh] | Data Sets and Variables This study was deemed to be exempt from review by the University of California, Irvine Institutional Review Board due to the public availability of the data sets. Data was obtained from the 2019 Medicare Provider Utilization and Payment Data: Physician and Other Supplier Detailed Data (detailed data set) , and Medicare Provider Utilization and Payment Data: Physician and Other Supplier Aggregate Data (aggregate data set). The aggregate data sets contain data about utilization, payment, and submitted charges organized by National Provider Identifier (NPI) and locale of service (office vs facility) for health care providers who have billed for >10 Medicare fee‐for‐service patients, inclusive of all Healthcare Common Procedure Coding System (HCPCS) codes. ORL physicians were selected from the aggregate data set by filtering by “Otolaryngology” as the provider type. To characterize the practice settings and fellowships of ORLs, we took a random sample of 10% (n = 897) of ORLs from the aggregate data set for further data searches. Providers were searched on the internet using name and location to classify them as academic versus community by determining if they were affiliated with a teaching hospital, and no fellowship (c‐ORLs) versus fellowship (f‐ORLs; including head and neck surgery, laryngology, neurotology, facial plastic, pediatrics, rhinology, and sleep). This characterization was used to calculate the subspecialty breakdown of reimbursement and services for general and fellowship‐trained ORLs. The internet search was performed in 2022. The aggregate data set was used to gather descriptive statistics for the number of unique Current Procedural Terminology (CPT) codes billed, the number of patients, the number of services, Medicare‐allowed payments (inclusive of copays and deductibles), submitted charges, patient age, hierarchical condition category (HCC) risk score, and location of practice (urban vs rural based on Rural‐Urban Commuting Area codes). Further data was generated through mathematical manipulation of variables (ie, the number of services per unique patient by dividing the 2 values). The markup ratio (MUR) was calculated by dividing the submitted charge amount by the Medicare‐allowed amount. The detailed data sets contain data about utilization, payment, and submitted charges organized by NPI, HCPCS code, and locale of service. HCPCS codes billed 10 or fewer times on 10 or fewer Medicare patients are excluded from the data set. This caveat of the aggregate and detailed data sets enabled us to characterize the percentage of codes and reimbursement from codes billed by a provider <10 times. NPI was used to isolate ORLs in this data set from our random sample. Evaluation and management (E&M) codes (99201‐99205, 99211‐99215) were treated separately from procedural codes. HCPCS codes billed by these providers were characterized in the following categories: general office codes, otology/neurotology, rhinology, laryngology, allergy, radiographic imaging, other non‐specialty specific codes, head and neck surgery, facial plastics, laboratory tests, and sleep. Statistical Analysis SPSS v27.0 (IBM Corp) was utilized to organize data, obtain a random sample (via the “Select Cases: Random Sample”), obtain descriptive statistics, and perform statistical analyses (Student's t tests and χ 2 analyses, as appropriate). Student's t test results are presented as mean (standard deviation), while χ 2 analyses are presented as percentages. Statistical significance was assessed using a 2‐sided α of .05 and 95% confidence intervals.
This study was deemed to be exempt from review by the University of California, Irvine Institutional Review Board due to the public availability of the data sets. Data was obtained from the 2019 Medicare Provider Utilization and Payment Data: Physician and Other Supplier Detailed Data (detailed data set) , and Medicare Provider Utilization and Payment Data: Physician and Other Supplier Aggregate Data (aggregate data set). The aggregate data sets contain data about utilization, payment, and submitted charges organized by National Provider Identifier (NPI) and locale of service (office vs facility) for health care providers who have billed for >10 Medicare fee‐for‐service patients, inclusive of all Healthcare Common Procedure Coding System (HCPCS) codes. ORL physicians were selected from the aggregate data set by filtering by “Otolaryngology” as the provider type. To characterize the practice settings and fellowships of ORLs, we took a random sample of 10% (n = 897) of ORLs from the aggregate data set for further data searches. Providers were searched on the internet using name and location to classify them as academic versus community by determining if they were affiliated with a teaching hospital, and no fellowship (c‐ORLs) versus fellowship (f‐ORLs; including head and neck surgery, laryngology, neurotology, facial plastic, pediatrics, rhinology, and sleep). This characterization was used to calculate the subspecialty breakdown of reimbursement and services for general and fellowship‐trained ORLs. The internet search was performed in 2022. The aggregate data set was used to gather descriptive statistics for the number of unique Current Procedural Terminology (CPT) codes billed, the number of patients, the number of services, Medicare‐allowed payments (inclusive of copays and deductibles), submitted charges, patient age, hierarchical condition category (HCC) risk score, and location of practice (urban vs rural based on Rural‐Urban Commuting Area codes). Further data was generated through mathematical manipulation of variables (ie, the number of services per unique patient by dividing the 2 values). The markup ratio (MUR) was calculated by dividing the submitted charge amount by the Medicare‐allowed amount. The detailed data sets contain data about utilization, payment, and submitted charges organized by NPI, HCPCS code, and locale of service. HCPCS codes billed 10 or fewer times on 10 or fewer Medicare patients are excluded from the data set. This caveat of the aggregate and detailed data sets enabled us to characterize the percentage of codes and reimbursement from codes billed by a provider <10 times. NPI was used to isolate ORLs in this data set from our random sample. Evaluation and management (E&M) codes (99201‐99205, 99211‐99215) were treated separately from procedural codes. HCPCS codes billed by these providers were characterized in the following categories: general office codes, otology/neurotology, rhinology, laryngology, allergy, radiographic imaging, other non‐specialty specific codes, head and neck surgery, facial plastics, laboratory tests, and sleep.
SPSS v27.0 (IBM Corp) was utilized to organize data, obtain a random sample (via the “Select Cases: Random Sample”), obtain descriptive statistics, and perform statistical analyses (Student's t tests and χ 2 analyses, as appropriate). Student's t test results are presented as mean (standard deviation), while χ 2 analyses are presented as percentages. Statistical significance was assessed using a 2‐sided α of .05 and 95% confidence intervals.
Comparing Comprehensive and Fellowship‐Trained Otolaryngologists Within our random sample of 897 ORLs, 554 (61.8%) were c‐ORLS. Of 343 f‐ORLs, 74 (21.6%) were head and neck oncologists, 18 (5.2%) laryngologists, 50 (14.6%) neurotologists, 123 (35.9%) facial plastic surgeons, 25 (7.3%) pediatric otolaryngologists, 41 (12.0%) rhinologists, and 12 (3.5%) sleep surgeons. There was no difference in gender between c‐ORLs and f‐ORLs ( P = .195) or rurality ( P = .114) with most of our sample of ORLs being male and located in urban areas. Both ORL groups practice with a large number of unique CPT codes billed (52.9 vs 51.8; P = .564). c‐ORLs, however, on average treat more patients (mean: 451.7 [SD: 296.9] vs 307.8 (255.7); P < .001) and perform a greater number of services (1982.2 [2614.7] vs 1412.2 [2508.7]; P = .001). However, both groups perform an equal number of services per unique patient (4.2 [4.8] vs 4.0 [3.4]; P = .447). c‐ORLs perform more of their service in the office setting than f‐ORLs (96.3% vs 92.2%; P < .001). With regards to reimbursement, Medicare paid an equal amount to both c‐ORLs and f‐ORLs, on average ($138,942 [$117,563] vs $132,551 [$147,886]; P = .474). Per patient, f‐ORLs were reimbursed at a higher rate ($453 [$384] vs $304 [$151]; P < .001). Ninety‐five percent of the reimbursement to c‐ORLs was in the office, as opposed to 89.7% of the reimbursements to f‐ORLs ( P < .001; ). MUR was significantly higher for f‐ORLs than for c‐ORLs (3.43 [1.70] vs 2.68 [1.27]; P < .001). On average, patients treated by f‐ORLs had a higher comorbidity burden (as defined by the HCC risk score; 1.44 [0.48] vs 1.29 [0.29]; P < .001; ). Characteristics of Comprehensive Otolaryngologists Broken Down by Setting, Gender, and Rurality Of the 554 c‐ORLs, 47 (8.5%) practiced in an academic setting, while the rest (91.5%) practiced in a community setting. There was no difference in gender and rurality ( P > .050 for both). Community c‐ORLs practiced with a broader number of unique CPT codes billed (54.7 [26.0] vs 40.2 [25.8]; P < .001) and a larger number of patients (465.1 [294.9] vs 372.3 [284.0]; P = .039). Although community c‐ORLs performed a nominally larger average number of services, this did not reach significance (2067.9 [2699.7] vs 1323.7 [1448.4]; P = .063). Total average reimbursement and reimbursement per patient also did not differ ( P > .05 for both), though academic c‐ORLs had a higher MUR (3.77 [2.25] vs 2.57 [1.10]; P < .001). There was no difference in the HCC risk score ( P = .790; ). Broken down by gender, male c‐ORLs practiced with a larger number of unique CPT codes (54.4 [26.9] vs 47.9 [21.4]; P = .035), had a larger population of Medicare patients (472.0 [302.0] vs 372.7 [234.9]; P = .004), and performed more services (2103.0 [2786.1] vs 1442 [1245.1]; P = .032). However, there was no difference in services per patient ( P = .478). As such, male c‐ORLs were reimbursed more ($146,913 [$122,344] vs $105,567 [$75,435]; P = .003), though not more per patient ( P = .117). There was no difference in MUR or HCC scores ( P > .050 for both; ). Broken down by rurality, rural c‐ORLs practiced with more unique CPT codes (63.9 [29.4] vs 52.3 [25.7]; P = .002) than their urban counterparts. Though the number of patients and services were equivalent ( P > .050 for both), rural c‐ORLs performed more services per patient (6.0 [3.6] vs 4.1 [3.0]; P < .001). There was no difference in total reimbursement, reimbursement per patient, or MUR ( P > .050 for all). Rural and urban c‐ORL patient populations had an equivalent HCC risk score ( P = .368; ). Practice Profiles Of the total reimbursements paid by Medicare to c‐ORLs (for the CPT codes detailed in the detailed data set), most of c‐ORLs reimbursement (52.8%) comes from E&M codes. Other significant factors in their reimbursement include procedures, specifically rhinology codes (17.8%), otology/neurotology codes (9.7%), laryngology codes (9.0%, and allergy codes (4.8%; ). In comparison, E&M codes only made up 43.0% of total reimbursements to f‐ORLs, while rhinology made up 28.3%, laryngology 8.0%, otology/neurotology 8.0%, and allergy 4.3% ( ).
Within our random sample of 897 ORLs, 554 (61.8%) were c‐ORLS. Of 343 f‐ORLs, 74 (21.6%) were head and neck oncologists, 18 (5.2%) laryngologists, 50 (14.6%) neurotologists, 123 (35.9%) facial plastic surgeons, 25 (7.3%) pediatric otolaryngologists, 41 (12.0%) rhinologists, and 12 (3.5%) sleep surgeons. There was no difference in gender between c‐ORLs and f‐ORLs ( P = .195) or rurality ( P = .114) with most of our sample of ORLs being male and located in urban areas. Both ORL groups practice with a large number of unique CPT codes billed (52.9 vs 51.8; P = .564). c‐ORLs, however, on average treat more patients (mean: 451.7 [SD: 296.9] vs 307.8 (255.7); P < .001) and perform a greater number of services (1982.2 [2614.7] vs 1412.2 [2508.7]; P = .001). However, both groups perform an equal number of services per unique patient (4.2 [4.8] vs 4.0 [3.4]; P = .447). c‐ORLs perform more of their service in the office setting than f‐ORLs (96.3% vs 92.2%; P < .001). With regards to reimbursement, Medicare paid an equal amount to both c‐ORLs and f‐ORLs, on average ($138,942 [$117,563] vs $132,551 [$147,886]; P = .474). Per patient, f‐ORLs were reimbursed at a higher rate ($453 [$384] vs $304 [$151]; P < .001). Ninety‐five percent of the reimbursement to c‐ORLs was in the office, as opposed to 89.7% of the reimbursements to f‐ORLs ( P < .001; ). MUR was significantly higher for f‐ORLs than for c‐ORLs (3.43 [1.70] vs 2.68 [1.27]; P < .001). On average, patients treated by f‐ORLs had a higher comorbidity burden (as defined by the HCC risk score; 1.44 [0.48] vs 1.29 [0.29]; P < .001; ).
Of the 554 c‐ORLs, 47 (8.5%) practiced in an academic setting, while the rest (91.5%) practiced in a community setting. There was no difference in gender and rurality ( P > .050 for both). Community c‐ORLs practiced with a broader number of unique CPT codes billed (54.7 [26.0] vs 40.2 [25.8]; P < .001) and a larger number of patients (465.1 [294.9] vs 372.3 [284.0]; P = .039). Although community c‐ORLs performed a nominally larger average number of services, this did not reach significance (2067.9 [2699.7] vs 1323.7 [1448.4]; P = .063). Total average reimbursement and reimbursement per patient also did not differ ( P > .05 for both), though academic c‐ORLs had a higher MUR (3.77 [2.25] vs 2.57 [1.10]; P < .001). There was no difference in the HCC risk score ( P = .790; ). Broken down by gender, male c‐ORLs practiced with a larger number of unique CPT codes (54.4 [26.9] vs 47.9 [21.4]; P = .035), had a larger population of Medicare patients (472.0 [302.0] vs 372.7 [234.9]; P = .004), and performed more services (2103.0 [2786.1] vs 1442 [1245.1]; P = .032). However, there was no difference in services per patient ( P = .478). As such, male c‐ORLs were reimbursed more ($146,913 [$122,344] vs $105,567 [$75,435]; P = .003), though not more per patient ( P = .117). There was no difference in MUR or HCC scores ( P > .050 for both; ). Broken down by rurality, rural c‐ORLs practiced with more unique CPT codes (63.9 [29.4] vs 52.3 [25.7]; P = .002) than their urban counterparts. Though the number of patients and services were equivalent ( P > .050 for both), rural c‐ORLs performed more services per patient (6.0 [3.6] vs 4.1 [3.0]; P < .001). There was no difference in total reimbursement, reimbursement per patient, or MUR ( P > .050 for all). Rural and urban c‐ORL patient populations had an equivalent HCC risk score ( P = .368; ).
Of the total reimbursements paid by Medicare to c‐ORLs (for the CPT codes detailed in the detailed data set), most of c‐ORLs reimbursement (52.8%) comes from E&M codes. Other significant factors in their reimbursement include procedures, specifically rhinology codes (17.8%), otology/neurotology codes (9.7%), laryngology codes (9.0%, and allergy codes (4.8%; ). In comparison, E&M codes only made up 43.0% of total reimbursements to f‐ORLs, while rhinology made up 28.3%, laryngology 8.0%, otology/neurotology 8.0%, and allergy 4.3% ( ).
With a growing number of graduating otolaryngologists choosing to pursue fellowship, the percentage of c‐ORLs practicing across the country is likely to fall in the coming years. In the setting of a shortage of ORLs across the country, , , , c‐ORLs would be the most equipped to meet anticipated shortfalls. The ideal model is likely one where patients have a “home” otolaryngologist who can manage more general complaints and who can then refer patients to specialists as needed. Having a doctor that can handle multiple otolaryngologic complaints is imperative for patients, given that the wait time to see an otolaryngologist is frequently above 20 days. In this study, we further investigated the practice and reimbursement patterns of c‐ORLs to better understand their role in the ORL landscape. It is generally accepted that there is a significant difference between rural‐urban health outcomes with higher age‐adjusted death rates across the 5 leading causes of death. Around 20% of the American population lives in what is considered a rural area, yet fewer than 10% of physicians as a whole live in these areas, and fewer than 8% of otolaryngologists. , This trend can have dire consequences for patients, as counties with 2 or fewer otolaryngologists tend to have worse head and neck cancer survival rates. Academic practice is the most commonly cited intended practice type among all ORL residents and is most commonly located in urban areas, , , but c‐ORLs are an exception and are far more likely to work in community or private practice. Due to this practice difference, we hypothesized that c‐ORLs would be more likely to practice in rural areas when compared to f‐ORLs. Nevertheless, we found no difference between f‐ORLs and c‐ORLs in terms of rurality. This seems to indicate that the lack of rural otolaryngologists stretches across all practice types and settings. The dearth of ORLs, and physicians more broadly, in rural areas does not seem to be driven by academia career goals or fellowship completion but rather seems to be a complex distribution problem driven by physician “preference” with no clear solutions. In regard to gender in ORL, women make up just 18% of the current ORL workforce; however, the percentage of women in the residency workforce is significantly higher making up 38% of resident otolaryngologists. , This data indicates that the gender gap in ORL is closing, although there is still significant room for improvement. There was no gender difference between f‐ORLs and c‐ORLs in our data indicating that women in ORL show no preference toward either practice type compared to men. While there was no difference in rurality or gender between c‐ORLs and f‐ORLs, we did find significant differences in terms of services offered and patients treated between the groups. c‐ORLs performed significantly more services and treated significantly more patients than their f‐ORL counterpart; however, importantly, there was no significant difference in the number of services offered per patient, suggesting patients are not being overtreated. This difference in volume can likely be explained by c‐ORLs tending to tackle more bread‐and‐butter complaints with f‐ORLs focusing on more complex pathologies that require more time. c‐ORLs were also found to provide more services in‐office, potentially contributing to their ability to see more patients. While f‐ORLs certainly treat complex pathology that is beyond the scope of a generalist, c‐ORLs see more patients and are equipped to handle a broader variety of pathologies, making them the ideal ORL to combat the workforce shortage. Conversely, while c‐ORLs may see a higher number of patients on average, there was no significant difference in Medicare reimbursement compared to f‐ORLs. This means that c‐ORLs are more dependent on patient volume for reimbursement. This may be concerning given a decline in reimbursements throughout the field of ORL. , , , With no evidence of reimbursements stabilizing, and with a limit on how high volume can go, this may lead to increased consolidation of practices and increasing speed of private equity entering the otolaryngologic space. Increased consolidation may further exacerbate the already urbanized/centralized distribution of ORLs. Furthermore, as c‐ORLs are already working on volume, there may be little room for growth in the setting of increased demand, further contributing to delays in care. This could force patients to wait longer or visit multiple specialists who offer fewer services to take care of their otolaryngologic issues that could have been handled by a generalist. Comparing rural and urban c‐ORLs, few important differences were seen. Though there were no differences in number of patients or services, rural c‐ORLs performed more services per patient. Furthermore, when looking at the number of unique CPT codes, rural c‐ORLs offered more varied services than those practicing in urban areas. Both details suggest rural c‐ORLs are filling an important gap in care with a wide practice range. This data is supported by other studies that suggest that rural providers have more procedural diversity than their urban counterparts. Despite rural c‐ORLs performing more volume with more varied services per patient than urban c‐ORLs, there was no difference in total reimbursement or reimbursement per patient. This may be secondary to performing more low‐reimbursing procedures and/or Medicare geographic index adjustments (ie, with rural areas getting reimbursed less due to lower costs of living). This may be an important contributing factor for the lack of rural ORLs. In our data, male c‐ORLs tended to practice with more unique CPT codes, had more Medicare patients, and performed more services per patient than their female counterparts. However, the importance of this data in practice has yet to be determined and there are several potential explanations. This may be due to the median age of men in ORL being significantly higher than the median age of women, and thus being more established and efficient in practice. Previous data also suggests that female ORLs tend to spend significantly more time with patients and this may contribute to fewer patients seen. Gender bias may also play a role in terms of the number of referrals female c‐ORLs receive and the types of practice opportunities available to them. A previous study in orthopedic surgery suggests that female surgeons were significantly less likely to receive referrals from male providers and a similar bias may exist in ORL. When considering the types of services c‐ORLs provided, outside of office visits, the next largest group of CPT codes for c‐ORLs was rhinology. This is important to note given that recent data suggests that Medicare reimbursements for rhinology have not been keeping up with inflation and may signal a future fall in reimbursements for c‐ORLs. Furthermore, reimbursements are falling in the next 4 most utilized CPT codes, including otology/neurotology, laryngology, allergy, and head and neck surgery. , , , , Although this study gives an overview of reimbursement and the practice patterns for c‐ORLs, there were some limitations. The biggest limitation is the use of the Medicare Provider Utilization and Payment Dataset which does not include provider‐HCPCS code combinations that were billed 10 or fewer times and/or on 10 or fewer Medicare patients. This is important for c‐ORLs, given that they are more likely to work in private practice and more likely to have a greater variety in the types of procedures they perform, and f‐ORLs, given the various niche codes for their complex patients; as such, practice patterns described in and may be somewhat skewed. Further, we are unable to characterize provider practice patterns when billing through private insurance. In addition, these codes cannot characterize disease severity or the quality of care, and how these may have differed depending on procedure when comparing c‐ORLs to f‐ORLs. There is also a 3‐year discrepancy between the data set year and the year of internet search, which may add some element of miscategorization with regards to academic versus community practice profiles, due to the possibility of a change in practice setting. Finally, the data set includes data from before the COVID‐19 pandemic. Thus, the impact of the pandemic on c‐ORL practice patterns is yet to be determined.
This is the first study aimed at determining reimbursements and practice patterns of c‐ORLs. While our data did not demonstrate any difference in rurality between c‐ORLs and f‐ORLs, c‐ORLs did offer a significantly higher number of services and treated more patients for a similar Medicare reimbursement. c‐ORLs also practiced with greater variety, with rural c‐ORLs providing the greatest variety of services. Regarding practice type, c‐ORLs show a significant predilection toward rhinology. Overall, with the increased patient volume and variety, our data demonstrates that c‐ORLs may likely serve a significant role in meeting anticipated shortages in ORL services.
Sina J. Torabi , conception and design, analysis and interpretation, drafting and critical revisions, final approval, accountable; Sagar Vasandani , interpretation, drafting, final approval, accountable Rahul A. Patel , conception and design, analysis and interpretation, drafting, final approval, accountable; R. Peter Manes , conception and design, interpretation, critical revisions, final approval, accountable; Edward C. Kuan , conception and design, interpretation, critical revisions, final approval, accountable.
Competing interests E.C.K. is a consultant for Stryker and 3‐D Matrix and receives royalties from Springer Books; these conflicts of interest are not relevant to the current study. The other authors declare no relevant conflict of interest. Funding source N/A.
E.C.K. is a consultant for Stryker and 3‐D Matrix and receives royalties from Springer Books; these conflicts of interest are not relevant to the current study. The other authors declare no relevant conflict of interest.
N/A.
|
Perspectives on a virtual student-led research conference in ophthalmology | 3786e996-deaf-4bd6-8ede-6bbe9f0e2f9a | 10961134 | Ophthalmology[mh] | The switch to virtual educational events during the COVID-19 pandemic has led reduced opportunities for clinical and educational opportunities. However, it also led to the emergence of student-led initiative such as the Canadian Ophthalmology Mentorship Program (COMP), and Post-Match webinars, which aimed to bridge gaps in access by offering virtual mentorship sessions and insights into the match process, respectively. The increased uptake of virtual, student-led pedagogy not only encourages medical students to expand their roles beyond mere learners, but also provides an opportunity to encourage wide participation by making conferences accessible to individuals with low-socioeconomic status, caregivers, and delegates with disabilities, thereby working towards reducing inequalities that persist in academia. An important consideration, given the growing emphasis on medical students' involvement in scholarly activities in ophthalmology and other surgical specialities. In response to these needs, the Canadian Ophthalmology Student Interest Group (COSIG) developed the first Canadian medical student-led ophthalmology conference, “COSIG Annual Meeting” (CAM), a one-day event with the theme of global health, with the goal of providing medical students with opportunities to create a sense of community, enhance knowledge about ophthalmology as a speciality, showcase medical student research in a supportive environment, network with peers and ophthalmology residents, and win prizes for trivia and research excellence. CAM took place on March 29, 2021 via the platform Zoom, and it was free and open to all medical students globally, thereby reducing barriers to participation such as cost, the need for professional memberships, geographic location, and competing personal commitments. It was organized by a team of five committees (logistics, marketing, mentorship, research, and sponsorships) consisting of medical student volunteers from 12 programs across Canada. CAM was largely marketed via the Canadian Ophthalmology Student Interest Group social media. In addition to keynote speakers, trivia with prizes, and research presentations, we implemented a novel one-hour Speed Mentorship Session in partnership with the COMP. During these sessions, nine ophthalmology residents acted as near-peer mentors in small group sessions, thereby addressing gaps in mentorship for medical students interested in ophthalmology while upholding psychological safety. To identify areas of improvement of the program, students completed pre- and post-conference questionnaires sent via the registration e-mails (one at registration, and one post-event) consisting of questions regarding the quality and impact of the event (64/123; 52% responded post-conference). The survey data was collected under quality assurance and quality improvement and so do not require Research Ethics Board. Survey results indicated that CAM organization and content was well-received by medical students ( ,B), and that many reported increased access to ophthalmology mentorship and interest in applying to the speciality. All medical schools were represented with the majority from McGill University (14%), University of British Columbia (14%), and Western University (12%) (123 participants). Most participants were in clerkship (57%). Tools used (i.e., ) were free, and the prizes for trivia and research presentations were obtained by e-mailing companies. CAM exemplified the potential of student-led virtual communities to provide more accessible, cost-effective, and inclusive reach than traditional conferences. Such flexible opportunities could help bridge the educational gap in the ever more crowded medical school curriculum. Future steps include collaborating directly with the Canadian Ophthalmological Society (COS), as this may not only increase conference awareness within the present ophthalmology community but also allow for the development of workshops and longitudinal mentorship opportunities. |
A survey of Canadian adult rheumatologists’ knowledge, comfort level, and barriers in assessing psychosocial needs of young adults with rheumatic diseases | a25d441b-f247-4dd2-aac3-1121d83cde20 | 10171159 | Internal Medicine[mh] | Chronic rheumatic diseases diagnosed in childhood often require long-term medical management, and the transition period from pediatric to adult rheumatology care is an important and often vulnerable time in patients’ lives . Adult care is more patient focused than family focused and requires more independence. Inadequate preparation for adult care combined with the vulnerable state patients can lead to gaps in care and loss to follow-up . Existing resources for transition include the American College of Rheumatology (ACR) pediatric transition toolkit and the European Alliance of Associations for Rheumatology (EULAR) Standards and Recommendations for the Transitional Care of Young People with Juvenile-Onset Rheumatic Diseases . However, in a recent publication, only 31% of both adult and pediatric health care providers reported using these resources to help with the transition process . For youth, gaps in care and loss to follow-up may occur for a multitude of reasons, many of which may be personal including psychosocial concerns, and important life milestones as well as disease relapses, medication side effects and difficulties accessing care . Disruptions in medical care may also be related to their initial experiences with their adult practitioner . There have been limited studies reporting the perspectives of adult rheumatologists in transition . A recent survey of adult rheumatologists, based out of the United States, self-identified as having inadequate training in transition issues of young adults, specifically psychosocial concerns . This contrasts with pediatric healthcare providers who have familiarity working with adolescent psychosocial concerns and behaviors . Adult rheumatologists reported a lack of comfort in managing patients with pediatric-onset disease and endorsed less familiarity with transition guidelines compared to pediatric healthcare providers . The objectives for this Canadian survey of adult rheumatologists and adult rheumatology trainees were to determine the comfort level and barriers to caring for young adult patients with pediatric-onset rheumatic disease after transfer to adult care. In addition, the management of psychosocial needs in the adolescent and young adult population including comfort level and frequency of discussions around these topics were assessed.
A combination of literature review, building on prior work from the United States-based study by Zisman et al. and information provided by a needs assessment conducted by the Canadian Rheumatology Association (CRA), formed the basis of our survey . The survey incorporated entrustable professional activities (EPAs) designed by the Royal College of Canada for Rheumatology trainees which are specific tasks that trainees can be trusted to perform independently in different contexts . Specifically, Core EPA #12 P: Supporting adolescents/young adults with rheumatologic disease in the transition from the pediatric to adult care setting and its individual milestones were used for each question. This EPA focuses on developmental readiness and the risks that can occur as patients develop autonomy regarding their health . The initial survey was drafted by two pediatric rheumatologists, a pediatric resident and a clinical researcher. It was then reviewed by two adult rheumatologists whose revisions were incorporated prior to sharing it with four adult rheumatologist members of the CRA Transition Working Group for final edits, comments, and feedback. The Consensus-based Checklist for reporting of Survey Studies (CROSS) and EQUATOR guidelines were followed during survey preparation and reporting . The survey collected information on demographics, current practice, comfort in managing psychosocial aspects of transition care and perceived barriers to providing care to this population. Responses to questions were graded on a five-point Likert scale. The project was approved by the Hamilton Integrated Research Ethics Board (Project #13568) on September 16th, 2021. The survey was distributed by email via the CRA using the Alchemer platform ( www.alchemer.com ) on November 3rd, 2021 to adult rheumatologists and trainees who were CRA members. Pediatric rheumatologists and trainees were not included. Data collection occurred until March 18 th , 2022 with monthly reminders sent by the CRA. Statistical analysis Descriptive statistics (frequencies and proportions) summarizing survey responses were determined using SPSS v.28 (IBM SPSS Statistics, USA). Responses from the five-point Likert scale were then translated to a dichotomous outcome to better analyze data due to the relatively small sample size in the study. For example, those who answered questions regarding comfort level with agree or strongly agree were designated as comfortable and those who answered strongly disagree, disagree, and neutral were designated not comfortable.
Descriptive statistics (frequencies and proportions) summarizing survey responses were determined using SPSS v.28 (IBM SPSS Statistics, USA). Responses from the five-point Likert scale were then translated to a dichotomous outcome to better analyze data due to the relatively small sample size in the study. For example, those who answered questions regarding comfort level with agree or strongly agree were designated as comfortable and those who answered strongly disagree, disagree, and neutral were designated not comfortable.
Demographics The survey was sent to 490 rheumatologists (451 in English and 39 in French) and 15.2% ( n = 69) completed the survey. More than half were female with ≥ 10 years of experience in practice and had trained in Canada. Most respondents were practicing in Ontario and had either a university-based practice or a combined university- and community-based practice (Table ). Approximately half reported involvement in providing transition care. Fifty-one rheumatologists (74%) reported that they had no formal training in transition care (Table ). Of those who had received training, it was primarily in fellowship and by continuing medical education at their own institution. Current transition practices Providers most commonly saw patients being transferred to them between 16 and 18 years of age and were referred by pediatric rheumatologists. The ideal age for transfer was felt to be between 16 and 20 years old. About 45% had a multidisciplinary team to help support transitioned patients which included nursing, physiotherapy, social work, and occupational therapy. On average, 40% had resources in their office for patients such as an orientation to the adult rheumatologists’ practice, self-management skills assessments and tools (pamphlets, phone apps, etc.). More than half of the participants reported no formal transition policy at their institution with one-third reporting an informal policy was being followed. Approximately 40% of individuals felt comfortable discussing psychosocial concerns such as gender identity, sexuality, contraception, drug and alcohol use, vaping, depression, and anxiety (Table ); however, about half reported discussing vaping and gender identity not at all or rarely. Barriers to transition More than two-thirds of respondents reported insufficient skills to address transition-related concerns while only 13% reported having sufficient resources and personnel to adequately address these concerns. Greater than 75% reported lack of time and renumeration for providing transition care. The most frequently reported barriers to providing optimal care to patients transitioning to adult care were: (1) lack of primary care providers, (2) lack of allied health support, (3) inadequate training in caring for this age group, and (4) not being able to be reimbursed or having time listed to provide transition care. Despite this, the majority (75%) expressed an interest in learning more about providing transition care to adolescents. Three quarters of respondents believed that the family physician should be the provider to discuss psychosocial concerns such as mental health, body image, drug use, gender identity and sexuality, and educational goals (Table ). About 40% felt that tele-rheumatology can be a helpful resource in transition care.
The survey was sent to 490 rheumatologists (451 in English and 39 in French) and 15.2% ( n = 69) completed the survey. More than half were female with ≥ 10 years of experience in practice and had trained in Canada. Most respondents were practicing in Ontario and had either a university-based practice or a combined university- and community-based practice (Table ). Approximately half reported involvement in providing transition care. Fifty-one rheumatologists (74%) reported that they had no formal training in transition care (Table ). Of those who had received training, it was primarily in fellowship and by continuing medical education at their own institution.
Providers most commonly saw patients being transferred to them between 16 and 18 years of age and were referred by pediatric rheumatologists. The ideal age for transfer was felt to be between 16 and 20 years old. About 45% had a multidisciplinary team to help support transitioned patients which included nursing, physiotherapy, social work, and occupational therapy. On average, 40% had resources in their office for patients such as an orientation to the adult rheumatologists’ practice, self-management skills assessments and tools (pamphlets, phone apps, etc.). More than half of the participants reported no formal transition policy at their institution with one-third reporting an informal policy was being followed. Approximately 40% of individuals felt comfortable discussing psychosocial concerns such as gender identity, sexuality, contraception, drug and alcohol use, vaping, depression, and anxiety (Table ); however, about half reported discussing vaping and gender identity not at all or rarely.
More than two-thirds of respondents reported insufficient skills to address transition-related concerns while only 13% reported having sufficient resources and personnel to adequately address these concerns. Greater than 75% reported lack of time and renumeration for providing transition care. The most frequently reported barriers to providing optimal care to patients transitioning to adult care were: (1) lack of primary care providers, (2) lack of allied health support, (3) inadequate training in caring for this age group, and (4) not being able to be reimbursed or having time listed to provide transition care. Despite this, the majority (75%) expressed an interest in learning more about providing transition care to adolescents. Three quarters of respondents believed that the family physician should be the provider to discuss psychosocial concerns such as mental health, body image, drug use, gender identity and sexuality, and educational goals (Table ). About 40% felt that tele-rheumatology can be a helpful resource in transition care.
The key findings from this survey highlight that most adult rheumatologists lack training in caring for the adolescent population and express a lack of supportive medical care, both family physicians and allied health, to help them care for this population. Further, while they may feel comfortable—specifically discussing psychosocial concerns—the majority believe that it is the family physician’s responsibility to do so. Compared to previous studies, there were similarities regarding the overall transition process and the barriers to a successful transition. For example, approximately one-third of respondents in our survey had no formal transition policy at their institution but an informal procedure that is followed, which is comparable to the findings in the Childhood Arthritis and Rheumatology Research Alliance (CARRA) survey completed in 2010 and the survey in 2014 by Zisman et al. . In addition, the most frequently encountered barriers such as lack of training in caring for this age group, allied health providers, primary care providers, not being able to be renumerated appropriately and time constraints were common themes that have been found in previous studies . Almost half of adult rheumatologists reported feeling comfortable discussing psychosocial concerns with patients which is higher than previous studies (Table ) . However, this was self-reported and patient perspectives were not sought. Topics such as alcohol and tobacco use, contraception, and fertility were reported to be discussed either most or all the time by the majority of rheumatologists. However, 45% reported discussing vaping either “not at all” or “rarely” despite the same number feeling comfortable discussing the topic. Given the increase in incidence and the detrimental effects on lung function in youth with chronic conditions, this topic should be addressed consistently in this population . Despite half of the participants reporting comfort in discussing psychosocial topics, more than 75% believed that the family physician should be the provider to discuss these issues (Table ). Many of these concerns are related to the patient’s rheumatic disease and/or medications and side effects leading to difficulties with body image, self-esteem and, ultimately, to mental health concerns . This can place patients in a difficult position as both health care providers feel as though it is the other’s responsibility to address these concerns. Unfortunately, many patients do not have a family doctor due to an overall shortage of primary care physicians in Canada or may have difficulty accessing their primary care provider . In addition, many adolescents with chronic rheumatic disease are otherwise generally healthy and medically stable, and their rheumatologist may be the only physician they see regularly. Ideally, conversations about psychosocial issues would happen jointly among health care providers and the primary care provider should be included in the transition process. The majority of transition research is targeted toward the pediatric healthcare provider’s care and perspective. Following a patient from childhood into adulthood establishes strong relationships between patients, their families, and the pediatric care team ; however, adult healthcare providers will ultimately spend a far longer time with patients over their life course. It is, therefore, essential to gain an understanding of their current practices and perceptions of gaps in care to ensure that patients receive optimal care. This survey adds to the body of research and provides a Canadian perspective on this topic. This survey design is unique in that it incorporated the competency by design framework. As more programs shift toward a competency by design curriculum, these milestones are now mandatory for trainees to complete their training . The questions are not only applicable to rheumatology but can be used to assess transition in other pediatric sub-specialities with chronic conditions. It also looked at multiple aspects of psychosocial care, which have not been previously evaluated, such as vaping and gender identity. Despite reminder emails, our study was limited by the relatively low response rate (15.2%) and small numbers of individuals in certain demographic groups (e.g., physicians 46–55 years old, practicing 6–15 years) making subgroup comparisons infeasible. In addition, many rheumatologists do not provide transition care and may not have completed the survey as a result. It is also probable that the survey was completed by adult rheumatologists who were already interested in transition care, leading to response bias and a limited perspective, thereby affecting the generalizability of results to the general adult rheumatology community. Although the survey was released in 2021, the impact of the COVID-19 pandemic upon transitional care and psychosocial concerns of patients was not included in the survey. These results demonstrate that adult rheumatologists would benefit from more training, multidisciplinary support, resources, and appropriate renumeration to optimize the provision of transition care to young adult patients. Knowledge and treatment of psychosocial factors related to the young adult population should be emphasized during training. Educational initiatives such as workshops and peer teaching sessions with patient engagement should be co-created to improve the transition knowledge and skills for adult health care providers. Tele-rheumatology is a resource that can be valuable for this population as it can be used when patients are away for university/college for follow-up appointments and make it easier to have joint appointments with health care providers. A transition navigator or coach may be a valuable resource to bridge the divide between pediatric and adult care, provide psychosocial support to young adult patients, and provide continuity to patients during this challenging time . Transition navigators have been found to improve adherence to medication, attendance at clinic, and reduce acute complications of disease in other chronic disease populations . Future research should be aimed toward assessing the impact of educational initiatives in subspecialty training of rheumatologists in their comfort supporting the young adult patient population with pediatric-onset disease. Furthermore, the effectiveness of implementing the above resources such as transition navigators upon the transition process from the patient, primary care provider and subspecialist perspective should also be obtained.
Below is the link to the electronic supplementary material. Supplementary file1 (DOCX 217 KB)
|
Determinants of dental care use in patients with rare diseases: a qualitative exploration | da47c181-5cbf-45fa-873c-0d95a6f03bbb | 10286462 | Dental[mh] | In the European Union, a disease is considered rare when it affects less than one person in 2000. Awareness of rare diseases has increased in the EU since 2009, when the Council of the European Union asked Member States to develop plans and strategies on rare diseases . Although the conditions are rare, the total number of patients affected is significant (3–4 million people in France, 27–36 million people in the EU and 25 million in the US). To date, 5,000 to 8,000 distinct rare diseases have been documented and newly discovered rare diseases (RD) are regularly reported in the literature. Most definitions seem to consider the prevalence of the disease, but sometimes other criteria apply, such as the severity of the disease, and whether it is hereditary . In France, the management of rare diseases has been institutionalized through different national plans - Plan National Maladies Rares (PNMR). The first one, included in the law on public health policy of 9 August 2004, was implemented from 2005 to 2008. Currently, the third PNMR (2018–2022) is a continuation of the two previous ones. These plans have set up the rare disease expert centres and networks covering the entire territory to give access to care and expertise for all affected individuals. Their clinical and biological parameters are collected as shared information to pilot patient care and prevention at the national and eventually European level and provide databanks for research on rare diseases . One common denominator of all patients affected by a rare disease is the relevance of their oral condition . Optimal oral health will limit the impact of nutritional disorders on their pathology and reduce the occurrence of microbial infection and inflammation. Sound oral health will also preserve the smile and facial form, thus maintaining function and limiting aesthetic damage and therefore preserving the individual’s self-esteem in social interactions. Rare disease patterns may include a constitutive oral phenotype or not. In these last cases, other pathophysiological traits may indirectly affect oral functionality and/or diseases such as impaired microbial defenses, motor capacities, eating or cleaning behavior. For instance, patients can have in their clinical picture an intellectual delay and an acquisition disorder, or not. All these constitutive and/or indirect oral features are important to consider analyzing in an accurate and efficient way, the oral care pathways and the resulting quality of life related to oral health [ – ]. Indeed, oral health is an inherent part of overall health as its physiological crossroads house basic functions such as mastication, swallowing or phonation, and oral health plays a central role in the life of relationships (aesthetic and social handicap, emotional feelings) and individual self-image and consideration . The general hypothesis was that in patients with rare diseases, access to dental care could be difficult because of the lack of professionals and expert structures in a situation to diagnose, follow and treat their disease in the necessary multi-disciplinary manner and/or because some patients, such as for instance the ones with cognitive disabilities, could not find the existing adequate structures to manage their oral health. For example: in the case of rare diseases involving oral disorders, especially in patients fed by enteral or gastrostomy methods during childhood, the practical question of the use of the oral sphere was questioned as it raises major difficulties . Our objective was to describe what could be the determinants of adequate access to dental care in a context of rare diseases, alongside organizational, technical, and human obstacles causing absent or inadequate management of the patients.
Design This study employed a qualitative descriptive design including semi-structured interviews using guiding themes. Participants In the specific context of rare diseases, it is realistic to exploit the number of patients available in a study and reach the point of saturated data than rather define the number of patients needed a priori . Given the diversity and the number of diseases encountered by the study and the systematic non-overlap of the problems encountered by patients, about thirty patients, over a 6-month period, were included in the study to ensure a wide range of responses that covered the guiding themes. Data saturation was achieved with 29 patients. At the beginning of the planned care consultation, the practitioner and the researcher briefly but clearly explained the study and obtained the assent of the person taking part in the research, or that of his/her legal representative (Table ). Ethics approval and consent to participate The study was approved by the Ethics committee for the protection of individual COMITÉ DE PROTECTION DES PERSONNES SUD-OUEST ET OUTRE MER III (ref APHP210890/IDRCB 2021-A00429-32). In France, ethics committees are randomly assigned throughout the national territory without any notion of the authors’ affiliation. All the patients and their parents (in case of minor’s or patient’s or mental disabilities) gave written informed consent for participation in the study and the publication of the study results: Patients were selected in coordination with the leader of each rare disease expertise centre in the Necker Hospital (Paris, France) respectively concerned during a usual consultation at the hospital and were initially informed orally about the study. When they were interested (patients and/or their parents), an information note was given to them to explain everything and to allow them to give the most informed consent possible. For minor patients, the parents, and their children each received an appropriate information note and the parents signed a consent document. Adult patients who are under the guardianship (legal document) of their parent because of mental disability or medical situation, the parents signed a consent document and received the information note. Patients of legal age who are not covered by the abovementioned, received the information note and signed the consent document. All methods were performed in accordance with the relevant guidelines and regulations. Eligibility criteria Patients were selected in coordination with the leader of each rare disease expertise centre in the Necker Hospital (Paris, France), respectively concerned. Inclusion criteria were: Patients with a rare disease (diagnosis of rare disease selected by the hospital medical teams). Patients seen between 1.1.2017 and 1.1.2020 in the concerned Rare Disease Expertise Center. Patients aged 6 years and over. Patients who are beneficiaries of one of the social security schemes in France (general scheme or special schemes). Patients seen at least once in their life in the medical genetics department of the Necker Hospital ((Paris, France) because of their rare disease. Non-inclusion criteria were: Patients not resident in France. Patients who do not speak French. Procedure A semi-structured interview guide was developed by the research group and discussed with researchers and clinicians beforehand to ensure that it had enough rigor to achieve credible data collection. The interviews all start on the disease itself and the overall care pathway. We gradually move into the topic of oral health by addressing the themes of access to care and past experiences. Most of the time, the topics intertwine, with patients and parents expressing their feelings about the overall care and dental care. It is in the analysis of the themes and sub-themes that the ideas are classified. Some patients have a significant intellectual delay, and, in this case, it was the parents who answered the interview. For the patients who did not have an intellectual delay, they answered the interview, certainly sometimes helped by their parents, but they had full voice to express absolutely everything they felt. Interview Data collection Patients who met the inclusion criteria were recruited from the expertise centres or during their usual consultation, with their non-objection and that of their legal guardians if they were minors. Face-to-face, semi-structured lasting interviews were held by the first author (L.F), trained in qualitative sociological and anthropological interview methods. All interviews took place in Necker Enfants Malades Hospital , (Paris, France), in a dedicated room after a standard consultation. We used an interview guide, structured by the following six themes: “Access to dental care “, “Oral health related quality of life,” “Orality disorders”, “Renouncement to dental care” and “School integration and daily life.” During these interviews, themes were discussed with the patients who were encouraged to talk freely. Interviews with duration between 30 and 60 min were conducted and recorded in the period of May 2021 until December 2021. Transcripts and other sensitive data were stored at Research Clinical Unit in Necker Enfants Malades Hospital , (Paris, France). Qualitative thematic analysis For the analysis of the qualitative data, all the recorded interviews were manually transcribed and then analyzed using a thematic data analysis software (NVIVO 12 on Windows). The thematic analysis of the questionnaires was carried out according to the following process and we are inspired in this analysis by the work of Vaismoradi, M. and all, who have been of great help to us and whom we thank [ – ]: Key themes, or big ideas, were identified by reading and re-reading the interview transcripts. Phrases or blocks of words that correspond directly to the research question were highlighted. These phrases were categorized so that they can be grouped together thematically. These categories gave rise to sub-themes which were then examined and analyzed. The transcripts were reviewed to identify key themes and coded by a single researcher. Interviews were performed until the data were saturated and no further themes emerged, as is the current standard for qualitative data analysis in health settings. 10% of the interviews (randomly chosen) were analyzed by a second coder to ensure consistency and quality assurance of the data. Key themes were organized in a table and those mentioned by three or more patients were considered common. Key common themes were converted into questions and grouped into general themes. Data credibility and rigor Data credibility was established through a triangulation strategy, which uses a combination of specialized teams to review and evaluate the results. In addition to semi-structured interviews, data credibility was ensured by note taking during the interviews. Data were verified by both peers outside the study and by research team members. The main findings were presented to some of the participants and their opinions, collected. In addition, the results were evaluated and verified on several occasions by supervisors . Furthermore, all the authors of this article are trained in social epidemiology using a mixed method (quantitative and qualitative).
This study employed a qualitative descriptive design including semi-structured interviews using guiding themes.
In the specific context of rare diseases, it is realistic to exploit the number of patients available in a study and reach the point of saturated data than rather define the number of patients needed a priori . Given the diversity and the number of diseases encountered by the study and the systematic non-overlap of the problems encountered by patients, about thirty patients, over a 6-month period, were included in the study to ensure a wide range of responses that covered the guiding themes. Data saturation was achieved with 29 patients. At the beginning of the planned care consultation, the practitioner and the researcher briefly but clearly explained the study and obtained the assent of the person taking part in the research, or that of his/her legal representative (Table ).
The study was approved by the Ethics committee for the protection of individual COMITÉ DE PROTECTION DES PERSONNES SUD-OUEST ET OUTRE MER III (ref APHP210890/IDRCB 2021-A00429-32). In France, ethics committees are randomly assigned throughout the national territory without any notion of the authors’ affiliation. All the patients and their parents (in case of minor’s or patient’s or mental disabilities) gave written informed consent for participation in the study and the publication of the study results: Patients were selected in coordination with the leader of each rare disease expertise centre in the Necker Hospital (Paris, France) respectively concerned during a usual consultation at the hospital and were initially informed orally about the study. When they were interested (patients and/or their parents), an information note was given to them to explain everything and to allow them to give the most informed consent possible. For minor patients, the parents, and their children each received an appropriate information note and the parents signed a consent document. Adult patients who are under the guardianship (legal document) of their parent because of mental disability or medical situation, the parents signed a consent document and received the information note. Patients of legal age who are not covered by the abovementioned, received the information note and signed the consent document. All methods were performed in accordance with the relevant guidelines and regulations.
Patients were selected in coordination with the leader of each rare disease expertise centre in the Necker Hospital (Paris, France), respectively concerned. Inclusion criteria were: Patients with a rare disease (diagnosis of rare disease selected by the hospital medical teams). Patients seen between 1.1.2017 and 1.1.2020 in the concerned Rare Disease Expertise Center. Patients aged 6 years and over. Patients who are beneficiaries of one of the social security schemes in France (general scheme or special schemes). Patients seen at least once in their life in the medical genetics department of the Necker Hospital ((Paris, France) because of their rare disease. Non-inclusion criteria were: Patients not resident in France. Patients who do not speak French.
A semi-structured interview guide was developed by the research group and discussed with researchers and clinicians beforehand to ensure that it had enough rigor to achieve credible data collection. The interviews all start on the disease itself and the overall care pathway. We gradually move into the topic of oral health by addressing the themes of access to care and past experiences. Most of the time, the topics intertwine, with patients and parents expressing their feelings about the overall care and dental care. It is in the analysis of the themes and sub-themes that the ideas are classified. Some patients have a significant intellectual delay, and, in this case, it was the parents who answered the interview. For the patients who did not have an intellectual delay, they answered the interview, certainly sometimes helped by their parents, but they had full voice to express absolutely everything they felt.
Data collection Patients who met the inclusion criteria were recruited from the expertise centres or during their usual consultation, with their non-objection and that of their legal guardians if they were minors. Face-to-face, semi-structured lasting interviews were held by the first author (L.F), trained in qualitative sociological and anthropological interview methods. All interviews took place in Necker Enfants Malades Hospital , (Paris, France), in a dedicated room after a standard consultation. We used an interview guide, structured by the following six themes: “Access to dental care “, “Oral health related quality of life,” “Orality disorders”, “Renouncement to dental care” and “School integration and daily life.” During these interviews, themes were discussed with the patients who were encouraged to talk freely. Interviews with duration between 30 and 60 min were conducted and recorded in the period of May 2021 until December 2021. Transcripts and other sensitive data were stored at Research Clinical Unit in Necker Enfants Malades Hospital , (Paris, France). Qualitative thematic analysis For the analysis of the qualitative data, all the recorded interviews were manually transcribed and then analyzed using a thematic data analysis software (NVIVO 12 on Windows). The thematic analysis of the questionnaires was carried out according to the following process and we are inspired in this analysis by the work of Vaismoradi, M. and all, who have been of great help to us and whom we thank [ – ]: Key themes, or big ideas, were identified by reading and re-reading the interview transcripts. Phrases or blocks of words that correspond directly to the research question were highlighted. These phrases were categorized so that they can be grouped together thematically. These categories gave rise to sub-themes which were then examined and analyzed. The transcripts were reviewed to identify key themes and coded by a single researcher. Interviews were performed until the data were saturated and no further themes emerged, as is the current standard for qualitative data analysis in health settings. 10% of the interviews (randomly chosen) were analyzed by a second coder to ensure consistency and quality assurance of the data. Key themes were organized in a table and those mentioned by three or more patients were considered common. Key common themes were converted into questions and grouped into general themes. Data credibility and rigor Data credibility was established through a triangulation strategy, which uses a combination of specialized teams to review and evaluate the results. In addition to semi-structured interviews, data credibility was ensured by note taking during the interviews. Data were verified by both peers outside the study and by research team members. The main findings were presented to some of the participants and their opinions, collected. In addition, the results were evaluated and verified on several occasions by supervisors . Furthermore, all the authors of this article are trained in social epidemiology using a mixed method (quantitative and qualitative).
Patients who met the inclusion criteria were recruited from the expertise centres or during their usual consultation, with their non-objection and that of their legal guardians if they were minors. Face-to-face, semi-structured lasting interviews were held by the first author (L.F), trained in qualitative sociological and anthropological interview methods. All interviews took place in Necker Enfants Malades Hospital , (Paris, France), in a dedicated room after a standard consultation. We used an interview guide, structured by the following six themes: “Access to dental care “, “Oral health related quality of life,” “Orality disorders”, “Renouncement to dental care” and “School integration and daily life.” During these interviews, themes were discussed with the patients who were encouraged to talk freely. Interviews with duration between 30 and 60 min were conducted and recorded in the period of May 2021 until December 2021. Transcripts and other sensitive data were stored at Research Clinical Unit in Necker Enfants Malades Hospital , (Paris, France).
For the analysis of the qualitative data, all the recorded interviews were manually transcribed and then analyzed using a thematic data analysis software (NVIVO 12 on Windows). The thematic analysis of the questionnaires was carried out according to the following process and we are inspired in this analysis by the work of Vaismoradi, M. and all, who have been of great help to us and whom we thank [ – ]: Key themes, or big ideas, were identified by reading and re-reading the interview transcripts. Phrases or blocks of words that correspond directly to the research question were highlighted. These phrases were categorized so that they can be grouped together thematically. These categories gave rise to sub-themes which were then examined and analyzed. The transcripts were reviewed to identify key themes and coded by a single researcher. Interviews were performed until the data were saturated and no further themes emerged, as is the current standard for qualitative data analysis in health settings. 10% of the interviews (randomly chosen) were analyzed by a second coder to ensure consistency and quality assurance of the data. Key themes were organized in a table and those mentioned by three or more patients were considered common. Key common themes were converted into questions and grouped into general themes.
Data credibility was established through a triangulation strategy, which uses a combination of specialized teams to review and evaluate the results. In addition to semi-structured interviews, data credibility was ensured by note taking during the interviews. Data were verified by both peers outside the study and by research team members. The main findings were presented to some of the participants and their opinions, collected. In addition, the results were evaluated and verified on several occasions by supervisors . Furthermore, all the authors of this article are trained in social epidemiology using a mixed method (quantitative and qualitative).
Sample Analysis In this study, 29 patients between the ages of 7 and 24 years of age were interviewed. Parents were present during the interviews when children or older patients wished. The pathologies, age, gender, and presence or absence of intellectual delay of the patients are described in Table . The pathologies found in the sample are for the most part syndromic, with the first signs appearing in antenatal or early childhood. The pathologies are here digestive, pulmonary, cardiac, renal, or neurodevelopmental. There are no rare diseases in the study sample that exclusively affect the orofacial sphere. Of these 29 patients, 15 patients had an intellectual delay. The exploration of themes and verbatim As envisaged in the preparation of the interview guide, four main themes were identified as keys: dental care course, mental disabilities, and orality disorders, and finally social impact. All analyses were carried out using exact transcripts of the interviews, whether the verbatims were those of the patients or of the patients’ relatives when they were unable to express themselves. Dental care course (concepts mentioned in all the 29 interviews, with 80 occurrences). In our study, patients affected by a rare disease with an either constitutive or indirect oral phenotype had difficulty to find a practitioner who would face the overall disease context and manage their oral health. They described difficulties to get an appointment and when they found a dentist, they expressed the feeling of being rejected and referred to another-one. Parents describe the feeling of not being listened to, even though they feel that they sometimes know their child’s illness better than the dentist. In our sample, we recruited patients in extreme situations with technically very difficult treatments. In our sample, this was exemplified in two contexts where the oral dysfunction is major: patients with fibrodysplasia ossificans progressiva (FOP), and the Di George syndrome. For FOP, the patients can have - and this was the case for 4 of the 7 patients in the study- a total ankylosis of the temporo-mandibular joints because of heterotopic ossification. This ankylosis makes almost impossible for them to open their mouths (opening of less than 1 cm) and complicates dental care. It is also known that these bone flare-ups can be caused by muscle trauma during dental care. Patients are therefore very apprehensive about dental care. These parents of a child with FOP described a rather chaotic oral health care journey: “The first dentist he saw was a friend of our pediatrician. He lived far away from us, but we needed someone we could trust, because the mouth is an extremely sensitive area for FOP Patients, so the dentist really had to understand that he shouldn’t be afraid of a FOP patient, but he shouldn’t open his mouth too much.“ “I basically write: FOP, etc., thinking a little bit like you stupidly: she’s going to ask about it. She arrived, didn’t even look at me and I was the one who had to say to her: “Wait, Madam, we’re going to stop, I’m going to explain the situation to you”, and so, that was still disturbing. “But it’s true that we arrive with a bit of annoyance because I had also taken the trouble to fill out her paperwork, she didn’t even look at it, it’s obvious. The theme of oral care and everything that affects the oral cavity appeared to all the patients and parents interviewed as a major and integral part of the overall care. Parents did report that when it was possible to have the child’s teeth treated in the same place as the overall management of the disease, the care pathways were extremely simplified. Professionals were able to work in better communication with each other and improve management.: “Last time at the dentist, I felt like they’re afraid to touch it. They’re hesitant. Like this one, Dr. xx she takes children, I know that, but in relation to him, she told me no and sent me to another one. “Already to get an appointment, it was very difficult to find someone who could do dental care” . Parents describe what they see as a lack of knowledge on the part of dentists, and sometimes what they interpret as fear of treating their children: “The dentist we had taken near our home, he didn’t want to do extractions that were a little complicated like that, so he told us to do it somewhere else. We started by going to a dental clinic. When we saw the estimates, we said to ourselves: “He doesn’t care about us”. “Yes, she didn’t know, you could tell.“ “That’s also the flaw when you’re a parent, is that you have your nose in the grindstone so much, I realize that with teachers, when we explain things, it seems so easy. It’s like, I think, with dentists, you don’t want to stretch, you don’t want to spread it too far, you don’t want to… and in fact, we don’t necessarily think to say it.“ “Diane, we did what we did for her brother. When she was very young, we took her to the dentist once a year for check-ups and the first few times, the dentist just looked at her and patted her teeth to reassure her. Then she got used to going to the dentist. You weren’t afraid.“ “Because at 18 months, the family dentist didn’t know what was wrong with her, she was losing enamel on the face of the teeth, and it was getting very white and not ivory at all.“ Mental disabilities (concepts mentioned in 15 interviews, with 45 occurrences). The parents who were interviewed and had a child with an intellectual disability emphasized the difficulty of access to dental care for their children. Parents reported inappropriate psychological behavior regarding the whole family with apparently untrained and stressed dentist regarding their children. “It’s true that disability is scary already, and I think they’re not trained.“ They may report comments that they feel have hurt them as parents and parents well describe the lack of therapeutic education and assistance in implementing effective dental hygiene protocols. “The last time we went there, 1 month ago, I had the reflection of the dentist who said to me: ‘it is up to you to do the oral hygiene of your daughter’. But it’s not easy with a child with a disability to brush her teeth, it’s not at all obvious. And when I explained that they got angry at me. They said, “Yes, you are responsible for your child. Do you realize, after 1 year, she will have cavities again”, I said I knew it very well.“ “It’s true that disability is already scary, and I think they are not trained.“ “In private practice, anyway, I think there’s the number to go behind. If they spend three quarters of an hour doing a treatment and they don’t succeed, it’s not profitable for them. I’m not saying this well, but afterwards, they have expenses, which is also understandable. So they don’t necessarily have the time or the desire, or the diplomacy, to take care of children with disabilities. “She was screaming so much that the dentist said that She was scaring all his clients.“ Orality disorders (concepts mentioned in 12 interviews, with 40 occurrences). Many oral problems were mentioned by parents. These difficulties with the oral sphere concerned all patient functions: feeding, cleaning, speaking. In this context, oral care was complicated, even in children who did not have intellectual delays. “She doesn’t like it when you put something in her mouth. Better now, but a year or two ago, she didn’t like anything in her mouth. Even the toothbrush was difficult. I was the one who had to brush, but that was hard too. She would gag every time you went in the back a little bit. It was very, very, very difficult. “Since baby, she always had trouble putting things in her mouth. Foods. Sometimes foods that were a little greasy or whatever, it was hard. It was always a heartache. “It’s always been hard on the food part since she was a little girl. Even now. It’s hard to identify something, to say she doesn’t like this texture, or she doesn’t like this consistency, etc., because it’s kind of like it’s diffuse and it’s very hard to identify the structuring. I say anything: she doesn’t like orange and she won’t eat carrots. It evolves and we’re not sure why.“ “She had hyper nausea. Chewing was very complicated, even now she can chew, but it shouldn’t be too…” . “From the moment it was to eat, she took her pacifier, I had a pacifier where we could put milk, medicine and she spit out the pacifier. It wasn’t the same function anymore; we had changed the function of the pacifier.“ “Care was impossible because you had to get into the mouth.“ “We were behind in getting good oral hygiene, because he had sensory dysorality syndrome.“ “As a result, nothing would fit in his mouth. He had his sensitivity threshold at the entrance to his mouth. In fact, he had bitten me once, so he never teethed on things. “He was gagging. The toothbrush, it wasn’t even worth it.“ “Chewing is not at its best. If he can swallow whole, he does. I have to remind him to chew.“ “To chew, it’s very complicated.“ “He had periods when he could eat, because we weren’t sure how to do it, whether to eat, not eat, infuse, enteral, eat a little bit. He was in occlusion very regularly and as a result, this really altered his orality in the sense that he did not develop his feeding as a toddler, where you gradually integrate milk, certain foods, textures, etc. In relation to that, he developed food phobias.“ “It was intrusive, and then because he wasn’t eating, it was intrusive to put in his mouth. It was complicated, I would say that he hasn’t been in the habit for very long, and you still have to be behind it, because it’s not something natural. “Yes. The mouth noises, she can’t stand, she doesn’t eat with us anymore. She already can’t stand herself, hearing her noises.“ “No, I think it’s subsiding. I think it’s people who are learning, who are less excited. Afterwards, it’s also Alexandre’s character that wants that, he’s a dynamic boy, who likes it when it pulsates, who likes it when it goes fast. Afterwards, I think that as we grow up we learn to settle down and the FOP forces us to settle down even more. Social impact (concepts mentioned in 10 interviews, with 30 occurrences). The patients and their parents did not return to the theme of social and school integration as much as to the themes of the dental care pathway. Nevertheless, the notions of the look of others, of appearance, of others were raised, especially by the older patients. “My teeth, I don’t have a nice smile like some of my friends. They have beautiful teeth, a beautiful smile, and I am too ashamed to smile in front of others. To laugh, I put my hand in front of my mouth, it has become a huge complex. “That’s also the flaw when you’re a parent, it’s that you have your nose in the grindstone so much, I realize that with the teachers, when we explain things, it seems so easy. It’s like, I think, with dentists, you don’t want to stretch, you don’t want to spread it too far, you don’t want to… and in fact, we don’t necessarily think to say it.“ “Because when you’re talking, people… sometimes you don’t realize, but sometimes there are people, they don’t look you in the eye, but they look at your lips or your teeth, how they look. And then, when you’re done talking, they’ll say: “But why are your teeth like that? “They are friends with me because of who I am and not because of what they see. They are friends with me for who I am, not for what they see. For my character, for my good mood, not for my disabilities, because I can’t go to Paris and I can’t take the metro. They know that, if they go to Paris, they won’t offer it to me, because it will be a direct “no”. But for example, if we go somewhere to eat, we go by bus, we’ve already thought about everything, we’re not going to do anything too complicated.“
In this study, 29 patients between the ages of 7 and 24 years of age were interviewed. Parents were present during the interviews when children or older patients wished. The pathologies, age, gender, and presence or absence of intellectual delay of the patients are described in Table . The pathologies found in the sample are for the most part syndromic, with the first signs appearing in antenatal or early childhood. The pathologies are here digestive, pulmonary, cardiac, renal, or neurodevelopmental. There are no rare diseases in the study sample that exclusively affect the orofacial sphere. Of these 29 patients, 15 patients had an intellectual delay.
As envisaged in the preparation of the interview guide, four main themes were identified as keys: dental care course, mental disabilities, and orality disorders, and finally social impact. All analyses were carried out using exact transcripts of the interviews, whether the verbatims were those of the patients or of the patients’ relatives when they were unable to express themselves. Dental care course (concepts mentioned in all the 29 interviews, with 80 occurrences). In our study, patients affected by a rare disease with an either constitutive or indirect oral phenotype had difficulty to find a practitioner who would face the overall disease context and manage their oral health. They described difficulties to get an appointment and when they found a dentist, they expressed the feeling of being rejected and referred to another-one. Parents describe the feeling of not being listened to, even though they feel that they sometimes know their child’s illness better than the dentist. In our sample, we recruited patients in extreme situations with technically very difficult treatments. In our sample, this was exemplified in two contexts where the oral dysfunction is major: patients with fibrodysplasia ossificans progressiva (FOP), and the Di George syndrome. For FOP, the patients can have - and this was the case for 4 of the 7 patients in the study- a total ankylosis of the temporo-mandibular joints because of heterotopic ossification. This ankylosis makes almost impossible for them to open their mouths (opening of less than 1 cm) and complicates dental care. It is also known that these bone flare-ups can be caused by muscle trauma during dental care. Patients are therefore very apprehensive about dental care. These parents of a child with FOP described a rather chaotic oral health care journey: “The first dentist he saw was a friend of our pediatrician. He lived far away from us, but we needed someone we could trust, because the mouth is an extremely sensitive area for FOP Patients, so the dentist really had to understand that he shouldn’t be afraid of a FOP patient, but he shouldn’t open his mouth too much.“ “I basically write: FOP, etc., thinking a little bit like you stupidly: she’s going to ask about it. She arrived, didn’t even look at me and I was the one who had to say to her: “Wait, Madam, we’re going to stop, I’m going to explain the situation to you”, and so, that was still disturbing. “But it’s true that we arrive with a bit of annoyance because I had also taken the trouble to fill out her paperwork, she didn’t even look at it, it’s obvious. The theme of oral care and everything that affects the oral cavity appeared to all the patients and parents interviewed as a major and integral part of the overall care. Parents did report that when it was possible to have the child’s teeth treated in the same place as the overall management of the disease, the care pathways were extremely simplified. Professionals were able to work in better communication with each other and improve management.: “Last time at the dentist, I felt like they’re afraid to touch it. They’re hesitant. Like this one, Dr. xx she takes children, I know that, but in relation to him, she told me no and sent me to another one. “Already to get an appointment, it was very difficult to find someone who could do dental care” . Parents describe what they see as a lack of knowledge on the part of dentists, and sometimes what they interpret as fear of treating their children: “The dentist we had taken near our home, he didn’t want to do extractions that were a little complicated like that, so he told us to do it somewhere else. We started by going to a dental clinic. When we saw the estimates, we said to ourselves: “He doesn’t care about us”. “Yes, she didn’t know, you could tell.“ “That’s also the flaw when you’re a parent, is that you have your nose in the grindstone so much, I realize that with teachers, when we explain things, it seems so easy. It’s like, I think, with dentists, you don’t want to stretch, you don’t want to spread it too far, you don’t want to… and in fact, we don’t necessarily think to say it.“ “Diane, we did what we did for her brother. When she was very young, we took her to the dentist once a year for check-ups and the first few times, the dentist just looked at her and patted her teeth to reassure her. Then she got used to going to the dentist. You weren’t afraid.“ “Because at 18 months, the family dentist didn’t know what was wrong with her, she was losing enamel on the face of the teeth, and it was getting very white and not ivory at all.“ Mental disabilities (concepts mentioned in 15 interviews, with 45 occurrences). The parents who were interviewed and had a child with an intellectual disability emphasized the difficulty of access to dental care for their children. Parents reported inappropriate psychological behavior regarding the whole family with apparently untrained and stressed dentist regarding their children. “It’s true that disability is scary already, and I think they’re not trained.“ They may report comments that they feel have hurt them as parents and parents well describe the lack of therapeutic education and assistance in implementing effective dental hygiene protocols. “The last time we went there, 1 month ago, I had the reflection of the dentist who said to me: ‘it is up to you to do the oral hygiene of your daughter’. But it’s not easy with a child with a disability to brush her teeth, it’s not at all obvious. And when I explained that they got angry at me. They said, “Yes, you are responsible for your child. Do you realize, after 1 year, she will have cavities again”, I said I knew it very well.“ “It’s true that disability is already scary, and I think they are not trained.“ “In private practice, anyway, I think there’s the number to go behind. If they spend three quarters of an hour doing a treatment and they don’t succeed, it’s not profitable for them. I’m not saying this well, but afterwards, they have expenses, which is also understandable. So they don’t necessarily have the time or the desire, or the diplomacy, to take care of children with disabilities. “She was screaming so much that the dentist said that She was scaring all his clients.“ Orality disorders (concepts mentioned in 12 interviews, with 40 occurrences). Many oral problems were mentioned by parents. These difficulties with the oral sphere concerned all patient functions: feeding, cleaning, speaking. In this context, oral care was complicated, even in children who did not have intellectual delays. “She doesn’t like it when you put something in her mouth. Better now, but a year or two ago, she didn’t like anything in her mouth. Even the toothbrush was difficult. I was the one who had to brush, but that was hard too. She would gag every time you went in the back a little bit. It was very, very, very difficult. “Since baby, she always had trouble putting things in her mouth. Foods. Sometimes foods that were a little greasy or whatever, it was hard. It was always a heartache. “It’s always been hard on the food part since she was a little girl. Even now. It’s hard to identify something, to say she doesn’t like this texture, or she doesn’t like this consistency, etc., because it’s kind of like it’s diffuse and it’s very hard to identify the structuring. I say anything: she doesn’t like orange and she won’t eat carrots. It evolves and we’re not sure why.“ “She had hyper nausea. Chewing was very complicated, even now she can chew, but it shouldn’t be too…” . “From the moment it was to eat, she took her pacifier, I had a pacifier where we could put milk, medicine and she spit out the pacifier. It wasn’t the same function anymore; we had changed the function of the pacifier.“ “Care was impossible because you had to get into the mouth.“ “We were behind in getting good oral hygiene, because he had sensory dysorality syndrome.“ “As a result, nothing would fit in his mouth. He had his sensitivity threshold at the entrance to his mouth. In fact, he had bitten me once, so he never teethed on things. “He was gagging. The toothbrush, it wasn’t even worth it.“ “Chewing is not at its best. If he can swallow whole, he does. I have to remind him to chew.“ “To chew, it’s very complicated.“ “He had periods when he could eat, because we weren’t sure how to do it, whether to eat, not eat, infuse, enteral, eat a little bit. He was in occlusion very regularly and as a result, this really altered his orality in the sense that he did not develop his feeding as a toddler, where you gradually integrate milk, certain foods, textures, etc. In relation to that, he developed food phobias.“ “It was intrusive, and then because he wasn’t eating, it was intrusive to put in his mouth. It was complicated, I would say that he hasn’t been in the habit for very long, and you still have to be behind it, because it’s not something natural. “Yes. The mouth noises, she can’t stand, she doesn’t eat with us anymore. She already can’t stand herself, hearing her noises.“ “No, I think it’s subsiding. I think it’s people who are learning, who are less excited. Afterwards, it’s also Alexandre’s character that wants that, he’s a dynamic boy, who likes it when it pulsates, who likes it when it goes fast. Afterwards, I think that as we grow up we learn to settle down and the FOP forces us to settle down even more. Social impact (concepts mentioned in 10 interviews, with 30 occurrences). The patients and their parents did not return to the theme of social and school integration as much as to the themes of the dental care pathway. Nevertheless, the notions of the look of others, of appearance, of others were raised, especially by the older patients. “My teeth, I don’t have a nice smile like some of my friends. They have beautiful teeth, a beautiful smile, and I am too ashamed to smile in front of others. To laugh, I put my hand in front of my mouth, it has become a huge complex. “That’s also the flaw when you’re a parent, it’s that you have your nose in the grindstone so much, I realize that with the teachers, when we explain things, it seems so easy. It’s like, I think, with dentists, you don’t want to stretch, you don’t want to spread it too far, you don’t want to… and in fact, we don’t necessarily think to say it.“ “Because when you’re talking, people… sometimes you don’t realize, but sometimes there are people, they don’t look you in the eye, but they look at your lips or your teeth, how they look. And then, when you’re done talking, they’ll say: “But why are your teeth like that? “They are friends with me because of who I am and not because of what they see. They are friends with me for who I am, not for what they see. For my character, for my good mood, not for my disabilities, because I can’t go to Paris and I can’t take the metro. They know that, if they go to Paris, they won’t offer it to me, because it will be a direct “no”. But for example, if we go somewhere to eat, we go by bus, we’ve already thought about everything, we’re not going to do anything too complicated.“
Discussion of results The results of this study can be summarized in three major several points. First point, in our study, patients with a rare disease including a behavioral disorder (neurodevelopmental disorder, delayed acquisition) have a lack of access to dental care in appropriate conditions such as sedation and a team accustomed to this care. The oral management of patients with cognitive and mental disabilities goes far beyond the notion of a rare disease. Barriers to care are permanent for patients and their families with care structures not always adapted to receive these patients. For these patients, a therapeutic arsenal of sedation, whether conscious (hypnosis, nitrous oxide) or general anesthesia, is necessary. This requires adapted care premises, adequate equipment, trained professionals, and care teams that are often larger than for patients without these disabilities . In this type of theme, where the feelings of the patients and their parents are major elements of understanding the problems, qualitative research provides a detailed understanding of patients’ perspectives and expectations and can be an essential first step in the development of future patient-centered measures. Qualitative research and its methods provide opportunities for a systemic and holistic understanding of the difficulties faced by patients. Qualitative methods are increasingly used in the medical literature to understand the issues of patients’ daily lives and care pathways . It seems interesting, and it is an originality of this study, to consider all the rare diseases as a whole and to give a global approach which is opposed to the usual literature which is interested in the diagnostic and therapeutic elements in each disease. This creates in these patients a renunciation of dental care, an oral health judged degraded by the parents and resentment expressed by the parents as well. The parents describe the very chaotic dental care pathway for these patients with difficult but often indispensable access to general anesthesia for care. In addition, in our study, the difficulties of oral management of patients with orality disorders, often caused by parenteral nutrition or gastrostomy, are an important point raised by the parents. Complications related to feeding, food choice, oral hygiene, and acceptance of oral care, even in children without neurodevelopmental delays, appear to be major. The second main point is that, overall, patients describe a lack of knowledge of their pathologies by oral health professionals. The complexity of management for professionals in terms of oral health is multiple. On the one hand, the rarity of the disease requires the patient to organize a care network that is often complex in terms of access to competent professionals trained in the management of this disease, and on the other hand, professionals who are not trained in the phenotypic specificity of this disease may have limitations in their management and direct patients in unsatisfactory therapeutic directions . These difficulties appear clearly in patients with oral disorders who have been parenterally fed, e.g., with biliary atresia. These patients have an altered relationship to all their oral sphere and inappropriate management will necessarily have consequences on the oral quality of life of patients and their general quality of life . The third important result of this study concerns the difficulties of access to adequate dental care for patients with rare diseases with a significant oral and dental component. The difficulties are more technical for the professionals than behavioral. For example, concerning patients with fibrodysplasia ossificans progressiva , an ultra-rare bone disease, where heterotopic bone growths can cause ankylosis of the temporo-mandibular joints, in case of dental care, all expressed fears about dental care and a great ignorance of their pathology by the dental surgeons [ – ]. These elements caused a delay in access to oral health care and a therapeutic wandering until the adequate professionals were found. Strengths Few studies have investigated determinants of dental care use of these patients, both children and adults. These studies use quantitative methods, the main measure being standardized questionnaires . These questionnaires, although very useful for guiding and highlighting certain aspects of the impact of the disease on oral quality of life, are nevertheless limited in their precision concerning the feelings and experiences expressed by the patients. For these diseases, there is no recent data in the literature from qualitative work based on interviews or focus groups. All the patients included in the study are followed in a hospital very specialized in the management of rare diseases with many centers of expertise. For most of them, the care network is nowadays rather well established, even in the field of oral health. We can consider this point as a bias in the analysis of the study, however, many of the patients’ verbatims concern the difficulties they may have had and still have in accessing adequate dental care, which shows that even if they are very well taken care of for their pathology in a global way, oral health remains an often very chaotic point. Limitations Necker Hospital in Paris, where this study was conducted, receives patients with very rare diseases such as Fibrodysplasia ossificans progressiva. We wanted to include these patients whose dental care pathway is quite singular given the specificity of this pathology (TMJ ankylosis can be caused by iatrogenic dental care). We are fully aware that, given the rarity of this disease but also the fact that teams very specialized in this pathology are present on site, we have a much higher recruitment than what we could have in the rest of France. This could be considered as a selection bias. Nevertheless, the observations we make about the complex oral care pathways for patients with mental disabilities, much more than the notion of a rare disease itself, depend on the territorial coverage of specialized services to receive this type of patient, potential geographical difficulties, and the possibility of accessing care under general anesthesia. We can see from this study, and the literature has already shown it in many aspects, that the “rarity” of a disease is not the most difficult point regarding the consideration of patients’ oral health. However, it appears that one of the levers for improving care paths must be through: A better knowledge of the oral care of patients with rare diseases. A real consideration and reflection on the oral care of patients with neurodevelopmental and intellectual delays whose access to care is clearly lacking. The promotion of oral health in the expertise center network not directly linked to the TETECOU network (TETECOU network” is a network of French centers of expertise taking care of rare diseases of the head, neck, and teeth): prevention, therapeutic education, awareness of professionals. Improvement of oral health care pathways: relationship between the TETECOU network and other rare disease health networks. To create new, inter rare disease disciplinary consultations (inter-channels): thanks to the collaboration between the different rare disease health channels, clinical research protocols can be developed as close as possible to the needs of patients: joint consultation between dentistry and other specialties, evaluation of oral rehabilitation for patients whose oral needs have not been considered until now. In France, and in some European countries, “national rare disease plans” follow one another and offer more and more territorial networking, centers of expertise and access to diagnosis and care for patients with rare diseases. As far as oral health is concerned, the stakes remain high as regards the most appropriate care for the different vulnerabilities of patients. It seems essential to enhance the care of patients with disabilities by improving the training of professionals, facilitating the reception of patients, and offering financial incentives to professionals. Inclusion efforts must be global and involve, to a large extent, access to care worthy of the rest of the health system. Through this study, we hope to contribute to the coordination of different care networks on a national scale to significantly improve patients’ daily lives. The identification of psychosocial repercussions of their disability should lead to a broadening of current care with the involvement of social workers and psychologists, but above all to the institutional organization of financial support for oral rehabilitation.
The results of this study can be summarized in three major several points. First point, in our study, patients with a rare disease including a behavioral disorder (neurodevelopmental disorder, delayed acquisition) have a lack of access to dental care in appropriate conditions such as sedation and a team accustomed to this care. The oral management of patients with cognitive and mental disabilities goes far beyond the notion of a rare disease. Barriers to care are permanent for patients and their families with care structures not always adapted to receive these patients. For these patients, a therapeutic arsenal of sedation, whether conscious (hypnosis, nitrous oxide) or general anesthesia, is necessary. This requires adapted care premises, adequate equipment, trained professionals, and care teams that are often larger than for patients without these disabilities . In this type of theme, where the feelings of the patients and their parents are major elements of understanding the problems, qualitative research provides a detailed understanding of patients’ perspectives and expectations and can be an essential first step in the development of future patient-centered measures. Qualitative research and its methods provide opportunities for a systemic and holistic understanding of the difficulties faced by patients. Qualitative methods are increasingly used in the medical literature to understand the issues of patients’ daily lives and care pathways . It seems interesting, and it is an originality of this study, to consider all the rare diseases as a whole and to give a global approach which is opposed to the usual literature which is interested in the diagnostic and therapeutic elements in each disease. This creates in these patients a renunciation of dental care, an oral health judged degraded by the parents and resentment expressed by the parents as well. The parents describe the very chaotic dental care pathway for these patients with difficult but often indispensable access to general anesthesia for care. In addition, in our study, the difficulties of oral management of patients with orality disorders, often caused by parenteral nutrition or gastrostomy, are an important point raised by the parents. Complications related to feeding, food choice, oral hygiene, and acceptance of oral care, even in children without neurodevelopmental delays, appear to be major. The second main point is that, overall, patients describe a lack of knowledge of their pathologies by oral health professionals. The complexity of management for professionals in terms of oral health is multiple. On the one hand, the rarity of the disease requires the patient to organize a care network that is often complex in terms of access to competent professionals trained in the management of this disease, and on the other hand, professionals who are not trained in the phenotypic specificity of this disease may have limitations in their management and direct patients in unsatisfactory therapeutic directions . These difficulties appear clearly in patients with oral disorders who have been parenterally fed, e.g., with biliary atresia. These patients have an altered relationship to all their oral sphere and inappropriate management will necessarily have consequences on the oral quality of life of patients and their general quality of life . The third important result of this study concerns the difficulties of access to adequate dental care for patients with rare diseases with a significant oral and dental component. The difficulties are more technical for the professionals than behavioral. For example, concerning patients with fibrodysplasia ossificans progressiva , an ultra-rare bone disease, where heterotopic bone growths can cause ankylosis of the temporo-mandibular joints, in case of dental care, all expressed fears about dental care and a great ignorance of their pathology by the dental surgeons [ – ]. These elements caused a delay in access to oral health care and a therapeutic wandering until the adequate professionals were found.
Few studies have investigated determinants of dental care use of these patients, both children and adults. These studies use quantitative methods, the main measure being standardized questionnaires . These questionnaires, although very useful for guiding and highlighting certain aspects of the impact of the disease on oral quality of life, are nevertheless limited in their precision concerning the feelings and experiences expressed by the patients. For these diseases, there is no recent data in the literature from qualitative work based on interviews or focus groups. All the patients included in the study are followed in a hospital very specialized in the management of rare diseases with many centers of expertise. For most of them, the care network is nowadays rather well established, even in the field of oral health. We can consider this point as a bias in the analysis of the study, however, many of the patients’ verbatims concern the difficulties they may have had and still have in accessing adequate dental care, which shows that even if they are very well taken care of for their pathology in a global way, oral health remains an often very chaotic point.
Necker Hospital in Paris, where this study was conducted, receives patients with very rare diseases such as Fibrodysplasia ossificans progressiva. We wanted to include these patients whose dental care pathway is quite singular given the specificity of this pathology (TMJ ankylosis can be caused by iatrogenic dental care). We are fully aware that, given the rarity of this disease but also the fact that teams very specialized in this pathology are present on site, we have a much higher recruitment than what we could have in the rest of France. This could be considered as a selection bias. Nevertheless, the observations we make about the complex oral care pathways for patients with mental disabilities, much more than the notion of a rare disease itself, depend on the territorial coverage of specialized services to receive this type of patient, potential geographical difficulties, and the possibility of accessing care under general anesthesia. We can see from this study, and the literature has already shown it in many aspects, that the “rarity” of a disease is not the most difficult point regarding the consideration of patients’ oral health. However, it appears that one of the levers for improving care paths must be through: A better knowledge of the oral care of patients with rare diseases. A real consideration and reflection on the oral care of patients with neurodevelopmental and intellectual delays whose access to care is clearly lacking. The promotion of oral health in the expertise center network not directly linked to the TETECOU network (TETECOU network” is a network of French centers of expertise taking care of rare diseases of the head, neck, and teeth): prevention, therapeutic education, awareness of professionals. Improvement of oral health care pathways: relationship between the TETECOU network and other rare disease health networks. To create new, inter rare disease disciplinary consultations (inter-channels): thanks to the collaboration between the different rare disease health channels, clinical research protocols can be developed as close as possible to the needs of patients: joint consultation between dentistry and other specialties, evaluation of oral rehabilitation for patients whose oral needs have not been considered until now. In France, and in some European countries, “national rare disease plans” follow one another and offer more and more territorial networking, centers of expertise and access to diagnosis and care for patients with rare diseases. As far as oral health is concerned, the stakes remain high as regards the most appropriate care for the different vulnerabilities of patients. It seems essential to enhance the care of patients with disabilities by improving the training of professionals, facilitating the reception of patients, and offering financial incentives to professionals. Inclusion efforts must be global and involve, to a large extent, access to care worthy of the rest of the health system. Through this study, we hope to contribute to the coordination of different care networks on a national scale to significantly improve patients’ daily lives. The identification of psychosocial repercussions of their disability should lead to a broadening of current care with the involvement of social workers and psychologists, but above all to the institutional organization of financial support for oral rehabilitation.
This study exceeded its objectives in many ways. We thought that we should once again insist on the “rare” side of the diseases concerned by our sample. We were almost surprised, even if we knew the existence of difficulties, the more than chaotic course of the care of patients with cognitive disabilities. It seems necessary to be able to promote oral health for all, in the best conditions, even when these are costly: sedations of all kinds, long sessions, general anesthesia, assistance by trained professionals (psychologists, speech therapists). Oral health care cannot be another reason for therapeutic wandering for these patients who have often already experienced diagnostic wandering for their pathology.
|
Acid Sphingomyelinase Regulates AdipoRon-Induced Differentiation of Arterial Smooth Muscle Cells via TFEB Activation | 707b2a62-a1fa-495f-81a7-f013c0cfa03b | 11899876 | Cardiovascular System[mh] | Vascular smooth muscle cells (SMCs) are the main cell type in the medial layer of healthy arteries, mainly in a quiescent state and highly differentiated . Vascular SMCs are essential for maintaining vascular integrity and homeostasis, and they are undoubtedly involved in physiological and pathological vascular remodeling [ , , ]. An increasing number of studies have shown that the aberrant proliferation and migration of vascular SMCs play a key role in the development of cardiovascular diseases (CVDs), such as atherosclerosis , inflammation , restenosis , and aortic aneurysm . Under these pathophysiological conditions, SMCs dedifferentiate from a contractile phenotype to a synthetic phenotype, resulting in enhanced proliferation and migration into the intima, thereby promoting neointimal formation . Hence, it is necessary to develop effective therapeutic approaches to inhibit SMC phenotypic dedifferentiation and explore its potential regulatory mechanisms. AdipoRon is a selective adiponectin receptor agonist that is widely suitable for clinical research and research that uses it to mimic the beneficial effects of adiponectin . For example, adipoRon may improve insulin resistance and lipotoxicity in type 2 diabetic mice by activating the AdipoR1/2 signaling pathway . AdipoRon exerts an anti-apoptotic effect during cardiac ischemia/reperfusion injury by partially activating AMPK . AdipoRon not only exhibits anti-atherosclerotic and anti-inflammatory effects through inhibiting ERK and p38 MAPK activation to reduce the proliferation/migration of vascular SMCs and the levels of pro-inflammatory factors but also suppresses vascular SMC proliferation and neointimal hyperplasia by inhibiting mTORC1/p70S6K signaling . Notably, our recent studies showed that adipoRon promotes SMC differentiation by activating transcription factor EB (TFEB), a master transcription regulator of autophagy–lysosomal signaling . TFEB activation leads to its nuclear translocation and binding to coordinated lysosomal expression and regulation (CLEAR) elements of autophagic genes, including microtubule-associated protein light chain 3 (LC3), ubiquitin-binding protein p62/sequestosome 1 (p62/SQSTM1), and lysosomal-associated membrane protein 1 (LAMP-1). This activation of TFEB–autophagy signaling has been shown to contribute to the inhibitory effects of adipoRon on vascular SMC proliferation and migration . Further studies have found that adipoRon-induced TFEB activation in SMCs was mainly dependent on intracellular calcium ions but not on kinases such as AMPK, ERK1/2, Akt, and mTORC1 . However, the specific mechanism of adipoRon-induced TFEB activation and how intracellular calcium ions are involved in this process remain unclear. Acid sphingomyelinase (ASM; gene symbol Smpd1 ) is a lysosomal hydrolase that metabolizes sphingomyelin to ceramide and phosphorylcholine, preferentially at an acidic pH . A genetic defect in ASM leads to the accumulation of sphingomyelin and a lysosomal storage disorder named Niemann–Pick disease. Our previous studies have demonstrated a protective role of ASM in maintaining SMC homeostasis by controlling autophagy signaling . It was found that ASM deficiency impaired autophagic flux by preventing TRPML1-Ca 2+ -dependent lysosome trafficking and its fusion with autophagosome to form autophagolysosomes . Impaired autophagic flux was also associated with increased SMC proliferation and their transformation to a myofibroblast-like phenotype . Interestingly, a recent study has shown that inhibition of ASM activates TFEB in endothelial cells . The present study aimed to explore the role of ASM in adipoRon-induced TFEB–autophagy signaling in SMCs and the associated regulatory mechanisms. Therefore, we first investigated the effects of ASM deficiency by genetic ablation of Smpd1 on adipoRon-induced TFEB activation in primary cultured arterial SMCs, as well as the associated inhibition of SMC proliferation and migration. Moreover, we tested the role of Ca 2+ -dependent phosphatase calcineurin and ceramide-activated protein phosphatase 2A (PP2A) in adipoRon-induced activation of TFEB and inhibition of SMC proliferation and migration. We also examined whether TRPML1-Ca 2+ channel activation could rescue the beneficial effects of adipoRon in ASM-deficient SMCs. The findings from this study provide novel mechanistic insights into the therapeutic effects of adipoRon on TFEB signaling and pathological vascular remodeling.
2.1. AdipoRon Induces ASM-Mediated Ceramide Signaling in SMCs Previous studies indicate that adipoRon promotes SMC differentiation through TFEB activation . Here, we investigated the mediating role of ASM in the effects of adipoRon on SMCs. Smpd1 +/+ and Smpd1 −/− SMCs were treated with or without adipoRon (50 μM) for 24 h, and changes in ASM expression and ceramide production were examined. As shown in A,B, immunofluorescent results showed that adipoRon significantly increased the protein expression of ASM in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. Consistently, ceramide production was significantly increased in Smpd1 +/+ SMCs treated with adipoRon, while no significant changes were observed in Smpd1 −/− SMCs ( D,E). Additionally, we examined adipoRon receptor 1 (AdipoR1, C) and AdipoR2 ( F) expressions and found no significant changes in Smpd1 +/+ and Smpd1 −/− SMCs after adipoRon treatment. Together, these results indicate that adipoRon regulates the expression of ASM and ceramide production in arterial SMCs, while Smpd1 gene ablation can eliminate this regulation. 2.2. ASM Deficiency Inhibits adipoRon-Induced TFEB Activation and Autophagy in SMCs Next, we investigated the role of ASM in adipoRon-induced TFEB activation and autophagy. We first examined the effect of adipoRon on TFEB nuclear translocation, a key event in TFEB activation, in Smpd1 +/+ and Smpd1 −/− SMCs. As shown in A,B, immunofluorescence studies showed that adipoRon significantly increased the nuclear translocation of TFEB in Smpd1 +/+ SMCs, whereas this TFEB nuclear translocation was absent in Smpd1 −/− SMCs. We further analyzed the mRNA levels of TFEB and its downstream autophagy and lysosomal genes. The results showed that adipoRon significantly increased the mRNA levels of TFEB ( C) and its target genes, including LC3 ( E), LAMP1 ( D), and p62 ( F) in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. Consistent with the increase in mRNA levels, adipoRon significantly increased the protein expression of LC3-II and p62, which were inhibited in Smpd1 −/− SMCs ( G–J). Together, these results suggest that ASM activity is required for adipoRon-induced TFEB activation and autophagy signaling in arterial SMC. 2.3. ASM Deficiency Prevents the Inhibitory Effects of adipoRon on SMC Proliferation and Migration Migration was first assessed by scratch assay in adipoRon-treated Smpd1 +/+ and Smpd1 −/− SMCs. As shown in A,B, adipoRon significantly decreased cell migration in Smpd1 +/+ SMCs, whereas this decrease was abolished in Smpd1 −/− SMCs. MMPs are peptidase enzymes involved in extracellular matrix degradation contributing to cell migration. AdipoRon inhibited MMP activity in Smpd1 +/+ SMCs, but not in Smpd1 −/− SMCs ( F), demonstrating ASM deficiency prevents adipoRon-induced inhibition of the migration in SMCs. The remodeling of F-actin, a filamentous actin in the cytoskeleton, is a marker event associated with cell migration. Both Smpd1 +/+ and Smpd1 −/− SMCs without adipoRon treatment exhibited a migratory phenotype with disassembled distribution and aggregation around the perinuclear region of actin filaments without clear filamentous organization ( E). However, Smpd1 +/+ SMCs treated with adipoRon showed a spindle-like shape and organization of the actin filaments ( E). This adipoRon-induced F-actin reorganization was not observed in Smpd1 −/− SMCs ( E). The anti-proliferative effect of adipoRon on SMCs was further examined by the immunostaining of proliferation marker Ki67 and by cell counting. As shown in C,D, adipoRon significantly decreased the expression of Ki67 in Smpd1 +/+ SMCs, which was not observed in Smpd1 −/− SMCs. Consistently, adipoRon significantly arrested the cell growth of reduced cell numbers in Smpd1 +/+ SMCs but had no further effect on Smpd1 −/− SMCs ( G). Additionally, the expression levels of SMC differentiation marker alpha-SMA ( H) and SM22 ( I) were also examined, and it was found that adipoRon significantly increased their expressions in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. 2.4. Effect of Calcineurin Inhibition on adipoRon-Induced TFEB–Autophagy in SMCs Our previous studies have shown that adipoRon-induced TFEB activation is dependent on intracellular Ca 2+ . As shown in A,B, immunoblotting results showed that adipoRon significantly increased the expression of the Ca 2+ -dependent phosphatase calcineurin in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. We then investigated the effects of two calcineurin inhibitors, FK506 and cyclosporin A, on adipoRon-induced TFEB nuclear translocation and LC3 expression in Smpd1 +/+ SMCs. Immunofluorescence studies showed that FK506 ( C,D) or cyclosporin A ( G,H) significantly attenuated adipoRon-induced TFEB nuclear translocation. Similarly, FK506 ( E,F) or cyclosporin A ( I,J) inhibited adipoRon-induced LC-3II expression. Together, these data suggest that calcineurin acts as a downstream effector of the ASM–ceramide pathway to activate TFEB–autophagy signaling in SMCs. 2.5. Effect of PP2A Inhibition on adipoRon-Induced TFEB–Autophagy in SMCs Ceramide has been shown to exert an anti-proliferative effect through PP2A activation . Interestingly, recent studies have reported that PP2A can dephosphorylate TFEB and mediate oxidative-stress-induced TFEB activation in epithelial cells . We sought to investigate whether PP2A plays a role in adipoRon-induced TFEB activation in SMCs. As shown in A,B, adipoRon significantly induced PP2A expression in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. Furthermore, inhibition of PP2A by okadaic acid significantly attenuated adipoRon-induced TFEB nuclear translocation ( C,D) and LC-3II expression in Smpd1 +/+ SMCs ( E,F). Notably, okadaic acid had no effect on the protein expression of p62 in SMCs ( E,G). Together, these data indicate that activation of PP2A may also contribute to adipoRon-induced TFEB activation and autophagy in SMCs. 2.6. Lysosomal Ca 2+ Release by ML-SA1 Rescues adipoRon-Induced Activation of Calcineurin and TFEB in Smpd1 −/− SMCs Lysosomal TRPML1-Ca 2+ release has been shown to activate the calcineurin–TFEB signaling axis . Here, Smpd1 −/− SMCs were treated with a TRPML1 channel activator, ML-SA1, and we examined its impact on adipoRon-induced effects. As shown in A,B, ML-SA1 significantly induced calcineurin expressions in Smpd1 −/− SMCs to a similar level in the presence or absence of adipoRon. Consistently, ML-SA1 significantly increased TFEB nuclear translocation ( C,D) and protein expression of LC3-II and p62 ( F–H) in Smpd1 −/− SMCs treated with or without adipoRon. Furthermore, wound scratch assay showed that ML-SA1 significantly decreased cell migration ( I,J) and inhibited cell proliferation ( E) in Smpd1 −/− SMCs treated with or without adipoRon. These results suggest that ML-SA1 activates calcineurin–TFEB–autophagy signaling and promotes differentiation in Smpd1 −/− SMCs regardless of adipoRon treatment.
Previous studies indicate that adipoRon promotes SMC differentiation through TFEB activation . Here, we investigated the mediating role of ASM in the effects of adipoRon on SMCs. Smpd1 +/+ and Smpd1 −/− SMCs were treated with or without adipoRon (50 μM) for 24 h, and changes in ASM expression and ceramide production were examined. As shown in A,B, immunofluorescent results showed that adipoRon significantly increased the protein expression of ASM in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. Consistently, ceramide production was significantly increased in Smpd1 +/+ SMCs treated with adipoRon, while no significant changes were observed in Smpd1 −/− SMCs ( D,E). Additionally, we examined adipoRon receptor 1 (AdipoR1, C) and AdipoR2 ( F) expressions and found no significant changes in Smpd1 +/+ and Smpd1 −/− SMCs after adipoRon treatment. Together, these results indicate that adipoRon regulates the expression of ASM and ceramide production in arterial SMCs, while Smpd1 gene ablation can eliminate this regulation.
Next, we investigated the role of ASM in adipoRon-induced TFEB activation and autophagy. We first examined the effect of adipoRon on TFEB nuclear translocation, a key event in TFEB activation, in Smpd1 +/+ and Smpd1 −/− SMCs. As shown in A,B, immunofluorescence studies showed that adipoRon significantly increased the nuclear translocation of TFEB in Smpd1 +/+ SMCs, whereas this TFEB nuclear translocation was absent in Smpd1 −/− SMCs. We further analyzed the mRNA levels of TFEB and its downstream autophagy and lysosomal genes. The results showed that adipoRon significantly increased the mRNA levels of TFEB ( C) and its target genes, including LC3 ( E), LAMP1 ( D), and p62 ( F) in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. Consistent with the increase in mRNA levels, adipoRon significantly increased the protein expression of LC3-II and p62, which were inhibited in Smpd1 −/− SMCs ( G–J). Together, these results suggest that ASM activity is required for adipoRon-induced TFEB activation and autophagy signaling in arterial SMC.
Migration was first assessed by scratch assay in adipoRon-treated Smpd1 +/+ and Smpd1 −/− SMCs. As shown in A,B, adipoRon significantly decreased cell migration in Smpd1 +/+ SMCs, whereas this decrease was abolished in Smpd1 −/− SMCs. MMPs are peptidase enzymes involved in extracellular matrix degradation contributing to cell migration. AdipoRon inhibited MMP activity in Smpd1 +/+ SMCs, but not in Smpd1 −/− SMCs ( F), demonstrating ASM deficiency prevents adipoRon-induced inhibition of the migration in SMCs. The remodeling of F-actin, a filamentous actin in the cytoskeleton, is a marker event associated with cell migration. Both Smpd1 +/+ and Smpd1 −/− SMCs without adipoRon treatment exhibited a migratory phenotype with disassembled distribution and aggregation around the perinuclear region of actin filaments without clear filamentous organization ( E). However, Smpd1 +/+ SMCs treated with adipoRon showed a spindle-like shape and organization of the actin filaments ( E). This adipoRon-induced F-actin reorganization was not observed in Smpd1 −/− SMCs ( E). The anti-proliferative effect of adipoRon on SMCs was further examined by the immunostaining of proliferation marker Ki67 and by cell counting. As shown in C,D, adipoRon significantly decreased the expression of Ki67 in Smpd1 +/+ SMCs, which was not observed in Smpd1 −/− SMCs. Consistently, adipoRon significantly arrested the cell growth of reduced cell numbers in Smpd1 +/+ SMCs but had no further effect on Smpd1 −/− SMCs ( G). Additionally, the expression levels of SMC differentiation marker alpha-SMA ( H) and SM22 ( I) were also examined, and it was found that adipoRon significantly increased their expressions in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs.
Our previous studies have shown that adipoRon-induced TFEB activation is dependent on intracellular Ca 2+ . As shown in A,B, immunoblotting results showed that adipoRon significantly increased the expression of the Ca 2+ -dependent phosphatase calcineurin in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. We then investigated the effects of two calcineurin inhibitors, FK506 and cyclosporin A, on adipoRon-induced TFEB nuclear translocation and LC3 expression in Smpd1 +/+ SMCs. Immunofluorescence studies showed that FK506 ( C,D) or cyclosporin A ( G,H) significantly attenuated adipoRon-induced TFEB nuclear translocation. Similarly, FK506 ( E,F) or cyclosporin A ( I,J) inhibited adipoRon-induced LC-3II expression. Together, these data suggest that calcineurin acts as a downstream effector of the ASM–ceramide pathway to activate TFEB–autophagy signaling in SMCs.
Ceramide has been shown to exert an anti-proliferative effect through PP2A activation . Interestingly, recent studies have reported that PP2A can dephosphorylate TFEB and mediate oxidative-stress-induced TFEB activation in epithelial cells . We sought to investigate whether PP2A plays a role in adipoRon-induced TFEB activation in SMCs. As shown in A,B, adipoRon significantly induced PP2A expression in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. Furthermore, inhibition of PP2A by okadaic acid significantly attenuated adipoRon-induced TFEB nuclear translocation ( C,D) and LC-3II expression in Smpd1 +/+ SMCs ( E,F). Notably, okadaic acid had no effect on the protein expression of p62 in SMCs ( E,G). Together, these data indicate that activation of PP2A may also contribute to adipoRon-induced TFEB activation and autophagy in SMCs.
2+ Release by ML-SA1 Rescues adipoRon-Induced Activation of Calcineurin and TFEB in Smpd1 −/− SMCs Lysosomal TRPML1-Ca 2+ release has been shown to activate the calcineurin–TFEB signaling axis . Here, Smpd1 −/− SMCs were treated with a TRPML1 channel activator, ML-SA1, and we examined its impact on adipoRon-induced effects. As shown in A,B, ML-SA1 significantly induced calcineurin expressions in Smpd1 −/− SMCs to a similar level in the presence or absence of adipoRon. Consistently, ML-SA1 significantly increased TFEB nuclear translocation ( C,D) and protein expression of LC3-II and p62 ( F–H) in Smpd1 −/− SMCs treated with or without adipoRon. Furthermore, wound scratch assay showed that ML-SA1 significantly decreased cell migration ( I,J) and inhibited cell proliferation ( E) in Smpd1 −/− SMCs treated with or without adipoRon. These results suggest that ML-SA1 activates calcineurin–TFEB–autophagy signaling and promotes differentiation in Smpd1 −/− SMCs regardless of adipoRon treatment.
The aim of this study was to determine the role of ASM in regulating adipoRon-induced TFEB–autophagy signaling and SMC differentiation. Our studies demonstrated that ASM deficiency by genetic ablation of the Smpd1 gene abolished adipoRon-induced TFEB–autophagy signaling and prevented its inhibitory effects on SMC proliferation and migration. The role of ASM in these adipoRon effects is associated with the upregulation of protein phosphatase calcineurin and PP2A. The lysosomal TRPML1-Ca 2+ channel agonist ML-SA1 effectively activated calcineurin and TFEB–autophagy signaling in Smpd1 −/− SMCs and rescued the inhibitory effects of adipoRon on the proliferation and migration of these cells. These results suggest that ASM regulates adipoRon-induced TFEB–autophagy signaling and SMC differentiation through activating protein phosphatases such as calcineurin and PP2A ( ). Obesity is closely associated with CVD mortality, with more than two-thirds of obese patients dying of CVDs . The high risk of CVD in obese patients is due to changes in cardiac and vascular structure and function . Adipokines are bioactive substances produced by adipose tissue . One of the main reasons for CVD development in obese patients is adipokine disorder, which results in insufficient production of beneficial adipokines or excessive production of deleterious adipokines. Adipokines can regulate the phenotype of SMCs between differentiated and dedifferentiated states upon various stimulations . Adiponectin is a major plasma adipokine that has beneficial effects on CVDs. However, the short half-life and large molecular weight of adiponectin limit its clinical application . AdipoRon is an active synthetic agonist of adiponectin receptor 1/2 (AdipoR1/2) that mimics the biological effects of adiponectin . Recent studies have shown that adipoRon induces SMC differentiation, inhibits SMC proliferation and migration, and attenuates neointima formation in femoral arteries in mice . However, in vascular SMCs, the downstream effectors that mediate the adipoRon agonism of AdipoR1/2 remain poorly defined. In the present study, we demonstrated for the first time that adipoRon increased the expression of ASM and ceramide levels in Smpd1 +/+ SMCs, and these adipoRon-induced changes were abolished in Smpd1 −/− SMCs. Therefore, our results suggest that adipoRon activates the ASM–ceramide signaling pathway in vascular SMCs. AdipoRon activates downstream signaling pathways by binding to AdipoR1 and AdipoR2, which are essential for the differentiation of SMCs . Our present result demonstrates that the mRNA expression levels of AdipoR1 and AdipoR2 are comparable across all groups, indicating that both receptors are expressed in SMCs. The present study did not attempt to identify the specific receptor subtype mediating adipoRon’s effects on SMCs or to delineate the mechanism by which adipoRon–receptor interaction leads to the activation of the ASM–ceramide signaling pathway. Our previous studies have highlighted the pivotal role of intracellular Ca 2+ in adipoRon-induced TFEB activation in vascular SMCs . Additionally, previous studies have demonstrated that elevated intracellular Ca2+ enhances ASM activation by promoting lysosomal exocytosis . Moreover, ASM activation can also occur through reactive oxygen species (ROS) and protein kinase Cδ (PKCδ) . It is plausible that these pathways contribute to adipoRon-induced ASM activation in SMCs, which deserve further investigation. The present study further explored the functional role of ASM in adipoRon-induced TFEB–autophagy signaling in SMCs. Here, we demonstrated that ablation of the Smpd1 gene in SMCs significantly blocked adipoRon-induced TFEB nuclear translocation and the increase in mRNA or protein levels of TFEB target genes. Notably, Smpd1 +/+ and Smpd1 −/− SMCs exhibited similar levels of TFEB nuclear translocation, suggesting that ablation of the Smpd1 gene does not induce TFEB activation at baseline. Consistent with our findings, a recent study has shown that ceramide increases TFEB expression and nuclear translocation and induces lysosomal formation and exocytosis in trophoblast cells . In contrast, our study contradicts a recent study that showed that ASM inhibition with imipramine or SMPD1 siRNA markedly increased TFEB activation in human lung endothelial cells under basal conditions. This effect of ASM inhibition on TFEB activation was further attributed to decreased levels of sphingosine and sphingosine-1-phosphate (S1P) and inhibition of mTOR kinase. Moreover, myriocin, an inhibitor of ceramide de novo synthesis, has been shown to activate TFEB, enhancing fatty acid oxidation and promoting autophagy in airway epithelial cells . The discrepancy between these previous studies and ours regarding the role of ASM/ceramide in TFEB activation or inhibition is unclear. Our previous studies have shown that adipoRon-induced TFEB activation is independent of mTOR kinase inhibition but is rather intracellular Ca 2+ -dependent . Therefore, it is plausible that adipoRon activates ASM–ceramide signaling in SMCs, but it may not affect mTOR kinase activity as it does in endothelial cells. Nonetheless, our results indicate that ASM-mediated ceramide promotes adipoRon-induced TFEB–autophagy in vascular SMCs. Given that the ceramide antibody used in the study detects multiple sphingolipid species, including ceramides (C16 and C24), dihydroceramide, sphingomyelin, and phosphatidylcholine, it is crucial to determine which specific sphingolipid is elevated following adipoRon treatment. Previous studies suggest that adipoRon can modulate sphingolipid pathways, particularly by altering ceramide metabolism . If ASM activity is upregulated, it could lead to increased ceramide production from sphingomyelin . Therefore, based on the available data and the role of adipoRon in TFEB activation, an increase in ceramide (particularly C16 and C24 ceramides) is a plausible outcome, potentially contributing to SMC differentiation. Under various pathological conditions, SMCs dedifferentiate into a more proliferative and migrative state, which causes pathological vascular remodeling and leads to CVDs such as atherosclerosis and restenosis . Autophagy is an evolutionarily conserved, repetitive, and dynamic process that degrades and recycles excessive proteins and fragmented organelles through lysosomes . Autophagy plays an important role in cell homeostasis and in maintaining the differentiated state of SMCs by preventing proliferation and migration . The physiological consequences of enhanced autophagy in vascular SMCs have profound implications for cardiovascular function and pathology, particularly in relation to vascular health, arterial remodeling, and cellular stress responses. Autophagy also plays a pivotal role in maintaining vascular SMC homeostasis by regulating survival and function, thereby influencing vascular elasticity, blood pressure, and overall circulatory health . During atherosclerosis progression, moderate autophagic activity facilitates the clearance of damaged cells, promoting vascular integrity. However, excessive autophagy may induce vascular SMC death, compromising arterial wall stability and increasing the risk of plaque rupture [ , , ]. Additionally, heightened autophagy is closely linked to arterial remodeling. Under chronic stress conditions, excessive autophagic activity in vascular SMCs can disrupt cellular function, altering arterial structure and function and promoting atherosclerosis . Autophagy-mediated changes also affect the extracellular matrix composition, further impacting vascular integrity and function . In vascular SMCs, enhanced autophagy plays a crucial role in cellular stress responses. By clearing reactive oxygen species, autophagy mitigates oxidative stress-induced damage [ , , ]. However, excessive autophagy can increase cellular sensitivity to oxidative stress, ultimately inducing cell death . Thus, elucidating the regulatory mechanisms of autophagy in vascular SMCs is essential for developing novel therapeutic strategies to improve cardiovascular disease prognosis. Genetic or pharmacological activation of the TFEB-mediated autophagy–lysosome system has been shown to reduce atherosclerosis in animal models . Our recent studies have also shown that suppression of TFEB promotes SMC dedifferentiation, while activation of TFEB by trehalose or adipoRon promotes SMC differentiation by inhibiting migration and proliferation . In the present study, we demonstrated that ablation of the Smpd1 gene in SMCs abolished the inhibitory effects of adipoRon on SMC proliferation and migration, MMP downregulation, and F-actin reorganization. These data from previous and present studies support the view that the ASM–TFEB autophagy signaling axis mediates adipoRon-induced SMC differentiation. Our previous studies also showed that ASM promotes autophagic flux by enhancing dynein-dependent lysosomal trafficking and fusion with autophagosomes in SMCs, while ablation of the Smpd1 gene impairs autophagic flux associated with enhanced SMC dedifferentiation . However, whether the ASM–TFEB axis regulates the adipoRon-induced differentiation by enhancing autophagic flux remains unclear. TFEB has been shown to upregulate lysosomal transmembrane protein TMEM55B, which recruits JIP4 and in turn activates dynein-dependent lysosomal trafficking . The ASM-TFEB pathway may coordinate lysosomal trafficking and fusion with autophagosomes to facilitate autophagic flux in vascular SMCs and deserves further investigation. Recent studies have shown that calcineurin, a Ca 2+ -activated protein phosphatase, is an important activator of TFEB, which dephosphorylates TFEB and triggers its nuclear translocation . We have reported that adipoRon-induced TFEB activation in SMCs is dependent on intracellular Ca 2+ . In the present study, we explored whether calcineurin is involved in adipoRon-induced TFEB activation in SMCs. The results showed that adipoRon increased the expression of calcineurin in Smpd1 +/+ SMCs but not in Smpd1 −/− SMCs. In addition, inhibition of calcineurin by FK506 or cyclosporin A blocked adipoRon-induced TFEB nuclear translocation and autophagy marker LC3-II expression. Therefore, our data suggest that ASM may control TFEB activation by regulating intracellular Ca 2+ and thus calcineurin activity. The present study did not further identify how the ASM–ceramide pathway regulates intracellular Ca 2+ in SMCs. In this regard, several mechanisms have been proposed. First, lysosomal TRPML1-mediated Ca 2+ release is known to activate calcineurin and TFEB . Sphingolipids such as sphingosine and sphingomyelin, but not ceramide, have been reported to regulate TRPML1 channel activity in various mammalian cells, including endothelial cells and podocytes . Sphingosine has been reported to enhance TRPML1 channel activity and lysosomal Ca 2+ release , whereas sphingomyelin has the opposite effect . It is possible that adipoRon activates ASM–ceramide, thereby increasing lysosomal sphingosine levels through ceramidase-mediated ceramide breakdown and then promoting TRPML1-Ca 2+ release . Conversely, ASM deficiency or inhibition leads to increased lysosomal sphingomyelin levels, which in turn inhibits TRPML1-Ca 2+ release. Second, increased ASM–ceramide may lead to increased production of S1P, which binds to membrane G protein-coupled receptors and triggers inositol triphosphate (IP3)-mediated Ca 2+ release . Third, increased ASM–ceramide may increase cytosolic Ca 2+ by inhibiting the sarco/endoplasmic reticulum (S/ER) Ca 2+ ATPase (SERCA) and depleting the S/ER Ca 2+ pool . Interestingly, inhibition of Ca 2+ ATPase of SERCA by thapsigargin has been shown to potently activate TFEB in SMCs . Thus, increased ASM–ceramide by adipoRon may inhibit SERCA-dependent Ca 2+ uptake into the ER lumen, thereby increasing cytosolic Ca 2+ concentrations. Therefore, ASM–ceramide may have multifactorial effects on intracellular Ca 2+ that controls calcineurin and TFEB activity and contribute to the effects of adipoRon on SMC homeostasis. Moreover, this study showed that ceramide-activated protein phosphatase PP2A is also involved in adipoRon-induced TFEB activation in SMCs. It has been well established that ceramide can activate PP2A, thereby exerting antiproliferation effects [ , , ]. Recent studies have shown that TFEB is activated upon induction of acute oxidative stress by sodium arsenite through an mTORC1-independent but PP2A-dependent process . In the present study, adipoRon increased PP2A expression in SMCs, while Smpd1 gene ablation suppressed this expression. Furthermore, inhibition of PP2A with okadaic acid significantly reduced TFEB nuclear translocation and attenuated LC-3II expression. These results suggest that the ASM–ceramide–PP2A axis plays a role in adipoRon-induced TFEB activation in SMCs. PP2A can directly dephosphorylate TFEB at several serine residues to facilitate TFEB activation . Instead, PP2A may decrease the activity of the Ca 2+ ATPase of the SERCA. SERCA activity is regulated by the inhibitory protein phospholamban in many cell types, including SMCs . Phospholamban can be phosphorylated by calmodulin-dependent protein kinase II (CaMKII) or protein kinase A (PKA) . Phospholamban inhibits SERCA activity, and its phosphorylation results in its dissociation from SERCA and release of inhibition . PP2A has been shown to dephosphorylate phospholamban, thereby inhibiting SERCA . Therefore, it is possible that PP2A inhibits SERCA, thereby increasing cytosolic Ca 2+ and enhancing calcineurin and TFEB activity. The present study further investigated the effects of TRPML1 channel activation on TFEB–autophagy and proliferation and migration in Smpd1 −/− SMCs. Previous studies have demonstrated that ML-SA1 as a TRPML1 channel agonist activated lysosomal Ca 2+ release in vascular SMCs and ECs [ , , ]. Moreover, ML-SA1 enhanced lysosome trafficking and its fusion with late endosomes/multivesicular bodies in these cells. The Ca 2+ bursts of lysosomes can activate global Ca 2+ release from the sarcoplasmic reticulum (SR) to increase cytosolic Ca 2+ , which may be sufficient to drive the dynein-dependent movement of lysosomes along microtubules to encounter other cellular vesicles such as multivesicular bodies or autophagosomes . In the present study, we demonstrated that ML-SA1 induced calcineurin expressions, triggered TFEB nuclear translocation, and increased autophagy signaling in Smpd1 −/− SMCs. ML-SA1-induced calcineurin–TFEB autophagy signaling in Smpd1 −/− SMCs was also accompanied by the inhibition of their proliferation and migration. Recently, sphingolipids were found to regulate TRPML1 channel activity . ASM deficiency may increase substrate lysosomal sphingomyelin but reduce lysosomal ceramide or its metabolites sphingosine and S1P. The ceramide metabolite sphingosine was found to directly enhance TRPML1 channel activity, but ceramide had no effect . Ablation of the Asah1 gene to inhibit acid ceramidase activity and sphingosine production remarkably decreased ML-SA1-induced Ca 2+ release through TRPML1 channels . Shen et al. showed that inhibited TRPML1 activity and a reduction in lysosomal Ca 2+ release observed in Niemann–Pick cells were associated with lysosomal sphingomyelin accumulation . These studies suggest that lysosomal TRPML1-mediated Ca 2+ release is a major contributor to the global Ca 2+ increase, thereby activating calcineurin and TFEB autophagy signaling and promoting SMC differentiation. In addition, these studies implicate that ASM/ceramide serves as an upstream regulator of lysosomal Ca 2+ release in vascular SMCs. Interestingly, the effects of TRPML1 channel activation on TFEB–autophagy and SMC differentiation were observed both at baseline and under adipoRon stimulation. Therefore, these findings support the view that TRPML1 activation by ML-SA1 is sufficient to activate the TFEB–autophagy signaling and induce differentiation in Smpd1 −/− SMCs, thereby rescuing the beneficial effects of adipoRon on SMC hemostasis in the absence of ASM.
4.1. Reagents and Antibodies Information on primary and secondary antibodies for immunoblotting and immunofluorescence staining is provided in . The following reagents were used: adipoRon (ab144867; Abcam, Cambridge, UK) and Triton X-100 (Sigma, Kawasaki, Japan, X100). 4.2. Mice All experimental protocols were reviewed and approved by the Animal Care Committee of the University of Houston. All animals were kept in a standardized manner in the animal center, University of Houston. 4.3. Primary Culture of Arterial SMCs from Mice Mouse arterial Smpd1 +/+ and Smpd1 −/− SMCs were isolated from the corresponding mice as previously described [ , , ]. Briefly, 6-week-old mice were deeply anesthetized with pentobarbital sodium (ip, 25 mg/kg). Mouse hearts were excised and immersed in an ice-cold Krebs–Henseleit (KH) solution. Then, a 25-gauge needle filled with Hanks’ buffered saline solution (HBSS) was inserted into the heart close to the aortic valve through the aortic lumen and tied when the needle tip reached the base of the heart. An infusion pump was started with a 20 mL syringe containing warm HBSS at a rate of 0.1 mL/min for 15 min. Then, the HBSS was replaced with a warm enzyme solution (containing 1 mg/mL collagenase type 1, 0.5 mg/mL soybean trypsin inhibitor, 3.0% BSA, and 2.0% antibiotics) which was flushed through the heart at a rate of 0.1 mL/min. The outflow perfusion fluid was collected at 30, 60, and 90 min intervals. The heart was cut to open the apex to flush out the cells inside the ventricle after collecting all outflow fluid at 90 min. The flushed cells were centrifuged at 1000 rpm for 10 min, and the pellets were resuspended in advanced Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS), 10% mouse serum, and 2.0% antibiotics. The isolated cells were plated on 2.0% gelatin-coated plates and incubated in 5.0% CO 2 at 37 °C. These isolated cells were confirmed as SMCs originated mainly from coronary arteries by positive staining with α-SMA antibodies and the SMC morphology. The culture medium was replaced 3 days after cell isolation and then twice each week until the cells grew to confluency. All studies were performed with cells at passages 5~8. In this study, SMCs were cultured under a dedifferentiation condition in full-serum medium (DMEM with 10% FBS), if not particularly mentioned. 4.4. Immunoblotting These SMCs were treated correspondingly and then collected for the following immunoblotting analysis as described previously . Briefly, the cell samples were lysed and denatured first, and 20 μg proteins were separated by SDS-PAGE before transferring onto a PVDF membrane. The membrane was blocked and then incubated with primary antibodies, followed by corresponding secondary antibodies and substrates by using LI-COR Odyssey Fc System. 4.5. Real-Time Quantitative PCR Quantitative RT-PCR was performed as previously described . Briefly, the total RNA was extracted by using the Aurum Total RNA Mini Kit (732-6820, Bio-Rad, Hercules, CA, USA), and then reverse transcribed into cDNA by using iScript Reverse Transcription Supermix (1708841, Bio-Rad). Quantitative RT-PCR was conducted by using iTaq Universal SYBR Green Supermix (1725121; Bio-Rad) on the Bio-Rad CFX Connect real-time system with the primers ( ) according to the manufacturer’s instructions. The expression of β-actin was used as the internal control, and the results were presented by the 2 −ΔΔCt method. 4.6. Immunofluorescence Staining Immunofluorescence staining was performed as previously described . Briefly, approximately 1.0 × 10 4 SMCs were cultured and treated with adipoRon for 24 h and then fixed for 15 min by using 4.0% paraformaldehyde at room temperature. Cells were blocked using 5.0% BSA in PBST for 1 h and then incubated with primary antibodies overnight at 4 °C. Secondary fluorescent antibodies and DAPI were incubated according to the corresponding primary antibody for 1 h and then mounted after washing. For phalloidin staining of F-actin in cultured cells, fixed and permeabilized cells were incubated with Alexa Fluor 568-conjugated phalloidin (1:50) for 30 min, washed with PBS, and mounted with an anti-fluorescence quenching agent. Results were visualized by the Olympus IX73 Imaging System and then analyzed with Image-Pro Plus 6.0 software for the Pearson correlation between colocalization efficiency and average fluorescence density. 4.7. Wound Scratch Assay Cell migration was assessed by using a wound scratch assay as previously described . Briefly, 90% of confluent SMCs were starved in low-serum media (0.1% FBS) overnight. Scratch wounds were created using a 2 mm-wide pipette tip. Cells were cultured in full-serum medium (10% FBS) with indicated treatment. After 24 h, the scratched area of cells was imaged using the Olympus IX73 imaging system. The average wounded area was quantified using Image-Pro Plus 6.0 software. 4.8. MMP Activity Assay The activity of MMP was determined by using an MMP Activity Kit (Abcam, ab112146) according to the manufacturer’s instructions. 4.9. Statistics Analysis Data are presented as mean ± SE. Experiment results were analyzed by Student’s t test or one/two-way ANOVA by GraphPad Prism 6.0 (GraphPad Software). p < 0.05 was considered statistically significant.
Information on primary and secondary antibodies for immunoblotting and immunofluorescence staining is provided in . The following reagents were used: adipoRon (ab144867; Abcam, Cambridge, UK) and Triton X-100 (Sigma, Kawasaki, Japan, X100).
All experimental protocols were reviewed and approved by the Animal Care Committee of the University of Houston. All animals were kept in a standardized manner in the animal center, University of Houston.
Mouse arterial Smpd1 +/+ and Smpd1 −/− SMCs were isolated from the corresponding mice as previously described [ , , ]. Briefly, 6-week-old mice were deeply anesthetized with pentobarbital sodium (ip, 25 mg/kg). Mouse hearts were excised and immersed in an ice-cold Krebs–Henseleit (KH) solution. Then, a 25-gauge needle filled with Hanks’ buffered saline solution (HBSS) was inserted into the heart close to the aortic valve through the aortic lumen and tied when the needle tip reached the base of the heart. An infusion pump was started with a 20 mL syringe containing warm HBSS at a rate of 0.1 mL/min for 15 min. Then, the HBSS was replaced with a warm enzyme solution (containing 1 mg/mL collagenase type 1, 0.5 mg/mL soybean trypsin inhibitor, 3.0% BSA, and 2.0% antibiotics) which was flushed through the heart at a rate of 0.1 mL/min. The outflow perfusion fluid was collected at 30, 60, and 90 min intervals. The heart was cut to open the apex to flush out the cells inside the ventricle after collecting all outflow fluid at 90 min. The flushed cells were centrifuged at 1000 rpm for 10 min, and the pellets were resuspended in advanced Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS), 10% mouse serum, and 2.0% antibiotics. The isolated cells were plated on 2.0% gelatin-coated plates and incubated in 5.0% CO 2 at 37 °C. These isolated cells were confirmed as SMCs originated mainly from coronary arteries by positive staining with α-SMA antibodies and the SMC morphology. The culture medium was replaced 3 days after cell isolation and then twice each week until the cells grew to confluency. All studies were performed with cells at passages 5~8. In this study, SMCs were cultured under a dedifferentiation condition in full-serum medium (DMEM with 10% FBS), if not particularly mentioned.
These SMCs were treated correspondingly and then collected for the following immunoblotting analysis as described previously . Briefly, the cell samples were lysed and denatured first, and 20 μg proteins were separated by SDS-PAGE before transferring onto a PVDF membrane. The membrane was blocked and then incubated with primary antibodies, followed by corresponding secondary antibodies and substrates by using LI-COR Odyssey Fc System.
Quantitative RT-PCR was performed as previously described . Briefly, the total RNA was extracted by using the Aurum Total RNA Mini Kit (732-6820, Bio-Rad, Hercules, CA, USA), and then reverse transcribed into cDNA by using iScript Reverse Transcription Supermix (1708841, Bio-Rad). Quantitative RT-PCR was conducted by using iTaq Universal SYBR Green Supermix (1725121; Bio-Rad) on the Bio-Rad CFX Connect real-time system with the primers ( ) according to the manufacturer’s instructions. The expression of β-actin was used as the internal control, and the results were presented by the 2 −ΔΔCt method.
Immunofluorescence staining was performed as previously described . Briefly, approximately 1.0 × 10 4 SMCs were cultured and treated with adipoRon for 24 h and then fixed for 15 min by using 4.0% paraformaldehyde at room temperature. Cells were blocked using 5.0% BSA in PBST for 1 h and then incubated with primary antibodies overnight at 4 °C. Secondary fluorescent antibodies and DAPI were incubated according to the corresponding primary antibody for 1 h and then mounted after washing. For phalloidin staining of F-actin in cultured cells, fixed and permeabilized cells were incubated with Alexa Fluor 568-conjugated phalloidin (1:50) for 30 min, washed with PBS, and mounted with an anti-fluorescence quenching agent. Results were visualized by the Olympus IX73 Imaging System and then analyzed with Image-Pro Plus 6.0 software for the Pearson correlation between colocalization efficiency and average fluorescence density.
Cell migration was assessed by using a wound scratch assay as previously described . Briefly, 90% of confluent SMCs were starved in low-serum media (0.1% FBS) overnight. Scratch wounds were created using a 2 mm-wide pipette tip. Cells were cultured in full-serum medium (10% FBS) with indicated treatment. After 24 h, the scratched area of cells was imaged using the Olympus IX73 imaging system. The average wounded area was quantified using Image-Pro Plus 6.0 software.
The activity of MMP was determined by using an MMP Activity Kit (Abcam, ab112146) according to the manufacturer’s instructions.
Data are presented as mean ± SE. Experiment results were analyzed by Student’s t test or one/two-way ANOVA by GraphPad Prism 6.0 (GraphPad Software). p < 0.05 was considered statistically significant.
In summary, for the first time, we elucidated the critical regulatory role of ASM in adipoRon-induced TFEB–autophagy signaling and the consequent inhibition of SMC proliferation and migration. Our study provides new mechanistic insights into how ASM controls the TFEB–autophagy signaling by regulating TRMPL1-mediated Ca 2+ and its implication for adipoRon-induced beneficial effects on SMC homeostasis.
|
Ethics in Pediatric Endocrinology: Curriculum for Fellows and Faculty | 1591d299-f319-40d6-b91b-9dfa61b7d0d6 | 6342438 | Pediatrics[mh] | By the end of this activity, learners will be able to: 1. Define the ethical pillars of clinical practice: beneficence, nonmaleficence, autonomy, and justice. 2. Analyze each case in the context of the ethical pillars. 3. Formulate an appropriate management based on the ethical dilemma in each case.
Over the past 5–10 years, the scope of pediatric endocrinology, like that of other clinical medical subspecialties, has expanded. With this expansion, practitioners are being increasingly called upon to use existing treatments to manage new patient populations, such as gonadotropin-releasing hormone analogues, testosterone, and estrogen in youth with gender dysphoria or hormone therapies to enhance the height of an otherwise healthy child, in addition to implementing new and sometimes experimental technologies such as fertility preservation. Such scenarios often raise ethical concerns. Cases of children with special needs who may be subjected to hormonal-based therapies may also be fraught with ethical issues. For instance, in 2007, a supratherapeutic dose of estrogen was administered along with bilateral mastectomy and hysterectomy to a 10-year-old girl with static encephalopathy who was named Ashley. Collectively, the aforementioned techniques have been termed the Ashley treatment. The parents argued that these interventions would allow them to more easily care for their daughter as she grew older, would preclude her from experiencing menstrual discomfort, and, in addition, would prevent the possible discomfort associated with having large breasts, reported to be common in the family. Similarly, there have been cases described of caregivers requesting hormonal suppression in children with autism and sexually inappropriate behaviors. Practitioners may face significant challenges in considering the interests of both the patient and the caregiver(s) in determining the best care plan for the patient. In addition, many practitioners believe that they do not have the background or training in clinical ethics with which a case may be dissected. The Pediatric Endocrine Society (PES) Ethics Committee recognized that while practitioners both during and after training may face such dilemmas as outlined above, the American Board of Pediatrics' content specification, despite outlining a curriculum for ethics in research, has not provided a curriculum outline for ethics in clinical practice. Hence, the current curriculum was designed to fulfill an unmet need. Case-based discussion modules designed to raise ethics awareness were first created approximately 5 years ago and progressively added to the PES website; however, they were underutilized. These modules included the following: Ethical Issues Regarding Endocrine Management in the Care of Disabled Children ( ), Ethical Issues in Prescribing Growth Hormone ( ), Ethical Issues Regarding Childhood Obesity , Ethical Issues Regarding the Poorly Adherent Patient With Type I Diabetes Mellitus , Ethical Issues in Children With Disorders of Sex Development , and Ethical Issues in Transgender Medicine . Recently, two new modules, Ethical Issues in Fertility Preservation in KIinefelter Syndrome and Ethical Issues in Fertility Preservation in Turner Syndrome , were created. All eight modules were edited and approved by the PES board of directors. Subsequently, we conducted a survey assessing whether pediatric endocrine fellows and faculty felt that there was adequate ethics training in place and whether these modules helped fulfill an unmet need in their ethics education.
Target Audience Practitioners of pediatric endocrinology (fellows at all levels of training, advanced practitioners, and attending physicians) were invited to participate. Practitioners from programs with which a member of the PES Ethics Committee was affiliated were invited to participate. Eight programs were invited, and six participated in the pilot testing (75% response rate). Participating programs included Children's Mercy Hospital, Nationwide Children's Hospital, Children's Hospital Colorado, Seattle Children's Hospital, Hassenfeld Children's Hospital of New York University, and the Children's Hospital of Philadelphia. For the pilot phase, each participant was sent a module (topic selected at random; – ), with instructions to complete a pretest survey prior to viewing the module and then a posttest survey . Using a 5-point Likert scale, eight paired questions were piloted in the pre- and postanalytical phase examining self-reported knowledge (K) of the ethical pillars (beneficence, nonmaleficence, autonomy, and justice), attitudes (A) regarding importance of these principles, and likelihood of applying these principles to clinical practice (P), as well as perceived need for/benefit of this curriculum. Though a KAP study has traditionally been a tool to explore changes in the KAP of communities and large populations, it was used for the pilot since it encompassed not only an assessment of the participant's knowledge but also an application of the ethical pillars that in effect represents a higher-order process, suitable for case studies. , Participants were encouraged to complete their module under the guidance of a faculty moderator; however, those who were unavailable for those sessions completed them individually. We chose this method rather than formal didactics on the ethical pillars of clinical practice (beneficence, nonmaleficence, autonomy, and justice) as case-based learning has been shown to be an effective strategy for optimizing clinical practice and in promoting knowledge retention among physicians. Preparation Though, in general, no acquaintance with clinical ethics is necessary prior to commencing a module, for faculty moderating these discussions, prior module review may facilitate more discussion and debate. This recommendation is based on feedback obtained during the pilot phase. In addition, we have added an introductory PowerPoint presentation ( ) to introduce users to the overall curriculum. Fellows at all training levels may benefit from these modules, which may also prove useful for individuals needing a refresher on applying the ethical pillars of beneficence, nonmaleficence, autonomy, and justice to cases.
Practitioners of pediatric endocrinology (fellows at all levels of training, advanced practitioners, and attending physicians) were invited to participate. Practitioners from programs with which a member of the PES Ethics Committee was affiliated were invited to participate. Eight programs were invited, and six participated in the pilot testing (75% response rate). Participating programs included Children's Mercy Hospital, Nationwide Children's Hospital, Children's Hospital Colorado, Seattle Children's Hospital, Hassenfeld Children's Hospital of New York University, and the Children's Hospital of Philadelphia. For the pilot phase, each participant was sent a module (topic selected at random; – ), with instructions to complete a pretest survey prior to viewing the module and then a posttest survey . Using a 5-point Likert scale, eight paired questions were piloted in the pre- and postanalytical phase examining self-reported knowledge (K) of the ethical pillars (beneficence, nonmaleficence, autonomy, and justice), attitudes (A) regarding importance of these principles, and likelihood of applying these principles to clinical practice (P), as well as perceived need for/benefit of this curriculum. Though a KAP study has traditionally been a tool to explore changes in the KAP of communities and large populations, it was used for the pilot since it encompassed not only an assessment of the participant's knowledge but also an application of the ethical pillars that in effect represents a higher-order process, suitable for case studies. , Participants were encouraged to complete their module under the guidance of a faculty moderator; however, those who were unavailable for those sessions completed them individually. We chose this method rather than formal didactics on the ethical pillars of clinical practice (beneficence, nonmaleficence, autonomy, and justice) as case-based learning has been shown to be an effective strategy for optimizing clinical practice and in promoting knowledge retention among physicians.
Though, in general, no acquaintance with clinical ethics is necessary prior to commencing a module, for faculty moderating these discussions, prior module review may facilitate more discussion and debate. This recommendation is based on feedback obtained during the pilot phase. In addition, we have added an introductory PowerPoint presentation ( ) to introduce users to the overall curriculum. Fellows at all training levels may benefit from these modules, which may also prove useful for individuals needing a refresher on applying the ethical pillars of beneficence, nonmaleficence, autonomy, and justice to cases.
Surveys were completed by fellows ( n = 29), faculty ( n = 7), and advanced practitioners ( n = 3) at six of the eight large pediatric endocrine programs (75% response rate), as stated above. Only 20.3% of the respondents felt that an effective ethics curriculum existed at the time of the survey. KAP scores improved after participants completed the modules, with knowledge scores showing the greatest improvement. As shown in the , 94.9% of respondents strongly agreed ( n = 26) or agreed ( n = 11) that the curriculum would be a helpful addition to pediatric endocrine fellowship training. Additionally, all faculty felt that the curriculum would be helpful for faculty to learn about ethical principles applicable to clinical practice. Feedback The feedback obtained at the time of conducting the posttest survey was positive, with comments in general attesting to the usefulness of the modules in facilitating discussion around their presented cases. The amount of time required varied (generally 30–60 minutes) based on the audience and level of discussion. Specific comments on the posttest survey included the following: • “Great discussion! We should have regular case scenario and discussions like this.” • “Good case!” • “Great module!” • “Nice job! Brought up good questions for discussion.” • “Would be helpful to have more discussion and debate within the session! Otherwise good topic!”
The feedback obtained at the time of conducting the posttest survey was positive, with comments in general attesting to the usefulness of the modules in facilitating discussion around their presented cases. The amount of time required varied (generally 30–60 minutes) based on the audience and level of discussion. Specific comments on the posttest survey included the following: • “Great discussion! We should have regular case scenario and discussions like this.” • “Good case!” • “Great module!” • “Nice job! Brought up good questions for discussion.” • “Would be helpful to have more discussion and debate within the session! Otherwise good topic!”
Given that ethical dilemmas are commonly encountered by pediatric endocrinologists, through this initiative we have identified and are addressing an important gap in training. The ethics modules that have been created can be used by learners at the graduate medical education level as well as by individuals posttraining. The latter includes faculty, physician assistants, and advanced practice nurses interested in advancing their knowledge of clinical ethics in pediatric endocrinology. Formatting of the modules conforms to the six criteria for scholarship: clear goals, adequate preparation, appropriate methods, significant results, effective presentation, and reflective critique (adapted from Scholarship Assessed ). The evaluations in the pilot phase of the curriculum were positive overall. Practitioners at all levels (fellows, advanced practitioners, attending physicians, and faculty) felt the modules were useful. They reinforced our initial perception that there is great need for more education of both trainees and physicians in practice on clinical ethics in pediatric endocrinology. Of note, learners completing the module without a faculty moderator felt that more discussion within the modules would have been helpful; thus, a facilitator's guide has been created. This will enhance the educational experience for the learner, facilitate group discussions, and standardize the clinical approach to such cases. Although the original six modules were created several years ago and located on the PES website (accessible to members), web usage data showed a relatively low number of views. In order to increase usage, we intend to send the curriculum to pediatric endocrinology fellowship program directors to facilitate utilization by trainees and faculty at their respective institutions. We will also disseminate information about this curriculum in the PES monthly newsletter. In its monthly Ethics Corner section, we will spread awareness of the curriculum and solicit feedback as to potential additional topics of interest in order to create new modules to add to the curriculum. Furthermore, we intend to send out an annual survey aiming to assess utilization, in addition to other aspects of the curriculum, such as perceived benefit and suggestions for improvement and for future topics to include. At this time, the existing modules cover only eight areas. Although this is not an exhaustive list of topics in which ethical dilemmas may arise in pediatric endocrinology, we believe that with the valuable knowledge gained from these modules, a learner can acquire some experience with which to approach other cases with similar ethical dilemmas in clinical practice. We plan to update the current modules as new medical information emerges and paradigms change. In summary, we believe that this formalized curriculum fulfills an unmet need and equips the learner to become more comfortable forming a management plan based on the pillars of clinical ethics. This is especially important as the scope of pediatric endocrinology expands with the incorporation of various topics in which ethical dilemmas arise in the provision of care.
A. Introduction to Ethics in Pediatric Endocrinology.pptx B. Ethical Issues Regarding Endocrine Management in the Care of Disabled Children.pptx C. Ethical Issues in Prescribing Growth Hormone.pptx D. Ethical Issues Regarding Childhood Obesity.ppt E. Ethical Issues Regarding the Poorly Adherent Patient With Type I Diabetes Mellitus.pptx F. Ethical Issues in Children With Disorders of Sex Development.pptx G. Ethical Issues in Transgender Medicine.pptx H. Ethical Issues in Fertility Preservation in Klinefelter Syndrome.pptx I. Ethical Issues in Fertility Preservation in Turner Syndrome.pptx J. Ethics in Pediatric Endocrinology Facilitator Resource Tool.docx K. Pre- and Posttest Questionnaire.doc All appendices are peer reviewed as integral parts of the Original Publication.
|
Reduced preoperative serum choline esterase levels and fecal peritoneal contamination as potential predictors for the leakage of intestinal sutures after source control in secondary peritonitis | 0bf011c1-9618-4f56-8492-c341d6ac9ecc | 11151556 | Suturing[mh] | Emergency laparotomy for secondary peritonitis is still associated with high mortality rates, albeit a substantial decrease was registered over the last decades as a result of improved perioperative care. By implementing evidence-based guidelines, as propagated by the Surving Sepsis Campaign (SSC), more patients with abdominal sepsis survive an otherwise fatal affliction [ – ]. The adverse effect of this development is an increase in morbidity following prolonged stays on intensive care units (ICUs), leading to debilitating chronical illness, poor clinical outcomes and poor quality of life . There remains an evident need to further optimize emergency care delivery. Adherence to the SSC recommendations benefits septic patients through the implementation of screening tools such as the SIRS (Systemic Inflammatory Response Syndrome) or MEWS (Modified Early Warning Score) scores for expediting diagnosis and through commitment to early treatment goals summarized in time-framed bundles [ , , , ]. Whilst scores for early sepsis recognition, prompt treatment initiation and post-operative intensive care protocols have thoroughly been investigated, there is little evidence-based guidance for the decision-making process during source control surgery (SCS) [ – ]. Key surgical decisions that highly impact patient outcome, such as anastomosis vs. stoma placement in a peritonitic abdomen are mostly based on the surgeon’s experience and appreciation of the patients’ severity of illness. The consideration of primary anastomosis during SCS for secondary peritonitis is fairly recent, as for many years the choice, including that of experienced surgeons, was to avoid bowel reconstruction and place stomata instead. An enterostomy negatively affects its carriers both on a physical and psychosocial level . While bound to a life-time risk of stoma-related complications, less than 50% of enterostomy-carriers undergo subsequent restoration of bowel continuity, a procedure with inherent morbidity [ , , ]. The tendency for enterostomy creation persists even in countries with well-developed public health-care systems. According to large-scale audits and observational multicentric studies, just about one quarter of patients undergoing emergency left-sided colonic resection receive a primary anastomosis . This occurs despite growing evidence that in many cases primary bowel reconstruction can be safely performed, even in patients with perforated diverticulitis and purulent or fecal peritonitis [ – ]. Addressing the same issue for small bowel perforation with peritonitis, a meta-analysis concluded that there is no sufficient data to issue evidence-based recommendations of whether and when an anastomosis can safely be placed . Even the recent Enhanced Recovery after Surgery (ERAS®) Society and the World Society of Emergency Surgery (WSES) guidelines for emergency laparotomy refrain from issuing detailed recommendations on the surgical approach due to lack of data or need to extrapolate from data derived from elective surgery, leaving the decision of primary anastomosis placement at the discretion of the operating surgeon . The surgical strategy needs of course to be tailored to the patients’ pre-existing conditions and pathophysiological response to the peritoneal contamination, ranging from compensated inflammation to septic shock, as well as to the intraoperative finding. While for elective surgery risk factors for anastomotic leakage have been identified , and scoring systems have been developed , these data remain scarce in the emergency setting. Among the preoperative tumor-unrelated parameters, the systematic review by McDermott et al. found male sex, American Society of Anesthesiologists (ASA) fitness grade, renal disease, obesity, hypoalbuminemia as a marker of a poor nutritional status, and an indication for emergency surgery to significantly increase the risk of colorectal anastomotic leaks . Two of the largest observational cohort studies analyzing bowel resection with or without primary reconstruction during emergency laparotomy identified fecal contamination as an independent predictor for suture leakage . Both patient cohorts were heterogenous with only 10–30% having peritonitis as an indication for emergency surgery. The authors acknowledged that the lack of data depicting preoperative nutritional deficits limited their risk assessment, as malnutrition has repeatedly been identified as an independent predictor of anastomotic leakage and sepsis [ – ]. Low serum albumin, a high C-reactive protein (CRP)-albumin ratio and low serum choline esterase (sCHE) as markers of malnutrition have been linked to a disturbed postoperative wound healing, including that of gastrointestinal (GI) sutures and to a poor prognosis in septic patients, highlighting the need of taking these factors into consideration when placing sutures in a septic surrounding [ – ]. Switching the focus towards the human, decision-making “surgeon factor”, the trend for on-going sub-specialization benefits patients undergoing elective oncologic surgery, but it has been shown to impair the outcomes of emergency surgery, when the required operation is not part of the surgeons’ usual procedural spectrum . Nevertheless, reality confronts all general surgeons on duty regardless of experience and sub-specialization with the risk/benefit assessment of primary bowel reconstruction in secondary peritonitis. Defining patient-associated factors and factors related to the intraabdominal pathology that might facilitate this decision remains of utmost importance. The aim of this retrospective data analysis is to identify quantifiable pre- and intraoperative parameters which might facilitate the surgeon’s decision for or against a primary bowel reconstruction in a peritonitic abdomen.
Patient selection All consecutive patients (≥ 18 years of age) who underwent emergency laparotomy for secondary peritonitis between January 2014 and December 2020 in the Department of General, Visceral, Thoracic and Transplant Surgery of the University Hospital of Giessen were retrospectively evaluated according to the following criteria defined for inclusion or exclusion from the study. Inclusion criterium The main inclusion criterium was primary bowel reconstruction through placement of intestinal sutures on the lower GI tract (below the ligament of Treitz) during emergency laparotomy for intraoperatively confirmed localized or generalized, purulent or fecal peritonitis. Exclusion criteria We excluded traumatic GI perforations due to blunt or penetrating trauma. Also excluded were patients with perforated acute appendicitis and cholecystitis, as morbidity and mortality rates in these cases are known to be significantly lower than in the case of hollow viscus perforation . Patients with chronic and contained enteric fistulae, repaired in an elective setting were not considered for inclusion in the study. Patients undergoing exclusive repair of the upper GI tract or of insufficient pancreatico-billiary reconstructions during emergency laparotomy for confirmed peritoneal contamination were excluded. Further exclusion criteria were discontinuity resections or enterostomy placement orally from the site of primary reconstructive sutures. All surgeries were either performed or supervised by a consultant surgeon who was primarily responsible for deciding the surgical strategy. All patients were treated according to the institutional standard of care. Study variables The preoperative parameters collected from the included patients were demographics: age, gender, body mass index (BMI) and pre-existing conditions: chronic pulmonary, liver or kidney disease, history of cardiovascular disease, diabetes, previously diagnosed malignancy as well as chronic inflammatory disease. ASA classification score was calculated based on the known comorbidities at the time of SCS. Previous medication that could influence postoperative morbidity in terms of bleeding or impaired wound healing, such as anticoagulant and antiplatelet agents, immunosuppressives, or chronic steroid therapy was also registered. The preoperative laboratory parameters were chosen to depict inflammation (leukocyte count, CRP), anemia (hemoglobin), liver function (sCHE, bilirubin) and kidney function (creatinine). These parameters were part of the standard blood analysis panel for surgical emergencies. SCHE was determined in the hospital’s central laboratory by the means of ultraviolet–visible (UV–VIS) spectrophotometry using the ADVIA assay kit from Siemens Healthineers (Erlangen, Germany). The collected intraoperative data included the condition identified during SCS as the cause or main contributor to the disruption of bowel integrity such as mesenteric ischemia, mechanical bowel obstruction, or bowel inflammation. Furthermore, the location of enteric sutures placed during SCS, intraoperative blood loss and procedure time were recorded. We calculated the Mannheim peritonitis index (MPI), as a validated score for predicting mortality from secondary peritonitis that takes into account the 8 parameters listed below (Fig. a.). For the sole purpose of documenting the extent of peritoneal contamination and quality of the peritoneal exudate we developed a simplified score, leaning on the MPI that we entitled peritonitis extent score (Fig. b.). Outcome measures The primary outcome parameter was the postoperative leakage rate of the intestinal sutures placed during SCS. Sutures were classified as insufficient either when leakage was directly confirmed during revision surgery or when computed tomography delivered strong proof of leakage, such as extra-enteric contrast medium spillage with the consequence of therapy limitation for patients deemed too critical for revision surgery. A secondary outcome measure was postoperative mortality, either in-hospital or within 100 days of the procedure if discharged. Also considered was the sequential organ failure assessment (SOFA) score upon ICU admittance and on the second postoperative day as well as surgical morbidity other than suture leakage. This included postoperative bleeding and superficial as well as deep surgical site infections. ICU length of stay and in-hospital length of stay were also recorded. Statistical analysis Data analysis was performed using GraphPad Prism (Version 9 for Windows, GraphPad Software, San Diego, CA, USA, www.graphpad.com ). Continuous variables are presented as median and interquartile range (IQR) and were analyzed using the Mann–Whitney U test. Categorical variables are shown as numbers with percentages, n (%), and were compared using a chi-squared test or Fisher’s exact test, as appropriate. Associations between preoperative as well as intraoperative parameters and suture leakage were investigated by univariate logistic regression. Variables with statistically significant association on univariate analysis were included in a multivariable logistic regression model. The multiple logistic regression model was tested for multicollinearity by calculating the variance inflation factors (VIF) for each variable included. Survival curves were generated using the Kaplan–Meier method and compared using a log-rank test. Spearman’s rho rank correlation was used to determine statistical dependence between preoperative parameters. Results are given as the Spearman’s rank correlation coefficient (r) and respective significances. P values of ≤ 0.05 (two-sided) were considered statistically significant.
All consecutive patients (≥ 18 years of age) who underwent emergency laparotomy for secondary peritonitis between January 2014 and December 2020 in the Department of General, Visceral, Thoracic and Transplant Surgery of the University Hospital of Giessen were retrospectively evaluated according to the following criteria defined for inclusion or exclusion from the study. Inclusion criterium The main inclusion criterium was primary bowel reconstruction through placement of intestinal sutures on the lower GI tract (below the ligament of Treitz) during emergency laparotomy for intraoperatively confirmed localized or generalized, purulent or fecal peritonitis. Exclusion criteria We excluded traumatic GI perforations due to blunt or penetrating trauma. Also excluded were patients with perforated acute appendicitis and cholecystitis, as morbidity and mortality rates in these cases are known to be significantly lower than in the case of hollow viscus perforation . Patients with chronic and contained enteric fistulae, repaired in an elective setting were not considered for inclusion in the study. Patients undergoing exclusive repair of the upper GI tract or of insufficient pancreatico-billiary reconstructions during emergency laparotomy for confirmed peritoneal contamination were excluded. Further exclusion criteria were discontinuity resections or enterostomy placement orally from the site of primary reconstructive sutures. All surgeries were either performed or supervised by a consultant surgeon who was primarily responsible for deciding the surgical strategy. All patients were treated according to the institutional standard of care.
The preoperative parameters collected from the included patients were demographics: age, gender, body mass index (BMI) and pre-existing conditions: chronic pulmonary, liver or kidney disease, history of cardiovascular disease, diabetes, previously diagnosed malignancy as well as chronic inflammatory disease. ASA classification score was calculated based on the known comorbidities at the time of SCS. Previous medication that could influence postoperative morbidity in terms of bleeding or impaired wound healing, such as anticoagulant and antiplatelet agents, immunosuppressives, or chronic steroid therapy was also registered. The preoperative laboratory parameters were chosen to depict inflammation (leukocyte count, CRP), anemia (hemoglobin), liver function (sCHE, bilirubin) and kidney function (creatinine). These parameters were part of the standard blood analysis panel for surgical emergencies. SCHE was determined in the hospital’s central laboratory by the means of ultraviolet–visible (UV–VIS) spectrophotometry using the ADVIA assay kit from Siemens Healthineers (Erlangen, Germany). The collected intraoperative data included the condition identified during SCS as the cause or main contributor to the disruption of bowel integrity such as mesenteric ischemia, mechanical bowel obstruction, or bowel inflammation. Furthermore, the location of enteric sutures placed during SCS, intraoperative blood loss and procedure time were recorded. We calculated the Mannheim peritonitis index (MPI), as a validated score for predicting mortality from secondary peritonitis that takes into account the 8 parameters listed below (Fig. a.). For the sole purpose of documenting the extent of peritoneal contamination and quality of the peritoneal exudate we developed a simplified score, leaning on the MPI that we entitled peritonitis extent score (Fig. b.).
The primary outcome parameter was the postoperative leakage rate of the intestinal sutures placed during SCS. Sutures were classified as insufficient either when leakage was directly confirmed during revision surgery or when computed tomography delivered strong proof of leakage, such as extra-enteric contrast medium spillage with the consequence of therapy limitation for patients deemed too critical for revision surgery. A secondary outcome measure was postoperative mortality, either in-hospital or within 100 days of the procedure if discharged. Also considered was the sequential organ failure assessment (SOFA) score upon ICU admittance and on the second postoperative day as well as surgical morbidity other than suture leakage. This included postoperative bleeding and superficial as well as deep surgical site infections. ICU length of stay and in-hospital length of stay were also recorded.
Data analysis was performed using GraphPad Prism (Version 9 for Windows, GraphPad Software, San Diego, CA, USA, www.graphpad.com ). Continuous variables are presented as median and interquartile range (IQR) and were analyzed using the Mann–Whitney U test. Categorical variables are shown as numbers with percentages, n (%), and were compared using a chi-squared test or Fisher’s exact test, as appropriate. Associations between preoperative as well as intraoperative parameters and suture leakage were investigated by univariate logistic regression. Variables with statistically significant association on univariate analysis were included in a multivariable logistic regression model. The multiple logistic regression model was tested for multicollinearity by calculating the variance inflation factors (VIF) for each variable included. Survival curves were generated using the Kaplan–Meier method and compared using a log-rank test. Spearman’s rho rank correlation was used to determine statistical dependence between preoperative parameters. Results are given as the Spearman’s rank correlation coefficient (r) and respective significances. P values of ≤ 0.05 (two-sided) were considered statistically significant.
Pre- and intraoperative characteristics of patients with lower GI sutures placed during SCS for secondary peritonitis A total number of 497 patients underwent SCS for secondary peritonitis caused either by hollow viscus perforation or insufficiency of electively placed GI sutures. 122 patients with source control interventions exclusively on the upper GI tract, and 44 patients with SCS consisting in the exclusive repair of insufficient pancreatico-billiary reconstructions were excluded. Of the 341 patients needing source control intervention on the lower GI tract, 154 received diverting or permanent enterostomies, leaving 187 patients with primary reconstructions of the lower GI tract during SCS for further analysis. These 187 patients were divided into two patient subgroups depending on whether the primary lower GI reconstructions performed during SCS remained intact (140 patients) or developed a leakage (47 patients) (Fig. ). Suture leakage was detected with a mean latency of 7.9 days from SCS and was confirmed either by revision surgery for 45 of the patients or by CT-scans showing direct extraluminal leakage of enteric contrast medium for the other two patients, whose therapy was limited prior to revision surgery due to poor overall prognosis. For the other subgroup of 140 patients the sutures placed during SCS remained intact. There was no significant difference in basal characteristics between the two patient subgroups. The subgroup of patients with intact sutures, however, had lower CRP and bilirubin levels as well as higher sCHE activity prior to SCS compared to the patients developing suture leakage (Table ). Secondary peritonitis requiring SCS was caused in similar proportions in both patient subgroups by gastrointestinal perforation (75.7% vs. 62.5%) or disruption of electively placed gastrointestinal sutures (24.3% vs. 37.5%). There was no significant difference in the location (small vs. large bowel) of the sutures placed during SCS. The length of source control procedures and intraoperative blood loss did not significantly differ in patients with intact vs. insufficient sutures (Table ). Postoperative outcomes of patients with lower GI sutures placed during SCS for secondary peritonitis While SOFA scores immediately upon postoperative ICU admittance were similarly elevated in both patient subgroups, the subgroup of patients with intact enteric sutures had a significantly lower SOFA score on the second postoperative day and therefore a significant improvement in organ functionality. Both incisional as well as intra-abdominal space infection were significantly higher in the subgroup of patients with insufficient enteric sutures. These patients also had a significantly prolonged stay on the ICU of a median of 8 days, almost three times longer than the intensive care period required by patients with intact sutures. The in-hospital mortality of 38.3% was also significantly higher in the subgroup of patients with suture leakage, of whom only 25.5% were released in their initial home environment (Table ). Univariate and multivariate analysis of preoperative and intraoperative factors associated with leakage of lower GI sutures placed during SCS for secondary peritonitis The following variables showed a statistically significant association with suture leakage in the univariate analysis (Table ): preoperative CRP levels ( p =0.0232), preoperative sCHE activity ( p =0.0019) and the peritonitis extent score ( p =0.0045). We chose not to include the MPI in our analysis as at least three of the parameters needed to calculate the MPI (age, sex, malignancy) showed no significant association with our primary outcome measure. Other parameters known to influence the outcome of colorectal sutures placed during elective surgery, such as BMI, chronic steroid intake and ASA-score were not significantly associated with the outcome (intact vs. insufficient) of sutures placed on the lower GI tract during SCS for secondary peritonitis. In the multivariate analysis sCHE activity and the peritonitis extent score remained independent predictors for suture outcome ( p =0.0472 and p =0.0234, respectively). Correlation of sCHE activity with suture outcome and patient survival after SCS for secondary peritonitis We analyzed the correlation of low preoperative sCHE activity and the development of suture leakage. As a cut-off value we took the lower end of the reference interval of 4.5 kU/L. Patients with a sCHE < 4.5 kU/L (n = 96) developed a significantly higher rate of suture insufficiency ( p =0.02) and had a significantly higher mortality ( p =0.001) than patients with sCHE activity within the normal range (Fig. ). Correlation of CRP/sCHE ratio with patient survival after SCS for secondary peritonitis, in dependence of suture outcome No multicollinearity issue was detected in the multiple logistic regression model, since the calculated variance inflation factors (VIF) for each independent variable were below 1.5. Nevertheless, there was a negative correlation detected between preoperative CRP and sCHE activity with a Spearman correlation coefficient of − 0.4046. The CRP/sCHE ratio was able to discriminate between death and survival following SCS for secondary peritonitis in both patient subgroups with intact ( p =0.0025) and insufficient ( p =0.0421) enteric sutures respectively (Fig. ). Correlation of peritonitis extent with suture outcome and patient survival after SCS for secondary peritonitis Patients with a peritonitis extent score of ≥ 18, implying a generalized fecal peritonitis, had a significantly higher incidence ( p =0.0014) of enteric suture leakage compared to patients with a less severe degree of peritonitis. There was no significant difference but a noticeable trend ( p =0.0788) in patient survival when taking the extent of peritoneal contamination into account (Fig. ).
A total number of 497 patients underwent SCS for secondary peritonitis caused either by hollow viscus perforation or insufficiency of electively placed GI sutures. 122 patients with source control interventions exclusively on the upper GI tract, and 44 patients with SCS consisting in the exclusive repair of insufficient pancreatico-billiary reconstructions were excluded. Of the 341 patients needing source control intervention on the lower GI tract, 154 received diverting or permanent enterostomies, leaving 187 patients with primary reconstructions of the lower GI tract during SCS for further analysis. These 187 patients were divided into two patient subgroups depending on whether the primary lower GI reconstructions performed during SCS remained intact (140 patients) or developed a leakage (47 patients) (Fig. ). Suture leakage was detected with a mean latency of 7.9 days from SCS and was confirmed either by revision surgery for 45 of the patients or by CT-scans showing direct extraluminal leakage of enteric contrast medium for the other two patients, whose therapy was limited prior to revision surgery due to poor overall prognosis. For the other subgroup of 140 patients the sutures placed during SCS remained intact. There was no significant difference in basal characteristics between the two patient subgroups. The subgroup of patients with intact sutures, however, had lower CRP and bilirubin levels as well as higher sCHE activity prior to SCS compared to the patients developing suture leakage (Table ). Secondary peritonitis requiring SCS was caused in similar proportions in both patient subgroups by gastrointestinal perforation (75.7% vs. 62.5%) or disruption of electively placed gastrointestinal sutures (24.3% vs. 37.5%). There was no significant difference in the location (small vs. large bowel) of the sutures placed during SCS. The length of source control procedures and intraoperative blood loss did not significantly differ in patients with intact vs. insufficient sutures (Table ).
While SOFA scores immediately upon postoperative ICU admittance were similarly elevated in both patient subgroups, the subgroup of patients with intact enteric sutures had a significantly lower SOFA score on the second postoperative day and therefore a significant improvement in organ functionality. Both incisional as well as intra-abdominal space infection were significantly higher in the subgroup of patients with insufficient enteric sutures. These patients also had a significantly prolonged stay on the ICU of a median of 8 days, almost three times longer than the intensive care period required by patients with intact sutures. The in-hospital mortality of 38.3% was also significantly higher in the subgroup of patients with suture leakage, of whom only 25.5% were released in their initial home environment (Table ).
The following variables showed a statistically significant association with suture leakage in the univariate analysis (Table ): preoperative CRP levels ( p =0.0232), preoperative sCHE activity ( p =0.0019) and the peritonitis extent score ( p =0.0045). We chose not to include the MPI in our analysis as at least three of the parameters needed to calculate the MPI (age, sex, malignancy) showed no significant association with our primary outcome measure. Other parameters known to influence the outcome of colorectal sutures placed during elective surgery, such as BMI, chronic steroid intake and ASA-score were not significantly associated with the outcome (intact vs. insufficient) of sutures placed on the lower GI tract during SCS for secondary peritonitis. In the multivariate analysis sCHE activity and the peritonitis extent score remained independent predictors for suture outcome ( p =0.0472 and p =0.0234, respectively).
We analyzed the correlation of low preoperative sCHE activity and the development of suture leakage. As a cut-off value we took the lower end of the reference interval of 4.5 kU/L. Patients with a sCHE < 4.5 kU/L (n = 96) developed a significantly higher rate of suture insufficiency ( p =0.02) and had a significantly higher mortality ( p =0.001) than patients with sCHE activity within the normal range (Fig. ).
No multicollinearity issue was detected in the multiple logistic regression model, since the calculated variance inflation factors (VIF) for each independent variable were below 1.5. Nevertheless, there was a negative correlation detected between preoperative CRP and sCHE activity with a Spearman correlation coefficient of − 0.4046. The CRP/sCHE ratio was able to discriminate between death and survival following SCS for secondary peritonitis in both patient subgroups with intact ( p =0.0025) and insufficient ( p =0.0421) enteric sutures respectively (Fig. ).
Patients with a peritonitis extent score of ≥ 18, implying a generalized fecal peritonitis, had a significantly higher incidence ( p =0.0014) of enteric suture leakage compared to patients with a less severe degree of peritonitis. There was no significant difference but a noticeable trend ( p =0.0788) in patient survival when taking the extent of peritoneal contamination into account (Fig. ).
This study aimed to identify quantifiable preoperative and intraoperative parameters associated with a high risk of leakage for sutures placed on the lower GI tract during SCS for secondary peritonitis. These parameters could serve as an everyday tool for surgeons to decide between a primary intestinal reconstruction vs. enterostomy placement. In our patient group, 25.1% of the sutures placed under these adverse emergency conditions developed a leakage in the early postoperative course, far surpassing the insufficiency rates of lower GI sutures placed under elective conditions . In fact, emergency surgery is a well-known independent risk factor for anastomotic leakage after colorectal surgery . After elective colorectal cancer surgery, the reported incidence of anastomotic leakage ranges between 1 and 19%, with higher leakage rates after left colonic and rectal resections compared to right colonic resections . In our patient collective 25% of the small to large bowel sutures and 27% of large to large bowel sutures developed a leakage as opposed to the reported insufficiency rates of 1–4% and 2–19%, respectively, under elective conditions . Most data concerning incidence and predisposing factors for lower GI suture leakage derive from elective colorectal surgery, leaving a marked paucity of information on the issue of primary suture placement during emergency laparotomy, with the exception of perforated diverticular disease. For perforated diverticulitis with purulent or fecal generalized peritonitis a series of randomized controlled trials (RCTs) triggered a shift in the indoctrinated non-restorative Hartmann approach by presenting primary anastomosis as a feasible alternative [ , – ]. In most of the mentioned RCTs, primary bowel reconstruction in the acute setting was accompanied by the placement of a diverting enterostomy by study design [ – ]. Only the LADIES trial allowed surgeons to decide whether or not to place a diverting enterostomy when performing primary reconstruction . In our study, placement of a diverting stoma was defined as an exclusion criterium because of the high incidence of non-clinical (asymptomatic) leakage of distal sutures reported in the literature . In an attempt to facilitate the choice of the appropriate surgical procedure in patients with generalized peritonitis due to perforated diverticulitis, a recent position paper defined septic shock, overall fitness to surgery and peritonitis severity as important factors to consider in the decision-making process . While the notion of septic shock is clearly defined by the SEPSIS-3 consensus definitions, no explicit easy-to-use, “surgeon-friendly” scoring system for pre- or intraoperative assessment could be recommended based on current evidence. Immunocompetence, ASA-Score and MPI were suggested as adjutants in choosing restorative or non-restorative resections in hemodynamically stable patients . As current guidelines and position papers ultimately leave the choice of the emergency operative procedure in the surgeons’ hands, the results of Karliczek et al. showing surgeons’ assessment to be a poor predictor for anastomotic leakage further consolidates the need of identifying objective criteria for selecting patients for primary bowel reconstruction under peritonitic conditions . By the a priori exclusion of non-restorative resections, hemodynamic instable patients for which damage control surgery is the only obvious and valid option are not included in the present study. The preoperative ASA score did not discriminate between patients developing suture leakage and those who did not in our patient cohort. Neither did the intake of immunosuppressives or the chronic use of corticosteroids. The extent of peritonitis was, however, an independent predictor of suture outcome in the multivariate analysis. We chose to evaluate a simplified form of the MPI, developed to solely assess the extent and quality of the intraoperatively determined peritonitis for a number of reasons. First, the MPI was originally developed in 1987 for predicting postoperative morbidity in a cohort that also included peritonitis due to upper GI perforation but excluded postoperative peritonitis and mesenteric infarction . Neither inclusion and exclusion criteria nor primary outcome matched the purpose of our study. Second, the MPI includes various parameters such as age, sex, preexisting malignancy that did not influence our primary outcome parameter in the univariate analysis. Third, it is easier for the operating surgeon to simply discriminate between purulent or fecal peritonitis and between localized or generalized peritonitis than to calculate a more intricate score. Our data show that patients with generalized fecal peritonitis developed a significantly higher rate of suture leakage ( p =0.0014) than patients with less extensive peritonitis, while also showing a trend ( p =0.0788) in the mortality rate. The pathophysiological events triggered within the peritoneal cavity by the spillage of intestinal content seem to critically impact the complex and incompletely understood healing process of the sutured intestinal wall. Altered peripheral blood perfusion, bowel distention and intestinal wall edema are just few of the macroscopic changes imposing a greater degree of difficulty for the surgeon attempting primary bowel repair. The alterations on a microscopic and molecular level are just as intricate, as inflammatory status, microbiome and genetics all seem to affect intestinal suture healing . In a histologic analysis of colonic tissue samples Stumpf et al. identified a preexisting impairment in collagen metabolism as a possible risk factor for the healing of enteric sutures . Polymorphisms in lipid signaling and metabolic pathways are also thought to predispose to altered intestinal suture healing, underlining the importance of the preoperative patient status . In our study, sCHE activity was the only relevant preoperative parameter identified as having a significant predictive value for suture outcome in the multivariate analysis. We deliberately chose to analyze sCHE activity instead of albumin in order to avoid data distortion by parenteral albumin infusions in patients that were hospitalized previous to emergency surgery. In support of sCHE as a predictor for anastomotic healing Antolovic et al. identified low preoperative sCHE levels as an independent risk factor for bile leakage in 519 patients who underwent hepaticojejunostomy . In an emergency setting, our study is one of the few approaching the issue of preoperative predictors for a successful primary bowel reconstruction. Various studies have validated sCHE as a marker of nutritional status, correlating low sCHE levels to sarcopenia and to a high nutritional risk in critically ill patients treated on ICUs [ , , ]. Beside the critically ill, oncologic patients are another group for which malnutrition importantly influenced postoperative morbidity and mortality . In patients with colorectal cancer, low sCHE levels were associated with poor 5-year overall and disease-specific survival rates , whereas nutritional support led to an increase in sCHE levels and in body weight . In an analysis of 453 prospectively recruited treatment-naïve cancer patients, without manifest hepatic involvement, Pavo et al. reported that decreased sCHE is associated with an increased all-cause mortality . Interestingly, an inverse correlation of sCHE with CRP was observed (r = − 0.21, p < 0.001) as in our study (r = − 0.40, p < 0.001). In another series of patients with non-malignant disease, sCHE was shown to negatively correlate with further parameters of inflammation, namely interleukin (IL)-6 and tumor necrosis factor alpha (TNF)-α . The observed association with inflammatory parameters is not surprising since the body of evidence linking sCHE to the inflammatory response to injury is continuously growing. SCHE is part of the non-neuronal cholinergic system (NNCS), a complex regulatory network including most immune cells and regulating their function in the setting of local and systemic inflammation . By targeting this system through intraperitoneal injection of CHE inhibitors in an experimental abdominal sepsis model, Hofer et al. showed that locally administered CHE inhibitors led to a reduced production of pro-inflammatory cytokines and improved survival, most probably by increasing acetylcholine levels that control cytokine production . This apparently beneficial effect of a lowered or inhibited CHE activity intuitively stands in contradiction with the clinical observation that a low sCHE activity measured at the clinical onset of sepsis is an independent predictor of worse outcome and higher mortality . However, the anti-inflammatory effect of increased acetylcholine levels is expected to impair host defense against infections, which most probably offsets its benefit [ – ]. Several other studies on collectives of critically ill patients requiring ICU care identified low sCHE activity as a relevant predictor of increased mortality [ , , ]. Peng et al. determined in an analysis of adult septic patients that every unit (kU/L) decrease in sCHE activity doubles the odds of death within 30 days from sepsis onset . The exact mechanisms through which a reduction in sCHE activity leads to the observed results are far from being elucidated but suggest complex and intertwined derangements of metabolic and inflammatory pathways. The two parameters, sCHE levels and the extent of peritonitis, put forward by our analysis to facilitate the intraoperative decision-making process during emergency surgery for secondary peritonitis require further prospective validation, as the retrospective and single center nature of the current study constitutes its major limitation. As discussed by previous authors, the recruitment of patients for RCTs in an emergency setting is challenging, as many patients are not able to give informed consent due to the severity of their condition. Several RCTs on emergency surgery for acute diverticulitis had to be prematurely terminated due to recruitment issues . Another shortcoming of the present study are the limited number of pre- and intraoperative variables considered for analysis. Although most of the standard laboratory parameters were accounted for in our study, further inflammatory and metabolic regulators that may affect postoperative wound healing need to be considered for a more accurate risk assessment. Nevertheless, the identified predictors of suture outcome have the benefit of being readily available at the time the decision on bowel reconstruction in SCS is due. A trial and error approach until having built one’s surgical experience is to be avoided at the expense of such a critical patient contingent. In the lack of data deriving from RCTs, the years of surgical experience compressed in the current study is a valuable stepping stone to further our understanding of intestinal suture healing in a peritonitic environment.
Low preoperative sCHE activity and a high extent of the intraoperatively determined peritonitis are two easily quantifiable parameters that significantly correlate with a poor outcome of enteric sutures placed during SCS for secondary peritonitis. An objective surgical decision tailored to the patients’ individual pathophysiological pattern helps the surgeons, as they are no longer dependent on subjective considerations alone, while also benefiting the patients through the choice of the appropriate procedure.
|
A multi-species benchmark for training and validating mass spectrometry proteomics machine learning models | 3c2d3b0f-3915-4f6d-af93-690d29b2a5af | 11549408 | Biochemistry[mh] | De novo sequencing of proteomics tandem mass spectrometry data, in which observed fragmentation spectra are translated into corresponding peptide sequences, has been an open challenge for more than 40 years . Recently, as in many other areas of science, considerable progress toward solving this challenge has been made using deep learning, in which multi-layer neural networks with millions of parameters are trained to generate peptide sequences from observed spectra. The first such deep learning method, DeepNovo , has been followed by at least 22 additional publications (reviewed in ). The standard method for evaluating these de novo sequencing methods is to use a gold standard produced via database search. In this approach, mass spectrometry data derived from a single species is searched against the reference proteome for that species, yielding a ranked list of peptide-spectrum matches (PSMs). Including in the peptide database a collection of reversed or shuffled “decoy” peptides provides a rigorous way to set a threshold in this list of PSMs while controlling the false discovery rate (FDR) among the PSMs above the threshold . The resulting set of high-confidence PSMs can be used either to train or evaluate a de novo sequencing model. Some version of the above protocol has been used to develop labeled training and validation data for essentially every published deep learning de novo sequencing method. One exception is methods that use spectra from synthesized peptide sequences for training – . However, even in these cases, a gold standard derived from database search is used for evaluation of the method. Unfortunately, creating a high quality gold standard set of labeled spectra can be tricky. One challenge is ensuring that the search strategy employs appropriate parameters. For instance, one widely used benchmark dataset used a search strategy that failed to account for missassigned isotopic peaks during the acquisition stage. This error led to frequently assigning a deamidation modification, when the observed mass shift was better explained by an isotopic mass shift on the precursor m/z . A second challenge relates to the notion of train/test leakage, in which information used to train the model leaks into the evaluation procedure. In the de novo setting, a common mistake is to randomly segregate a given set of labeled spectra into training and test sets, without regard to the associated peptides. As a result, spectra generated by the same peptide sequence may occur in both the training and test sets. Such duplicated peptides give an unfair advantage to the sequencing method, and the leakage will be even more useful to parameter-rich methods that are capable of memorizing many features of the training data. In this work, we revisit the nine-species benchmark dataset that was employed in the first deep learning de novo sequencing method, DeepNovo . This is a widely used dataset, which has been employed for training or evaluation in at least 15 subsequent studies , – . The setup is quite straightforward. The authors downloaded nine publicly available datasets, all of which were generated on a Thermo Scientific Q Exactive mass spectrometer, and each of which was carried out in a different species. Each dataset was searched against the corresponding reference proteome, using a target-decoy strategy to accept a set of PSMs subject to a PSM-level FDR threshold of 1%. Because the data are derived from different species, the peptides in each set are largely (but not entirely) disjoint. To use the benchmark, it is typical to apply a cross-validation strategy, in which a model is trained on eight species and tested on the held-out species, and the procedure is repeated nine different ways. In developing our Casanovo de novo sequencing model, we identified several problems with the nine-species benchmark . These included the deamidation problem mentioned above, as well as some uncertainty regarding how the FDR was controlled. Perhaps most importantly, we recognized that a non-negligible proportion of peptides are shared among the different species, with the highest overlap between human and mouse. In light of these difficulties, we downloaded the same datasets from the PRIDE repository and systematically reanalyzed all of the data, using a standard search procedure—the Tide search engine followed by Percolator with PSM-level FDR control at 1%. We then filtered the PSMs to prevent any peptide sequence from appearing in more than one species. The resulting data set was used to evaluate Casanovo . Finally, because some of the single-species datasets are markedly larger than others, we produced a more balanced version of the dataset. Hence, we make publicly available both versions of this dataset: the peptide-disjoint dataset that can be used to avoid train/test leakage (“main”), and the reduced peptide-disjoint dataset if you want your analysis to run more quickly (“balanced”). In addition, we make available all of the intermediate files, for use in validating the benchmark.
Data sets For our benchmark, we used the same nine studies originally identified by Tran et al . . Paiva et al . investigated the protein expression response of the cowpea plant ( Vigna unguiculata ) to infection by Cowpea severe mosaic virus (CPSMV) by carrying out label-free proteomic analysis of cowpea leaves that were inoculated with CPSMV compared to mock inoculation controls . Nevo et al . studied a rare autosomal recessive lysosomal storage disorder, cystinosis, by carrying out SILAC proteomic analysis of engineered mouse cell lines that harbor a known pathogenic mutation of the causative gene, CTNS . Cassidy et al . evaluated two different analytical approaches for carrying out full proteome analysis while identifying short open reading frames: a high/low pH reversed phase LC-MS bottom-up approach and a semi-top-down strategy involving separation of proteins in a GelFree system followed by digestion and LC-MS analysis . The experiments were carried out using the methane producing archaeon Methanosarcina mazei . Reuss et al . carried out proteomic analyses on a series of minimized strains of the model bacterium, Bacillus subtilis , with genomes reduced by ~ 36% . Petersen et al . performed proteomic analysis of Candidatus endoloripes , which are bacterial symbionts of the Lucinidae family of marine bivalves . Mata et al . characterized the proteome of the tomato pericarp at its ripe red stage . Seidel et al . analyzed the global proteomic stress response in wildtype and two yeast knockout strains for the gene PBP1 . Hu et al . studied honeybees that exhibit a suite of behaviors ( Varroa sensitive hygiene—VSH) associated with infection with the Varroa destructor virus . Proteomic analysis was carried out on mushroom bodies and antennae of adult honeybees with and without VSH. Cypryck et al . characterized extracellular vesicles released from human primary macrophages after infection with influenza A viruses . All nine studies were performed using a Thermo Scientific Q Exactive mass spectrometer. We downloaded the RAW files from the corresponding PRIDE projects (Table ) and converted them to MGF format using the ThermoRawFileParser v1.3.4 . We downloaded the corresponding nine UniProt reference proteomes and constructed a Tide index for each one, using Crux version 4.2. Note that, for one species ( Vigna mungo ) no reference proteome is available, so we used the proteome of the closely related species Vigna radiata . Database search and FDR control We assigned peptide labels to spectra using the Tide search engine followed by post-processing with Percolator. In creating the Tide index, we specified Cys carbamidomethylation as a static modification and allowed for the following variable modifications: Met oxidation, Asn deamidation, Gln deamidation, N-term acetylation, N-term carbamylation, N-term NH 3 loss, and the combination of N-term carbamylation and NH 3 loss by using the tide-index options --mods-spec 1M+15.994915, 1N+0.984016, 1Q+0.984016 --nterm-peptide-mods-spec 1X+42.010565, 1X+43.005814, 1X-17.026549, 1X+25.980265 --max-mods 3. Note that one of the nine experiments ( Mus musculus ) was performed using SILAC labeling, but we searched without SILAC modifications and hence include in the benchmark only PSMs from unlabeled peptides. Tide automatically added to each index a shuffled decoy peptide corresponding to each target peptide. Thereafter, each MGF file was searched against the corresponding index using the precursor window size and fragment bin tolerance specified in the original study (Table ). The search engine employed XCorr scoring with Tailor calibration , and we allowed for 1 isotope error in the selection of candidate peptides. All search results were then analyzed jointly per species using the Crux implementation of Percolator, with default parameters. For the benchmark, we retained all PSMs with Percolator q value < 0.01. We identified 13 MGF files with fewer than 100 accepted PSMs, and we eliminated all of these PSMs from the benchmark. At this point in the processing pipeline, the dataset contains 2,898,611 annotated spectra (PSMs) drawn from 343 RAW files and associated with 168,422 distinct peptides. Avoiding train/test leakage To avoid train/test leakage, we post-processed the PSMs to eliminate peptides that are shared between species. Among the 168,422 distinct peptides, we identified 4121 (2.4%) that occur in more than one species. For each such peptide, we selected one of the associated species at random and then eliminated all PSMs containing that peptide in other species. Note that when identifying shared peptides between species, we considered all modified forms of a given peptide sequence to be the same, and we converted all isoleucines to leucines. Hence, if a given peptide appears in more than one species, then that peptide, including all its modified forms, is randomly assigned to a single species and eliminated from the others. The final, non-redundant benchmark dataset (“main”) consists of 2,838,117 PSMs corresponding to 168,422 distinct peptides. Balancing the benchmark At this stage, the benchmark was quite imbalanced, in the sense that some species had a much larger number of associated PSMs. We therefore used a random downsampling procedure to produce a benchmark that is more evenly balanced across species. Among the nine species, the one with the fewest PSMs is Mus musculus , with 25,522. Downsampling all of the other eight species to have 25,000 PSMs would reduce the size of the dataset from 2.8 million PSMs to 225,000—a reduction of 92%. To avoid producing such a small dataset, we therefore opted to downsample each dataset to approximately 100,000 PSMs. This approach yields a slight imbalance, because three species have fewer than 100,000 PSMs (44,555 for H. sapiens and 82,514 for Candidatus endoloripes ), while retaining a larger percentage of the original data. Our downsampling procedure involved randomly permuting the order of the MGF files for each species and then selecting the files in order until at least 100,000 PSMs have been accepted. The final, balanced benchmark dataset is approximately one quarter the size of the main benchmark, consisting of 779,879 PSMs from 133,232 distinct peptides.
For our benchmark, we used the same nine studies originally identified by Tran et al . . Paiva et al . investigated the protein expression response of the cowpea plant ( Vigna unguiculata ) to infection by Cowpea severe mosaic virus (CPSMV) by carrying out label-free proteomic analysis of cowpea leaves that were inoculated with CPSMV compared to mock inoculation controls . Nevo et al . studied a rare autosomal recessive lysosomal storage disorder, cystinosis, by carrying out SILAC proteomic analysis of engineered mouse cell lines that harbor a known pathogenic mutation of the causative gene, CTNS . Cassidy et al . evaluated two different analytical approaches for carrying out full proteome analysis while identifying short open reading frames: a high/low pH reversed phase LC-MS bottom-up approach and a semi-top-down strategy involving separation of proteins in a GelFree system followed by digestion and LC-MS analysis . The experiments were carried out using the methane producing archaeon Methanosarcina mazei . Reuss et al . carried out proteomic analyses on a series of minimized strains of the model bacterium, Bacillus subtilis , with genomes reduced by ~ 36% . Petersen et al . performed proteomic analysis of Candidatus endoloripes , which are bacterial symbionts of the Lucinidae family of marine bivalves . Mata et al . characterized the proteome of the tomato pericarp at its ripe red stage . Seidel et al . analyzed the global proteomic stress response in wildtype and two yeast knockout strains for the gene PBP1 . Hu et al . studied honeybees that exhibit a suite of behaviors ( Varroa sensitive hygiene—VSH) associated with infection with the Varroa destructor virus . Proteomic analysis was carried out on mushroom bodies and antennae of adult honeybees with and without VSH. Cypryck et al . characterized extracellular vesicles released from human primary macrophages after infection with influenza A viruses . All nine studies were performed using a Thermo Scientific Q Exactive mass spectrometer. We downloaded the RAW files from the corresponding PRIDE projects (Table ) and converted them to MGF format using the ThermoRawFileParser v1.3.4 . We downloaded the corresponding nine UniProt reference proteomes and constructed a Tide index for each one, using Crux version 4.2. Note that, for one species ( Vigna mungo ) no reference proteome is available, so we used the proteome of the closely related species Vigna radiata .
We assigned peptide labels to spectra using the Tide search engine followed by post-processing with Percolator. In creating the Tide index, we specified Cys carbamidomethylation as a static modification and allowed for the following variable modifications: Met oxidation, Asn deamidation, Gln deamidation, N-term acetylation, N-term carbamylation, N-term NH 3 loss, and the combination of N-term carbamylation and NH 3 loss by using the tide-index options --mods-spec 1M+15.994915, 1N+0.984016, 1Q+0.984016 --nterm-peptide-mods-spec 1X+42.010565, 1X+43.005814, 1X-17.026549, 1X+25.980265 --max-mods 3. Note that one of the nine experiments ( Mus musculus ) was performed using SILAC labeling, but we searched without SILAC modifications and hence include in the benchmark only PSMs from unlabeled peptides. Tide automatically added to each index a shuffled decoy peptide corresponding to each target peptide. Thereafter, each MGF file was searched against the corresponding index using the precursor window size and fragment bin tolerance specified in the original study (Table ). The search engine employed XCorr scoring with Tailor calibration , and we allowed for 1 isotope error in the selection of candidate peptides. All search results were then analyzed jointly per species using the Crux implementation of Percolator, with default parameters. For the benchmark, we retained all PSMs with Percolator q value < 0.01. We identified 13 MGF files with fewer than 100 accepted PSMs, and we eliminated all of these PSMs from the benchmark. At this point in the processing pipeline, the dataset contains 2,898,611 annotated spectra (PSMs) drawn from 343 RAW files and associated with 168,422 distinct peptides.
To avoid train/test leakage, we post-processed the PSMs to eliminate peptides that are shared between species. Among the 168,422 distinct peptides, we identified 4121 (2.4%) that occur in more than one species. For each such peptide, we selected one of the associated species at random and then eliminated all PSMs containing that peptide in other species. Note that when identifying shared peptides between species, we considered all modified forms of a given peptide sequence to be the same, and we converted all isoleucines to leucines. Hence, if a given peptide appears in more than one species, then that peptide, including all its modified forms, is randomly assigned to a single species and eliminated from the others. The final, non-redundant benchmark dataset (“main”) consists of 2,838,117 PSMs corresponding to 168,422 distinct peptides.
At this stage, the benchmark was quite imbalanced, in the sense that some species had a much larger number of associated PSMs. We therefore used a random downsampling procedure to produce a benchmark that is more evenly balanced across species. Among the nine species, the one with the fewest PSMs is Mus musculus , with 25,522. Downsampling all of the other eight species to have 25,000 PSMs would reduce the size of the dataset from 2.8 million PSMs to 225,000—a reduction of 92%. To avoid producing such a small dataset, we therefore opted to downsample each dataset to approximately 100,000 PSMs. This approach yields a slight imbalance, because three species have fewer than 100,000 PSMs (44,555 for H. sapiens and 82,514 for Candidatus endoloripes ), while retaining a larger percentage of the original data. Our downsampling procedure involved randomly permuting the order of the MGF files for each species and then selecting the files in order until at least 100,000 PSMs have been accepted. The final, balanced benchmark dataset is approximately one quarter the size of the main benchmark, consisting of 779,879 PSMs from 133,232 distinct peptides.
The dataset contains files resulting from various steps in the generation of the benchmark: Spectrum files in MGF format, produced by ThermoRawFileParser. Reference proteome files in FASTA format, downloaded from UniProt. Search results files for both targets and decoys, in tab-delimited format, produced by Tide. PSM-level Percolator results files for targets, in tab-delimited format. Annotated MGF and corresponding mzSpecLib files for both versions of the benchmark (main and balanced). Also included are log files for the steps of the analysis pipeline carried out using Crux (Tide indexing, Tide search, and Percolator). The data is available at 10.5281/zenodo.13685813.
Data quality and interpretability varies dramatically from study to study, due to differences in sample type, sample preparation protocols, chromatography and instrument settings, and database size. To assess the overall rate of successful identification of spectra in each data set, we plotted the number of accepted PSMs as a function of PSM-level FDR threshold (Fig. ). As is typical in proteomics database search, the curves go up rapidly before leaving the y-axis, corresponding to the many spectra with highly confident peptide assignments. To better understand the relative quality of the datasets, we also computed the proportion of spectra that were accepted at 1% PSM-level FDR per species (Fig. ). Here we observe that some datasets yield much higher rates of accepted PSMs than others, up to 39.7% for Saccharomyces cerevisiae and down to 3.6% for Candidatus endoloripes . Despite this large variance in the rate of accepted PSMs, characterizing the proportion of the total peak intensities that is explained by matched b- and y-ions (Fig. ) suggests that the quality of the accepted PSMs is high. Notably, the proportion of matched b- and y-ions does not appear to be strongly correlated with the rate of accepted PSMs per species.
|
First impressions: A prospective evaluation of patient–physician concordance and satisfaction following the initial medical oncology consultation | 8938b667-1358-4b27-8ff8-57f6c9ff35fd | 10757128 | Internal Medicine[mh] | INTRODUCTION Gastrointestinal (GI) malignancies account for 26% of all new cancer cases and 35% of all cancer‐related deaths globally. Systemic chemotherapy is an integral component of the multidisciplinary treatment of GI cancers and most patients are seen by a medical oncologist at some point in their cancer journey in addition to assessments by surgical and/or radiation oncologists. Due to the complex nature of the disease and the involvement of multiple specialties, it is critical that patients and families understand their cancer diagnosis, the purpose of each treatment modality, and prognosis, so that they can make informed decisions regarding their care. In previous studies that measured comprehension and satisfaction with the information given to newly diagnosed lung cancer patients, concordance between patients and physicians were highest in diagnosis (86%–90%) and type of treatment (81%–83%) but relatively low in treatment intent (42%–49%). , The agreement between patients and physicians on cancer curability in other studies ranged between 29% and 76%, although patient cohorts were variable in disease site and stage in cancer journey. , , , Furthermore, physician cohorts were heterogeneous with respect to medical specialty and level of experience dealing with serious illnesses. , , , , , Prognostic disclosure following a cancer diagnosis has been a more difficult topic for physicians to address across all parts of the world. , , , , , , , , In a multicenter study, over 70% of patients diagnosed with metastatic solid malignancies in the United States wanted to be told their life expectancy. However, only 17.6% recalled a prognosis being disclosed by their oncologists. This discordance is concerning, as an accurate understanding of prognosis may help patients and caregivers with the decision‐making process and manage their expectations as they begin treatment. Effective communication skills and patient‐centered care have been key focus areas in improving health literacy, decision‐making, and clinical outcomes in cancer patients. , , , , , , In fact, many consultation aids have been developed for both physicians and patients to optimize resources and improve communication at the initial visit in an oncology setting. , , , , , , , , , In this study, we aimed to evaluate the effectiveness of physician‐patient communication during a patient's first medical oncology consultation at the Gastrointestinal Oncology Clinic at the Princess Margaret Cancer Centre (PMCC), a tertiary referral center in Toronto, Canada, to identify key areas in which we can improve clinical care based on the needs and values of our patients and caregivers. We chose to focus on the first consultation as it is often an emotionally overwhelming and an information‐heavy visit. We also wanted to assess a more homogenous group of physicians such as medical oncologists, who would be more adept in delivering the type of sensitive news specific to cancer patients. The primary objective of this study was to quantify patient–physician concordance in understanding of diagnosis, treatment modality and intent, and prognosis of GI malignancies. The secondary objective was to investigate whether patient or physician characteristics were associated with patient satisfaction on communication delivered at the initial consultation.
METHODS 2.1 Study design and population This was a prospective study that recruited consecutive patients (≥18 years old) with a confirmed diagnosis of a GI malignancy (i.e., gastroesophageal, small intestinal, colorectal, anal canal, hepatic, pancreatic, and biliary cancers) during their first encounter with a medical oncologist at the PMCC. Medical oncologists who participated in this study all subspecialized in GI malignancies. Between January and August 2021, patients without prior systemic treatment for their GI malignancy were approached within 24 h of their initial consultation for voluntary participation in a paper‐based questionnaire assessing understanding of their disease and satisfaction with the communication surrounding their cancer. Patients were excluded if they met one or more of the following criteria: (1) Non‐English speaking patients who were not accompanied by an English‐speaking caregiver at the consultation (as the study questionnaire was provided only in English); (2) patients already receiving chemotherapy or have received chemotherapy for their GI malignancy; (3) enrolled patients who were unable to complete the questionnaire by their second clinic visit; and (4) patients with poor performance status (i.e., Eastern Cooperative Oncology Group (ECOG) score 3–4) as judged by the patient's health‐care team. 2.2 Survey instrument and measures Informed consent was obtained from each patient. Patients were required to complete the study questionnaire after their consultation in clinic or at home and return it to the study coordinator prior to the second clinic visit. 2.2.1 Patient understanding Patient understanding of the information delivered during the initial consultation was measured with a questionnaire adapted from previous patient recall studies , and input from the clinical team. The questionnaire composed of multiple choice and open‐ended questions on demographics (age, sex, educational level, primary language, and whether a family member or friend was present during the initial consultation) and the information discussed during the consultation (cancer diagnosis, treatment options, treatment intent [curative vs. palliative], and prognosis) (Data ). Staff medical oncologists completed a similar version of the questionnaire for each patient who consented to the study. The physician questionnaire was also composed of open‐ended and multiple choice questions on demographics (years in practice, whether a medical trainee saw the patient first) and the information discussed during the consultation (Data ). Patient and physician questionnaires were completed separately and responses were not shared with either group. The responses were then compared and the degree of congruence were defined as fully concordant, partially concordant, and fully discordant. Each category was scored either fully concordant or fully discordant, except for treatment plan, in which patients and physicians were allowed to select more than one option (e.g., chemotherapy and radiotherapy). A response was partially concordant if the patient and physician did not select all the same options but had at least one common answer selected. 2.2.2 Patient satisfaction with communication The satisfaction survey consisted of 10 statements adapted from Schofield et al., which assessed how clearly diagnosis, treatment plan, treatment intent, and prognosis were communicated. A 5‐point Likert scale was employed (strongly disagree, disagree, neither agree or disagree, agree, or strongly agree). If all 10 statements were completed by the patient, a total satisfaction score was calculated for a maximum of 50 points (10 questions × 5 points/question). Total satisfaction scores were stratified by patient response to what information was given for treatment intent and prognosis (items 7 and 8 on Data , respectively). 2.3 Data collection and analysis Study data were de‐identified, collected, and managed using REDCap electronic data capture tools hosted at the University Health Network. REDCap (Research Electronic Data Capture) is a secure, web‐based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages; and (4) procedures for data integration and interoperability with external sources. Data were reported as means and standard deviations for continuous variables and as counts and percentages for categorical variables. A univariable logistic regression analysis was performed to assess patient‐ and physician‐based factors associated with concordance on treatment intent and prognosis. Patient‐based factors included age, sex, prior cancer history, educational level, primary language, whether the patient completed the study questionnaire at home, whether a family member or friend accompanied the patient, whether a patient has had a consultation with an external medical oncologist or surgical/radiation oncologist, and patient perception of treatment intent, and prognosis. Physician‐based factors included years of practice in medical oncology and whether the patient was first assessed by a medical trainee (fellow, resident, physician assistant). Similarly, a uni/multivariable linear regression analysis was conducted using the same variables on patient satisfaction levels. Statistically significant results were defined with p ≤0.05. Statistical analyses were performed using R v4.1.2.
Study design and population This was a prospective study that recruited consecutive patients (≥18 years old) with a confirmed diagnosis of a GI malignancy (i.e., gastroesophageal, small intestinal, colorectal, anal canal, hepatic, pancreatic, and biliary cancers) during their first encounter with a medical oncologist at the PMCC. Medical oncologists who participated in this study all subspecialized in GI malignancies. Between January and August 2021, patients without prior systemic treatment for their GI malignancy were approached within 24 h of their initial consultation for voluntary participation in a paper‐based questionnaire assessing understanding of their disease and satisfaction with the communication surrounding their cancer. Patients were excluded if they met one or more of the following criteria: (1) Non‐English speaking patients who were not accompanied by an English‐speaking caregiver at the consultation (as the study questionnaire was provided only in English); (2) patients already receiving chemotherapy or have received chemotherapy for their GI malignancy; (3) enrolled patients who were unable to complete the questionnaire by their second clinic visit; and (4) patients with poor performance status (i.e., Eastern Cooperative Oncology Group (ECOG) score 3–4) as judged by the patient's health‐care team.
Survey instrument and measures Informed consent was obtained from each patient. Patients were required to complete the study questionnaire after their consultation in clinic or at home and return it to the study coordinator prior to the second clinic visit. 2.2.1 Patient understanding Patient understanding of the information delivered during the initial consultation was measured with a questionnaire adapted from previous patient recall studies , and input from the clinical team. The questionnaire composed of multiple choice and open‐ended questions on demographics (age, sex, educational level, primary language, and whether a family member or friend was present during the initial consultation) and the information discussed during the consultation (cancer diagnosis, treatment options, treatment intent [curative vs. palliative], and prognosis) (Data ). Staff medical oncologists completed a similar version of the questionnaire for each patient who consented to the study. The physician questionnaire was also composed of open‐ended and multiple choice questions on demographics (years in practice, whether a medical trainee saw the patient first) and the information discussed during the consultation (Data ). Patient and physician questionnaires were completed separately and responses were not shared with either group. The responses were then compared and the degree of congruence were defined as fully concordant, partially concordant, and fully discordant. Each category was scored either fully concordant or fully discordant, except for treatment plan, in which patients and physicians were allowed to select more than one option (e.g., chemotherapy and radiotherapy). A response was partially concordant if the patient and physician did not select all the same options but had at least one common answer selected. 2.2.2 Patient satisfaction with communication The satisfaction survey consisted of 10 statements adapted from Schofield et al., which assessed how clearly diagnosis, treatment plan, treatment intent, and prognosis were communicated. A 5‐point Likert scale was employed (strongly disagree, disagree, neither agree or disagree, agree, or strongly agree). If all 10 statements were completed by the patient, a total satisfaction score was calculated for a maximum of 50 points (10 questions × 5 points/question). Total satisfaction scores were stratified by patient response to what information was given for treatment intent and prognosis (items 7 and 8 on Data , respectively).
Patient understanding Patient understanding of the information delivered during the initial consultation was measured with a questionnaire adapted from previous patient recall studies , and input from the clinical team. The questionnaire composed of multiple choice and open‐ended questions on demographics (age, sex, educational level, primary language, and whether a family member or friend was present during the initial consultation) and the information discussed during the consultation (cancer diagnosis, treatment options, treatment intent [curative vs. palliative], and prognosis) (Data ). Staff medical oncologists completed a similar version of the questionnaire for each patient who consented to the study. The physician questionnaire was also composed of open‐ended and multiple choice questions on demographics (years in practice, whether a medical trainee saw the patient first) and the information discussed during the consultation (Data ). Patient and physician questionnaires were completed separately and responses were not shared with either group. The responses were then compared and the degree of congruence were defined as fully concordant, partially concordant, and fully discordant. Each category was scored either fully concordant or fully discordant, except for treatment plan, in which patients and physicians were allowed to select more than one option (e.g., chemotherapy and radiotherapy). A response was partially concordant if the patient and physician did not select all the same options but had at least one common answer selected.
Patient satisfaction with communication The satisfaction survey consisted of 10 statements adapted from Schofield et al., which assessed how clearly diagnosis, treatment plan, treatment intent, and prognosis were communicated. A 5‐point Likert scale was employed (strongly disagree, disagree, neither agree or disagree, agree, or strongly agree). If all 10 statements were completed by the patient, a total satisfaction score was calculated for a maximum of 50 points (10 questions × 5 points/question). Total satisfaction scores were stratified by patient response to what information was given for treatment intent and prognosis (items 7 and 8 on Data , respectively).
Data collection and analysis Study data were de‐identified, collected, and managed using REDCap electronic data capture tools hosted at the University Health Network. REDCap (Research Electronic Data Capture) is a secure, web‐based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages; and (4) procedures for data integration and interoperability with external sources. Data were reported as means and standard deviations for continuous variables and as counts and percentages for categorical variables. A univariable logistic regression analysis was performed to assess patient‐ and physician‐based factors associated with concordance on treatment intent and prognosis. Patient‐based factors included age, sex, prior cancer history, educational level, primary language, whether the patient completed the study questionnaire at home, whether a family member or friend accompanied the patient, whether a patient has had a consultation with an external medical oncologist or surgical/radiation oncologist, and patient perception of treatment intent, and prognosis. Physician‐based factors included years of practice in medical oncology and whether the patient was first assessed by a medical trainee (fellow, resident, physician assistant). Similarly, a uni/multivariable linear regression analysis was conducted using the same variables on patient satisfaction levels. Statistically significant results were defined with p ≤0.05. Statistical analyses were performed using R v4.1.2.
RESULTS A total of 184 matched patient–physician surveys were completed. The mean age of patients was 64.5 ± 12.3 years (Table ). One fifth of patients had a prior history of a second primary cancer. Based on the oncologists' answers, there was a higher proportion of patients seen for palliative versus curative intent therapy ( n = 85 [46.7%] vs. n = 58 [31.8%], respectively). In our patient group, 84 (45.6%) were female, 112 (60.9%) had a college degree or higher, and 143 (77.7%) selected English as their primary language. Thirty‐six (19.6%) patients had already seen an external medical oncologist and were referred for either a second opinion and/or potential participation in clinical trials. One hundred twenty‐three (66.8%) patients had also seen a radiation and/or surgical oncologist prior to their medical oncology consultation at PMCC. Nearly 75% ( n = 137) of patients had a family member or friend that accompanied the patient at their consultation and was able to assist in completing the study questionnaire. Nine staff medical oncologists participated in the study. The mean number of years of experience as a staff medical oncologist was 13.1 ± 11.0 years (range: 2–34 years). More than half of the study patients ( n = 115; 62.5%) were seen by a medical oncologist within 10 years of appointment. 3.1 Patient–physician concordance Patient–physician concordance was divided into four categories: (1) diagnosis, (2) treatment plan, (3) treatment intent, and (4) prognosis (Table ). More than 90% ( n = 171) of patients agreed with their medical oncologist on the cancer disease site and 100% ( n = 184) of patients were either fully ( n = 107; 59.2%) or partially ( n = 75; 40.8%) concordant on their understanding of their treatment plan. The concordance rate for treatment intent was 66.8% ( n = 123). Of the 55 patients whose response for treatment intent was discordant from the physician's, 42 (76.4%) patients believed the intent of treatment to be curative or unclear when in fact the physician reported them to be palliative. The concordance rate for prognosis was 59.8% ( n = 110). Of the 65 discordant cases, 22 (33.8%) patients reported that prognosis was not discussed with them when physicians reported otherwise and six (9.2%) patients overestimated their prognosis to ≥1 year when physicians reported <1 year. Based on physician‐reported responses, the topic of prognosis was not addressed with 113 out of 184 (61.4%) patients, with an additional five (2.7%) patients requesting not to discuss it. Of the patients with whom prognosis was discussed ( n = 65, as per the physician's response), the majority were those who had a prognosis of ≥1 year ( n = 45; 69.2%) as opposed to patients who were given a prognosis of <1 year ( n = 20; 30.8%). As the understanding of treatment intent and prognosis had the highest discordant rates between patients and physicians, we conducted separate univariable logistic regression analyses to determine which factors were associated with concordance rates of treatment intent and prognosis (Table ). In regard to treatment intent, we did not find any statistically significant associations. Our analysis of prognostic concordance, however, found male patients (OR = 3.03; 95% CI: 1.61–5.72) to be more likely in agreement with their oncologist. Conversely, the duration of practice as an attending medical oncologist (OR = 0.95; 95% CI: 0.92–0.98) was inversely associated with prognostic concordance. 3.2 Patient satisfaction The mean scores for each satisfaction item can be found in Table . Based on the 10 items (Data ) included in the satisfaction section of the questionnaire, the two areas that showed the lowest satisfaction scores (i.e., 0 = strongly disagree, 1 = disagree, or 2 = neither agree/disagree) were communication on (1) prognosis (12.5%) and (2) the way that their physician explored the impact of their diagnosis on their life (15.2%). Patients were most satisfied (i.e., 4 = agree or 5 = strongly agree) with the way their doctor answered their questions (95.7%) and 95.1% of patients felt assured that their doctor had their best interests in mind. They were also satisfied with how they were included in the decisions made around their care (94.5%). Overall, satisfaction with the quality of communication that patients received during their first medical oncology visit was high (total mean score out of 50 = 46.5 ± 6.5). Internal reliability of the satisfaction questionnaire was high (Cronbach's α = 0.947). In Figure , total satisfaction scores were stratified based on patient‐reported prognosis which showed statistically significant differences between the subgroups (prognosis <1 year; prognosis ≥1 year; prognosis not discussed) ( p = 0.043). We did not find any statistically significant differences between patient‐reported treatment intent subgroups (results not shown). Uni‐ and multivariable analyses were conducted to assess potential factors associated with patient satisfaction on patient–physician communication. As shown in Table , patients completing their questionnaire immediately after their consultation were significantly more likely to give a higher total satisfaction score than those who took their questionnaire home (adjusted estimate = 2.76; 95% CI: 0.72–4.81). However, total patient satisfaction scores were significantly lower if patients reported that the Intent of their treatment was unclear at time of consultation (adjusted estimate = −3.32; 95% CI: ‐5.93 ‐ ‐0.70). There was no statistical significance between patient‐reported prognosis and their satisfaction in communication.
Patient–physician concordance Patient–physician concordance was divided into four categories: (1) diagnosis, (2) treatment plan, (3) treatment intent, and (4) prognosis (Table ). More than 90% ( n = 171) of patients agreed with their medical oncologist on the cancer disease site and 100% ( n = 184) of patients were either fully ( n = 107; 59.2%) or partially ( n = 75; 40.8%) concordant on their understanding of their treatment plan. The concordance rate for treatment intent was 66.8% ( n = 123). Of the 55 patients whose response for treatment intent was discordant from the physician's, 42 (76.4%) patients believed the intent of treatment to be curative or unclear when in fact the physician reported them to be palliative. The concordance rate for prognosis was 59.8% ( n = 110). Of the 65 discordant cases, 22 (33.8%) patients reported that prognosis was not discussed with them when physicians reported otherwise and six (9.2%) patients overestimated their prognosis to ≥1 year when physicians reported <1 year. Based on physician‐reported responses, the topic of prognosis was not addressed with 113 out of 184 (61.4%) patients, with an additional five (2.7%) patients requesting not to discuss it. Of the patients with whom prognosis was discussed ( n = 65, as per the physician's response), the majority were those who had a prognosis of ≥1 year ( n = 45; 69.2%) as opposed to patients who were given a prognosis of <1 year ( n = 20; 30.8%). As the understanding of treatment intent and prognosis had the highest discordant rates between patients and physicians, we conducted separate univariable logistic regression analyses to determine which factors were associated with concordance rates of treatment intent and prognosis (Table ). In regard to treatment intent, we did not find any statistically significant associations. Our analysis of prognostic concordance, however, found male patients (OR = 3.03; 95% CI: 1.61–5.72) to be more likely in agreement with their oncologist. Conversely, the duration of practice as an attending medical oncologist (OR = 0.95; 95% CI: 0.92–0.98) was inversely associated with prognostic concordance.
Patient satisfaction The mean scores for each satisfaction item can be found in Table . Based on the 10 items (Data ) included in the satisfaction section of the questionnaire, the two areas that showed the lowest satisfaction scores (i.e., 0 = strongly disagree, 1 = disagree, or 2 = neither agree/disagree) were communication on (1) prognosis (12.5%) and (2) the way that their physician explored the impact of their diagnosis on their life (15.2%). Patients were most satisfied (i.e., 4 = agree or 5 = strongly agree) with the way their doctor answered their questions (95.7%) and 95.1% of patients felt assured that their doctor had their best interests in mind. They were also satisfied with how they were included in the decisions made around their care (94.5%). Overall, satisfaction with the quality of communication that patients received during their first medical oncology visit was high (total mean score out of 50 = 46.5 ± 6.5). Internal reliability of the satisfaction questionnaire was high (Cronbach's α = 0.947). In Figure , total satisfaction scores were stratified based on patient‐reported prognosis which showed statistically significant differences between the subgroups (prognosis <1 year; prognosis ≥1 year; prognosis not discussed) ( p = 0.043). We did not find any statistically significant differences between patient‐reported treatment intent subgroups (results not shown). Uni‐ and multivariable analyses were conducted to assess potential factors associated with patient satisfaction on patient–physician communication. As shown in Table , patients completing their questionnaire immediately after their consultation were significantly more likely to give a higher total satisfaction score than those who took their questionnaire home (adjusted estimate = 2.76; 95% CI: 0.72–4.81). However, total patient satisfaction scores were significantly lower if patients reported that the Intent of their treatment was unclear at time of consultation (adjusted estimate = −3.32; 95% CI: ‐5.93 ‐ ‐0.70). There was no statistical significance between patient‐reported prognosis and their satisfaction in communication.
DISCUSSION A heavy emphasis has been placed on patient‐centered care and informed decision‐making across all medical specialties in the last few decades. Not only was this study an internal quality assurance check in a single center, we objectively measured patient understanding and satisfaction of the information delivered specifically to patients with a GI malignancy at the initial medical oncology consultation. Patient–physician concordance was highest in cancer diagnosis (92.9%). Furthermore, all patients were able to select at least one treatment modality that was concordant with the physicians' response (59.2% fully concordant; 40.8% partially concordant). However, lower concordance rates were observed in patients' understanding of the intent of their treatment (66.8%) and their prognosis (59.8%). Patient satisfaction was generally high, although there were a few factors that were independently associated with lower satisfaction scores. 4.1 Patient understanding of treatment intent The terms “curative” versus “palliative” intent in the context of cancer treatment are often misunderstood by patients, caregivers, and even among health‐care providers. , While active anticancer treatment in the palliative (i.e., non‐curative) setting is given to control the progression of cancer, optimize symptom control, and hopefully, prolong survival, many patients perceive any systemic chemotherapy and/or radiation as a means to cure their cancer. On the other hand, patients may misinterpret “palliative intent” as being ineligible for any type of anticancer treatment and may cause additional distress. In this study, 123 (66.8%) patients selected the same answer as their oncologist for treatment intent (curative vs palliative). Over 75% of discordant cases were patients who believed their cancer to be curable or that more tests were necessary to establish a treatment plan, but in fact, were deemed incurable by their medical oncologist. Concordance rates on the understanding of treatment goals in Western and non‐Western countries range from 29% to 76%, suggesting that the challenges physicians face in communicating treatment intent are universal. , , , , , Focusing on the distinction between “curative” versus “palliative” intent and the role of palliative care services in the context of cancer is important early on, as it allows future conversations with patients to be based on an accurate understanding of the purpose of their treatment. 4.2 Patient understanding of prognosis There is significant variability in the attitudes and practices of oncologists giving “bad news” to patients. Although geographical, cultural, and familial factors may influence the disclosure of unfavorable medical information, , , there is also an inherent hesitation from physicians, who, understandably, worry about the emotional well‐being and relationship with their patients. Not surprisingly, the lowest concordance rate in this study was cancer prognosis (59.8%) (Table ). Based on physician responses, prognosis was not discussed during the initial consultation in 113 (62%) of the cases, in addition to the five (3%) participants who specifically requested not to discuss prognosis. Interestingly, nearly 70% of patients with which prognosis was discussed (physician‐reported) were given a prognosis of ≥1 year, whereas the remaining patients had <1 year, suggesting that oncologists may be subconsciously more likely to disclose prognosis if it was favorable. We did not perform a multivariable logistic regression analysis in patient–physician concordance rates as nearly 75% of our patients were assisted by their caregivers in the completion of the research questionnaire and hence, not an accurate depiction of how patient factors (e.g., education level, primary language) influence concordance. Weeks et al., found that the risk of reporting inaccurate beliefs about chemotherapy were dependent on cancer site (lung vs. colorectal), ethnicity, and patients' satisfaction with physician communication—regardless of education level, functional status, and patients' role in the decision‐making process. In our univariable logistic regression analysis, we showed that the odds of achieving patient‐physician concordance on prognosis was significantly higher if the patient was seen by a physician with less years of medical oncology experience (Table ). This may reflect the impact of an evolving medical curriculum that emphasizes patient‐centered care and strategies for high‐quality goals of care conversations for trainees in recent decades. 4.3 Patient satisfaction on physician–patient communication This study's secondary objective was to assess patients' satisfaction with the communication with their physician based on the first medical oncology consultation as it sets the tone throughout a patient's cancer journey. In our multivariable linear regression analysis, independent predictors of lower overall satisfaction in our patients included: (1) those who completed the questionnaire at home rather than immediately after the initial consultation; and (2) those who reported that treatment intent was unclear at time of the consultation (i.e., more diagnostic tests were needed) (Table ). However, what was more insightful was the absence of statistical significance for patient‐reported prognosis. Compared to patients who reported that prognosis was disclosed, patients were neither less nor more satisfied if prognosis was not disclosed (i.e., physician did not address it or patient requested not to discuss it). Although prognosis should be given at the patient's request, physician hesitancy to address the topic due to concerns of patient well‐being and their relationship with the patient should be alleviated, as we and many other studies showed that discussing prognosis was not associated with worse patient–physician relationship ratings, sadness, or anxiety. , , , , , , 4.4 Implications for practice This study revealed present challenges that medical oncologists encounter when delivering important and sensitive information to patients during a consultation. The application to the current practice of cancer medicine is evident—patient understanding of treatment intent and prognosis were areas that need the most improvement—similar to what has been previously published. , , , , , , , , , , , , , , Small but impactful changes such as focusing the discussion on defining key phrases like “curative vs. palliative treatment intent” and “median survival” can be made relatively quickly. In terms of patients who reported lower satisfaction scores if treatment intent was unclear at the time of the consultation, perhaps holding multidisciplinary cancer conferences prior to the consultation may help the clinical team agree on a general treatment plan/intent and avoid giving contradictory information to their patients. Physician‐targeted interventions such as OncoTalk designed by Back et al. may be considered to strengthen the communication between oncologists and patients. This experimental curriculum demonstrated that the 4‐day workshop significantly improved skills in “giving bad news.” Other aids that are less time‐demanding such as the SPIKES protocol developed by Baile et al. have been found to improve learner satisfaction, knowledge, and performance. However, patient outcomes following physician‐targeted interventions are not well studied and warrants further investigation. Consultation aids developed for patients may also be considered. Previous studies have found that patients asked significantly more questions about prognosis if they were provided a question prompt list than those with a general fact sheet prior to the consultation. , , , , However, they were also significantly more anxious than the control group and were less likely to achieve their preferred decision‐making style. This was alleviated if the aids were endorsed by the oncologist during the consultation, , indicating that both patients' and physicians' behavior must be targeted to improve the experience. Real‐world implementation of patient‐ and physician‐based interventions will depend on patient‐, provider‐, and system‐level barriers. 4.5 Limitations There were some limitations to this study. The first limitation was selection bias as non‐English speaking patients who were not accompanied by an English‐speaking caregiver were not approached. Thus, the concordance rates may be overestimated. Along these lines, as we did not provide questionnaires in languages other than English, participants who speak English as a second language may have had issues fully understanding certain questions. This was also a single‐center study taking place in the GI oncology clinic at an academic hospital where patients would see a medical trainee prior to the staff oncologist and thus, may not be generalizable to other disease sites nor community‐based centers. Secondly, recall bias from the patient as well as the physician may affect concordance rates if the questionnaire was completed immediately or a few days after the consultation. The direction of which this affects concordance is unknown as completing the patient questionnaire at home may allow the patient to talk to family and friends and research their illness through other resources, but it can also reduce the accuracy of recalling information discussed with their medical oncologist. Of note, total patient satisfaction scores were significantly higher when patients completed the questionnaire in clinic compared to completing it at home, which suggests that patients may have felt more inclined to give a higher rating when their oncologist was in closer proximity. Lastly, this study did not address whether patients' level of understanding of treatment intent and prognosis improved over consecutive visits. This may not be something we can “perfect” in the first visit, given the emotional toll of being told one has cancer and we acknowledge that as the patient–physician relationship evolves, the concordance in understanding evolves as well.
Patient understanding of treatment intent The terms “curative” versus “palliative” intent in the context of cancer treatment are often misunderstood by patients, caregivers, and even among health‐care providers. , While active anticancer treatment in the palliative (i.e., non‐curative) setting is given to control the progression of cancer, optimize symptom control, and hopefully, prolong survival, many patients perceive any systemic chemotherapy and/or radiation as a means to cure their cancer. On the other hand, patients may misinterpret “palliative intent” as being ineligible for any type of anticancer treatment and may cause additional distress. In this study, 123 (66.8%) patients selected the same answer as their oncologist for treatment intent (curative vs palliative). Over 75% of discordant cases were patients who believed their cancer to be curable or that more tests were necessary to establish a treatment plan, but in fact, were deemed incurable by their medical oncologist. Concordance rates on the understanding of treatment goals in Western and non‐Western countries range from 29% to 76%, suggesting that the challenges physicians face in communicating treatment intent are universal. , , , , , Focusing on the distinction between “curative” versus “palliative” intent and the role of palliative care services in the context of cancer is important early on, as it allows future conversations with patients to be based on an accurate understanding of the purpose of their treatment.
Patient understanding of prognosis There is significant variability in the attitudes and practices of oncologists giving “bad news” to patients. Although geographical, cultural, and familial factors may influence the disclosure of unfavorable medical information, , , there is also an inherent hesitation from physicians, who, understandably, worry about the emotional well‐being and relationship with their patients. Not surprisingly, the lowest concordance rate in this study was cancer prognosis (59.8%) (Table ). Based on physician responses, prognosis was not discussed during the initial consultation in 113 (62%) of the cases, in addition to the five (3%) participants who specifically requested not to discuss prognosis. Interestingly, nearly 70% of patients with which prognosis was discussed (physician‐reported) were given a prognosis of ≥1 year, whereas the remaining patients had <1 year, suggesting that oncologists may be subconsciously more likely to disclose prognosis if it was favorable. We did not perform a multivariable logistic regression analysis in patient–physician concordance rates as nearly 75% of our patients were assisted by their caregivers in the completion of the research questionnaire and hence, not an accurate depiction of how patient factors (e.g., education level, primary language) influence concordance. Weeks et al., found that the risk of reporting inaccurate beliefs about chemotherapy were dependent on cancer site (lung vs. colorectal), ethnicity, and patients' satisfaction with physician communication—regardless of education level, functional status, and patients' role in the decision‐making process. In our univariable logistic regression analysis, we showed that the odds of achieving patient‐physician concordance on prognosis was significantly higher if the patient was seen by a physician with less years of medical oncology experience (Table ). This may reflect the impact of an evolving medical curriculum that emphasizes patient‐centered care and strategies for high‐quality goals of care conversations for trainees in recent decades.
Patient satisfaction on physician–patient communication This study's secondary objective was to assess patients' satisfaction with the communication with their physician based on the first medical oncology consultation as it sets the tone throughout a patient's cancer journey. In our multivariable linear regression analysis, independent predictors of lower overall satisfaction in our patients included: (1) those who completed the questionnaire at home rather than immediately after the initial consultation; and (2) those who reported that treatment intent was unclear at time of the consultation (i.e., more diagnostic tests were needed) (Table ). However, what was more insightful was the absence of statistical significance for patient‐reported prognosis. Compared to patients who reported that prognosis was disclosed, patients were neither less nor more satisfied if prognosis was not disclosed (i.e., physician did not address it or patient requested not to discuss it). Although prognosis should be given at the patient's request, physician hesitancy to address the topic due to concerns of patient well‐being and their relationship with the patient should be alleviated, as we and many other studies showed that discussing prognosis was not associated with worse patient–physician relationship ratings, sadness, or anxiety. , , , , , ,
Implications for practice This study revealed present challenges that medical oncologists encounter when delivering important and sensitive information to patients during a consultation. The application to the current practice of cancer medicine is evident—patient understanding of treatment intent and prognosis were areas that need the most improvement—similar to what has been previously published. , , , , , , , , , , , , , , Small but impactful changes such as focusing the discussion on defining key phrases like “curative vs. palliative treatment intent” and “median survival” can be made relatively quickly. In terms of patients who reported lower satisfaction scores if treatment intent was unclear at the time of the consultation, perhaps holding multidisciplinary cancer conferences prior to the consultation may help the clinical team agree on a general treatment plan/intent and avoid giving contradictory information to their patients. Physician‐targeted interventions such as OncoTalk designed by Back et al. may be considered to strengthen the communication between oncologists and patients. This experimental curriculum demonstrated that the 4‐day workshop significantly improved skills in “giving bad news.” Other aids that are less time‐demanding such as the SPIKES protocol developed by Baile et al. have been found to improve learner satisfaction, knowledge, and performance. However, patient outcomes following physician‐targeted interventions are not well studied and warrants further investigation. Consultation aids developed for patients may also be considered. Previous studies have found that patients asked significantly more questions about prognosis if they were provided a question prompt list than those with a general fact sheet prior to the consultation. , , , , However, they were also significantly more anxious than the control group and were less likely to achieve their preferred decision‐making style. This was alleviated if the aids were endorsed by the oncologist during the consultation, , indicating that both patients' and physicians' behavior must be targeted to improve the experience. Real‐world implementation of patient‐ and physician‐based interventions will depend on patient‐, provider‐, and system‐level barriers.
Limitations There were some limitations to this study. The first limitation was selection bias as non‐English speaking patients who were not accompanied by an English‐speaking caregiver were not approached. Thus, the concordance rates may be overestimated. Along these lines, as we did not provide questionnaires in languages other than English, participants who speak English as a second language may have had issues fully understanding certain questions. This was also a single‐center study taking place in the GI oncology clinic at an academic hospital where patients would see a medical trainee prior to the staff oncologist and thus, may not be generalizable to other disease sites nor community‐based centers. Secondly, recall bias from the patient as well as the physician may affect concordance rates if the questionnaire was completed immediately or a few days after the consultation. The direction of which this affects concordance is unknown as completing the patient questionnaire at home may allow the patient to talk to family and friends and research their illness through other resources, but it can also reduce the accuracy of recalling information discussed with their medical oncologist. Of note, total patient satisfaction scores were significantly higher when patients completed the questionnaire in clinic compared to completing it at home, which suggests that patients may have felt more inclined to give a higher rating when their oncologist was in closer proximity. Lastly, this study did not address whether patients' level of understanding of treatment intent and prognosis improved over consecutive visits. This may not be something we can “perfect” in the first visit, given the emotional toll of being told one has cancer and we acknowledge that as the patient–physician relationship evolves, the concordance in understanding evolves as well.
CONCLUSION This study was one that focused specifically on GI malignancies and the only one to assess patient understanding and satisfaction after the first consultation with their medical oncologist. It objectively assessed a more homogeneous group of specialists (i.e., medical oncologists) at communicating complex ideas with patients during and identified areas for improvement in our clinic. We observed that concordance in the understanding of treatment intent and prognostic disclosure were suboptimal. It is reasonable to have high‐quality conversations regarding goals of care and prognosis early on in the oncologist–patient relationship.
Yvonne Bach: Data curation (lead); formal analysis (supporting); investigation (equal); methodology (supporting); project administration (equal); writing – original draft (lead); writing – review and editing (lead). Elan Panov: Conceptualization (lead); data curation (equal); formal analysis (supporting); investigation (equal); methodology (lead); project administration (equal); writing – review and editing (lead). Osvaldo Espin‐Garcia: Formal analysis (lead); writing – review and editing (supporting). Eric Chen: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (equal). Monika Krzyzanowska: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (equal). Grainne O'Kane: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (equal). Malcolm Moore: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (supporting). Rebecca M. Prince: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (supporting). Jennifer Knox: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (supporting). Robert Grant: Project administration (equal); resources (equal); supervision (equal); writing – review and editing (supporting). Lucy X. Ma: Project administration (supporting); writing – review and editing (equal). Michael J. Allen: Project administration (supporting); writing – review and editing (equal). Lawson Eng: Project administration (supporting); writing – review and editing (supporting). Ekaterina Kosyachkova: Project administration (supporting); writing – review and editing (equal). Thais Baccili Cury Megid: Writing – review and editing (equal). Carly Barron: Writing – review and editing (equal). Xin Wang: Writing – review and editing (equal). Marie‐Philippe Saltiel: Writing – review and editing (equal). Abdul Rehman Rehman Farooq: Writing – review and editing (equal). Raymond W. Jang: Conceptualization (equal); methodology (equal); project administration (equal); resources (equal); supervision (lead); writing – review and editing (equal). Elena Elimova: Conceptualization (equal); funding acquisition (lead); investigation (lead); methodology (equal); project administration (equal); resources (lead); supervision (lead); writing – review and editing (equal).
This study was funded by a generous donation through the Princess Margaret Cancer Foundation.
There are no disclosures to report.
This study received ethics approval from the University Health Network Review Ethics Board (CAPCR ID: 20‐6119).
Data S1. Click here for additional data file. Data S2. Click here for additional data file.
|
Living Well With Uncertainty in Advanced, Metastatic or Incurable Cancers: A Pragmatic Feasibility Study of the Adapting to Life With Cancer Cognitive ExisteNtial Therapy (ACCENT) | 988d813a-1cdd-42c7-9fb4-fc2298a2a1cb | 11953013 | Neoplasms[mh] | Introduction Forty‐six percent of Canadians will be diagnosed with cancer in their lifetime . The overall age‐standardized incidence rates for cancer have slowly declined over the last 3 decades; however, the number of individuals being diagnosed with and treated for cancer is rising owing to a growing and aging population . Individuals living with cancer are also surviving longer. In 2015–2017, the predicted 5‐year net survival rate for all cancers was 64% compared to 55% in the 1990s and 25% in the 1940s . This is due to improvements in early detection and treatments, especially recent advances in targeted‐, immuno‐, and hormone therapies . Many of these new treatments aim to increase length of life, especially in individuals with advanced, metastatic, or incurable (AMI) cancers . As more individuals are being diagnosed with cancer and living longer, there has been an appropriate shift in attention to the clinical needs of those with AMI cancers . Individuals living with AMI cancers report a range of psychosocial challenges related to the uncertainty inherent in their illness, including anxiety, fear of disease progression, fears about death or dying, and worries about the impact of illness on loved ones and social life . The ability to live well with uncertainty is known to vary considerably across individuals and greater difficulty tolerating uncertainty is associated with poorer psychological adjustment [ , , ]. Systematic reviews including thematic synthesis in AMI cancers suggest that experiences of uncertainty can impact financial, emotional, and social domains, and create unmet practical, informational, psychological, and communication needs in these individuals . Living well with uncertainty thus presents a unique challenge that must be addressed in this population [ , , ]. Interventions that address uncertainty among cancer patients have been researched. A recent review by Guan et al. identified 26 studies that reported on interventions to manage uncertainty in cancer for patients and caregivers. Eighteen of these had positive effects on uncertainty outcomes; however, only five included patients with advanced cancers [ , , , , ]. Two of these 5 studies involved patients with new diagnoses and disease recurrences, and both reported no significant changes in illness uncertainty following their respective interventions . Of the 3 studies focused exclusively on advanced cancer, one was a purely informational intervention that showed a reduction in decision making uncertainty after patients watched a video decision aid about palliative radiation therapy . The second examined FOCUS, a dyadic intervention for patients with advanced cancer and their family caregivers that addresses themes of f amily engagement, o ptimistic attitude, effective c oping strategies, dealing with u ncertainty, and s ymptom management . There were no effects of the intervention on primary outcomes including illness uncertainty and quality of life compared to usual care . The third study examined the effect of a Cognitive Behavioral Therapy‐Acceptance and Commitment Therapy (CBT‐ACT) intervention targeting sleep difficulties, worry, depression, and fatigue, with intolerance of uncertainty as a secondary outcome that was not significantly reduced . Thus, to our knowledge, there are no reports of interventions that comprehensively and effectively address the unique challenges associated with the experience of uncertainty for individuals living with AMI cancers. Adapting current interventions that have been successful at addressing uncertainty in early‐stage cancer survivors to the reality of AMI cancers may be a fruitful venue . The Fear Of Recurrence Therapy (FORT) [ , , ], that was developed to address fear of cancer recurrence (FCR) in cancer survivors (stages I‐III) may be particularly suited to adaptation in the AMI population. In a randomized controlled trial of early‐stage cancer survivors, FORT demonstrated its efficacy at reducing FCR and other secondary outcomes (e.g., intolerance of uncertainty, uncertainty in illness, avoidance) . FORT is based on a blended theoretical model of FCR aiming to target key vulnerability factors such as internal and external triggers, exaggerated perceived risk of recurrence, hyper‐focus on ambiguous physical sensations, maladaptive coping, uncertainty around cancer and its treatments or care, intolerance of uncertainty, and beliefs about the benefits of worrying about one's health. The development of FORT was guided by Leventhal's Common Sense Model , Mishel's Uncertainty in Illness Theory , and the cognitive model of worry . The intervention integrates cognitive and existential based techniques to specifically target FCR during six consecutive weekly sessions of 90–120 min in a group format led by two healthcare professionals trained in psychotherapy. To meet the clinical need of AMI patients presenting for psychological services within the Psychosocial Oncology Program (PSOP) at a regional cancer center, CH adapted FORT to those with AMI cancers. In doing so, the findings of systematic reviews suggesting that uncertainty is the overarching psychological concern of this patient population were key to determining content. The potential energy limitations of patients was considered in deciding the number of sessions, duration of each session, and the online format. A group format was selected given the benefits derived by patients through group process, and for reasons of cost‐effectiveness . The result is a novel intervention called Adapting to life with Cancer Cognitive ExisteNtial Therapy, or ACCENT, a manual‐based group psychotherapy aimed at addressing key psychological concerns of AMI cancer patients: intolerance of uncertainty, anxiety, and cancer‐specific distress. The purpose of this exploratory study is to improve the quality of this clinical service by assessing the feasibility, acceptability, and preliminary efficacy of ACCENT in AMI cancers using data collected as part of routine clinical practice. If preliminary results indicate that ACCENT is feasible, acceptable, and shows a clinical signal, we will conduct a pilot study to assess the potential clinical utility of this intervention in a more systematic way.
Methods 2.1 Design and Outcomes Using the data collected as part of routine patient care, we framed our evaluation within a pragmatic feasibility study design . Pragmatic feasibility studies focus on the potential benefits of an intervention offered in routine clinical practice. Primary outcomes were feasibility and acceptability of ACCENT in individuals living with AMI cancer. Secondary outcomes were estimates of preliminary efficacy for intolerance of uncertainty, anxiety, and cancer‐specific distress. 2.2 Participants As part of routine referral practices, patients living with heterogeneous AMI cancers were referred for potential participation in the ACCENT group by their treating social worker at The Ottawa Hospital Cancer Center Psychosocial Oncology Program in Ottawa, Canada. Participation required the patient to (a) have a diagnosis of AMI cancer; (b) express interest in participating in a group aimed at managing cancer‐related uncertainty; (c) be age 18 or older; (d) speak English; and (e) have access to internet in a private setting. Exclusion criteria were (a) self‐reported major psychological disorder (e.g., cognitive impairment, severe depression, personality issues): that would interfere with group participation. This study was deemed consistent with a quality improvement (QI) effort by the Ottawa Health Science Network Research Ethics Board (OHSN‐REB); thus, REB exemption was provided. For the purpose of outcome monitoring, patients were invited to complete questionnaires pre‐ and post‐intervention, but this was not a requirement of participation in the group. 2.3 Procedures Patients who met the inclusion requirements were referred from January 2022 to October 2023. Identified participants were contacted by the lead psychologist (CH) over the telephone for screening. Screening consisted of assessing commitment to and suitability of the 6‐week group therapy, openness to the emotional content and existential components, and comfort with the virtual group format. Patients were added to a waiting list and the intervention began as the psychologist's clinical time allowed. Membership was closed once the group was on‐going. Patients provided verbal consent and were invited by CH to complete questionnaires via Microsoft Forms prior to the start of the group and immediately post intervention. The intervention was delivered as part of routine psychological care via The Ottawa Hospital Epic Zoom medical software application by the lead psychologist alone, or with a trainee (i.e., psychology resident or psychiatry resident). Participant attendance was documented, and individual make‐up sessions were offered by the lead psychologist or her trainee to participants who missed a group session. 2.4 Intervention Adapting to life with Cancer Cognitive ExisteNtial Therapy (ACCENT) is a virtual group psychotherapy delivered by up to two group facilitators that consists of 6 consecutive weekly sessions of 90 min and between session assignments (see Table for a description of each session). The focus of ACCENT is to address uncertainty, anxiety, and distress in AMI cancers. Like FORT, the ACCENT intervention is based on Cognitive Behavioral Therapy (CBT) and Existential Psychotherapy but was adapted to include components of Acceptance and Commitment Therapy (ACT) as ACT has demonstrated efficacy in AMI cancer patients . Table includes a description of the 6 sessions of ACCENT with CBT interventions (e.g., psychoeducation, worry management, cognitive restructuring, coping statements, exposure, relaxation), ACT interventions (e.g., values‐based living, mindfulness), and existential interventions (e.g., exploring and processing fears related to death and dying, prioritizing activities that are perceived as meaningful). Before starting the group, participants are sent an electronic patient workbook describing session activities and assignments. 2.5 Outcomes Measures were administered as part of standard of care to evaluate the project's primary and secondary outcomes. All measures were administered in English. 2.5.1 Primary Outcomes Feasibility was assessed by considering the number of referrals and the refusal rate. Acceptability was evaluated by calculating the attrition rate, adherence rate (attendance), and reported satisfaction (quantitative and qualitative) and perceptions of improvements and usefulness. 2.5.1.1 Satisfaction With Therapy and Therapist Scale‐Revised (STTS‐R) The STTS‐R was used to assess participants' satisfaction with ACCENT and the therapist(s) post‐intervention. The STTS‐R is comprised of 12 items rated on a 5‐point Likert Scale from 1 (“Strongly disagree”) to 5 (“Strongly agree”). The STTS‐R has strong psychometric properties . 2.5.1.2 Open‐Ended Questions About ACCENT Post‐intervention, participants were asked to write about which program components they liked best and what they would change to make the group better. 2.5.1.3 Post‐Intervention Perceptions of Improvement and Usefulness Post‐intervention, participants used a 5‐point scale to rate the impact of the group on their ability to manage uncertainty (made things a lot worse, made things somewhat worse, made no difference, made things somewhat better, made things a lot better) and a 5‐point scale to rate their plan to continue using the strategies introduced in the group (definitely not, I don't think so, not sure, I think so, definitely). 2.5.2 Secondary Outcomes 2.5.2.1 Intolerance of Uncertainty Scale—Short Form (IUS‐12) The IUS‐12 was used to assess intolerance of uncertainty by measuring reactions to uncertainty, ambiguous situations, and the future. It is comprised of two factors, prospective anxiety (seven items) and inhibitory anxiety (five items). All items are rated on a 5‐point Likert scale from 1 (“Not at all characteristic of me”) to 5 (“Entirely characteristic of me”). The IUS‐12 has good psychometric properties and has been used in the cancer population . 2.5.2.2 Generalized Anxiety Disorder‐7 (GAD‐7) The GAD‐7 was used to assess generalized anxiety and frequency of symptoms. It is comprised of seven items rated on a 4‐point Likert scale from 0 (“Not at all”) to 3 (“Nearly every day”). The GAD‐7 has strong psychometric properties including in individuals with cancer . 2.5.2.3 Impact of Events Scale (IES) The IES was used to assess subjective distress associated with cancer. The IES is comprised of 15 items rated on a 5‐point Likert Scale from 0 (“Not at all”) to 5 (“Often”) which reflect intrusive thoughts (seven items) and avoidance symptoms (eight items). The IES has strong psychometric properties and is frequently used in patients with cancer . 2.6 Data Analysis Qualitative data was summarized by co‐author LG rather than undergoing content analysis due to limited responses. Frequent comments were summarized as one while more specific, constructive and actionable feedback was directly reported. Paired samples two‐tailed ( p < 0.05) t ‐tests were used to compare pre‐ and post‐scores using SPSS. One t ‐test per outcome was performed using participants' total scores on each measure. The assumptions for normality and outliers were met: The data were relatively normal as assessed by the Shapiro‐Wilks test for normality and paired t ‐tests are robust to the small deviations identified. Outliers were too few to remove and not significant. Due to the exploratory nature of this study, only complete pre‐ and post‐intervention data were used. Analyses explored the clinical signal of the intervention to determine if future, more rigorous, studies of ACCENT are justified.
Design and Outcomes Using the data collected as part of routine patient care, we framed our evaluation within a pragmatic feasibility study design . Pragmatic feasibility studies focus on the potential benefits of an intervention offered in routine clinical practice. Primary outcomes were feasibility and acceptability of ACCENT in individuals living with AMI cancer. Secondary outcomes were estimates of preliminary efficacy for intolerance of uncertainty, anxiety, and cancer‐specific distress.
Participants As part of routine referral practices, patients living with heterogeneous AMI cancers were referred for potential participation in the ACCENT group by their treating social worker at The Ottawa Hospital Cancer Center Psychosocial Oncology Program in Ottawa, Canada. Participation required the patient to (a) have a diagnosis of AMI cancer; (b) express interest in participating in a group aimed at managing cancer‐related uncertainty; (c) be age 18 or older; (d) speak English; and (e) have access to internet in a private setting. Exclusion criteria were (a) self‐reported major psychological disorder (e.g., cognitive impairment, severe depression, personality issues): that would interfere with group participation. This study was deemed consistent with a quality improvement (QI) effort by the Ottawa Health Science Network Research Ethics Board (OHSN‐REB); thus, REB exemption was provided. For the purpose of outcome monitoring, patients were invited to complete questionnaires pre‐ and post‐intervention, but this was not a requirement of participation in the group.
Procedures Patients who met the inclusion requirements were referred from January 2022 to October 2023. Identified participants were contacted by the lead psychologist (CH) over the telephone for screening. Screening consisted of assessing commitment to and suitability of the 6‐week group therapy, openness to the emotional content and existential components, and comfort with the virtual group format. Patients were added to a waiting list and the intervention began as the psychologist's clinical time allowed. Membership was closed once the group was on‐going. Patients provided verbal consent and were invited by CH to complete questionnaires via Microsoft Forms prior to the start of the group and immediately post intervention. The intervention was delivered as part of routine psychological care via The Ottawa Hospital Epic Zoom medical software application by the lead psychologist alone, or with a trainee (i.e., psychology resident or psychiatry resident). Participant attendance was documented, and individual make‐up sessions were offered by the lead psychologist or her trainee to participants who missed a group session.
Intervention Adapting to life with Cancer Cognitive ExisteNtial Therapy (ACCENT) is a virtual group psychotherapy delivered by up to two group facilitators that consists of 6 consecutive weekly sessions of 90 min and between session assignments (see Table for a description of each session). The focus of ACCENT is to address uncertainty, anxiety, and distress in AMI cancers. Like FORT, the ACCENT intervention is based on Cognitive Behavioral Therapy (CBT) and Existential Psychotherapy but was adapted to include components of Acceptance and Commitment Therapy (ACT) as ACT has demonstrated efficacy in AMI cancer patients . Table includes a description of the 6 sessions of ACCENT with CBT interventions (e.g., psychoeducation, worry management, cognitive restructuring, coping statements, exposure, relaxation), ACT interventions (e.g., values‐based living, mindfulness), and existential interventions (e.g., exploring and processing fears related to death and dying, prioritizing activities that are perceived as meaningful). Before starting the group, participants are sent an electronic patient workbook describing session activities and assignments.
Outcomes Measures were administered as part of standard of care to evaluate the project's primary and secondary outcomes. All measures were administered in English. 2.5.1 Primary Outcomes Feasibility was assessed by considering the number of referrals and the refusal rate. Acceptability was evaluated by calculating the attrition rate, adherence rate (attendance), and reported satisfaction (quantitative and qualitative) and perceptions of improvements and usefulness. 2.5.1.1 Satisfaction With Therapy and Therapist Scale‐Revised (STTS‐R) The STTS‐R was used to assess participants' satisfaction with ACCENT and the therapist(s) post‐intervention. The STTS‐R is comprised of 12 items rated on a 5‐point Likert Scale from 1 (“Strongly disagree”) to 5 (“Strongly agree”). The STTS‐R has strong psychometric properties . 2.5.1.2 Open‐Ended Questions About ACCENT Post‐intervention, participants were asked to write about which program components they liked best and what they would change to make the group better. 2.5.1.3 Post‐Intervention Perceptions of Improvement and Usefulness Post‐intervention, participants used a 5‐point scale to rate the impact of the group on their ability to manage uncertainty (made things a lot worse, made things somewhat worse, made no difference, made things somewhat better, made things a lot better) and a 5‐point scale to rate their plan to continue using the strategies introduced in the group (definitely not, I don't think so, not sure, I think so, definitely). 2.5.2 Secondary Outcomes 2.5.2.1 Intolerance of Uncertainty Scale—Short Form (IUS‐12) The IUS‐12 was used to assess intolerance of uncertainty by measuring reactions to uncertainty, ambiguous situations, and the future. It is comprised of two factors, prospective anxiety (seven items) and inhibitory anxiety (five items). All items are rated on a 5‐point Likert scale from 1 (“Not at all characteristic of me”) to 5 (“Entirely characteristic of me”). The IUS‐12 has good psychometric properties and has been used in the cancer population . 2.5.2.2 Generalized Anxiety Disorder‐7 (GAD‐7) The GAD‐7 was used to assess generalized anxiety and frequency of symptoms. It is comprised of seven items rated on a 4‐point Likert scale from 0 (“Not at all”) to 3 (“Nearly every day”). The GAD‐7 has strong psychometric properties including in individuals with cancer . 2.5.2.3 Impact of Events Scale (IES) The IES was used to assess subjective distress associated with cancer. The IES is comprised of 15 items rated on a 5‐point Likert Scale from 0 (“Not at all”) to 5 (“Often”) which reflect intrusive thoughts (seven items) and avoidance symptoms (eight items). The IES has strong psychometric properties and is frequently used in patients with cancer .
Primary Outcomes Feasibility was assessed by considering the number of referrals and the refusal rate. Acceptability was evaluated by calculating the attrition rate, adherence rate (attendance), and reported satisfaction (quantitative and qualitative) and perceptions of improvements and usefulness. 2.5.1.1 Satisfaction With Therapy and Therapist Scale‐Revised (STTS‐R) The STTS‐R was used to assess participants' satisfaction with ACCENT and the therapist(s) post‐intervention. The STTS‐R is comprised of 12 items rated on a 5‐point Likert Scale from 1 (“Strongly disagree”) to 5 (“Strongly agree”). The STTS‐R has strong psychometric properties . 2.5.1.2 Open‐Ended Questions About ACCENT Post‐intervention, participants were asked to write about which program components they liked best and what they would change to make the group better. 2.5.1.3 Post‐Intervention Perceptions of Improvement and Usefulness Post‐intervention, participants used a 5‐point scale to rate the impact of the group on their ability to manage uncertainty (made things a lot worse, made things somewhat worse, made no difference, made things somewhat better, made things a lot better) and a 5‐point scale to rate their plan to continue using the strategies introduced in the group (definitely not, I don't think so, not sure, I think so, definitely).
Satisfaction With Therapy and Therapist Scale‐Revised (STTS‐R) The STTS‐R was used to assess participants' satisfaction with ACCENT and the therapist(s) post‐intervention. The STTS‐R is comprised of 12 items rated on a 5‐point Likert Scale from 1 (“Strongly disagree”) to 5 (“Strongly agree”). The STTS‐R has strong psychometric properties .
Open‐Ended Questions About ACCENT Post‐intervention, participants were asked to write about which program components they liked best and what they would change to make the group better.
Post‐Intervention Perceptions of Improvement and Usefulness Post‐intervention, participants used a 5‐point scale to rate the impact of the group on their ability to manage uncertainty (made things a lot worse, made things somewhat worse, made no difference, made things somewhat better, made things a lot better) and a 5‐point scale to rate their plan to continue using the strategies introduced in the group (definitely not, I don't think so, not sure, I think so, definitely).
Secondary Outcomes 2.5.2.1 Intolerance of Uncertainty Scale—Short Form (IUS‐12) The IUS‐12 was used to assess intolerance of uncertainty by measuring reactions to uncertainty, ambiguous situations, and the future. It is comprised of two factors, prospective anxiety (seven items) and inhibitory anxiety (five items). All items are rated on a 5‐point Likert scale from 1 (“Not at all characteristic of me”) to 5 (“Entirely characteristic of me”). The IUS‐12 has good psychometric properties and has been used in the cancer population . 2.5.2.2 Generalized Anxiety Disorder‐7 (GAD‐7) The GAD‐7 was used to assess generalized anxiety and frequency of symptoms. It is comprised of seven items rated on a 4‐point Likert scale from 0 (“Not at all”) to 3 (“Nearly every day”). The GAD‐7 has strong psychometric properties including in individuals with cancer . 2.5.2.3 Impact of Events Scale (IES) The IES was used to assess subjective distress associated with cancer. The IES is comprised of 15 items rated on a 5‐point Likert Scale from 0 (“Not at all”) to 5 (“Often”) which reflect intrusive thoughts (seven items) and avoidance symptoms (eight items). The IES has strong psychometric properties and is frequently used in patients with cancer .
Intolerance of Uncertainty Scale—Short Form (IUS‐12) The IUS‐12 was used to assess intolerance of uncertainty by measuring reactions to uncertainty, ambiguous situations, and the future. It is comprised of two factors, prospective anxiety (seven items) and inhibitory anxiety (five items). All items are rated on a 5‐point Likert scale from 1 (“Not at all characteristic of me”) to 5 (“Entirely characteristic of me”). The IUS‐12 has good psychometric properties and has been used in the cancer population .
Generalized Anxiety Disorder‐7 (GAD‐7) The GAD‐7 was used to assess generalized anxiety and frequency of symptoms. It is comprised of seven items rated on a 4‐point Likert scale from 0 (“Not at all”) to 3 (“Nearly every day”). The GAD‐7 has strong psychometric properties including in individuals with cancer .
Impact of Events Scale (IES) The IES was used to assess subjective distress associated with cancer. The IES is comprised of 15 items rated on a 5‐point Likert Scale from 0 (“Not at all”) to 5 (“Often”) which reflect intrusive thoughts (seven items) and avoidance symptoms (eight items). The IES has strong psychometric properties and is frequently used in patients with cancer .
Data Analysis Qualitative data was summarized by co‐author LG rather than undergoing content analysis due to limited responses. Frequent comments were summarized as one while more specific, constructive and actionable feedback was directly reported. Paired samples two‐tailed ( p < 0.05) t ‐tests were used to compare pre‐ and post‐scores using SPSS. One t ‐test per outcome was performed using participants' total scores on each measure. The assumptions for normality and outliers were met: The data were relatively normal as assessed by the Shapiro‐Wilks test for normality and paired t ‐tests are robust to the small deviations identified. Outliers were too few to remove and not significant. Due to the exploratory nature of this study, only complete pre‐ and post‐intervention data were used. Analyses explored the clinical signal of the intervention to determine if future, more rigorous, studies of ACCENT are justified.
Results 3.1 Participants Five groups (six to seven participants in each) were completed from January 2022 to November 2023. In total, 32 patients were interested in participating in ACCENT. Eligible participants were included in the groups on a first come first served basis. Of the 28 who took part in the group (4 dropped out), 25 completed both pre‐ and post‐intervention assessments. Due to clerical errors, the IES was not included in the post‐intervention assessment of the first group; therefore, the data for this measure counts five more missing entries than that of the GAD‐7 or IUS‐12. Twenty‐four women and one man provided data. Breast cancer ( n = 8) and colorectal cancer ( n = 5) were the most common primary cancer types. All participants had AMI cancer. The sociodemographic and medical characteristics of the sample are presented in Table . 3.2 Feasibility and Acceptability All 32 participants who were referred for the group were contacted, and all agreed to participate in ACCENT. One participant dropped out after completing pre‐intervention questionnaires but before starting the group due to scheduling conflicts. Their data were removed from the analyses. In addition, one participant did not provide data. Three participants dropped out during the intervention after attending two sessions or less. The reasons for these dropouts were: unexpected rapid health decline ( n = 2) and too difficult to hear stories of others ( n = 1). Thus, the attrition rate was 4/32 = 12.5%. Of the 28 patients who took part in the intervention, 17 attended all six group sessions (60.7%), 8 attended five group sessions (28.6%), 2 attended four group sessions (7.1%), and 1 attended three group sessions (3.6%). Seven participants attended at least one make‐up session offered on a one‐to‐one basis. The make‐up sessions were not included in the completion statistics listed above. Reasons for missing sessions included health concerns (i.e., fever, treatment side effects, pain) and other medical appointments conflicting with the group time. Figure shows attendance and drop‐out rates. 3.2.1 Self‐Reported Satisfaction Of 30 participants, 25 completed pre‐ and post‐questionnaires. Overall satisfaction with ACCENT and the therapist(s) was high as indicated by subscale scores (satisfaction with therapy mean = 27.9/30, range = 23–30; satisfaction with therapist(s) mean = 28.8/30, range = 24–30) and by the overall mean of 56.6/60 on the STTS‐R (range = 48–60). Twenty‐five ( n = 25) participants reported qualitatively on what they liked best and what they would change about ACCENT. Participants appreciated being able to share their worries with a group of individuals with similar experiences ( n = 20). The size of the group was also acceptable ( n = 1). The atmosphere of the group was reportedly safe and respectful ( n = 8): I was able to express myself in a safe place with people who are going through the same experience. Participants appreciated learning new strategies to live with uncertainty ( n = 9). Some participants shared their appreciation for the facilitators' empathy and their expertise ( n = 8). One participant appreciated the online format: Strangely enough, the zoom format was conducive to therapy ‐ regardless of energy level, could assist; allows for private moments when discussion is too intense. Generally, they felt heard, understood, supported, and experienced a sense of community: Hearing other people's concerns and stories; not feeling alone; being able to share and know that what is shared is understood and felt by others in the group. The suggested changes included increasing the number of sessions for the intervention (e.g., from 6 to 10; n = 7) and increasing the length of each session (e.g., 1.5–2 h; n = 1). Some participants asked to add a session or two that included caregivers ( n = 2), but another wanted the group to be offered only to patients while still including a session on how to communicate with caregivers ( n = 1). There were also some participants who wished to separate the groups according to age or cancer type to facilitate more relatable discussions ( n = 2). Maintain criteria of similar cancer condition for group participants (ex. women with breast cancer and metastasis). Maintain focus on cancer patients (keep caregivers therapy separate), but allow for discussion about relations with close caregivers/family members. Finally, some participants asked to have a mix of in‐person and online sessions ( n = 3). Other comments were asking for more explanations on the worst‐case scenario exercise and its utility ( n = 2), and having more roleplay exercises ( n = 1): Of all the strategies, the facing your worst fear scenario is the one which I didn't feel helpful. Maybe more explanation would have been good. When asked if the intervention helped them manage uncertainty, all participants ( n = 25) answered that ACCENT “made things a lot better” or “made things somewhat better.” In addition, all ( n = 25) reported that they “think” they will or “definitely” will use the strategies learned during the intervention. 3.3 Preliminary Efficacy Out of 30 participants, a total of 25 complete responses for assessments of anxiety and intolerance of uncertainty and 20 complete responses for cancer‐specific distress were provided. T ‐tests revealed a non‐significant decrease in intolerance of uncertainty ( n = 25, t (24) = 1.77, p = 0.089), and a statistically significant decrease in anxiety ( n = 25, t (24) = 3.61, p = 0.001), and cancer‐specific distress ( n = 20, t (19) = 2.23, p = 0.038). Effect sizes were small for intolerance of uncertainty (Cohen's d = 0.36), and moderate for anxiety (Cohen's d = 0.72) and cancer‐specific distress (Cohen's d = 0.50) (Table ).
Participants Five groups (six to seven participants in each) were completed from January 2022 to November 2023. In total, 32 patients were interested in participating in ACCENT. Eligible participants were included in the groups on a first come first served basis. Of the 28 who took part in the group (4 dropped out), 25 completed both pre‐ and post‐intervention assessments. Due to clerical errors, the IES was not included in the post‐intervention assessment of the first group; therefore, the data for this measure counts five more missing entries than that of the GAD‐7 or IUS‐12. Twenty‐four women and one man provided data. Breast cancer ( n = 8) and colorectal cancer ( n = 5) were the most common primary cancer types. All participants had AMI cancer. The sociodemographic and medical characteristics of the sample are presented in Table .
Feasibility and Acceptability All 32 participants who were referred for the group were contacted, and all agreed to participate in ACCENT. One participant dropped out after completing pre‐intervention questionnaires but before starting the group due to scheduling conflicts. Their data were removed from the analyses. In addition, one participant did not provide data. Three participants dropped out during the intervention after attending two sessions or less. The reasons for these dropouts were: unexpected rapid health decline ( n = 2) and too difficult to hear stories of others ( n = 1). Thus, the attrition rate was 4/32 = 12.5%. Of the 28 patients who took part in the intervention, 17 attended all six group sessions (60.7%), 8 attended five group sessions (28.6%), 2 attended four group sessions (7.1%), and 1 attended three group sessions (3.6%). Seven participants attended at least one make‐up session offered on a one‐to‐one basis. The make‐up sessions were not included in the completion statistics listed above. Reasons for missing sessions included health concerns (i.e., fever, treatment side effects, pain) and other medical appointments conflicting with the group time. Figure shows attendance and drop‐out rates. 3.2.1 Self‐Reported Satisfaction Of 30 participants, 25 completed pre‐ and post‐questionnaires. Overall satisfaction with ACCENT and the therapist(s) was high as indicated by subscale scores (satisfaction with therapy mean = 27.9/30, range = 23–30; satisfaction with therapist(s) mean = 28.8/30, range = 24–30) and by the overall mean of 56.6/60 on the STTS‐R (range = 48–60). Twenty‐five ( n = 25) participants reported qualitatively on what they liked best and what they would change about ACCENT. Participants appreciated being able to share their worries with a group of individuals with similar experiences ( n = 20). The size of the group was also acceptable ( n = 1). The atmosphere of the group was reportedly safe and respectful ( n = 8): I was able to express myself in a safe place with people who are going through the same experience. Participants appreciated learning new strategies to live with uncertainty ( n = 9). Some participants shared their appreciation for the facilitators' empathy and their expertise ( n = 8). One participant appreciated the online format: Strangely enough, the zoom format was conducive to therapy ‐ regardless of energy level, could assist; allows for private moments when discussion is too intense. Generally, they felt heard, understood, supported, and experienced a sense of community: Hearing other people's concerns and stories; not feeling alone; being able to share and know that what is shared is understood and felt by others in the group. The suggested changes included increasing the number of sessions for the intervention (e.g., from 6 to 10; n = 7) and increasing the length of each session (e.g., 1.5–2 h; n = 1). Some participants asked to add a session or two that included caregivers ( n = 2), but another wanted the group to be offered only to patients while still including a session on how to communicate with caregivers ( n = 1). There were also some participants who wished to separate the groups according to age or cancer type to facilitate more relatable discussions ( n = 2). Maintain criteria of similar cancer condition for group participants (ex. women with breast cancer and metastasis). Maintain focus on cancer patients (keep caregivers therapy separate), but allow for discussion about relations with close caregivers/family members. Finally, some participants asked to have a mix of in‐person and online sessions ( n = 3). Other comments were asking for more explanations on the worst‐case scenario exercise and its utility ( n = 2), and having more roleplay exercises ( n = 1): Of all the strategies, the facing your worst fear scenario is the one which I didn't feel helpful. Maybe more explanation would have been good. When asked if the intervention helped them manage uncertainty, all participants ( n = 25) answered that ACCENT “made things a lot better” or “made things somewhat better.” In addition, all ( n = 25) reported that they “think” they will or “definitely” will use the strategies learned during the intervention.
Self‐Reported Satisfaction Of 30 participants, 25 completed pre‐ and post‐questionnaires. Overall satisfaction with ACCENT and the therapist(s) was high as indicated by subscale scores (satisfaction with therapy mean = 27.9/30, range = 23–30; satisfaction with therapist(s) mean = 28.8/30, range = 24–30) and by the overall mean of 56.6/60 on the STTS‐R (range = 48–60). Twenty‐five ( n = 25) participants reported qualitatively on what they liked best and what they would change about ACCENT. Participants appreciated being able to share their worries with a group of individuals with similar experiences ( n = 20). The size of the group was also acceptable ( n = 1). The atmosphere of the group was reportedly safe and respectful ( n = 8): I was able to express myself in a safe place with people who are going through the same experience. Participants appreciated learning new strategies to live with uncertainty ( n = 9). Some participants shared their appreciation for the facilitators' empathy and their expertise ( n = 8). One participant appreciated the online format: Strangely enough, the zoom format was conducive to therapy ‐ regardless of energy level, could assist; allows for private moments when discussion is too intense. Generally, they felt heard, understood, supported, and experienced a sense of community: Hearing other people's concerns and stories; not feeling alone; being able to share and know that what is shared is understood and felt by others in the group. The suggested changes included increasing the number of sessions for the intervention (e.g., from 6 to 10; n = 7) and increasing the length of each session (e.g., 1.5–2 h; n = 1). Some participants asked to add a session or two that included caregivers ( n = 2), but another wanted the group to be offered only to patients while still including a session on how to communicate with caregivers ( n = 1). There were also some participants who wished to separate the groups according to age or cancer type to facilitate more relatable discussions ( n = 2). Maintain criteria of similar cancer condition for group participants (ex. women with breast cancer and metastasis). Maintain focus on cancer patients (keep caregivers therapy separate), but allow for discussion about relations with close caregivers/family members. Finally, some participants asked to have a mix of in‐person and online sessions ( n = 3). Other comments were asking for more explanations on the worst‐case scenario exercise and its utility ( n = 2), and having more roleplay exercises ( n = 1): Of all the strategies, the facing your worst fear scenario is the one which I didn't feel helpful. Maybe more explanation would have been good. When asked if the intervention helped them manage uncertainty, all participants ( n = 25) answered that ACCENT “made things a lot better” or “made things somewhat better.” In addition, all ( n = 25) reported that they “think” they will or “definitely” will use the strategies learned during the intervention.
Preliminary Efficacy Out of 30 participants, a total of 25 complete responses for assessments of anxiety and intolerance of uncertainty and 20 complete responses for cancer‐specific distress were provided. T ‐tests revealed a non‐significant decrease in intolerance of uncertainty ( n = 25, t (24) = 1.77, p = 0.089), and a statistically significant decrease in anxiety ( n = 25, t (24) = 3.61, p = 0.001), and cancer‐specific distress ( n = 20, t (19) = 2.23, p = 0.038). Effect sizes were small for intolerance of uncertainty (Cohen's d = 0.36), and moderate for anxiety (Cohen's d = 0.72) and cancer‐specific distress (Cohen's d = 0.50) (Table ).
Discussion This study reports on the feasibility, acceptability, and preliminary efficacy of ACCENT, a manual‐based psychological intervention designed to help AMI cancer patients live well with uncertainty. To our knowledge, there are no reports of interventions that comprehensively and effectively address the unique challenges associated with the experience of uncertainty for individuals living with AMI cancers. Strengths of ACCENT are that it was adapted from evidence‐based, theoretically driven interventions, based on the identified psychological needs of this population, developed in keeping with the potential energy limitations of those with AMI cancers, and offered in a group format. Specifically, ACCENT adapted FORT, a cognitive existential intervention with demonstrated efficacy in reducing fear of recurrence and uncertainty in early‐stage cancer survivors to focus on the broader construct of uncertainty, which is a key source of distress in patients with AMI cancers . ACCENT also incorporates elements of third‐wave CBT interventions (i.e., ACT) that promote value‐based living. This resulted in a brief, 6‐week manual‐based psychological intervention grounded in theory and explicating the group processes and exercises that clinicians can use to address uncertainty and associated distress in a virtual format (see Table ). Findings demonstrated evidence of feasibility, acceptability, and preliminary clinical efficacy. We were able to offer ACCENT to 31 participants in 22 months from a single recruitment site. Anecdotally, the limiting factor to number of groups provided was the psychologist's clinical time rather than patient demand for the service. All individuals who were offered the intervention agreed to participate. ACCENT appears acceptable as few participants (12.5%) dropped out. Of note, one of the dropouts occurred prior to the group beginning due to a scheduling conflict and half of those who dropped out were unable to continue because their health deteriorated rapidly. Rapid health decline may be expected with this population and will be considered for recruitment efforts in future ACCENT efficacy studies. Dropouts tended to occur early in the intervention, after the first or second session, which is typical of patients seeking mental health services . The attendance rate was high with 89% of participants attending five or all of the group sessions. The reasons for missed sessions most commonly included being unwell or attending another medical appointment. Patients were offered individual make‐up sessions and attended them. This suggests that clinicians will need to extend this flexibility when working with the AMI population. Satisfaction ratings were high and qualitative data suggested that the group format and content were valued. Participants expressed feelings of cohesiveness and shared experience. In terms of preliminary efficacy, ACCENT may be successful in increasing the ability to live with the uncertainty inherent in AMI cancers. The non‐significant decrease in intolerance to uncertainty was unexpected. This may have been due to lower base rates and a non‐clinical baseline level of uncertainty at the start of the intervention. Because of the potential mediating role of intolerance of uncertainty between uncertainty in illness and fear of disease progression, it may have been useful to assess for uncertainty in illness in addition to intolerance of uncertainty . Relatedly, uncertainty in illness being an inherent component of AMI cancers, this may have influenced this result . This should be further investigated in future studies and more systematically measured. The intervention may also reduce anxiety and cancer‐specific distress. The preliminary effect sizes of the observed changes were, for the most part, in the medium range and consistent with interventions for uncertainty, fear of cancer recurrence, and anxiety in early‐stage cancer patients [ , , ].
Conclusion This preliminary evaluation of ACCENT was anchored in a pragmatic approach that favors external generalizability over internal validity. There are advantages to this approach, the main one being that ACCENT was evaluated in the context in which it was delivered as a clinical service, which may facilitate future implementation efforts. Another advantage is that ACCENT was offered to groups of patients with mixed primary tumor sites. This will facilitate its future testing and implementation in settings where offering cancer site specific groups is not feasible or desirable. ACCENT was delivered in an online, group format which potentially increases access to the intervention and cost‐effectiveness. 5.1 Limitations There are also disadvantages to the pragmatic approach: ACCENT was delivered at a single site without the strict methods that are typical of a clinical trial. Lack of a control group and randomization precludes us from determining that the intervention is responsible for the changes observed amongst participants. In addition, inclusion criteria did not involve meeting minimum symptoms scores on self‐report measures which may have contributed to the statistically insignificant finding on the Intolerance of Uncertainty scale. We have no information about compliance of homework or rate of practice of between session exercises. Data were collected by the study psychologist which introduces the likelihood of demand effects. Participation of a single man limits generalizability of the findings, although this is not unique to our study and likely a reflection of gender differences in seeking psychological support Additionally, absence of data on the degree to which ACCENT was delivered as intended and on the ease of using the intervention from the therapists involved could be a limitation. Finally, using a complete‐cases analysis may have introduced bias to the observed changes in anxiety, intolerance of uncertainty and cancer‐specific distress levels. Results from this pragmatic feasibility study should therefore be interpreted with some caution. 5.2 Future Directions A future pilot study is needed to thoroughly examine feasibility and acceptability outcomes with greater methodological rigor to address the limitations noted above. This would include the feasibility of randomizing people living with AMI cancers to a control group. Additional psychological outcomes that have been reported in this population such as fear of disease progression, uncertainty in illness, and death anxiety could be included as potential outcome variables. Possible modifications include increasing the number of sessions based on participant feedback and having additional sessions with caregivers, for which validated dyadic interventions may be relevant . 5.3 Implications for Individuals With Cancer This pragmatic feasibility study suggests that ACCENT, a 6‐week, manual‐based group psychotherapy to address uncertainty, shows promise for the growing population of individuals living longer with AMI cancers. Future studies are needed to more rigorously test ACCENT.
Limitations There are also disadvantages to the pragmatic approach: ACCENT was delivered at a single site without the strict methods that are typical of a clinical trial. Lack of a control group and randomization precludes us from determining that the intervention is responsible for the changes observed amongst participants. In addition, inclusion criteria did not involve meeting minimum symptoms scores on self‐report measures which may have contributed to the statistically insignificant finding on the Intolerance of Uncertainty scale. We have no information about compliance of homework or rate of practice of between session exercises. Data were collected by the study psychologist which introduces the likelihood of demand effects. Participation of a single man limits generalizability of the findings, although this is not unique to our study and likely a reflection of gender differences in seeking psychological support Additionally, absence of data on the degree to which ACCENT was delivered as intended and on the ease of using the intervention from the therapists involved could be a limitation. Finally, using a complete‐cases analysis may have introduced bias to the observed changes in anxiety, intolerance of uncertainty and cancer‐specific distress levels. Results from this pragmatic feasibility study should therefore be interpreted with some caution.
Future Directions A future pilot study is needed to thoroughly examine feasibility and acceptability outcomes with greater methodological rigor to address the limitations noted above. This would include the feasibility of randomizing people living with AMI cancers to a control group. Additional psychological outcomes that have been reported in this population such as fear of disease progression, uncertainty in illness, and death anxiety could be included as potential outcome variables. Possible modifications include increasing the number of sessions based on participant feedback and having additional sessions with caregivers, for which validated dyadic interventions may be relevant .
Implications for Individuals With Cancer This pragmatic feasibility study suggests that ACCENT, a 6‐week, manual‐based group psychotherapy to address uncertainty, shows promise for the growing population of individuals living longer with AMI cancers. Future studies are needed to more rigorously test ACCENT.
The authors declare no conflicts of interest.
|
Revised ISHAM-ABPA working group clinical practice guidelines for diagnosing, classifying and treating allergic bronchopulmonary aspergillosis/mycoses | fb982c29-d174-4191-9385-f9eb4fdee786 | 10991853 | Microbiology[mh] | Allergic bronchopulmonary mycoses are complex pulmonary disorders caused by immune reactions mounted against fungi, most often Aspergillus fumigatus , which colonise the airways of patients with chronic lung disease, most commonly asthma or cystic fibrosis (CF) . Allergic bronchopulmonary aspergillosis (ABPA) may occasionally occur in the absence of any predisposing condition and other chronic lung conditions, including bronchiectasis and COPD . Conventionally, the term ABPA is used when the causative pathogen is A. fumigatus . In contrast, allergic bronchopulmonary mycosis (ABPM) is an ABPA-like syndrome caused by fungi other than A. fumigatus . Among the allergens involved in asthma, no other allergen generates as much interest as A. fumigatus because the fungus is growing in the airways. Also, ABPA responds exceptionally well to a specific form of therapy. Accordingly, ABPA is considered an asthma endotype and a treatable trait in CF and non-CF bronchiectasis . Early identification and treatment of ABPA is crucial to prevent the progression of bronchiectasis. The diagnostic criteria proposed by the International Society for Human and Animal Mycology (ISHAM)-ABPA working group (AWG) in 2013 are widely used for diagnosing ABPA . Since the inception of these guidelines a decade ago, newer evidence has emerged concerning diagnostic test performance for ABPA. For instance, the skin test was found inferior to serum A. fumigatus -specific IgE assay , serum A. fumigatus -specific IgG detection by enzyme immunoassay proved superior to immunoprecipitation , a lateral flow assay is now available for A. fumigatus -IgG and the minimal diagnostic level of 500 IU·mL −1 is more sensitive than 1000 IU·mL −1 for serum total IgE . Also, several randomised controlled trials (RCTs) in ABPA therapy have been published in the last decade . In light of the above evidence, several international groups have proposed modifications to the ISHAM-AWG criteria . The diagnosis of ABPA in CF and non-CF bronchiectasis is even more challenging as manifestations such as bronchiectasis, Aspergillus bronchitis and mucus plugging are seen in these entities independently of ABPA . Given the emergence of novel yet occasionally conflicting findings and the lack of evidence in certain areas, new guidelines are needed to assist clinicians and researchers in managing ABPA. With this end in view, an expert group was constituted to develop a statement on diagnosing and treating allergic bronchopulmonary mycoses for clinical practice and research in adults and children.
As a first step, a core committee was formed with five authors of this statement (R. Agarwal, I.S. Sehgal, V. Muthu, D.W. Denning and A. Chakrabarti). Two authors (R. Agarwal and V. Muthu) performed a systematic literature review of the PubMed and Embase databases (to 15 March 2023) to support the guidelines statement and identify the current gaps in managing ABPA. The following search terms were used: “allergic bronchopulmonary aspergillosis” OR “abpa” OR “allergic bronchopulmonary mycosis” OR “fungal sensit*” OR “fungal allerg*” OR “mould allerg*” OR “mold allerg*” OR “mould sensit*” OR “mold sensit*” OR “fungal asthma” OR “aspergillus sensit*” OR “aspergillus hypersensitivity” OR aspergillosis, allergic bronchopulmonary [MeSH]. The core committee then framed the questions for the first round of the Delphi consensus and searched the Scopus researcher discovery database ( www.scopus.com/search/form.uri#researcher-discovery ) to identify participants for the Delphi expert consensus group (DECG). The DECG included specialists from adult and paediatric pulmonary medicine, infectious diseases, clinical mycology, and radiodiagnosis who were actively involved in the clinical or laboratory aspects of managing ABPA ( supplementary table S1 ). We followed a modified Delphi method . The experts were briefed about the objectives and methodology of the Delphi process. The questions were initially circulated to the experts by e-mail and additional (or modification of) questions were invited. The questions were modified after receiving opinions from the expert group. The first-round questionnaire contained topics spanning various domains (individual diagnostic tests, optimal cut-offs, diagnostic and classification criteria, and treatment options). The questionnaire was circulated online using the commercially available Delphi platform ( www.edelphi.org ) and anonymous responses were obtained from the participants. We refined and recirculated the questions for the second round. Reminders were sent by e-mail before concluding each Delphi round to ensure participation. We defined consensus as ≥70% of experts agreeing or disagreeing on a statement. The statements and questions that did not achieve consensus online were discussed in a hybrid meeting of all experts (face-to-face or virtual participation; 7 September 2023, Pune, India). The answers to the entire set of questions were also refined wherever required. The guidelines were formulated by the responses and comments received during the two online rounds and the subsequent in-person discussion. The draft was then circulated among the experts for further comments and suggestions. We used the terms “recommend” and “suggest” where the consensus was ≥70% and <70%, respectively. Finally, we provided the level of consensus (LoC) for important summary statements based on rounds 1 and 2 ( supplementary table S2 ). We used the LoC achieved during the final round for statements not achieving consensus in rounds 1 and 2.
The online surveys were conducted between 15 June and 15 August 2023. We sent invitations to 43 experts, of whom 39 participated. The 39 experts represented 13 countries across six continents ( supplementary table S1 ). Most experts had managed asthma with ABPA for at least 5 years and 51.3% also reported caring for patients with CF-ABPA. Adult (49%) and paediatric (5%) pulmonologists accounted for over half of the experts. The results of the Delphi process are presented in supplementary table S2 . Nomenclature of allergic bronchopulmonary mycoses We first deliberated on the nomenclature of ABPA and ABPM. The most common form of allergic airway mycoses is ABPA , while ABPM is far less common . Given the considerable overlap of the antigen repertoire of the Aspergillus species, the DECG recommended using the term ABPA when allergic mycoses are caused by any Aspergillus spp. (not A. fumigatus only) and ABPM when attributable to fungi other than Aspergillus spp. The most common fungi responsible for ABPM include Bipolaris spp., Schizophyllum commune and Curvularia spp. . Candida albicans has been implicated in several cases of ABPM ; however, its pathogenicity remains uncertain. Diagnosis of fungal sensitisation As sensitisation represents the first diagnostic step in allergic mycoses , we discussed a few questions regarding fungal sensitisation. However, we do not provide detailed guidance on fungal asthma without ABPA, which can be found elsewhere . A. fumigatus is the most common fungus associated with allergic sensitisation and ABPA . In a recent meta-analysis, the pooled prevalence of A. fumigatus sensitisation in asthmatic adults was 25% in tertiary care. Of the Aspergillus -sensitised individuals, nearly 37% could develop ABPA . The prevalence of Aspergillus sensitisation was high (16–17%) even in population-based studies . While most patients with ABPA have moderate-to-severe asthma, some have mild asthma and thus screening solely based on symptoms or asthma control may miss several cases . Given the high prevalence of A. fumigatus sensitisation (and ABPA in A. fumigatus sensitisation), all asthmatic adults seeking tertiary care should be evaluated for sensitisation against A. fumigatus . Screening is essential since ABPA can occur even in mild asthmatic subjects and there is a high risk of progression to bronchiectasis if ABPA is undetected. Other fungi (other Aspergillus spp., Candida , Penicillium , Alternaria , Cladosporium and Trichophyton ) are also implicated in allergic sensitisation; however, they rarely cause allergic airway mycoses . Thus, evaluation for sensitisation to other fungi may be reserved for difficult-to-treat asthma patients who do not have A. fumigatus sensitisation. The literature on fungal sensitisation in children is predominantly for A. fumigatus and data on other fungi are scarce . The experts agreed that among children, only those with difficult-to-treat asthma require screening for A. fumigatus sensitisation rather than all asthmatic children . The IgE immunoassay (cut-off 0.35 kUA·L −1 , fluorescent enzyme immunoassay (FEIA)) is the most widely used test to diagnose Aspergillus sensitisation . The DECG accepted A. fumigatus -specific IgE as the preferred screening tool for Aspergillus sensitisation (and ABPA), given its higher sensitivity (99–100%) than the Aspergillus skin test (88–94%) . Also, A. fumigatus -IgE can detect sensitisation against other Aspergillus spp., especially Aspergillus flavus . A skin prick test may be performed additionally or if fungus-specific IgE is unavailable. In asthmatic subjects without known A. fumigatus sensitisation, sensitisation may be re-evaluated if there is unexplained deterioration in asthma control. While a few studies have investigated repeated evaluation for sensitisation , more evidence is required on the frequency of periodic evaluation in those with previously negative A. fumigatus -specific IgE and uncontrolled asthma. Recommendations We recommend evaluation for A. fumigatus sensitisation (LoC: 94.9%) rather than all fungi. Assessment of sensitisation to other fungi is suggested in difficult-to-treat asthmatic subjects with negative A. fumigatus sensitisation (LoC: 61.5%). We recommend fungus-specific IgE in preference to a skin prick test for documenting fungal sensitisation in asthmatic subjects (LoC: 76.5%). We recommend evaluating Aspergillus sensitisation in all newly diagnosed asthmatic adults in tertiary care settings (LoC: 71.4%). For children, we recommend evaluating Aspergillus sensitisation only in those with difficult-to-treat asthma (LoC: 73.0%). We are unable to recommend the periodicity of screening for A. fumigatus sensitisation in those with a negative test at the first screening. Investigations for ABPA/M and the diagnostic cut-offs Asthmatic subjects with A. fumigatus sensitisation need further evaluation to exclude ABPA . Notably, the methodology of performing the various immunological tests and the different cut-offs are important sources of variation in practice across different centres , with the cut-off values varying with the method used. There was consensus for performing the following immunological tests in suspected ABPA: A. fumigatus -specific IgE and IgG, serum total IgE, and peripheral blood eosinophil count. We could not reach a consensus for recommending the Aspergillus skin test and serum precipitins against Aspergillus , partly because access to these different test formats varies widely, and they have varying diagnostic accuracy. However, these tests can be used when automated immunoassays are unavailable. Serum total IgE is a non-specific marker of immunological activity, with a broad differential diagnosis when elevated . However, it reflects disease activity and is an essential monitoring tool in ABPA . The serum total IgE values decrease during treatment and the last recorded value during clinical stability is termed the “new baseline” . An increase of ≥50% of this new baseline serum total IgE is used for diagnosing exacerbation. A value ≥500 IU·mL −1 (by enzyme immunoassay) was recommended as the IgE cut-off to diagnose ABPA. This recommendation deviates from the previous ISHAM-AWG guidelines (≥1000 IU·mL −1 ) , as the lower cut-off offers higher sensitivity (98% versus 91%) . Immunoassay and immunoprecipitation (precipitins) are standard methods to detect IgG against A. fumigatus . A recent meta-analysis found the pooled sensitivity of immunoassays better than immunoprecipitation . Automated immunoassays are easier to implement and more sensitive than immunoprecipitation. On the other hand, immunoprecipitation allows in-house methods to vary the antigens tested and is useful in diagnosing ABPM . The cut-off of A. fumigatus -IgG for automated immunoassays differs from the manufacturer's recommendation and between assays and different populations . For instance, the cut-off values for A. fumigatus -IgG used in India (≥27 mgA·L −1 ) and Japan (≥60 mgA·L −1 ) differ from the UK cut-off (≥40 mgA·L −1 ; manufacturer's recommendation) . The experts stressed the need for data on the optimal cut-offs for A. fumigatus -IgG in different populations and using different immunoassays. Until such data are available, other population-specific cut-offs or the manufacturer's recommendation should be used. Eosinophils primarily drive ABPA pathogenesis; thus, lung or blood eosinophilia is a common feature of ABPA . However, eosinophilia may also be present in asthma, fungal-sensitised asthma and several other disorders . Also, overlap between different eosinophilia-associated diseases is frequent and contributes to higher levels of eosinophilia . Despite a modest diagnostic performance for differentiating ABPA from asthma , blood eosinophilia can guide therapy, such as initiating anti-type 2 biological agents or a need for combination therapy (with prednisolone and itraconazole) . The DECG thus recommended blood eosinophil count to evaluate ABPA (cut-off 500 cells·µL −1 ). Sputum eosinophilia may be a more accurate marker of eosinophilic inflammation and can guide therapy , although the experts felt that in many practice settings it may be difficult to obtain quality sputum differential cell counts. The experts acknowledged the underutilised potential of sputum eosinophil count and identified this as an unmet research need in ABPA . Sputum differential cell counts could also guide therapy in patients with ABPA exacerbations. One suggested algorithm that needs further research is provided in supplementary figure S1 . Airway colonisation by Aspergillus spp. (or other fungi in ABPM) is crucial in initiating and sustaining immunological responses against the fungi . Unfortunately, the sensitivity and specificity of sputum fungal culture are low in diagnosing ABPA. Further, it is difficult to assign causality to the isolated fungi in ABPA, and dissociation between colonising and sensitising fungi is known . Thus, the DECG did not recommend sputum fungal culture for diagnosing ABPA but recommended its use in ABPM. Unlike ABPA, repeated isolation of a fungus is crucial for diagnosing ABPM . Sputum fungal cultures are essential to assess for azole resistance and could be obtained before starting antifungal treatment and later to characterise treatment failures better . Galactomannan is a vital component of the Aspergillus cell wall and detecting serum galactomannan antigen has been approved to diagnose invasive pulmonary aspergillosis. However, given the poor accuracy of serum galactomannan testing in ABPA , the DECG recommended against its use for diagnosing ABPA. Immunological tests for ABPA currently utilise crude A. fumigatus extracts . Several A. fumigatus -specific antigens (f1, f2, f3, f4 and f6) are commercially available through recombinant technology . Recombinant A. fumigatus (rAsp) antigens can identify true A. fumigatus sensitisation . IgE against rAsp antigens (f1, f2 and f4) was found specific for ABPA in two different studies , and is particularly helpful in cases where there is a mismatch between the colonising and sensitising fungi . While the elevation of blood eosinophil count, serum total IgE and A. fumigatus -specific IgG can have several other causes, the IgE against rAsp antigens is highly specific . Despite these advantages, the group recommended against the routine use of rAsp antigens for diagnosing ABPA as they are not widely available . However, the experts suggested that IgE against rAsp f1, f2 and f4 may be used for specific purposes, such as differentiating ABPA from ABPM and clinical research . IgE against rAsp f6 has been found helpful in diagnosing ABPA in systematic reviews ; however, it lacks specificity and can be falsely positive in subjects with atopic dermatitis, possibly due to Malassezia cross-sensitisation . Thus, the expert guidance based on prospective studies suggests that only IgE against rAsp f1 and f2 (followed by f4) consistently differentiates asthmatic subjects with and without ABPA. The manufacturer-recommended cut-offs may be suboptimal and appropriate cut-offs should be derived for different populations . Imaging the lungs is critical in diagnosing ABPA and the DECG recommended using thin-section computed tomography (CT) (1.25–1.5 mm) . We have provided the technical details of the CT acquisition protocol for ABPA in supplementary table S3 . The higher sensitivity, identification of the type and distribution of bronchiectasis, and recognition of mucus plugs are advantages of CT over a chest radiograph . High-attenuation mucus (HAM), i.e. mucus visually denser than the paraspinal muscles on non-contrast thorax CT, is a pathognomonic feature found in a subset of patients with ABPA . The sensitivity and specificity of HAM are 35% and 100%, respectively . The DECG recommended performing chest CT at baseline for diagnosis, assessment of bronchiectasis and prognostication. A chest radiograph, not a chest CT, should be used during follow-up. While magnetic resonance imaging is radiation-free, the DECG did not routinely recommend its use as it has no significant diagnostic advantage over the readily available chest CT . Flexible bronchoscopy is used to obtain respiratory samples for fungal culture . However, considering the invasive nature of the procedure, most experts did not recommend the routine use of bronchoscopy in diagnosing ABPA. Instead, the DECG suggested performing bronchoscopy in suspected ABPA/M patients in the following situations: 1) uncertain diagnosis, 2) in those with suspected ABPM where sputum cultures are uninformative or cannot be obtained, 3) unexplained haemoptysis, or 4) in patients with suspicion of chronic infection (tuberculous or non-tuberculous mycobacterial infection) before initiating systemic glucocorticoids. Infrequently, therapeutic bronchoscopy is required in ABPA patients to remove mucus plugs in the setting of respiratory failure or recalcitrant mucus plugs despite systemic therapy . Recommendations In asthmatic subjects with Aspergillus sensitisation, we recommend performing serum total IgE (LoC: 89.7%), A. fumigatus -specific IgG (LoC: 82.1%) and peripheral blood eosinophil count (LoC: 87.2%). We recommend using population-specific cut-offs to interpret Aspergillus -specific IgG. When data are unavailable, we recommend using manufacturer-recommended cut-offs (LoC: 82.8%). We recommend the following cut-offs: serum total IgE ≥500 IU·mL −1 (LoC: 71.8%) and blood eosinophil count ≥500 cells·µL −1 for diagnosing ABPA (LoC: 73.0%). We do not recommend using serum galactomannan for diagnosing ABPA (LoC: 92.3%). Sputum fungal culture is suggested during the evaluation of ABPA and may help identify the species or guide therapy (LoC: 61.5%). Sputum fungal culture is recommended during the evaluation of ABPM (LoC: 100%). We recommend a thin-section chest CT at baseline to identify and characterise bronchiectasis, mucus plugging, HAM and other abnormalities in patients with suspected ABPA (LoC: 92.3%). We suggest using a chest radiograph to assess treatment response in ABPA (LoC: 62.3%). Bronchoscopy is not routinely recommended for diagnosing ABPA (LoC: 86.1%). Diagnostic criteria ABPA was first described in 1952 by H inson et al . . However, the first attempt to formulate diagnostic criteria was made in 1977 by R osenberg et al . . Subsequently, several criteria have been proposed, including the 2013 ISHAM-AWG criteria . The group suggested modifying the existing ISHAM-AWG criteria. In both rounds, consensus could not be reached (LoC: 48.7% and 53.8%). Most experts felt that the criteria must be simple and allow identification and differentiation of ABPA and ABPM. Finally, after achieving consensus, we recommend separate criteria for diagnosing ABPA and ABPM ( and ). The diagnosis of ABPA/M should be suspected in patients with predisposing conditions or a compatible clinico-radiological presentation (expectoration of mucus plugs, fleeting opacities on chest imaging, finger-in-glove opacities and lung collapse). Thus, the revised criteria include a compatible presentation to enable diagnoses of ABPA/M in those without predisposing conditions . Additionally, two components are essential to make a diagnosis. The first is to document sensitisation against the implicated fungus (using fungus-specific IgE), while the other is to demonstrate immunological activity (raised serum total IgE). However, these two tests can also be positive in patients with fungal sensitisation without ABPA. Here, besides the essential components, the presence of other features, including fungal-specific IgG, peripheral blood eosinophilia and consistent imaging, confirms the diagnosis of ABPA/M. Importantly, the presence of HAM on chest CT is pathognomonic and diagnoses ABPA/M, even when a few other criteria components are missing . We have added another radiological finding, namely “fleeting opacities consistent with ABPA” on chest radiographs, given its high specificity for diagnosing ABPA . In most ABPA/M patients, serum total IgE is ≥500 IU·mL −1 . Uncommonly, serum total IgE could be <500 IU·mL −1 despite the presence of all other components. Low serum total IgE can be seen in those with prior glucocorticoid treatment , the elderly or when the patient has constitutively low IgE before developing ABPA . Also, any range only covers 95% of the population and all individuals will not meet a specific cut-off . If available, IgE against rAsp antigens (f1, f2 and f4) may be used to diagnose ABPA. While investigating a patient for ABPA, we recommend performing A. fumigatus -specific IgE . If the value is ≥0.35 kUA·L −1 , serum total IgE levels should be measured. If the value is ≥500 IU·mL −1 , other tests for ABPA, including A. fumigatus -specific IgG, peripheral blood eosinophil count, chest CT and lung function tests, should be done to characterise the disease . The basic framework for diagnosing ABPM is similar to ABPA, with a few differences . ABPM should be considered in patients with possible ABPA, but A. fumigatus -specific IgE is <0.35 kUA·L −1 . ABPM can be suspected when a causative fungus is isolated in at least two sputum culture specimens or bronchoalveolar lavage fluid culture. ABPM is then confirmed by demonstrating allergic sensitisation (skin test or fungus-specific IgE), combined with a raised serum total IgE and consistent radiological features . Unfortunately, commercial assays for detecting IgE and IgG against fungi other than Aspergillus spp. are available only for a few species ( Alternaria , Cladosporium , Candida , Mucor , Trichophyton and Penicillium ). For other fungi, including S. commune , Bipolaris and others, in-house assays are required for detecting IgE and IgG. A skin test or immunoprecipitation would be required when fungus-specific IgE or IgG is unavailable. There is also a high probability of misclassifying ABPA as ABPM if IgE and IgG against Aspergillus spp. are performed using non-standardised assays. The rest of the workup for ABPM is similar to ABPA . Notably, the absence of elevated IgE against rAsp f1, f2 and f4 strongly supports the diagnosis of ABPM over ABPA in a patient with allergic pulmonary mycoses . In settings where fungus-specific serology is not available, ABPM may be pragmatically diagnosed if there is repeated and consistent culture growth, serum total IgE ≥500 IU·mL −1 , peripheral blood eosinophilia and radiological features of ABPM, provided the Aspergillus -specific serology is negative. The differential diagnosis of ABPA/M is broad and caution is advised in making the diagnosis in patients without either asthma or CF. A. fumigatus -specific IgE and IgG can be elevated in COPD, pulmonary tuberculosis and bronchiectasis, and some of these patients can develop ABPA. Patients with chronic pulmonary aspergillosis may have raised serum A. fumigatus -IgE and total IgE in addition to A. fumigatus -IgG . Aspergillus (and fungal) bronchitis is associated with at least two positive respiratory samples yielding the same fungus and may be associated with a raised A. fumigatus -IgG and bronchiectasis, but without fulfilling the diagnostic criteria for ABPA/M. Patients with severe asthma may be sensitised to A. fumigatus or multiple other fungi with a raised total IgE. They are classified as severe asthma with fungal sensitisation under the umbrella of fungal asthma when they do not fulfil the ABPA/M criteria. Patients with ABPA may also have an additional underlying aetiology for bronchiectasis . Therefore, a search for other causes of bronchiectasis (immunodeficiencies, ciliary disorders and mycobacterial infection) is prudent . The diagnostic workup of bronchiectasis includes complete blood count, A. fumigatus -specific IgE, sweat chloride test, immunoglobulin levels and mycobacterial cultures from sputum . If the initial workup is negative, whole-exome sequencing can be performed (to identify aetiologies such as primary ciliary dyskinesias, primary immunodeficiency and atypical CF), especially in those with extensive bronchiectasis and recurrent infections since childhood. Clinical classification and treatment response criteria A clinical framework for classifying ABPA is essential due to the chronic relapsing nature of the illness and the propensity for developing severe complications. Also, an objective treatment response criterion is useful for monitoring therapy during routine care and in clinical trials. The first classification attempt categorised ABPA into five stages . As the stages were imprecise, the ISHAM-AWG previously proposed a modified staging with more detailed definitions . However, there were several unresolved issues. Most importantly, the stages were labelled 0–6, but a patient does not necessarily progress from one to another. The previous classification also did not reflect progressive severity since stage 4 (remission) is a more stable clinical state than stage 3 (exacerbation), which is counterintuitive. To overcome these limitations, we proposed modifications that achieved consensus in the second round (LoC: 85.3%). In the new ABPA/M classification, we have removed the numbered stages and retained five categories: acute ABPA, response, remission, treatment-dependent ABPA and advanced ABPA . We have removed the asymptomatic stage and glucocorticoid-dependent asthma as they had no clear treatment implications in ABPA. Also, we have included newly diagnosed ABPA and exacerbation together as acute ABPA. To diagnose ABPA exacerbation, asthma or bronchiectasis (infective) exacerbations need to be excluded, and we provide definitions for the two entities in . Remission, as in asthma , is diagnosed when the patient has no asthma or ABPA exacerbations, is not dependent on oral glucocorticoid therapy and has the best possible lung function. Remission may be achieved spontaneously after treatment or with antifungal azoles or biological agents. Finally, advanced ABPA is defined in patients with extensive bronchiectasis and type 2 respiratory failure or secondary pulmonary hypertension . Radiological classification of ABPA CT of the thorax is crucial in diagnosing ABPA. However, due consideration should be given to the radiation dosage when CT scans are ordered, especially in children. Chest CT is also prognostic. For instance, the extent of bronchiectasis, HAM and any fungal ball are independent predictors of recurrent ABPA exacerbations . Central bronchiectasis (usually bilateral) is the predominant pattern seen in ABPA, although it is not uncommon to find both central and peripheral bronchiectasis . Isolated central bronchiectasis is encountered only in a few conditions, including ABPA and tracheobronchomegaly, and is thus a helpful distinguishing feature . Previously, Greenberger's group classified ABPA as ABPA with central bronchiectasis (ABPA-CB) or serological ABPA (ABPA-S) based on the presence or absence of bronchiectasis . Subsequently, K umar classified ABPA into three groups: ABPA-S, ABPA-CB and ABPA-CB with other radiological findings (ABPA-CB-ORF). In a study involving 234 patients, A garwal et al . categorised ABPA into ABPA-S (mild), ABPA-CB (moderate) and ABPA-CB-HAM (severe). Based on all the evidence, the ISHAM-AWG had previously classified ABPA radiologically into four categories: ABPA-S, ABPA with bronchiectasis (ABPA-B), ABPA-HAM and ABPA with chronic pleuropulmonary fibrosis (ABPA-CPF) . The term “bronchiectasis” (B) was used instead of “central bronchiectasis” (CB) as bronchiectasis in ABPA can extend to the periphery in up to 40% of the lobes . Mucus plugging without HAM is another common radiological finding in ABPA. Mucus plugs are consistently associated with eosinophilic inflammation and immunologically severe ABPA . The DECG discussed several radiological classifications. Finally, the scheme presented in achieved consensus (LoC: 88.2%). The new classification includes five classes: ABPA-S, ABPA-B, ABPA with mucus plugging (ABPA-MP), ABPA-HAM and ABPA-CPF . ABPA-S refers to patients of ABPA without bronchiectasis, while ABPA-B includes patients with bronchiectasis. ABPA-HAM has been retained, as HAM is an independent and pathognomonic diagnostic feature of ABPA . ABPA-MP includes patients with non-hyperattenuating mucus plugs. Patients with bronchiectasis and mucus plugging are labelled as ABPA-MP, given the greater immunological severity in those with mucus plugging. Other radiological findings frequently observed in ABPA include centrilobular nodules (with a tree-in-bud appearance), atelectasis, mosaic attenuation and consolidation. These findings can be seen in isolation or with ABPA-B, ABPA-MP and ABPA-HAM. In patients with ABPA-CPF, a vital consideration is the exclusion of chronic pulmonary aspergillosis . Treatment of ABPA The principles of treating ABPA involve using anti-inflammatory agents (glucocorticoids or biological agents targeting type 2 immune response) to control immune responses or antifungal agents to decrease airway fungal colonisation. The treatment goals are symptom relief, improving asthma control, preventing asthma and ABPA exacerbations, abrogating bronchiectasis progression, and minimising therapy-related adverse events. The treatment principles of ABPM are like ABPA, except that the implicated fungus guides the choice of antifungal drugs. The DECG reviewed the RCTs and the therapies available for treating ABPA patients ( and ) . Initiating treatment for newly diagnosed ABPA Patients with acute ABPA require treatment with systemic therapies. Glucocorticoids are the most effective treatment for acute ABPA . An RCT involving 92 ABPA patients compared two glucocorticoid dosing protocols (low dose (prednisolone 0.5 mg·kg −1 ·day −1 for 2 weeks, then on alternate days for 8 weeks; then tapered by 5 mg every 2 weeks and discontinued after 3–5 months) versus high dose (prednisolone 0.75 and 0.5 mg·kg −1 ·day −1 for 6 weeks each; subsequently, tapered by 5 mg every 6 weeks and discontinued after 8–10 months)). The frequency of ABPA exacerbations was similar in the two groups and the lower dose resulted in fewer adverse events. However, there was a lower clinico-radiological and immunological response at 6 weeks with the lower dose . Several centres use doses intermediate between the low and high doses (prednisolone 0.5, 0.25 and 0.125 mg·kg −1 ·day −1 for 4 weeks each, then tapered by 5 mg every 2 weeks till discontinuation). The DECG recommended using a 4-month course of low-to-moderate dose oral prednisolone (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) for acute ABPA . Care should be taken while using methylprednisolone because when combined with oral itraconazole, there is a higher risk of exogenous Cushing's syndrome and adrenal insufficiency . Notably, the experts suggested the need for trials with even shorter duration of glucocorticoids, as the 4-month duration was derived from the need to randomise against longer-term azole therapy. Many clinicians administer an initial 2-week course of glucocorticoids in those started on an oral azole, and as symptoms are controlled, transition to high-dose inhaled corticosteroids (ICS). Importantly, a combination of inhaled budesonide or fluticasone and itraconazole can also cause exogenous Cushing's syndrome . While asymptomatic ABPA patients do not routinely require systemic therapy, the treatment decision needs to be individualised. For instance, patients can have well-controlled asthma on high-dose inhaled steroids and may benefit from treatment of underlying ABPA, especially if the chest CT shows bronchiectasis or mucus plugging. Also, asymptomatic patients with prolonged mucus plugging can progress to irreversible bronchiectasis. Thus, optimisation of asthma treatment and close observation with a clinical review, chest radiograph and serum total IgE every 3–6 months is required if a decision is made not to treat patients with asymptomatic ABPA. Oral antifungal triazoles, especially itraconazole, have similar effects as glucocorticoids but a slower trajectory to improvement and a better safety profile than glucocorticoids . Although the evidence was limited to a single RCT, the DECG recommended using oral itraconazole (for 4 months) as an alternative initial therapy for acute ABPA, given the considerable clinical experience with itraconazole. While voriconazole has similar efficacy as glucocorticoids for treating acute ABPA , the experts expressed concerns with its use as first-line therapy due to poorer patient tolerance. Also, prednisolone decreases the plasma concentration of voriconazole in a dose-dependent fashion . A recent RCT of 191 patients found no significant reduction in ABPA exacerbations with a combination of prednisolone and itraconazole compared to prednisolone alone and a propensity for higher adverse events . However, patients with blood eosinophil count ≥1000 cells·µL −1 and extensive bronchiectasis (≥10 segments) had a reduced 1-year exacerbation rate with the combination therapy . Similarly, there is little evidence for posaconazole, isavuconazole or a combination of newer azoles and glucocorticoids as the initial therapy for ABPA. There are no data supporting the use of biological agents as first-line therapy in ABPA. Adjunctive vitamin D supplementation was not helpful in managing ABPA , but vitamin D deficiency should be corrected as it aggravates osteopenia due to long-term glucocorticoid usage. High doses of ICS alone do not achieve immunological control or reduce exacerbations when used as therapy for ABPA . The efficacy of nebulised amphotericin B in acute ABPA is also poor . Notably, most studies on ABPA therapy have included mainly patients with ABPA-B, with little data on ABPA-S . Most experts do not routinely treat ABPA-S with systemic ABPA-specific treatment. Instead, patients with ABPA-S are managed like asthma. However, oral glucocorticoids or azoles may be required in those with poor asthma control or recurrent exacerbations despite optimal asthma management. Recommendations We do not recommend treating asymptomatic ABPA patients with systemic therapy (LoC: 85.7%). ABPA-S should be treated with systemic therapy only if there is poor asthma control (LoC: 79.4%) or recurrent exacerbations despite asthma therapy (LoC: 85.3%). We recommend a low-to-moderate dose (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) of oral prednisolone (LoC: 78.1%) or oral itraconazole (LoC: 73.5%) for 4 months as the initial therapy for treating acute ABPA. We recommend oral itraconazole as the initial therapy where systemic glucocorticoids are contraindicated (LoC: 84.6%). We do not recommend using a combination of itraconazole and glucocorticoids as first-line therapy for acute ABPA (LoC: 71.9%). However, a short course of glucocorticoids (<2 weeks) may be used as initial therapy along with oral itraconazole. Oral voriconazole, posaconazole and isavuconazole should not be used as first-line agents for treating acute ABPA (LoC: 78.1–96.9%). They may be used if there are contraindications to systemic glucocorticoids and intolerance, failure or resistance to itraconazole therapy (LoC: 12.8–64.1%). High-dose ICS should not be used as primary therapy for acute ABPA (LoC: 100%). We do not recommend using biological agents as first-line therapy for acute ABPA (LoC: 96.9%). Treatment of ABPA exacerbation After treatment cessation, almost 50% of patients experience ABPA exacerbations . In a patient with ABPA, worsening of respiratory symptoms may occur due to asthma exacerbation, immunologically exacerbated ABPA and infective exacerbation of bronchiectasis, apart from unrelated causes. The three types of exacerbations can usually be differentiated using chest radiographs, serum total IgE and sputum bacterial cultures. Occasionally, there can be an overlap of more than one cause of exacerbation in an individual patient. ABPA exacerbations are characterised by sustained worsening (≥2 weeks) of clinical symptoms or the appearance of new infiltrates on chest imaging, along with an increase in serum total IgE by ≥50% above the “new baseline” IgE (during clinical stability) . Asthma exacerbations are not associated with increased serum total IgE or new infiltrates on chest imaging and should be managed with a short course of oral glucocorticoids. Bronchiectasis (infective) exacerbations are diagnosed with clinical worsening without elevation in serum total IgE ≥50% compared to baseline. Sputum cultures frequently show bacterial growth in bronchiectasis exacerbations. There are no RCTs for managing ABPA exacerbations and the options include using prednisolone, itraconazole or their combination . In clinical practice, ABPA exacerbation is managed like newly diagnosed ABPA using either prednisolone or itraconazole. Many experts use a combination of oral prednisolone and oral itraconazole in patients with recurrent exacerbations (≥2 in the last 1–2 years), especially in those with extensive bronchiectasis. Also, nebulised amphotericin B has poor efficacy for treating ABPA exacerbations . Pulse doses of methylprednisolone have been used for ABPA exacerbations refractory to oral glucocorticoids . Recommendations We recommend treating acute ABPA exacerbations in the same manner as newly diagnosed ABPA (LoC: 100%). A combination of oral prednisolone and itraconazole should be used for treating recurrent (≥2 in the last 1–2 years) ABPA exacerbations (LoC: 71.4%). We do not recommend using biological agents (LoC: 94.3%) or nebulised amphotericin B (LoC: 100%) for treating acute ABPA exacerbations. Monitoring treatment response After treatment initiation, patients should be monitored for response after 8–12 weeks using clinical symptoms, serum total IgE and chest radiographs . Instead of subjective assessment, the DECG suggested that symptom monitoring be done using a semiquantitative Likert scale as no improvement (or worsening), mild (<25% of baseline), moderate (25–50% of baseline) or significant improvement (>50% of baseline) for routine clinical care. The experts suggested using a more quantitative scale like a visual analogue scale (VAS) for clinical trials. A good response is indicated by a significant improvement in symptoms (Likert score or VAS ≥50%) and imaging , along with at least a 20% reduction in the serum total IgE levels . Spirometry may be used to monitor treatment response. A recent study reported the minimal clinically important difference (MCID) of forced expiratory volume in 1 s of 158 mL for treatment response in ABPA . Quality-of-life questionnaires are cumbersome and were discouraged unanimously by experts for routine patient care . However, for clinical trials, quality-of-life questionnaires could be used. An important consideration is that the MCID for the quality-of-life questionnaires could differ in ABPA from asthma. For instance, in one study, the MCID of the St George's Respiratory Questionnaire was 8 points, rather than the 4 points used in asthma . The serum A. fumigatus -specific IgE and IgG levels do not consistently fall following treatment . No studies have evaluated blood eosinophil count or concentrations of IgE/IgG against rAsp antigens for assessing treatment response. While using oral azoles, therapeutic drug monitoring is recommended initially at 2-week and 3-month intervals or during clinical worsening. Significant variations in the bioavailability of different itraconazole preparations compromise response . Likewise, drug–drug interactions are frequent with itraconazole, with common issues being ICS exposure boosting and undetectable levels of itraconazole with rifampicin . The recommended minimum therapeutic levels are >0.5, >1 and >1 µg·mL −1 for itraconazole, voriconazole and posaconazole, respectively . Monitoring for treatment-related adverse events is paramount. When using systemic glucocorticoids, the DECG recommended monitoring plasma glucose, blood pressure, body weight and mental status at the least. Monitoring liver function is necessary for patients receiving oral azoles . Recommendations We recommend assessing the initial treatment response after 8–12 weeks using a combination of clinical, immunological and imaging findings (LoC: 100%). A good response is indicated by a major improvement in symptoms (Likert score or VAS ≥50%) and chest radiographs, along with at least a 20% reduction in serum total IgE. A. fumigatus -specific IgE and IgG (LoC: 100%), peripheral blood eosinophil count (LoC: 73.6%), and IgE against recombinant antigens are not recommended for response assessment (LoC: 100%). We recommend therapeutic drug monitoring while using antifungal azoles (LoC: 91.4%). Management of treatment-dependent ABPA Almost 10–25% of ABPA patients become treatment-dependent . Two RCTs (84 patients) have evaluated the efficacy of itraconazole in treatment-dependent ABPA . Itraconazole reduced the oral glucocorticoid dose, sputum eosinophil count and ABPA exacerbations . A significant limitation was that neither study reported outcomes beyond 8 months regarding ABPA exacerbations. In the last two decades, several trials of monoclonal antibodies against IgE, interleukin (IL)-5, IL-5 receptor (IL-5R), IL-4 receptor α (IL-4Rα) and thymic stromal lymphopoietin (TSLP) have demonstrated clinical benefit in severe eosinophilic asthma . Biological agents are likely helpful in stable treatment-dependent ABPA based on case reports and small case series . Omalizumab is a potential therapeutic approach since ABPA is associated with elevated IgE levels. Mepolizumab (anti-IL-5), benralizumab (anti-IL-5R), dupilumab (anti-IL-4Rα) and tezepelumab (anti-TSLP) have also been used in ABPA patients . Most experience of biological agents in ABPA is with omalizumab. Using omalizumab in ABPA led to improvement in symptoms, reduction in exacerbations and asthma hospitalisations, improvement in lung function, and reduction in the dose of oral steroids . In the only crossover RCT of 13 patients, omalizumab use ( versus placebo) was associated with less frequent exacerbations and decreased basophil reactivity to A. fumigatus . Biological agents should also be considered for maintenance therapy for underlying asthma . Finally, continuous low-dose glucocorticoids should be the last option in managing treatment-dependent ABPA. Recommendation Long-term itraconazole (100%), nebulised amphotericin B (LoC: 100%) or biological agents (LoC: 71%) are recommended options for managing treatment-dependent ABPA. Management of ABPA in remission During remission (stable disease), ABPA patients should be managed for underlying asthma and bronchiectasis (ICS and long-acting bronchodilators, nebulised saline, antibiotics, and others) per the existing guidelines . Patients should be monitored with clinical review, serum total IgE levels and lung function test every 3–6 months for the first year and then every 6–12 months. Remission can be prolonged by using long-term itraconazole, nebulised amphotericin B (LoC: 100%) and biological agents (LoC: 71%), especially in those with treatment-dependent ABPA. Two RCTs have also evaluated the role of nebulised amphotericin B as maintenance to prevent future ABPA exacerbations . In both studies, patients with ABPA exacerbation were treated for 4 months with oral glucocorticoids or prednisolone and itraconazole. After that, they were randomised to receive nebulised amphotericin B versus placebo. In the smaller pilot study (21 patients), using nebulised amphotericin B deoxycholate (10 mg twice daily, three times a week) reduced ABPA exacerbations at 1 year . In the larger NEBULAMB study (139 patients), while the primary outcome was not significant, the time-to-first exacerbation was significantly longer with nebulised liposomal amphotericin B (25 mg weekly) compared to the control group . Recommendations We recommend that patients in remission be managed for underlying asthma and bronchiectasis per the existing guidelines (LoC: 100%). For patients achieving remission with antifungal azoles or biological agents, we recommend periodic assessments to determine the need for these therapies (LoC: 100%). Management of extensive bronchiectasis and advanced ABPA There are no specific studies on advanced ABPA . Nebulised hypertonic saline (3–7%, 4–5 mL) can reduce sputum viscosity and ease the expectoration of mucus plugs in bronchiectasis patients . Treatment should be preceded by nebulised salbutamol to minimise the risk of bronchospasm. Also, the first dose of nebulised hypertonic saline should be administered under supervision . Nebulised antibiotics and long-term azithromycin therapy can improve outcomes in ABPA patients with bronchiectasis and frequent infective exacerbations . Caution is advised when using azithromycin in those receiving itraconazole, as it can cause QTc prolongation. In some ABPA patients, widespread bronchiectasis can eventually lead to chronic type 2 respiratory failure and pulmonary hypertension . The DECG recommended extrapolating guidance from other pulmonary disorders (chronic obstructive lung disease, interstitial lung disease and bronchiectasis) for long-term oxygen therapy (LTOT), vaccination and lung transplantation . LTOT in those with resting hypoxaemia (arterial oxygen tension ( P aO 2 ) ≤55 mmHg) reduces pulmonary hypertension and improves survival in patients with chronic obstructive lung disease. There is no role for LTOT in mild hypoxaemia ( P aO 2 >55 mmHg at rest) or nocturnal oxygen desaturation . In one study, patients with chronic and allergic aspergillosis responded poorly to the 23-valent pneumococcal polysaccharide vaccine compared to healthy adults . Thus, influenza or pneumococcal vaccine should be administered before initiating glucocorticoid therapy or delayed until the underlying condition is better controlled ( e.g. after treatment with antifungals or glucocorticoids). The International Society for Heart and Lung Transplantation criteria should be used to refer and list patients with advanced ABPA . Recommendation We recommend using standard guidelines from other pulmonary disorders for LTOT, vaccination and lung transplantation (LoC: 100%). Treatment in special conditions ABPA during pregnancy The principles of managing ABPA during pregnancy are to optimise asthma and ABPA control while preventing harm to the fetus. Most ABPA trials have excluded pregnant patients and data are extrapolated. Oral glucocorticoids for the duration and dose required in ABPA are safe in pregnancy , while the use of oral itraconazole is associated with a higher risk of premature births and abortions . Experience with biological agents in ABPA during pregnancy is limited. When used with oral glucocorticoids, omalizumab increases the risk of pre-term delivery . In a systematic review of biological agent use during pregnancy with atopic disease, major fetal defects, low birthweight, pre-term delivery and stillbirths were reported . Recommendations We recommended using systemic glucocorticoids (in the same doses as for non-pregnant) for managing acute-stage or treatment-dependent ABPA in pregnancy (LoC: 73.0%). We recommend avoiding the use of biological agents (LoC: 86.5%) or oral azoles (LoC: 100%) for managing acute-stage or treatment-dependent ABPA in pregnancy. ABPA in CF There are no RCTs of treatment in CF-ABPA . Glucocorticoids or azoles are the preferred initial agents. Glucocorticoids can induce diabetes mellitus, with grave consequences in CF. Monthly doses of intravenous methylprednisolone therapy alone or with azoles have been used to limit the toxicity associated with daily glucocorticoids . The combination of glucocorticoids with azoles is also increasingly being used in CF-ABPA. In one survey, combination therapy was preferred by most respondents for treating newly diagnosed ABPA and first exacerbation . A 3-week course of oral corticosteroids combined with oral itraconazole for 12 months effectively treated CF-ABPA . While using itraconazole capsules, therapeutic drug monitoring is essential as the drug is poorly absorbed in CF. The super bioavailable itraconazole formulation was well-tolerated and achieved therapeutic response in 81% of patients with CF-ABPA . Voriconazole leads to photosensitivity and should be used with caution. Posaconazole achieved therapeutic drug levels and good clinical response in CF-ABPA . While a previous review suggested that omalizumab might benefit patients with ABPA , a recent case series found no benefit with omalizumab . Like in asthma, pulmonary exacerbations in a patient with CF-ABPA could be related to ABPA or pulmonary infection . Recommendation We recommend diagnosing and treating CF-ABPA using the same treatment tenets outlined for ABPA in asthma; however, exercising due caution for issues specific to CF (LoC: 100%). ABPA in children Growth retardation with the use of systemic corticosteroids and the erratic bioavailability of azoles due to age-related changes in cytochrome P450 enzymes are of primary concern . There are no randomised trials on the treatment of ABPA in children . Omalizumab and mepolizumab are approved for asthma therapy in the age group ≥6 years , while other biological agents can be used in those ≥12 years old. The DECG recommended using systemic glucocorticoids or oral itraconazole for treating acute ABPA in children as in adults. The experts felt more research was needed to address the therapeutic issues affecting ABPA in children. Recommendation We recommend treating ABPA in children using the same treatment principles outlined for ABPA in asthmatic adults; however, exercising due caution for issues specific to children (LoC: 100%). Environmental control Aspergillus is a common mould found in various environments, including soil, decaying vegetation and indoor environments. Evidence suggests that environmental exposures to Aspergillus spores could drive exacerbations . The DECG, therefore, suggested minimising activities that could result in the inhalation of large numbers of Aspergillus conidia . If such activities are unavoidable, surgical masks or the more effective N95 respirators may minimise spore inhalation. Further, regular cleaning and maintenance of heating, ventilation and air conditioning (HVAC) systems may prevent mould growth . Identifying and promptly addressing water leaks is essential, as damp environments promote mould growth. Similarly, ensuring proper ventilation in bathrooms, kitchens and other areas prone to moisture accumulation may prevent mould growth . Cleaning and dusting living spaces should be performed regularly using damp cloths to minimise the accumulation and dispersion of spores into the air. Indoor air quality monitoring represents a valuable tool to evaluate the fungal exposome .
We first deliberated on the nomenclature of ABPA and ABPM. The most common form of allergic airway mycoses is ABPA , while ABPM is far less common . Given the considerable overlap of the antigen repertoire of the Aspergillus species, the DECG recommended using the term ABPA when allergic mycoses are caused by any Aspergillus spp. (not A. fumigatus only) and ABPM when attributable to fungi other than Aspergillus spp. The most common fungi responsible for ABPM include Bipolaris spp., Schizophyllum commune and Curvularia spp. . Candida albicans has been implicated in several cases of ABPM ; however, its pathogenicity remains uncertain.
As sensitisation represents the first diagnostic step in allergic mycoses , we discussed a few questions regarding fungal sensitisation. However, we do not provide detailed guidance on fungal asthma without ABPA, which can be found elsewhere . A. fumigatus is the most common fungus associated with allergic sensitisation and ABPA . In a recent meta-analysis, the pooled prevalence of A. fumigatus sensitisation in asthmatic adults was 25% in tertiary care. Of the Aspergillus -sensitised individuals, nearly 37% could develop ABPA . The prevalence of Aspergillus sensitisation was high (16–17%) even in population-based studies . While most patients with ABPA have moderate-to-severe asthma, some have mild asthma and thus screening solely based on symptoms or asthma control may miss several cases . Given the high prevalence of A. fumigatus sensitisation (and ABPA in A. fumigatus sensitisation), all asthmatic adults seeking tertiary care should be evaluated for sensitisation against A. fumigatus . Screening is essential since ABPA can occur even in mild asthmatic subjects and there is a high risk of progression to bronchiectasis if ABPA is undetected. Other fungi (other Aspergillus spp., Candida , Penicillium , Alternaria , Cladosporium and Trichophyton ) are also implicated in allergic sensitisation; however, they rarely cause allergic airway mycoses . Thus, evaluation for sensitisation to other fungi may be reserved for difficult-to-treat asthma patients who do not have A. fumigatus sensitisation. The literature on fungal sensitisation in children is predominantly for A. fumigatus and data on other fungi are scarce . The experts agreed that among children, only those with difficult-to-treat asthma require screening for A. fumigatus sensitisation rather than all asthmatic children . The IgE immunoassay (cut-off 0.35 kUA·L −1 , fluorescent enzyme immunoassay (FEIA)) is the most widely used test to diagnose Aspergillus sensitisation . The DECG accepted A. fumigatus -specific IgE as the preferred screening tool for Aspergillus sensitisation (and ABPA), given its higher sensitivity (99–100%) than the Aspergillus skin test (88–94%) . Also, A. fumigatus -IgE can detect sensitisation against other Aspergillus spp., especially Aspergillus flavus . A skin prick test may be performed additionally or if fungus-specific IgE is unavailable. In asthmatic subjects without known A. fumigatus sensitisation, sensitisation may be re-evaluated if there is unexplained deterioration in asthma control. While a few studies have investigated repeated evaluation for sensitisation , more evidence is required on the frequency of periodic evaluation in those with previously negative A. fumigatus -specific IgE and uncontrolled asthma. Recommendations We recommend evaluation for A. fumigatus sensitisation (LoC: 94.9%) rather than all fungi. Assessment of sensitisation to other fungi is suggested in difficult-to-treat asthmatic subjects with negative A. fumigatus sensitisation (LoC: 61.5%). We recommend fungus-specific IgE in preference to a skin prick test for documenting fungal sensitisation in asthmatic subjects (LoC: 76.5%). We recommend evaluating Aspergillus sensitisation in all newly diagnosed asthmatic adults in tertiary care settings (LoC: 71.4%). For children, we recommend evaluating Aspergillus sensitisation only in those with difficult-to-treat asthma (LoC: 73.0%). We are unable to recommend the periodicity of screening for A. fumigatus sensitisation in those with a negative test at the first screening.
We recommend evaluation for A. fumigatus sensitisation (LoC: 94.9%) rather than all fungi. Assessment of sensitisation to other fungi is suggested in difficult-to-treat asthmatic subjects with negative A. fumigatus sensitisation (LoC: 61.5%). We recommend fungus-specific IgE in preference to a skin prick test for documenting fungal sensitisation in asthmatic subjects (LoC: 76.5%). We recommend evaluating Aspergillus sensitisation in all newly diagnosed asthmatic adults in tertiary care settings (LoC: 71.4%). For children, we recommend evaluating Aspergillus sensitisation only in those with difficult-to-treat asthma (LoC: 73.0%). We are unable to recommend the periodicity of screening for A. fumigatus sensitisation in those with a negative test at the first screening.
Asthmatic subjects with A. fumigatus sensitisation need further evaluation to exclude ABPA . Notably, the methodology of performing the various immunological tests and the different cut-offs are important sources of variation in practice across different centres , with the cut-off values varying with the method used. There was consensus for performing the following immunological tests in suspected ABPA: A. fumigatus -specific IgE and IgG, serum total IgE, and peripheral blood eosinophil count. We could not reach a consensus for recommending the Aspergillus skin test and serum precipitins against Aspergillus , partly because access to these different test formats varies widely, and they have varying diagnostic accuracy. However, these tests can be used when automated immunoassays are unavailable. Serum total IgE is a non-specific marker of immunological activity, with a broad differential diagnosis when elevated . However, it reflects disease activity and is an essential monitoring tool in ABPA . The serum total IgE values decrease during treatment and the last recorded value during clinical stability is termed the “new baseline” . An increase of ≥50% of this new baseline serum total IgE is used for diagnosing exacerbation. A value ≥500 IU·mL −1 (by enzyme immunoassay) was recommended as the IgE cut-off to diagnose ABPA. This recommendation deviates from the previous ISHAM-AWG guidelines (≥1000 IU·mL −1 ) , as the lower cut-off offers higher sensitivity (98% versus 91%) . Immunoassay and immunoprecipitation (precipitins) are standard methods to detect IgG against A. fumigatus . A recent meta-analysis found the pooled sensitivity of immunoassays better than immunoprecipitation . Automated immunoassays are easier to implement and more sensitive than immunoprecipitation. On the other hand, immunoprecipitation allows in-house methods to vary the antigens tested and is useful in diagnosing ABPM . The cut-off of A. fumigatus -IgG for automated immunoassays differs from the manufacturer's recommendation and between assays and different populations . For instance, the cut-off values for A. fumigatus -IgG used in India (≥27 mgA·L −1 ) and Japan (≥60 mgA·L −1 ) differ from the UK cut-off (≥40 mgA·L −1 ; manufacturer's recommendation) . The experts stressed the need for data on the optimal cut-offs for A. fumigatus -IgG in different populations and using different immunoassays. Until such data are available, other population-specific cut-offs or the manufacturer's recommendation should be used. Eosinophils primarily drive ABPA pathogenesis; thus, lung or blood eosinophilia is a common feature of ABPA . However, eosinophilia may also be present in asthma, fungal-sensitised asthma and several other disorders . Also, overlap between different eosinophilia-associated diseases is frequent and contributes to higher levels of eosinophilia . Despite a modest diagnostic performance for differentiating ABPA from asthma , blood eosinophilia can guide therapy, such as initiating anti-type 2 biological agents or a need for combination therapy (with prednisolone and itraconazole) . The DECG thus recommended blood eosinophil count to evaluate ABPA (cut-off 500 cells·µL −1 ). Sputum eosinophilia may be a more accurate marker of eosinophilic inflammation and can guide therapy , although the experts felt that in many practice settings it may be difficult to obtain quality sputum differential cell counts. The experts acknowledged the underutilised potential of sputum eosinophil count and identified this as an unmet research need in ABPA . Sputum differential cell counts could also guide therapy in patients with ABPA exacerbations. One suggested algorithm that needs further research is provided in supplementary figure S1 . Airway colonisation by Aspergillus spp. (or other fungi in ABPM) is crucial in initiating and sustaining immunological responses against the fungi . Unfortunately, the sensitivity and specificity of sputum fungal culture are low in diagnosing ABPA. Further, it is difficult to assign causality to the isolated fungi in ABPA, and dissociation between colonising and sensitising fungi is known . Thus, the DECG did not recommend sputum fungal culture for diagnosing ABPA but recommended its use in ABPM. Unlike ABPA, repeated isolation of a fungus is crucial for diagnosing ABPM . Sputum fungal cultures are essential to assess for azole resistance and could be obtained before starting antifungal treatment and later to characterise treatment failures better . Galactomannan is a vital component of the Aspergillus cell wall and detecting serum galactomannan antigen has been approved to diagnose invasive pulmonary aspergillosis. However, given the poor accuracy of serum galactomannan testing in ABPA , the DECG recommended against its use for diagnosing ABPA. Immunological tests for ABPA currently utilise crude A. fumigatus extracts . Several A. fumigatus -specific antigens (f1, f2, f3, f4 and f6) are commercially available through recombinant technology . Recombinant A. fumigatus (rAsp) antigens can identify true A. fumigatus sensitisation . IgE against rAsp antigens (f1, f2 and f4) was found specific for ABPA in two different studies , and is particularly helpful in cases where there is a mismatch between the colonising and sensitising fungi . While the elevation of blood eosinophil count, serum total IgE and A. fumigatus -specific IgG can have several other causes, the IgE against rAsp antigens is highly specific . Despite these advantages, the group recommended against the routine use of rAsp antigens for diagnosing ABPA as they are not widely available . However, the experts suggested that IgE against rAsp f1, f2 and f4 may be used for specific purposes, such as differentiating ABPA from ABPM and clinical research . IgE against rAsp f6 has been found helpful in diagnosing ABPA in systematic reviews ; however, it lacks specificity and can be falsely positive in subjects with atopic dermatitis, possibly due to Malassezia cross-sensitisation . Thus, the expert guidance based on prospective studies suggests that only IgE against rAsp f1 and f2 (followed by f4) consistently differentiates asthmatic subjects with and without ABPA. The manufacturer-recommended cut-offs may be suboptimal and appropriate cut-offs should be derived for different populations . Imaging the lungs is critical in diagnosing ABPA and the DECG recommended using thin-section computed tomography (CT) (1.25–1.5 mm) . We have provided the technical details of the CT acquisition protocol for ABPA in supplementary table S3 . The higher sensitivity, identification of the type and distribution of bronchiectasis, and recognition of mucus plugs are advantages of CT over a chest radiograph . High-attenuation mucus (HAM), i.e. mucus visually denser than the paraspinal muscles on non-contrast thorax CT, is a pathognomonic feature found in a subset of patients with ABPA . The sensitivity and specificity of HAM are 35% and 100%, respectively . The DECG recommended performing chest CT at baseline for diagnosis, assessment of bronchiectasis and prognostication. A chest radiograph, not a chest CT, should be used during follow-up. While magnetic resonance imaging is radiation-free, the DECG did not routinely recommend its use as it has no significant diagnostic advantage over the readily available chest CT . Flexible bronchoscopy is used to obtain respiratory samples for fungal culture . However, considering the invasive nature of the procedure, most experts did not recommend the routine use of bronchoscopy in diagnosing ABPA. Instead, the DECG suggested performing bronchoscopy in suspected ABPA/M patients in the following situations: 1) uncertain diagnosis, 2) in those with suspected ABPM where sputum cultures are uninformative or cannot be obtained, 3) unexplained haemoptysis, or 4) in patients with suspicion of chronic infection (tuberculous or non-tuberculous mycobacterial infection) before initiating systemic glucocorticoids. Infrequently, therapeutic bronchoscopy is required in ABPA patients to remove mucus plugs in the setting of respiratory failure or recalcitrant mucus plugs despite systemic therapy . Recommendations In asthmatic subjects with Aspergillus sensitisation, we recommend performing serum total IgE (LoC: 89.7%), A. fumigatus -specific IgG (LoC: 82.1%) and peripheral blood eosinophil count (LoC: 87.2%). We recommend using population-specific cut-offs to interpret Aspergillus -specific IgG. When data are unavailable, we recommend using manufacturer-recommended cut-offs (LoC: 82.8%). We recommend the following cut-offs: serum total IgE ≥500 IU·mL −1 (LoC: 71.8%) and blood eosinophil count ≥500 cells·µL −1 for diagnosing ABPA (LoC: 73.0%). We do not recommend using serum galactomannan for diagnosing ABPA (LoC: 92.3%). Sputum fungal culture is suggested during the evaluation of ABPA and may help identify the species or guide therapy (LoC: 61.5%). Sputum fungal culture is recommended during the evaluation of ABPM (LoC: 100%). We recommend a thin-section chest CT at baseline to identify and characterise bronchiectasis, mucus plugging, HAM and other abnormalities in patients with suspected ABPA (LoC: 92.3%). We suggest using a chest radiograph to assess treatment response in ABPA (LoC: 62.3%). Bronchoscopy is not routinely recommended for diagnosing ABPA (LoC: 86.1%).
In asthmatic subjects with Aspergillus sensitisation, we recommend performing serum total IgE (LoC: 89.7%), A. fumigatus -specific IgG (LoC: 82.1%) and peripheral blood eosinophil count (LoC: 87.2%). We recommend using population-specific cut-offs to interpret Aspergillus -specific IgG. When data are unavailable, we recommend using manufacturer-recommended cut-offs (LoC: 82.8%). We recommend the following cut-offs: serum total IgE ≥500 IU·mL −1 (LoC: 71.8%) and blood eosinophil count ≥500 cells·µL −1 for diagnosing ABPA (LoC: 73.0%). We do not recommend using serum galactomannan for diagnosing ABPA (LoC: 92.3%). Sputum fungal culture is suggested during the evaluation of ABPA and may help identify the species or guide therapy (LoC: 61.5%). Sputum fungal culture is recommended during the evaluation of ABPM (LoC: 100%). We recommend a thin-section chest CT at baseline to identify and characterise bronchiectasis, mucus plugging, HAM and other abnormalities in patients with suspected ABPA (LoC: 92.3%). We suggest using a chest radiograph to assess treatment response in ABPA (LoC: 62.3%). Bronchoscopy is not routinely recommended for diagnosing ABPA (LoC: 86.1%).
ABPA was first described in 1952 by H inson et al . . However, the first attempt to formulate diagnostic criteria was made in 1977 by R osenberg et al . . Subsequently, several criteria have been proposed, including the 2013 ISHAM-AWG criteria . The group suggested modifying the existing ISHAM-AWG criteria. In both rounds, consensus could not be reached (LoC: 48.7% and 53.8%). Most experts felt that the criteria must be simple and allow identification and differentiation of ABPA and ABPM. Finally, after achieving consensus, we recommend separate criteria for diagnosing ABPA and ABPM ( and ). The diagnosis of ABPA/M should be suspected in patients with predisposing conditions or a compatible clinico-radiological presentation (expectoration of mucus plugs, fleeting opacities on chest imaging, finger-in-glove opacities and lung collapse). Thus, the revised criteria include a compatible presentation to enable diagnoses of ABPA/M in those without predisposing conditions . Additionally, two components are essential to make a diagnosis. The first is to document sensitisation against the implicated fungus (using fungus-specific IgE), while the other is to demonstrate immunological activity (raised serum total IgE). However, these two tests can also be positive in patients with fungal sensitisation without ABPA. Here, besides the essential components, the presence of other features, including fungal-specific IgG, peripheral blood eosinophilia and consistent imaging, confirms the diagnosis of ABPA/M. Importantly, the presence of HAM on chest CT is pathognomonic and diagnoses ABPA/M, even when a few other criteria components are missing . We have added another radiological finding, namely “fleeting opacities consistent with ABPA” on chest radiographs, given its high specificity for diagnosing ABPA . In most ABPA/M patients, serum total IgE is ≥500 IU·mL −1 . Uncommonly, serum total IgE could be <500 IU·mL −1 despite the presence of all other components. Low serum total IgE can be seen in those with prior glucocorticoid treatment , the elderly or when the patient has constitutively low IgE before developing ABPA . Also, any range only covers 95% of the population and all individuals will not meet a specific cut-off . If available, IgE against rAsp antigens (f1, f2 and f4) may be used to diagnose ABPA. While investigating a patient for ABPA, we recommend performing A. fumigatus -specific IgE . If the value is ≥0.35 kUA·L −1 , serum total IgE levels should be measured. If the value is ≥500 IU·mL −1 , other tests for ABPA, including A. fumigatus -specific IgG, peripheral blood eosinophil count, chest CT and lung function tests, should be done to characterise the disease . The basic framework for diagnosing ABPM is similar to ABPA, with a few differences . ABPM should be considered in patients with possible ABPA, but A. fumigatus -specific IgE is <0.35 kUA·L −1 . ABPM can be suspected when a causative fungus is isolated in at least two sputum culture specimens or bronchoalveolar lavage fluid culture. ABPM is then confirmed by demonstrating allergic sensitisation (skin test or fungus-specific IgE), combined with a raised serum total IgE and consistent radiological features . Unfortunately, commercial assays for detecting IgE and IgG against fungi other than Aspergillus spp. are available only for a few species ( Alternaria , Cladosporium , Candida , Mucor , Trichophyton and Penicillium ). For other fungi, including S. commune , Bipolaris and others, in-house assays are required for detecting IgE and IgG. A skin test or immunoprecipitation would be required when fungus-specific IgE or IgG is unavailable. There is also a high probability of misclassifying ABPA as ABPM if IgE and IgG against Aspergillus spp. are performed using non-standardised assays. The rest of the workup for ABPM is similar to ABPA . Notably, the absence of elevated IgE against rAsp f1, f2 and f4 strongly supports the diagnosis of ABPM over ABPA in a patient with allergic pulmonary mycoses . In settings where fungus-specific serology is not available, ABPM may be pragmatically diagnosed if there is repeated and consistent culture growth, serum total IgE ≥500 IU·mL −1 , peripheral blood eosinophilia and radiological features of ABPM, provided the Aspergillus -specific serology is negative. The differential diagnosis of ABPA/M is broad and caution is advised in making the diagnosis in patients without either asthma or CF. A. fumigatus -specific IgE and IgG can be elevated in COPD, pulmonary tuberculosis and bronchiectasis, and some of these patients can develop ABPA. Patients with chronic pulmonary aspergillosis may have raised serum A. fumigatus -IgE and total IgE in addition to A. fumigatus -IgG . Aspergillus (and fungal) bronchitis is associated with at least two positive respiratory samples yielding the same fungus and may be associated with a raised A. fumigatus -IgG and bronchiectasis, but without fulfilling the diagnostic criteria for ABPA/M. Patients with severe asthma may be sensitised to A. fumigatus or multiple other fungi with a raised total IgE. They are classified as severe asthma with fungal sensitisation under the umbrella of fungal asthma when they do not fulfil the ABPA/M criteria. Patients with ABPA may also have an additional underlying aetiology for bronchiectasis . Therefore, a search for other causes of bronchiectasis (immunodeficiencies, ciliary disorders and mycobacterial infection) is prudent . The diagnostic workup of bronchiectasis includes complete blood count, A. fumigatus -specific IgE, sweat chloride test, immunoglobulin levels and mycobacterial cultures from sputum . If the initial workup is negative, whole-exome sequencing can be performed (to identify aetiologies such as primary ciliary dyskinesias, primary immunodeficiency and atypical CF), especially in those with extensive bronchiectasis and recurrent infections since childhood.
A clinical framework for classifying ABPA is essential due to the chronic relapsing nature of the illness and the propensity for developing severe complications. Also, an objective treatment response criterion is useful for monitoring therapy during routine care and in clinical trials. The first classification attempt categorised ABPA into five stages . As the stages were imprecise, the ISHAM-AWG previously proposed a modified staging with more detailed definitions . However, there were several unresolved issues. Most importantly, the stages were labelled 0–6, but a patient does not necessarily progress from one to another. The previous classification also did not reflect progressive severity since stage 4 (remission) is a more stable clinical state than stage 3 (exacerbation), which is counterintuitive. To overcome these limitations, we proposed modifications that achieved consensus in the second round (LoC: 85.3%). In the new ABPA/M classification, we have removed the numbered stages and retained five categories: acute ABPA, response, remission, treatment-dependent ABPA and advanced ABPA . We have removed the asymptomatic stage and glucocorticoid-dependent asthma as they had no clear treatment implications in ABPA. Also, we have included newly diagnosed ABPA and exacerbation together as acute ABPA. To diagnose ABPA exacerbation, asthma or bronchiectasis (infective) exacerbations need to be excluded, and we provide definitions for the two entities in . Remission, as in asthma , is diagnosed when the patient has no asthma or ABPA exacerbations, is not dependent on oral glucocorticoid therapy and has the best possible lung function. Remission may be achieved spontaneously after treatment or with antifungal azoles or biological agents. Finally, advanced ABPA is defined in patients with extensive bronchiectasis and type 2 respiratory failure or secondary pulmonary hypertension .
CT of the thorax is crucial in diagnosing ABPA. However, due consideration should be given to the radiation dosage when CT scans are ordered, especially in children. Chest CT is also prognostic. For instance, the extent of bronchiectasis, HAM and any fungal ball are independent predictors of recurrent ABPA exacerbations . Central bronchiectasis (usually bilateral) is the predominant pattern seen in ABPA, although it is not uncommon to find both central and peripheral bronchiectasis . Isolated central bronchiectasis is encountered only in a few conditions, including ABPA and tracheobronchomegaly, and is thus a helpful distinguishing feature . Previously, Greenberger's group classified ABPA as ABPA with central bronchiectasis (ABPA-CB) or serological ABPA (ABPA-S) based on the presence or absence of bronchiectasis . Subsequently, K umar classified ABPA into three groups: ABPA-S, ABPA-CB and ABPA-CB with other radiological findings (ABPA-CB-ORF). In a study involving 234 patients, A garwal et al . categorised ABPA into ABPA-S (mild), ABPA-CB (moderate) and ABPA-CB-HAM (severe). Based on all the evidence, the ISHAM-AWG had previously classified ABPA radiologically into four categories: ABPA-S, ABPA with bronchiectasis (ABPA-B), ABPA-HAM and ABPA with chronic pleuropulmonary fibrosis (ABPA-CPF) . The term “bronchiectasis” (B) was used instead of “central bronchiectasis” (CB) as bronchiectasis in ABPA can extend to the periphery in up to 40% of the lobes . Mucus plugging without HAM is another common radiological finding in ABPA. Mucus plugs are consistently associated with eosinophilic inflammation and immunologically severe ABPA . The DECG discussed several radiological classifications. Finally, the scheme presented in achieved consensus (LoC: 88.2%). The new classification includes five classes: ABPA-S, ABPA-B, ABPA with mucus plugging (ABPA-MP), ABPA-HAM and ABPA-CPF . ABPA-S refers to patients of ABPA without bronchiectasis, while ABPA-B includes patients with bronchiectasis. ABPA-HAM has been retained, as HAM is an independent and pathognomonic diagnostic feature of ABPA . ABPA-MP includes patients with non-hyperattenuating mucus plugs. Patients with bronchiectasis and mucus plugging are labelled as ABPA-MP, given the greater immunological severity in those with mucus plugging. Other radiological findings frequently observed in ABPA include centrilobular nodules (with a tree-in-bud appearance), atelectasis, mosaic attenuation and consolidation. These findings can be seen in isolation or with ABPA-B, ABPA-MP and ABPA-HAM. In patients with ABPA-CPF, a vital consideration is the exclusion of chronic pulmonary aspergillosis .
The principles of treating ABPA involve using anti-inflammatory agents (glucocorticoids or biological agents targeting type 2 immune response) to control immune responses or antifungal agents to decrease airway fungal colonisation. The treatment goals are symptom relief, improving asthma control, preventing asthma and ABPA exacerbations, abrogating bronchiectasis progression, and minimising therapy-related adverse events. The treatment principles of ABPM are like ABPA, except that the implicated fungus guides the choice of antifungal drugs. The DECG reviewed the RCTs and the therapies available for treating ABPA patients ( and ) . Initiating treatment for newly diagnosed ABPA Patients with acute ABPA require treatment with systemic therapies. Glucocorticoids are the most effective treatment for acute ABPA . An RCT involving 92 ABPA patients compared two glucocorticoid dosing protocols (low dose (prednisolone 0.5 mg·kg −1 ·day −1 for 2 weeks, then on alternate days for 8 weeks; then tapered by 5 mg every 2 weeks and discontinued after 3–5 months) versus high dose (prednisolone 0.75 and 0.5 mg·kg −1 ·day −1 for 6 weeks each; subsequently, tapered by 5 mg every 6 weeks and discontinued after 8–10 months)). The frequency of ABPA exacerbations was similar in the two groups and the lower dose resulted in fewer adverse events. However, there was a lower clinico-radiological and immunological response at 6 weeks with the lower dose . Several centres use doses intermediate between the low and high doses (prednisolone 0.5, 0.25 and 0.125 mg·kg −1 ·day −1 for 4 weeks each, then tapered by 5 mg every 2 weeks till discontinuation). The DECG recommended using a 4-month course of low-to-moderate dose oral prednisolone (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) for acute ABPA . Care should be taken while using methylprednisolone because when combined with oral itraconazole, there is a higher risk of exogenous Cushing's syndrome and adrenal insufficiency . Notably, the experts suggested the need for trials with even shorter duration of glucocorticoids, as the 4-month duration was derived from the need to randomise against longer-term azole therapy. Many clinicians administer an initial 2-week course of glucocorticoids in those started on an oral azole, and as symptoms are controlled, transition to high-dose inhaled corticosteroids (ICS). Importantly, a combination of inhaled budesonide or fluticasone and itraconazole can also cause exogenous Cushing's syndrome . While asymptomatic ABPA patients do not routinely require systemic therapy, the treatment decision needs to be individualised. For instance, patients can have well-controlled asthma on high-dose inhaled steroids and may benefit from treatment of underlying ABPA, especially if the chest CT shows bronchiectasis or mucus plugging. Also, asymptomatic patients with prolonged mucus plugging can progress to irreversible bronchiectasis. Thus, optimisation of asthma treatment and close observation with a clinical review, chest radiograph and serum total IgE every 3–6 months is required if a decision is made not to treat patients with asymptomatic ABPA. Oral antifungal triazoles, especially itraconazole, have similar effects as glucocorticoids but a slower trajectory to improvement and a better safety profile than glucocorticoids . Although the evidence was limited to a single RCT, the DECG recommended using oral itraconazole (for 4 months) as an alternative initial therapy for acute ABPA, given the considerable clinical experience with itraconazole. While voriconazole has similar efficacy as glucocorticoids for treating acute ABPA , the experts expressed concerns with its use as first-line therapy due to poorer patient tolerance. Also, prednisolone decreases the plasma concentration of voriconazole in a dose-dependent fashion . A recent RCT of 191 patients found no significant reduction in ABPA exacerbations with a combination of prednisolone and itraconazole compared to prednisolone alone and a propensity for higher adverse events . However, patients with blood eosinophil count ≥1000 cells·µL −1 and extensive bronchiectasis (≥10 segments) had a reduced 1-year exacerbation rate with the combination therapy . Similarly, there is little evidence for posaconazole, isavuconazole or a combination of newer azoles and glucocorticoids as the initial therapy for ABPA. There are no data supporting the use of biological agents as first-line therapy in ABPA. Adjunctive vitamin D supplementation was not helpful in managing ABPA , but vitamin D deficiency should be corrected as it aggravates osteopenia due to long-term glucocorticoid usage. High doses of ICS alone do not achieve immunological control or reduce exacerbations when used as therapy for ABPA . The efficacy of nebulised amphotericin B in acute ABPA is also poor . Notably, most studies on ABPA therapy have included mainly patients with ABPA-B, with little data on ABPA-S . Most experts do not routinely treat ABPA-S with systemic ABPA-specific treatment. Instead, patients with ABPA-S are managed like asthma. However, oral glucocorticoids or azoles may be required in those with poor asthma control or recurrent exacerbations despite optimal asthma management. Recommendations We do not recommend treating asymptomatic ABPA patients with systemic therapy (LoC: 85.7%). ABPA-S should be treated with systemic therapy only if there is poor asthma control (LoC: 79.4%) or recurrent exacerbations despite asthma therapy (LoC: 85.3%). We recommend a low-to-moderate dose (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) of oral prednisolone (LoC: 78.1%) or oral itraconazole (LoC: 73.5%) for 4 months as the initial therapy for treating acute ABPA. We recommend oral itraconazole as the initial therapy where systemic glucocorticoids are contraindicated (LoC: 84.6%). We do not recommend using a combination of itraconazole and glucocorticoids as first-line therapy for acute ABPA (LoC: 71.9%). However, a short course of glucocorticoids (<2 weeks) may be used as initial therapy along with oral itraconazole. Oral voriconazole, posaconazole and isavuconazole should not be used as first-line agents for treating acute ABPA (LoC: 78.1–96.9%). They may be used if there are contraindications to systemic glucocorticoids and intolerance, failure or resistance to itraconazole therapy (LoC: 12.8–64.1%). High-dose ICS should not be used as primary therapy for acute ABPA (LoC: 100%). We do not recommend using biological agents as first-line therapy for acute ABPA (LoC: 96.9%). Treatment of ABPA exacerbation After treatment cessation, almost 50% of patients experience ABPA exacerbations . In a patient with ABPA, worsening of respiratory symptoms may occur due to asthma exacerbation, immunologically exacerbated ABPA and infective exacerbation of bronchiectasis, apart from unrelated causes. The three types of exacerbations can usually be differentiated using chest radiographs, serum total IgE and sputum bacterial cultures. Occasionally, there can be an overlap of more than one cause of exacerbation in an individual patient. ABPA exacerbations are characterised by sustained worsening (≥2 weeks) of clinical symptoms or the appearance of new infiltrates on chest imaging, along with an increase in serum total IgE by ≥50% above the “new baseline” IgE (during clinical stability) . Asthma exacerbations are not associated with increased serum total IgE or new infiltrates on chest imaging and should be managed with a short course of oral glucocorticoids. Bronchiectasis (infective) exacerbations are diagnosed with clinical worsening without elevation in serum total IgE ≥50% compared to baseline. Sputum cultures frequently show bacterial growth in bronchiectasis exacerbations. There are no RCTs for managing ABPA exacerbations and the options include using prednisolone, itraconazole or their combination . In clinical practice, ABPA exacerbation is managed like newly diagnosed ABPA using either prednisolone or itraconazole. Many experts use a combination of oral prednisolone and oral itraconazole in patients with recurrent exacerbations (≥2 in the last 1–2 years), especially in those with extensive bronchiectasis. Also, nebulised amphotericin B has poor efficacy for treating ABPA exacerbations . Pulse doses of methylprednisolone have been used for ABPA exacerbations refractory to oral glucocorticoids . Recommendations We recommend treating acute ABPA exacerbations in the same manner as newly diagnosed ABPA (LoC: 100%). A combination of oral prednisolone and itraconazole should be used for treating recurrent (≥2 in the last 1–2 years) ABPA exacerbations (LoC: 71.4%). We do not recommend using biological agents (LoC: 94.3%) or nebulised amphotericin B (LoC: 100%) for treating acute ABPA exacerbations. Monitoring treatment response After treatment initiation, patients should be monitored for response after 8–12 weeks using clinical symptoms, serum total IgE and chest radiographs . Instead of subjective assessment, the DECG suggested that symptom monitoring be done using a semiquantitative Likert scale as no improvement (or worsening), mild (<25% of baseline), moderate (25–50% of baseline) or significant improvement (>50% of baseline) for routine clinical care. The experts suggested using a more quantitative scale like a visual analogue scale (VAS) for clinical trials. A good response is indicated by a significant improvement in symptoms (Likert score or VAS ≥50%) and imaging , along with at least a 20% reduction in the serum total IgE levels . Spirometry may be used to monitor treatment response. A recent study reported the minimal clinically important difference (MCID) of forced expiratory volume in 1 s of 158 mL for treatment response in ABPA . Quality-of-life questionnaires are cumbersome and were discouraged unanimously by experts for routine patient care . However, for clinical trials, quality-of-life questionnaires could be used. An important consideration is that the MCID for the quality-of-life questionnaires could differ in ABPA from asthma. For instance, in one study, the MCID of the St George's Respiratory Questionnaire was 8 points, rather than the 4 points used in asthma . The serum A. fumigatus -specific IgE and IgG levels do not consistently fall following treatment . No studies have evaluated blood eosinophil count or concentrations of IgE/IgG against rAsp antigens for assessing treatment response. While using oral azoles, therapeutic drug monitoring is recommended initially at 2-week and 3-month intervals or during clinical worsening. Significant variations in the bioavailability of different itraconazole preparations compromise response . Likewise, drug–drug interactions are frequent with itraconazole, with common issues being ICS exposure boosting and undetectable levels of itraconazole with rifampicin . The recommended minimum therapeutic levels are >0.5, >1 and >1 µg·mL −1 for itraconazole, voriconazole and posaconazole, respectively . Monitoring for treatment-related adverse events is paramount. When using systemic glucocorticoids, the DECG recommended monitoring plasma glucose, blood pressure, body weight and mental status at the least. Monitoring liver function is necessary for patients receiving oral azoles . Recommendations We recommend assessing the initial treatment response after 8–12 weeks using a combination of clinical, immunological and imaging findings (LoC: 100%). A good response is indicated by a major improvement in symptoms (Likert score or VAS ≥50%) and chest radiographs, along with at least a 20% reduction in serum total IgE. A. fumigatus -specific IgE and IgG (LoC: 100%), peripheral blood eosinophil count (LoC: 73.6%), and IgE against recombinant antigens are not recommended for response assessment (LoC: 100%). We recommend therapeutic drug monitoring while using antifungal azoles (LoC: 91.4%). Management of treatment-dependent ABPA Almost 10–25% of ABPA patients become treatment-dependent . Two RCTs (84 patients) have evaluated the efficacy of itraconazole in treatment-dependent ABPA . Itraconazole reduced the oral glucocorticoid dose, sputum eosinophil count and ABPA exacerbations . A significant limitation was that neither study reported outcomes beyond 8 months regarding ABPA exacerbations. In the last two decades, several trials of monoclonal antibodies against IgE, interleukin (IL)-5, IL-5 receptor (IL-5R), IL-4 receptor α (IL-4Rα) and thymic stromal lymphopoietin (TSLP) have demonstrated clinical benefit in severe eosinophilic asthma . Biological agents are likely helpful in stable treatment-dependent ABPA based on case reports and small case series . Omalizumab is a potential therapeutic approach since ABPA is associated with elevated IgE levels. Mepolizumab (anti-IL-5), benralizumab (anti-IL-5R), dupilumab (anti-IL-4Rα) and tezepelumab (anti-TSLP) have also been used in ABPA patients . Most experience of biological agents in ABPA is with omalizumab. Using omalizumab in ABPA led to improvement in symptoms, reduction in exacerbations and asthma hospitalisations, improvement in lung function, and reduction in the dose of oral steroids . In the only crossover RCT of 13 patients, omalizumab use ( versus placebo) was associated with less frequent exacerbations and decreased basophil reactivity to A. fumigatus . Biological agents should also be considered for maintenance therapy for underlying asthma . Finally, continuous low-dose glucocorticoids should be the last option in managing treatment-dependent ABPA. Recommendation Long-term itraconazole (100%), nebulised amphotericin B (LoC: 100%) or biological agents (LoC: 71%) are recommended options for managing treatment-dependent ABPA. Management of ABPA in remission During remission (stable disease), ABPA patients should be managed for underlying asthma and bronchiectasis (ICS and long-acting bronchodilators, nebulised saline, antibiotics, and others) per the existing guidelines . Patients should be monitored with clinical review, serum total IgE levels and lung function test every 3–6 months for the first year and then every 6–12 months. Remission can be prolonged by using long-term itraconazole, nebulised amphotericin B (LoC: 100%) and biological agents (LoC: 71%), especially in those with treatment-dependent ABPA. Two RCTs have also evaluated the role of nebulised amphotericin B as maintenance to prevent future ABPA exacerbations . In both studies, patients with ABPA exacerbation were treated for 4 months with oral glucocorticoids or prednisolone and itraconazole. After that, they were randomised to receive nebulised amphotericin B versus placebo. In the smaller pilot study (21 patients), using nebulised amphotericin B deoxycholate (10 mg twice daily, three times a week) reduced ABPA exacerbations at 1 year . In the larger NEBULAMB study (139 patients), while the primary outcome was not significant, the time-to-first exacerbation was significantly longer with nebulised liposomal amphotericin B (25 mg weekly) compared to the control group . Recommendations We recommend that patients in remission be managed for underlying asthma and bronchiectasis per the existing guidelines (LoC: 100%). For patients achieving remission with antifungal azoles or biological agents, we recommend periodic assessments to determine the need for these therapies (LoC: 100%). Management of extensive bronchiectasis and advanced ABPA There are no specific studies on advanced ABPA . Nebulised hypertonic saline (3–7%, 4–5 mL) can reduce sputum viscosity and ease the expectoration of mucus plugs in bronchiectasis patients . Treatment should be preceded by nebulised salbutamol to minimise the risk of bronchospasm. Also, the first dose of nebulised hypertonic saline should be administered under supervision . Nebulised antibiotics and long-term azithromycin therapy can improve outcomes in ABPA patients with bronchiectasis and frequent infective exacerbations . Caution is advised when using azithromycin in those receiving itraconazole, as it can cause QTc prolongation. In some ABPA patients, widespread bronchiectasis can eventually lead to chronic type 2 respiratory failure and pulmonary hypertension . The DECG recommended extrapolating guidance from other pulmonary disorders (chronic obstructive lung disease, interstitial lung disease and bronchiectasis) for long-term oxygen therapy (LTOT), vaccination and lung transplantation . LTOT in those with resting hypoxaemia (arterial oxygen tension ( P aO 2 ) ≤55 mmHg) reduces pulmonary hypertension and improves survival in patients with chronic obstructive lung disease. There is no role for LTOT in mild hypoxaemia ( P aO 2 >55 mmHg at rest) or nocturnal oxygen desaturation . In one study, patients with chronic and allergic aspergillosis responded poorly to the 23-valent pneumococcal polysaccharide vaccine compared to healthy adults . Thus, influenza or pneumococcal vaccine should be administered before initiating glucocorticoid therapy or delayed until the underlying condition is better controlled ( e.g. after treatment with antifungals or glucocorticoids). The International Society for Heart and Lung Transplantation criteria should be used to refer and list patients with advanced ABPA . Recommendation We recommend using standard guidelines from other pulmonary disorders for LTOT, vaccination and lung transplantation (LoC: 100%).
Patients with acute ABPA require treatment with systemic therapies. Glucocorticoids are the most effective treatment for acute ABPA . An RCT involving 92 ABPA patients compared two glucocorticoid dosing protocols (low dose (prednisolone 0.5 mg·kg −1 ·day −1 for 2 weeks, then on alternate days for 8 weeks; then tapered by 5 mg every 2 weeks and discontinued after 3–5 months) versus high dose (prednisolone 0.75 and 0.5 mg·kg −1 ·day −1 for 6 weeks each; subsequently, tapered by 5 mg every 6 weeks and discontinued after 8–10 months)). The frequency of ABPA exacerbations was similar in the two groups and the lower dose resulted in fewer adverse events. However, there was a lower clinico-radiological and immunological response at 6 weeks with the lower dose . Several centres use doses intermediate between the low and high doses (prednisolone 0.5, 0.25 and 0.125 mg·kg −1 ·day −1 for 4 weeks each, then tapered by 5 mg every 2 weeks till discontinuation). The DECG recommended using a 4-month course of low-to-moderate dose oral prednisolone (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) for acute ABPA . Care should be taken while using methylprednisolone because when combined with oral itraconazole, there is a higher risk of exogenous Cushing's syndrome and adrenal insufficiency . Notably, the experts suggested the need for trials with even shorter duration of glucocorticoids, as the 4-month duration was derived from the need to randomise against longer-term azole therapy. Many clinicians administer an initial 2-week course of glucocorticoids in those started on an oral azole, and as symptoms are controlled, transition to high-dose inhaled corticosteroids (ICS). Importantly, a combination of inhaled budesonide or fluticasone and itraconazole can also cause exogenous Cushing's syndrome . While asymptomatic ABPA patients do not routinely require systemic therapy, the treatment decision needs to be individualised. For instance, patients can have well-controlled asthma on high-dose inhaled steroids and may benefit from treatment of underlying ABPA, especially if the chest CT shows bronchiectasis or mucus plugging. Also, asymptomatic patients with prolonged mucus plugging can progress to irreversible bronchiectasis. Thus, optimisation of asthma treatment and close observation with a clinical review, chest radiograph and serum total IgE every 3–6 months is required if a decision is made not to treat patients with asymptomatic ABPA. Oral antifungal triazoles, especially itraconazole, have similar effects as glucocorticoids but a slower trajectory to improvement and a better safety profile than glucocorticoids . Although the evidence was limited to a single RCT, the DECG recommended using oral itraconazole (for 4 months) as an alternative initial therapy for acute ABPA, given the considerable clinical experience with itraconazole. While voriconazole has similar efficacy as glucocorticoids for treating acute ABPA , the experts expressed concerns with its use as first-line therapy due to poorer patient tolerance. Also, prednisolone decreases the plasma concentration of voriconazole in a dose-dependent fashion . A recent RCT of 191 patients found no significant reduction in ABPA exacerbations with a combination of prednisolone and itraconazole compared to prednisolone alone and a propensity for higher adverse events . However, patients with blood eosinophil count ≥1000 cells·µL −1 and extensive bronchiectasis (≥10 segments) had a reduced 1-year exacerbation rate with the combination therapy . Similarly, there is little evidence for posaconazole, isavuconazole or a combination of newer azoles and glucocorticoids as the initial therapy for ABPA. There are no data supporting the use of biological agents as first-line therapy in ABPA. Adjunctive vitamin D supplementation was not helpful in managing ABPA , but vitamin D deficiency should be corrected as it aggravates osteopenia due to long-term glucocorticoid usage. High doses of ICS alone do not achieve immunological control or reduce exacerbations when used as therapy for ABPA . The efficacy of nebulised amphotericin B in acute ABPA is also poor . Notably, most studies on ABPA therapy have included mainly patients with ABPA-B, with little data on ABPA-S . Most experts do not routinely treat ABPA-S with systemic ABPA-specific treatment. Instead, patients with ABPA-S are managed like asthma. However, oral glucocorticoids or azoles may be required in those with poor asthma control or recurrent exacerbations despite optimal asthma management. Recommendations We do not recommend treating asymptomatic ABPA patients with systemic therapy (LoC: 85.7%). ABPA-S should be treated with systemic therapy only if there is poor asthma control (LoC: 79.4%) or recurrent exacerbations despite asthma therapy (LoC: 85.3%). We recommend a low-to-moderate dose (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) of oral prednisolone (LoC: 78.1%) or oral itraconazole (LoC: 73.5%) for 4 months as the initial therapy for treating acute ABPA. We recommend oral itraconazole as the initial therapy where systemic glucocorticoids are contraindicated (LoC: 84.6%). We do not recommend using a combination of itraconazole and glucocorticoids as first-line therapy for acute ABPA (LoC: 71.9%). However, a short course of glucocorticoids (<2 weeks) may be used as initial therapy along with oral itraconazole. Oral voriconazole, posaconazole and isavuconazole should not be used as first-line agents for treating acute ABPA (LoC: 78.1–96.9%). They may be used if there are contraindications to systemic glucocorticoids and intolerance, failure or resistance to itraconazole therapy (LoC: 12.8–64.1%). High-dose ICS should not be used as primary therapy for acute ABPA (LoC: 100%). We do not recommend using biological agents as first-line therapy for acute ABPA (LoC: 96.9%).
We do not recommend treating asymptomatic ABPA patients with systemic therapy (LoC: 85.7%). ABPA-S should be treated with systemic therapy only if there is poor asthma control (LoC: 79.4%) or recurrent exacerbations despite asthma therapy (LoC: 85.3%). We recommend a low-to-moderate dose (0.5 mg·kg −1 ·day −1 for 2–4 weeks, tapered and completed over 4 months) of oral prednisolone (LoC: 78.1%) or oral itraconazole (LoC: 73.5%) for 4 months as the initial therapy for treating acute ABPA. We recommend oral itraconazole as the initial therapy where systemic glucocorticoids are contraindicated (LoC: 84.6%). We do not recommend using a combination of itraconazole and glucocorticoids as first-line therapy for acute ABPA (LoC: 71.9%). However, a short course of glucocorticoids (<2 weeks) may be used as initial therapy along with oral itraconazole. Oral voriconazole, posaconazole and isavuconazole should not be used as first-line agents for treating acute ABPA (LoC: 78.1–96.9%). They may be used if there are contraindications to systemic glucocorticoids and intolerance, failure or resistance to itraconazole therapy (LoC: 12.8–64.1%). High-dose ICS should not be used as primary therapy for acute ABPA (LoC: 100%). We do not recommend using biological agents as first-line therapy for acute ABPA (LoC: 96.9%).
After treatment cessation, almost 50% of patients experience ABPA exacerbations . In a patient with ABPA, worsening of respiratory symptoms may occur due to asthma exacerbation, immunologically exacerbated ABPA and infective exacerbation of bronchiectasis, apart from unrelated causes. The three types of exacerbations can usually be differentiated using chest radiographs, serum total IgE and sputum bacterial cultures. Occasionally, there can be an overlap of more than one cause of exacerbation in an individual patient. ABPA exacerbations are characterised by sustained worsening (≥2 weeks) of clinical symptoms or the appearance of new infiltrates on chest imaging, along with an increase in serum total IgE by ≥50% above the “new baseline” IgE (during clinical stability) . Asthma exacerbations are not associated with increased serum total IgE or new infiltrates on chest imaging and should be managed with a short course of oral glucocorticoids. Bronchiectasis (infective) exacerbations are diagnosed with clinical worsening without elevation in serum total IgE ≥50% compared to baseline. Sputum cultures frequently show bacterial growth in bronchiectasis exacerbations. There are no RCTs for managing ABPA exacerbations and the options include using prednisolone, itraconazole or their combination . In clinical practice, ABPA exacerbation is managed like newly diagnosed ABPA using either prednisolone or itraconazole. Many experts use a combination of oral prednisolone and oral itraconazole in patients with recurrent exacerbations (≥2 in the last 1–2 years), especially in those with extensive bronchiectasis. Also, nebulised amphotericin B has poor efficacy for treating ABPA exacerbations . Pulse doses of methylprednisolone have been used for ABPA exacerbations refractory to oral glucocorticoids . Recommendations We recommend treating acute ABPA exacerbations in the same manner as newly diagnosed ABPA (LoC: 100%). A combination of oral prednisolone and itraconazole should be used for treating recurrent (≥2 in the last 1–2 years) ABPA exacerbations (LoC: 71.4%). We do not recommend using biological agents (LoC: 94.3%) or nebulised amphotericin B (LoC: 100%) for treating acute ABPA exacerbations.
We recommend treating acute ABPA exacerbations in the same manner as newly diagnosed ABPA (LoC: 100%). A combination of oral prednisolone and itraconazole should be used for treating recurrent (≥2 in the last 1–2 years) ABPA exacerbations (LoC: 71.4%). We do not recommend using biological agents (LoC: 94.3%) or nebulised amphotericin B (LoC: 100%) for treating acute ABPA exacerbations.
After treatment initiation, patients should be monitored for response after 8–12 weeks using clinical symptoms, serum total IgE and chest radiographs . Instead of subjective assessment, the DECG suggested that symptom monitoring be done using a semiquantitative Likert scale as no improvement (or worsening), mild (<25% of baseline), moderate (25–50% of baseline) or significant improvement (>50% of baseline) for routine clinical care. The experts suggested using a more quantitative scale like a visual analogue scale (VAS) for clinical trials. A good response is indicated by a significant improvement in symptoms (Likert score or VAS ≥50%) and imaging , along with at least a 20% reduction in the serum total IgE levels . Spirometry may be used to monitor treatment response. A recent study reported the minimal clinically important difference (MCID) of forced expiratory volume in 1 s of 158 mL for treatment response in ABPA . Quality-of-life questionnaires are cumbersome and were discouraged unanimously by experts for routine patient care . However, for clinical trials, quality-of-life questionnaires could be used. An important consideration is that the MCID for the quality-of-life questionnaires could differ in ABPA from asthma. For instance, in one study, the MCID of the St George's Respiratory Questionnaire was 8 points, rather than the 4 points used in asthma . The serum A. fumigatus -specific IgE and IgG levels do not consistently fall following treatment . No studies have evaluated blood eosinophil count or concentrations of IgE/IgG against rAsp antigens for assessing treatment response. While using oral azoles, therapeutic drug monitoring is recommended initially at 2-week and 3-month intervals or during clinical worsening. Significant variations in the bioavailability of different itraconazole preparations compromise response . Likewise, drug–drug interactions are frequent with itraconazole, with common issues being ICS exposure boosting and undetectable levels of itraconazole with rifampicin . The recommended minimum therapeutic levels are >0.5, >1 and >1 µg·mL −1 for itraconazole, voriconazole and posaconazole, respectively . Monitoring for treatment-related adverse events is paramount. When using systemic glucocorticoids, the DECG recommended monitoring plasma glucose, blood pressure, body weight and mental status at the least. Monitoring liver function is necessary for patients receiving oral azoles . Recommendations We recommend assessing the initial treatment response after 8–12 weeks using a combination of clinical, immunological and imaging findings (LoC: 100%). A good response is indicated by a major improvement in symptoms (Likert score or VAS ≥50%) and chest radiographs, along with at least a 20% reduction in serum total IgE. A. fumigatus -specific IgE and IgG (LoC: 100%), peripheral blood eosinophil count (LoC: 73.6%), and IgE against recombinant antigens are not recommended for response assessment (LoC: 100%). We recommend therapeutic drug monitoring while using antifungal azoles (LoC: 91.4%).
We recommend assessing the initial treatment response after 8–12 weeks using a combination of clinical, immunological and imaging findings (LoC: 100%). A good response is indicated by a major improvement in symptoms (Likert score or VAS ≥50%) and chest radiographs, along with at least a 20% reduction in serum total IgE. A. fumigatus -specific IgE and IgG (LoC: 100%), peripheral blood eosinophil count (LoC: 73.6%), and IgE against recombinant antigens are not recommended for response assessment (LoC: 100%). We recommend therapeutic drug monitoring while using antifungal azoles (LoC: 91.4%).
Almost 10–25% of ABPA patients become treatment-dependent . Two RCTs (84 patients) have evaluated the efficacy of itraconazole in treatment-dependent ABPA . Itraconazole reduced the oral glucocorticoid dose, sputum eosinophil count and ABPA exacerbations . A significant limitation was that neither study reported outcomes beyond 8 months regarding ABPA exacerbations. In the last two decades, several trials of monoclonal antibodies against IgE, interleukin (IL)-5, IL-5 receptor (IL-5R), IL-4 receptor α (IL-4Rα) and thymic stromal lymphopoietin (TSLP) have demonstrated clinical benefit in severe eosinophilic asthma . Biological agents are likely helpful in stable treatment-dependent ABPA based on case reports and small case series . Omalizumab is a potential therapeutic approach since ABPA is associated with elevated IgE levels. Mepolizumab (anti-IL-5), benralizumab (anti-IL-5R), dupilumab (anti-IL-4Rα) and tezepelumab (anti-TSLP) have also been used in ABPA patients . Most experience of biological agents in ABPA is with omalizumab. Using omalizumab in ABPA led to improvement in symptoms, reduction in exacerbations and asthma hospitalisations, improvement in lung function, and reduction in the dose of oral steroids . In the only crossover RCT of 13 patients, omalizumab use ( versus placebo) was associated with less frequent exacerbations and decreased basophil reactivity to A. fumigatus . Biological agents should also be considered for maintenance therapy for underlying asthma . Finally, continuous low-dose glucocorticoids should be the last option in managing treatment-dependent ABPA. Recommendation Long-term itraconazole (100%), nebulised amphotericin B (LoC: 100%) or biological agents (LoC: 71%) are recommended options for managing treatment-dependent ABPA.
Long-term itraconazole (100%), nebulised amphotericin B (LoC: 100%) or biological agents (LoC: 71%) are recommended options for managing treatment-dependent ABPA.
During remission (stable disease), ABPA patients should be managed for underlying asthma and bronchiectasis (ICS and long-acting bronchodilators, nebulised saline, antibiotics, and others) per the existing guidelines . Patients should be monitored with clinical review, serum total IgE levels and lung function test every 3–6 months for the first year and then every 6–12 months. Remission can be prolonged by using long-term itraconazole, nebulised amphotericin B (LoC: 100%) and biological agents (LoC: 71%), especially in those with treatment-dependent ABPA. Two RCTs have also evaluated the role of nebulised amphotericin B as maintenance to prevent future ABPA exacerbations . In both studies, patients with ABPA exacerbation were treated for 4 months with oral glucocorticoids or prednisolone and itraconazole. After that, they were randomised to receive nebulised amphotericin B versus placebo. In the smaller pilot study (21 patients), using nebulised amphotericin B deoxycholate (10 mg twice daily, three times a week) reduced ABPA exacerbations at 1 year . In the larger NEBULAMB study (139 patients), while the primary outcome was not significant, the time-to-first exacerbation was significantly longer with nebulised liposomal amphotericin B (25 mg weekly) compared to the control group . Recommendations We recommend that patients in remission be managed for underlying asthma and bronchiectasis per the existing guidelines (LoC: 100%). For patients achieving remission with antifungal azoles or biological agents, we recommend periodic assessments to determine the need for these therapies (LoC: 100%).
We recommend that patients in remission be managed for underlying asthma and bronchiectasis per the existing guidelines (LoC: 100%). For patients achieving remission with antifungal azoles or biological agents, we recommend periodic assessments to determine the need for these therapies (LoC: 100%).
There are no specific studies on advanced ABPA . Nebulised hypertonic saline (3–7%, 4–5 mL) can reduce sputum viscosity and ease the expectoration of mucus plugs in bronchiectasis patients . Treatment should be preceded by nebulised salbutamol to minimise the risk of bronchospasm. Also, the first dose of nebulised hypertonic saline should be administered under supervision . Nebulised antibiotics and long-term azithromycin therapy can improve outcomes in ABPA patients with bronchiectasis and frequent infective exacerbations . Caution is advised when using azithromycin in those receiving itraconazole, as it can cause QTc prolongation. In some ABPA patients, widespread bronchiectasis can eventually lead to chronic type 2 respiratory failure and pulmonary hypertension . The DECG recommended extrapolating guidance from other pulmonary disorders (chronic obstructive lung disease, interstitial lung disease and bronchiectasis) for long-term oxygen therapy (LTOT), vaccination and lung transplantation . LTOT in those with resting hypoxaemia (arterial oxygen tension ( P aO 2 ) ≤55 mmHg) reduces pulmonary hypertension and improves survival in patients with chronic obstructive lung disease. There is no role for LTOT in mild hypoxaemia ( P aO 2 >55 mmHg at rest) or nocturnal oxygen desaturation . In one study, patients with chronic and allergic aspergillosis responded poorly to the 23-valent pneumococcal polysaccharide vaccine compared to healthy adults . Thus, influenza or pneumococcal vaccine should be administered before initiating glucocorticoid therapy or delayed until the underlying condition is better controlled ( e.g. after treatment with antifungals or glucocorticoids). The International Society for Heart and Lung Transplantation criteria should be used to refer and list patients with advanced ABPA . Recommendation We recommend using standard guidelines from other pulmonary disorders for LTOT, vaccination and lung transplantation (LoC: 100%).
We recommend using standard guidelines from other pulmonary disorders for LTOT, vaccination and lung transplantation (LoC: 100%).
ABPA during pregnancy The principles of managing ABPA during pregnancy are to optimise asthma and ABPA control while preventing harm to the fetus. Most ABPA trials have excluded pregnant patients and data are extrapolated. Oral glucocorticoids for the duration and dose required in ABPA are safe in pregnancy , while the use of oral itraconazole is associated with a higher risk of premature births and abortions . Experience with biological agents in ABPA during pregnancy is limited. When used with oral glucocorticoids, omalizumab increases the risk of pre-term delivery . In a systematic review of biological agent use during pregnancy with atopic disease, major fetal defects, low birthweight, pre-term delivery and stillbirths were reported . Recommendations We recommended using systemic glucocorticoids (in the same doses as for non-pregnant) for managing acute-stage or treatment-dependent ABPA in pregnancy (LoC: 73.0%). We recommend avoiding the use of biological agents (LoC: 86.5%) or oral azoles (LoC: 100%) for managing acute-stage or treatment-dependent ABPA in pregnancy. ABPA in CF There are no RCTs of treatment in CF-ABPA . Glucocorticoids or azoles are the preferred initial agents. Glucocorticoids can induce diabetes mellitus, with grave consequences in CF. Monthly doses of intravenous methylprednisolone therapy alone or with azoles have been used to limit the toxicity associated with daily glucocorticoids . The combination of glucocorticoids with azoles is also increasingly being used in CF-ABPA. In one survey, combination therapy was preferred by most respondents for treating newly diagnosed ABPA and first exacerbation . A 3-week course of oral corticosteroids combined with oral itraconazole for 12 months effectively treated CF-ABPA . While using itraconazole capsules, therapeutic drug monitoring is essential as the drug is poorly absorbed in CF. The super bioavailable itraconazole formulation was well-tolerated and achieved therapeutic response in 81% of patients with CF-ABPA . Voriconazole leads to photosensitivity and should be used with caution. Posaconazole achieved therapeutic drug levels and good clinical response in CF-ABPA . While a previous review suggested that omalizumab might benefit patients with ABPA , a recent case series found no benefit with omalizumab . Like in asthma, pulmonary exacerbations in a patient with CF-ABPA could be related to ABPA or pulmonary infection . Recommendation We recommend diagnosing and treating CF-ABPA using the same treatment tenets outlined for ABPA in asthma; however, exercising due caution for issues specific to CF (LoC: 100%). ABPA in children Growth retardation with the use of systemic corticosteroids and the erratic bioavailability of azoles due to age-related changes in cytochrome P450 enzymes are of primary concern . There are no randomised trials on the treatment of ABPA in children . Omalizumab and mepolizumab are approved for asthma therapy in the age group ≥6 years , while other biological agents can be used in those ≥12 years old. The DECG recommended using systemic glucocorticoids or oral itraconazole for treating acute ABPA in children as in adults. The experts felt more research was needed to address the therapeutic issues affecting ABPA in children. Recommendation We recommend treating ABPA in children using the same treatment principles outlined for ABPA in asthmatic adults; however, exercising due caution for issues specific to children (LoC: 100%).
The principles of managing ABPA during pregnancy are to optimise asthma and ABPA control while preventing harm to the fetus. Most ABPA trials have excluded pregnant patients and data are extrapolated. Oral glucocorticoids for the duration and dose required in ABPA are safe in pregnancy , while the use of oral itraconazole is associated with a higher risk of premature births and abortions . Experience with biological agents in ABPA during pregnancy is limited. When used with oral glucocorticoids, omalizumab increases the risk of pre-term delivery . In a systematic review of biological agent use during pregnancy with atopic disease, major fetal defects, low birthweight, pre-term delivery and stillbirths were reported . Recommendations We recommended using systemic glucocorticoids (in the same doses as for non-pregnant) for managing acute-stage or treatment-dependent ABPA in pregnancy (LoC: 73.0%). We recommend avoiding the use of biological agents (LoC: 86.5%) or oral azoles (LoC: 100%) for managing acute-stage or treatment-dependent ABPA in pregnancy.
We recommended using systemic glucocorticoids (in the same doses as for non-pregnant) for managing acute-stage or treatment-dependent ABPA in pregnancy (LoC: 73.0%). We recommend avoiding the use of biological agents (LoC: 86.5%) or oral azoles (LoC: 100%) for managing acute-stage or treatment-dependent ABPA in pregnancy.
There are no RCTs of treatment in CF-ABPA . Glucocorticoids or azoles are the preferred initial agents. Glucocorticoids can induce diabetes mellitus, with grave consequences in CF. Monthly doses of intravenous methylprednisolone therapy alone or with azoles have been used to limit the toxicity associated with daily glucocorticoids . The combination of glucocorticoids with azoles is also increasingly being used in CF-ABPA. In one survey, combination therapy was preferred by most respondents for treating newly diagnosed ABPA and first exacerbation . A 3-week course of oral corticosteroids combined with oral itraconazole for 12 months effectively treated CF-ABPA . While using itraconazole capsules, therapeutic drug monitoring is essential as the drug is poorly absorbed in CF. The super bioavailable itraconazole formulation was well-tolerated and achieved therapeutic response in 81% of patients with CF-ABPA . Voriconazole leads to photosensitivity and should be used with caution. Posaconazole achieved therapeutic drug levels and good clinical response in CF-ABPA . While a previous review suggested that omalizumab might benefit patients with ABPA , a recent case series found no benefit with omalizumab . Like in asthma, pulmonary exacerbations in a patient with CF-ABPA could be related to ABPA or pulmonary infection . Recommendation We recommend diagnosing and treating CF-ABPA using the same treatment tenets outlined for ABPA in asthma; however, exercising due caution for issues specific to CF (LoC: 100%).
We recommend diagnosing and treating CF-ABPA using the same treatment tenets outlined for ABPA in asthma; however, exercising due caution for issues specific to CF (LoC: 100%).
Growth retardation with the use of systemic corticosteroids and the erratic bioavailability of azoles due to age-related changes in cytochrome P450 enzymes are of primary concern . There are no randomised trials on the treatment of ABPA in children . Omalizumab and mepolizumab are approved for asthma therapy in the age group ≥6 years , while other biological agents can be used in those ≥12 years old. The DECG recommended using systemic glucocorticoids or oral itraconazole for treating acute ABPA in children as in adults. The experts felt more research was needed to address the therapeutic issues affecting ABPA in children. Recommendation We recommend treating ABPA in children using the same treatment principles outlined for ABPA in asthmatic adults; however, exercising due caution for issues specific to children (LoC: 100%).
We recommend treating ABPA in children using the same treatment principles outlined for ABPA in asthmatic adults; however, exercising due caution for issues specific to children (LoC: 100%).
Aspergillus is a common mould found in various environments, including soil, decaying vegetation and indoor environments. Evidence suggests that environmental exposures to Aspergillus spores could drive exacerbations . The DECG, therefore, suggested minimising activities that could result in the inhalation of large numbers of Aspergillus conidia . If such activities are unavoidable, surgical masks or the more effective N95 respirators may minimise spore inhalation. Further, regular cleaning and maintenance of heating, ventilation and air conditioning (HVAC) systems may prevent mould growth . Identifying and promptly addressing water leaks is essential, as damp environments promote mould growth. Similarly, ensuring proper ventilation in bathrooms, kitchens and other areas prone to moisture accumulation may prevent mould growth . Cleaning and dusting living spaces should be performed regularly using damp cloths to minimise the accumulation and dispersion of spores into the air. Indoor air quality monitoring represents a valuable tool to evaluate the fungal exposome .
Our guidelines have a few limitations. We used a modified Delphi methodology and the third round was non-anonymised. However, this was by design, as we wanted a consensus on most questions, especially the diagnostic criteria. A consensus might not have been achieved if we followed the conventional Delphi for the third round. We tried to ensure the representation of experts from various disciplines. However, nearly half of the experts were from India. One can argue that the agreement for several questions was due to a significant representation from a few countries (India and the UK). However, this was not the case, as we can note that we did not reach even 50% consensus for nearly half of the questions circulated in the first round of Delphi ( supplementary table S2 ), including the most critical aspects of the guidelines (the diagnostic criteria and therapeutics), thus indicating a significant heterogeneity in practice. Finally, most recommendations were consensus-based rather than evidence-based due to the lack of evidence in many areas. However, the Delphi process is ideally suited for such situations.
More studies are needed on the community prevalence of fungal sensitisation and ABPA/M to evaluate the need for screening asthmatic subjects for ABPA in primary or secondary care. The prevalent fungus in ABPM must also be evaluated in various geographic locations. The pathogenesis of ABPA needs to be better understood, such as research on airway mucus biology, genetic predisposition and detailed elucidation of the host–pathogen interaction. In the diagnosis of ABPA, most studies have used the FEIA platform to perform immunological investigations. Hence, more studies using different platforms and their comparisons are needed. Increasingly, newer automated platforms are being introduced and would require fresh performance evaluations. ABPM is under-recognised because of a lack of standardised commercial assays, which need to be developed. The practical application of component-resolved diagnostics requires the development of diagnostic cut-offs and validation from various countries to enable their use in routine care. While we have proposed guidance for differentiating asthma, bronchiectasis and ABPA exacerbations, this may not always be possible. More research is required to evaluate the role of rAsp antigens, sputum Aspergillus PCR and sputum eosinophil count to differentiate between these entities accurately ( supplementary figure S1 ). The treatment of ABPA/M requires detailed evaluation. For instance, the radiological categorisation of ABPA has no clear therapeutic implication. However, future trials should investigate personalised therapy for ABPA, depending on the specific imaging subgroups. Randomised trials are necessary to define the role of various biological agents (as maintenance therapy for glucocorticoid-dependent ABPA and acute ABPA), inhaled antifungals (for acute ABPA as well as for maintenance therapy) and safer glucocorticoids in different patient populations, including CF, children and others. Future studies must also address indoor air quality control using air purifiers, dehumidifiers and others to reduce fungal airway colonisation and improve ABPA outcomes.
We present Delphi consensus guidelines on diagnosing, classifying and treating allergic bronchopulmonary mycoses from the ISHAM-AWG. These guidelines will help bring uniformity in diagnosis and simplify the management of ABPA patients in both clinical care and research.
10.1183/13993003.00061-2024.Supp1 Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author. Supplementary material ERJ-00061-2024.Supplement
10.1183/13993003.00061-2024.Shareable1 This one-page PDF can be shared freely online. Shareable PDF ERJ-00061-2024.Shareable
|
Inter-specialty collaboration in the formalization of a new foregut subspecialty | 1afddbb7-869f-4566-b580-9fec3415fe8d | 8718094 | Internal Medicine[mh] | Diseases of the foregut represent a large and growing burden in the United States (US) and a significant source of controversy among treating providers. The term “foregut” is typically used by clinicians to reference the beginning portions of the gastrointestinal tract, including the esophagus, stomach, and first portion of the small intestines. For example, one such disease, gastroesophageal reflux disease (GERD), affects an estimated 10–20% of the adult population in the US . Persistent foregut complaints and disease progression result in millions of visits annually to primary care providers, gastroenterologists, surgeons, and emergency rooms in the US , representing significant cost to the system . Such disease trends have spurred significant innovation in this disease space over the past decade by both venture capitalists and large medical device companies . Consequently, providers can select from a variety of changing tools in their management of foregut diseases and must acquire new skills to keep pace with the market. New technologies include advanced diagnostic therapies that detect early disease progression, videoendoscopic therapies that mitigate surgical intervention, and minimally invasive surgical therapies to alleviate advanced symptoms or cure malignant disease . Both medical and surgical societies publish patient management guidelines at regular intervals , while expert panels and working groups release their own consensus statements on clinical algorithms in various journals ; not all are in concordance. In the spring of 2018, the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) established a new Foregut Task Force to “determine standards to optimize patient safety” in the management and treatment of foregut diseases . In 2019, the same society published The SAGES Manual of Foregut Surgery , with an online introduction describing its comprehensive review and expert commentary aimed “to clarify controversies in the field” . Concurrently, a new multi-disciplinary society, the American Foregut Society (AFS), emerged, recruiting membership throughout 2018 and hosting its inaugural meeting in March 2019 . The advocacy for this new, multidisciplinary specialty society coincides with other noteworthy provider trends, such as the advent of a new “foregut” fellowship designation and the implementation of esophageal and foregut disease service lines, or “Centers of Excellence” in hospitals . These evolving provider and organizational trends have afforded a unique opportunity to examine the development of the new foregut surgical specialty parallel to its emergence. Understanding the context, provider attitudes, as well as historical social patterns may allow healthcare administrators to influence subspecialty progress toward provider collaboration rather than competition. Professional competition over medical turf, procedures, and specialties has been described in the sociology literature for decades, with each new theory reflecting the evolution of healthcare during that period . While many of the early theories focus on the client-professional relationship and common characteristics that define a professional, theories have since shifted toward understanding the motivations of professionals that might result in conflict . Eliot Freidson, a key influencer in this movement, describes the “privileged position” of professionals, including physicians, which contributes to competition over the rewards associated with such status . Freidson contends that physicians traditionally exert collegial control via collective, self-regulation because of the level of autonomy expected of professional decision-making, and in so doing, they ensure their continued professional dominance . Critics of Freidson’s theories point out that physicians have become increasingly beholden to bureaucratic processes, such as credentialing or employment by hospitals, because of the expansion of capitalism in the medical profession . When examining the profiles of professionals in the hospital, organizational theorists describe the profile of physicians as competitive, occasionally to the point of sabotage, which can be reinforced by training systems that emphasize individual performance and by the social reward of prestige . Competition over disease space ownership has historically resulted in the emergence of new, regulated specialty identities, as in the case of obstetrics , or persistent conflict between competing specialists, as with gastroenterologists and surgeons . The dynamic relationship between the latter was particularly exacerbated by gastroenterologists’ development and champion of the videoendoscope, which disrupted the traditional workflow and dominant position previously held by surgeons . Sudden innovation can trigger intra-specialty and inter-specialty turf warring by causing a professions system disturbance , which then leads to formal attempts to establish control over certain tasks . In such times of heightened conflict, emerging specialists use claims of superiority in knowledge, experience, or outcomes to legitimize their movement toward a position of dominance . Structural position in the workflow is a critical factor in determining the outcomes of jurisdictional disputes, or attempts to control certain tasks, procedures, or anatomy . Referral-dependent professionals, like surgeons, may have trouble exerting regulative or normative control over their profession ; in such cases, timing can be the deciding factor for turf war outcomes . Based on our understanding of these theories, we hypothesized that the quickly evolving advancements in the foregut disease space are driving a faction of surgeons to seek regulative control over the foregut subspecialty. Our primary research question was as follows: What individual provider attitudes and organizational behaviors are contributing to the formalization of a foregut sub-specialty?
Study design This was a qualitative, cross-sectional study consisting of interviews, archival and survey data, and non-participant observation, intending to explore the provider attitudes, behaviors, and market conditions that are contributing to the creation of a new foregut subspecialty. The primary author, a clinician with experience in the subject matter, led all interviews. Her background includes work as a certified physician assistant in general and bariatric surgery, a Master of Business Administration with healthcare emphasis, and consultant for reflux and endosurgery device manufacturers. Due to her role in the medical device industry working with hospital administrators, the primary author had a preexisting collegial relationship with many of the interviewees; they were formally notified of the purpose of the research as related to graduate level studies in healthcare management and pursuit of publication of these findings, independent of her professional employment. The second author is an Associate Professor of Healthcare Administration at Suffolk University with a PhD in Health Services and Policy Analysis from UC-Berkeley. He served as the first author’s independent study preceptor, overseeing development of the interview guide, participating in limited data collection, reviewing all transcripts, and supporting data analysis and manuscript development. Participants were healthcare professionals purposively selected based on their self-identification as esophageal or foregut specialists, known participation in an esophageal or foregut focused practice, or direct professional involvement with such providers in a healthcare setting. They were approached face-to-face at a medical conference or via email script by the lead author. Interview data was coupled with field notes from observation of the inaugural AFS conference, archival review of working documents, and descriptive analysis of data from the first AFS member survey, made available to the researchers as a secondary data source by AFS ( ). Semi-structured interviews were conducted with 30 professionals from October 2018 to June 2019, ceasing at theoretical saturation ( and ). Interviews followed an interview guide ( ) and lasted 15–35 minutes each, with no repeat interviews required and only researchers and participants present. Interviews were audio-recorded and transcribed verbatim; they were not returned for comment or correction to participants. Research procedures were approved by the Institutional Review Board at Suffolk University (Protocol #1295907–2), and informed written consent was obtained from all participants. No interviewees withdrew after providing written consent to participate, and participant information was anonymized. Data analysis Content analysis of the qualitative interview data was performed using N*Vivo v12. With the help of two research assistants, the authors generated broad codes based on the social theory of professions and independently assigned them to a subset of five transcripts. This subset was then closely reviewed to reconcile coding assignments, generate focused codes, and develop supplementary codes from emergent concepts ( ). Coding of the full set of transcripts was performed by the lead author, and a second subset was fully coded by both authors to reduce bias, demonstrating high interrater reliability (Cohen’s kappa = 0.8487). Themes in the coded transcripts were developed and compared with observational field notes and archival data; participants were not approached for feedback on the findings.
This was a qualitative, cross-sectional study consisting of interviews, archival and survey data, and non-participant observation, intending to explore the provider attitudes, behaviors, and market conditions that are contributing to the creation of a new foregut subspecialty. The primary author, a clinician with experience in the subject matter, led all interviews. Her background includes work as a certified physician assistant in general and bariatric surgery, a Master of Business Administration with healthcare emphasis, and consultant for reflux and endosurgery device manufacturers. Due to her role in the medical device industry working with hospital administrators, the primary author had a preexisting collegial relationship with many of the interviewees; they were formally notified of the purpose of the research as related to graduate level studies in healthcare management and pursuit of publication of these findings, independent of her professional employment. The second author is an Associate Professor of Healthcare Administration at Suffolk University with a PhD in Health Services and Policy Analysis from UC-Berkeley. He served as the first author’s independent study preceptor, overseeing development of the interview guide, participating in limited data collection, reviewing all transcripts, and supporting data analysis and manuscript development. Participants were healthcare professionals purposively selected based on their self-identification as esophageal or foregut specialists, known participation in an esophageal or foregut focused practice, or direct professional involvement with such providers in a healthcare setting. They were approached face-to-face at a medical conference or via email script by the lead author. Interview data was coupled with field notes from observation of the inaugural AFS conference, archival review of working documents, and descriptive analysis of data from the first AFS member survey, made available to the researchers as a secondary data source by AFS ( ). Semi-structured interviews were conducted with 30 professionals from October 2018 to June 2019, ceasing at theoretical saturation ( and ). Interviews followed an interview guide ( ) and lasted 15–35 minutes each, with no repeat interviews required and only researchers and participants present. Interviews were audio-recorded and transcribed verbatim; they were not returned for comment or correction to participants. Research procedures were approved by the Institutional Review Board at Suffolk University (Protocol #1295907–2), and informed written consent was obtained from all participants. No interviewees withdrew after providing written consent to participate, and participant information was anonymized.
Content analysis of the qualitative interview data was performed using N*Vivo v12. With the help of two research assistants, the authors generated broad codes based on the social theory of professions and independently assigned them to a subset of five transcripts. This subset was then closely reviewed to reconcile coding assignments, generate focused codes, and develop supplementary codes from emergent concepts ( ). Coding of the full set of transcripts was performed by the lead author, and a second subset was fully coded by both authors to reduce bias, demonstrating high interrater reliability (Cohen’s kappa = 0.8487). Themes in the coded transcripts were developed and compared with observational field notes and archival data; participants were not approached for feedback on the findings.
The surgeon interviewees articulate a professional need to qualify foregut expertise based on superior knowledge and outcomes, to foster collaboration across specialties, and to define the allocation of certain tasks and procedures in this field. Non-surgeon participants confirm these themes in their descriptions of surgeon colleagues and evolution of the space. Current specific market conditions are discussed as precipitating individual and organizational decisions to pursue formal specialization. Claims of superiority: Qualifying expertise The respondents suggest that only a small number of general, thoracic, and bariatric surgeons receive specific training in the diagnosis and treatment of esophageal disease and have the surgical volumes to back their claim of superior experience. Every procedure is technically demanding. It requires precision.–R7 It’s not something I just do once a month, so I think volumes speak volumes in this case.–R9 Similarly, one faction of gastroenterologists is noted to have particular skill in physiologic testing and interventional therapy for esophageal disease. It requires dedication to learn how to interpret the testing. Right? It’s not something you can read about and then the following day you’re good to go.–R6 They have additional training in the specialty trainings, diagnostics… they’re committed to the disease in a way that, that propels us forward, I think.–R19 Many respondents clarify that completion of a minimally invasive surgery (MIS) fellowship does not necessarily equate to foregut expertise. In the AFS membership survey, 52% of respondents believed that greater than half of practice volume should be devoted to foregut surgery to “really be a specialist” ( ). Conversely, the SAGES proposed criteria for Advanced GI/MIS Fellowship requires only 20 foregut cases during fellowship , yet a published survey of MIS trainees reveals that 52% of MIS fellows will identify themselves as “foregut surgeons” upon graduation . Second to specialized knowledge, surgeons and gastroenterologists emphasize better outcomes as evidence of superiority, suggesting a collective altruistic concern for patient welfare and drawing on the literature to back claims. I believe in this day of super-specialization, you can’t dabble. If you dabble and do less than a certain number of cases a year, your outcomes are not going to be as good.–R1 There is increasing evidence on the surgical side that foregut disease care is better if it’s specialized. And that not every Tom, Dick, or Harry ought to be doing this.–R2 Professional dominance: Protecting reputation In conjunction with superior outcome claims arise themes of reputation and restriction of professional labels. The outcomes have been poor, the complication rates have been high, and it has gotten foregut a very bad name. Both in the medical community as well as, more importantly, the patient community.–R4 Not surprisingly, many surgeons suggest the role of a new society to restrict the labeling of “Specialists” and “Centers of Excellence” to those who meet certain criteria. We’ll have to get to a point where if you want to be part of a national system of being recognized as being excellent, then you’re going to have to prove it and you’ll have to set up criteria for people to do so.–R10 In fact, the secondary goal listed in the mission statement of the American Foregut Society is “to foster research that will culminate in the development of benchmarks for excellence” . The majority of membership confirmed the necessity of credentialing criteria in the AFS survey ( ). Task structure: Altering workflow and dependency The dependence of the surgeons, both on referrals and expensive healthcare resources, is discussed in 60% of interviews, with one surgeon jokingly dubbing their position as “the end of the food chain” (R18). Their position within the workflow seems particularly dependent on the gastroenterologists, which at times affects which procedures surgeons are willing to offer. I’m competing against the people who are going to send me the reflux patients. So basically, I don’t scope them because they’ll scope them. Alright so in other words, don’t bite the hand that feeds you.–R8 This same respondent also suggests that despite his dependence, surgeons and gastroenterologists can structure a symbiotic collegial relationship benefiting both parties. This sentiment of a mutually beneficial working relationship is noted by 57% of respondents, who describe sharing patients, complementary services, “common goals”, and “collaboration” in developing their local program. Collaboration and its synonyms are used in over one third of the interviews. Collaboration is particularly noteworthy in connection with AFS, whose mission statement begins with the aim “to help guide both the diagnosis and management of Foregut disease through collaboration between Gastroenterologists and Foregut Surgeons” . The term was also used frequently at the inaugural AFS meeting by both gastroenterologist and surgeon speakers, along with the terms “align”, “consensus”, and “team”. Intentions to improve collegiality are noted frequently throughout the interviews. There’s a real attempt to… create a collegial environment between gastroenterology and general or foregut surgeons.–R12 While some respondents suggest leveraging the influence of other surgical societies in accomplishing goals, most draw distinction between the AFS and existing societies in its aim to influence the collaboration between medical and surgical foregut specialists. There isn’t any way that SAGES can bring the same type of focus to those particular problems that I just mentioned, as this society can.–R16 Task jurisdiction: Defining turf Despite the advocacy for collaboration, however, remnants of the historical inter-specialty turf war between surgeons and gastroenterologists persist, with more senior surgeons referencing it specifically. I believe that general surgery and gastroenterology have been slow to get out of silos… and there’s too much turf. Endoscopy is a great example.–R8 This historical turf war over endoscopy, ongoing disagreements in disease management, or professional bias are referenced in 70% of interviews, and the resulting conflicting guidelines between surgical and gastrointestinal societies is mentioned as an area to be addressed by a multispecialty society. They evaluate the diseases of the foregut, and address its treatment from a medical perspective, whereas we have a tendency to address it from a surgical perspective. Those are the relative tools that we have.–R16 Evidence of intra-specialty turf warring appears in the interviews as well, with references to the delineation of general surgeons versus foregut surgeons. Just because you’re a surgeon, just because you’re a gastroenterologist doesn’t mean you’re qualified to be a foregut surgeon and be a specialist in that field.–R12 Surgical turf warring is highlighted in discussions of societies, particularly when respondents are prompted to describe the role of SAGES compared to the new AFS. But SAGES still wants to consider foregut to be part of general surgery…they don’t promote it as a specialty. They can’t…because they represent general surgery.–R2 Ultimately, the conflict between specialties, within specialties, and among societies regards qualified ownership of tests or procedures in the continuum of care. Surgeons need to do this. But if they can’t get the experience, and that means accreditations, that means credentialing, that means then proving that they can, you know, safely perform these things in their hospitals throughout the United States.–R10 Professional interests and collegiality In qualifying their colleagues, the interviewed surgeons were quick to distinguish those who have the “interest” to specialize in esophageal or foregut care. In fact, “interest” and its synonyms are used in 63% of the interviews. I’ve always had an interest in anti-reflux operations and my practice sort of was headed that way in any case and so I decided to just stop performing general surgery and focusing on foregut.–R7 Particular “interest” is often discussed in conjunction with commitment to the disease space in some form, whether in procedural volume, research, or training; “dedication” is mentioned in 9/30 (30%) of interviews, often as a requirement for labeling oneself a foregut or esophageal specialist. A dedicated interest in expertise in adopting new technologies, pioneering new treatments in this area would be the main thing I would look for.–R26 Common interest appears to be a prompt for collegiality and partnerships in the creation of AFS and new Centers of Excellence. Results from the AFS survey indicate that membership highly rank “a forum to share thoughts among like-minded people” as “Critical” in the goals of AFS ( ). Common interest and a desire for collegiality is also evidenced in the success of the SAGES online social media forum dedicated to foregut surgery . Market timing Some economic drivers in the market, such as seeking insurance approval for new procedures, generating hospital revenue, and acquiring a volume monopoly over certain procedures, present as themes in the interviews. Financial concerns, however, are not discussed as elaborately as qualifying expertise when examining the needs of space; instead, these themes often arise only when specifically prompted by the interviewer. INT: Is, is the hospital invested in having those tests done? I mean, is it lucrative? RES: I think so. Uh, I don’t know the ins and outs of the cost and you know the, uh, reimbursement.–R4 Local market conditions appear to influence individual decisions to specialize or to open a center with special foregut dedication, often in combination with professional interest. And my focus has certainly over the last two years evolved from general surgery into primarily foregut specialization…but with the departure of [another surgeon], there was a huge patient volume that was in the midst of a workup and also needed follow up. So I inherited all of those patients.–R15 Recent advancements in diagnostic and therapeutic technologies are identified as the major catalysts for maturation of the space and drivers for specialization. “Technology” is referenced by 36% of interviewees. Four new therapeutic procedures are mentioned by name 67 times in the interviews; robotic surgery is also mentioned by five interviewed surgeons. You need specialization because there are more tools. Both in the surgical arena with different anti-reflux procedures.–R2 The described market conditions, as well as references to historical dynamics between surgeons and gastroenterologists, suggest a critical juncture in cross-disciplinary relations as these providers pursue the formation of this subspecialty.
The respondents suggest that only a small number of general, thoracic, and bariatric surgeons receive specific training in the diagnosis and treatment of esophageal disease and have the surgical volumes to back their claim of superior experience. Every procedure is technically demanding. It requires precision.–R7 It’s not something I just do once a month, so I think volumes speak volumes in this case.–R9 Similarly, one faction of gastroenterologists is noted to have particular skill in physiologic testing and interventional therapy for esophageal disease. It requires dedication to learn how to interpret the testing. Right? It’s not something you can read about and then the following day you’re good to go.–R6 They have additional training in the specialty trainings, diagnostics… they’re committed to the disease in a way that, that propels us forward, I think.–R19 Many respondents clarify that completion of a minimally invasive surgery (MIS) fellowship does not necessarily equate to foregut expertise. In the AFS membership survey, 52% of respondents believed that greater than half of practice volume should be devoted to foregut surgery to “really be a specialist” ( ). Conversely, the SAGES proposed criteria for Advanced GI/MIS Fellowship requires only 20 foregut cases during fellowship , yet a published survey of MIS trainees reveals that 52% of MIS fellows will identify themselves as “foregut surgeons” upon graduation . Second to specialized knowledge, surgeons and gastroenterologists emphasize better outcomes as evidence of superiority, suggesting a collective altruistic concern for patient welfare and drawing on the literature to back claims. I believe in this day of super-specialization, you can’t dabble. If you dabble and do less than a certain number of cases a year, your outcomes are not going to be as good.–R1 There is increasing evidence on the surgical side that foregut disease care is better if it’s specialized. And that not every Tom, Dick, or Harry ought to be doing this.–R2
In conjunction with superior outcome claims arise themes of reputation and restriction of professional labels. The outcomes have been poor, the complication rates have been high, and it has gotten foregut a very bad name. Both in the medical community as well as, more importantly, the patient community.–R4 Not surprisingly, many surgeons suggest the role of a new society to restrict the labeling of “Specialists” and “Centers of Excellence” to those who meet certain criteria. We’ll have to get to a point where if you want to be part of a national system of being recognized as being excellent, then you’re going to have to prove it and you’ll have to set up criteria for people to do so.–R10 In fact, the secondary goal listed in the mission statement of the American Foregut Society is “to foster research that will culminate in the development of benchmarks for excellence” . The majority of membership confirmed the necessity of credentialing criteria in the AFS survey ( ).
The dependence of the surgeons, both on referrals and expensive healthcare resources, is discussed in 60% of interviews, with one surgeon jokingly dubbing their position as “the end of the food chain” (R18). Their position within the workflow seems particularly dependent on the gastroenterologists, which at times affects which procedures surgeons are willing to offer. I’m competing against the people who are going to send me the reflux patients. So basically, I don’t scope them because they’ll scope them. Alright so in other words, don’t bite the hand that feeds you.–R8 This same respondent also suggests that despite his dependence, surgeons and gastroenterologists can structure a symbiotic collegial relationship benefiting both parties. This sentiment of a mutually beneficial working relationship is noted by 57% of respondents, who describe sharing patients, complementary services, “common goals”, and “collaboration” in developing their local program. Collaboration and its synonyms are used in over one third of the interviews. Collaboration is particularly noteworthy in connection with AFS, whose mission statement begins with the aim “to help guide both the diagnosis and management of Foregut disease through collaboration between Gastroenterologists and Foregut Surgeons” . The term was also used frequently at the inaugural AFS meeting by both gastroenterologist and surgeon speakers, along with the terms “align”, “consensus”, and “team”. Intentions to improve collegiality are noted frequently throughout the interviews. There’s a real attempt to… create a collegial environment between gastroenterology and general or foregut surgeons.–R12 While some respondents suggest leveraging the influence of other surgical societies in accomplishing goals, most draw distinction between the AFS and existing societies in its aim to influence the collaboration between medical and surgical foregut specialists. There isn’t any way that SAGES can bring the same type of focus to those particular problems that I just mentioned, as this society can.–R16
Despite the advocacy for collaboration, however, remnants of the historical inter-specialty turf war between surgeons and gastroenterologists persist, with more senior surgeons referencing it specifically. I believe that general surgery and gastroenterology have been slow to get out of silos… and there’s too much turf. Endoscopy is a great example.–R8 This historical turf war over endoscopy, ongoing disagreements in disease management, or professional bias are referenced in 70% of interviews, and the resulting conflicting guidelines between surgical and gastrointestinal societies is mentioned as an area to be addressed by a multispecialty society. They evaluate the diseases of the foregut, and address its treatment from a medical perspective, whereas we have a tendency to address it from a surgical perspective. Those are the relative tools that we have.–R16 Evidence of intra-specialty turf warring appears in the interviews as well, with references to the delineation of general surgeons versus foregut surgeons. Just because you’re a surgeon, just because you’re a gastroenterologist doesn’t mean you’re qualified to be a foregut surgeon and be a specialist in that field.–R12 Surgical turf warring is highlighted in discussions of societies, particularly when respondents are prompted to describe the role of SAGES compared to the new AFS. But SAGES still wants to consider foregut to be part of general surgery…they don’t promote it as a specialty. They can’t…because they represent general surgery.–R2 Ultimately, the conflict between specialties, within specialties, and among societies regards qualified ownership of tests or procedures in the continuum of care. Surgeons need to do this. But if they can’t get the experience, and that means accreditations, that means credentialing, that means then proving that they can, you know, safely perform these things in their hospitals throughout the United States.–R10
In qualifying their colleagues, the interviewed surgeons were quick to distinguish those who have the “interest” to specialize in esophageal or foregut care. In fact, “interest” and its synonyms are used in 63% of the interviews. I’ve always had an interest in anti-reflux operations and my practice sort of was headed that way in any case and so I decided to just stop performing general surgery and focusing on foregut.–R7 Particular “interest” is often discussed in conjunction with commitment to the disease space in some form, whether in procedural volume, research, or training; “dedication” is mentioned in 9/30 (30%) of interviews, often as a requirement for labeling oneself a foregut or esophageal specialist. A dedicated interest in expertise in adopting new technologies, pioneering new treatments in this area would be the main thing I would look for.–R26 Common interest appears to be a prompt for collegiality and partnerships in the creation of AFS and new Centers of Excellence. Results from the AFS survey indicate that membership highly rank “a forum to share thoughts among like-minded people” as “Critical” in the goals of AFS ( ). Common interest and a desire for collegiality is also evidenced in the success of the SAGES online social media forum dedicated to foregut surgery .
Some economic drivers in the market, such as seeking insurance approval for new procedures, generating hospital revenue, and acquiring a volume monopoly over certain procedures, present as themes in the interviews. Financial concerns, however, are not discussed as elaborately as qualifying expertise when examining the needs of space; instead, these themes often arise only when specifically prompted by the interviewer. INT: Is, is the hospital invested in having those tests done? I mean, is it lucrative? RES: I think so. Uh, I don’t know the ins and outs of the cost and you know the, uh, reimbursement.–R4 Local market conditions appear to influence individual decisions to specialize or to open a center with special foregut dedication, often in combination with professional interest. And my focus has certainly over the last two years evolved from general surgery into primarily foregut specialization…but with the departure of [another surgeon], there was a huge patient volume that was in the midst of a workup and also needed follow up. So I inherited all of those patients.–R15 Recent advancements in diagnostic and therapeutic technologies are identified as the major catalysts for maturation of the space and drivers for specialization. “Technology” is referenced by 36% of interviewees. Four new therapeutic procedures are mentioned by name 67 times in the interviews; robotic surgery is also mentioned by five interviewed surgeons. You need specialization because there are more tools. Both in the surgical arena with different anti-reflux procedures.–R2 The described market conditions, as well as references to historical dynamics between surgeons and gastroenterologists, suggest a critical juncture in cross-disciplinary relations as these providers pursue the formation of this subspecialty.
The results of this study indicate that converging market conditions have created volatility among professions, leading to formal jurisdictional claims by a cohort of physicians. Based on the themes analyzed, the surgeon specialists aim to protect the reputation of and to exert professional dominance over their evolving field by collaborating with a select group of their traditional competitors, with whom they identify as having common interests. Precipitous innovation in this disease space seems to motivate the self-described specialists to seek exclusive control over certain procedures, thereby enhancing their claims to superiority in this arena. This method of professional differentiation based on unique knowledge is a concept described by Andrew Abbott as “cognitive abstraction” in his discussion of professional competition . The secondary emphasis on superior outcomes suggests a concern over the dilution of reputation that could occur with poor patient outcomes, harkening to Zetka’s description of the “delegitimation threat” in the early days of the surgical laparoscope . Promotion of improved outcomes also implies a collective concern for the patient’s well-being, perhaps an effort to fulfill their tacit professional contract with society while protecting professional reputation . Using superiority claims, these professionals seek formal, regulative distinction between foregut specialists and others. Intra-disciplinary competition, evidenced by use of terms like “dabblers” and references to inadequate training, seems a significant impetus for the creation of a new specialty society. The results from the AFS survey indicate that most members want a professional society to outline foregut credentials, perhaps a preemptive move to self-regulate and maintain collective autonomy ( ) . Legal definition or differentiation of foregut expertise will be difficult to establish on the collective rhetoric of superior knowledge alone . Well-established surgical societies may resist formalization of foregut credentials under the abstract claim of superior knowledge, as evidenced in the society tensions alluded to by respondents. Zetka’s theories suggest that soliciting support of other organizations, other professionals, or the patients themselves will be more efficient than claims of superiority in establishing turf dominance. Thus, “bridging” their movement with organizations like SAGES and the American College of Surgeons may be required to legitimize the efforts of the new society and its constituents . The self-described foregut surgeons are, however, deploying bridging strategies with gastroenterologists in effort to exert some control over their environment . While it is possible that some surgeons have learned to espouse the importance of collaboration to curry favor with colleagues, the behaviors manifested by these interviewees, such as partnership in the creation of Centers of Excellence and inclusion of gastroenterologists in a new society development, suggest a legitimate interest in promoting team-based care within their profession, albeit with some benefit to their dependent position. Unlike the jurisdictional contests typically discussed in the literature , foregut surgeons stake claim over specific surgical tasks while relinquishing jurisdiction over others in order to maintain positive relationships with a key referral source and traditional competitor. Given the surgeons’ structural dependence in the referral pathway, downstream from gastroenterologists, purposeful collaboration with gastroenterologists may prove more successful than claims to cognitive superiority in establishing dominance over the surgical component of the subspecialty, effectively cutting off referrals for general surgeons. Interestingly, while their structural dependence is readily apparent and admitted, foregut surgeons also claim collegiality and a symbiotically beneficial relationship with GIs. Propagating this rhetoric may enhance the more formal professional bridging strategies of including gastroenterologists in their newly forming society and in multi-specialty Centers of Excellence. The interviewed surgeons are quick to distinguish their gastroenterological colleagues who have foregut interest from those who do not, elevating them to a similar professional status within the boundaries of the foregut subspecialty. In the absence of a formal “foregut” specialty in prior years, and few programs devoted to specific training, many interviewed surgeons and gastroenterologists have chosen this area of focus through evolving, real-world preference; a desire to protect its reputation may foster comradery amongst traditionally competing medical and surgical providers, rather than overt antagonism. The creation of multi-specialty Centers of Excellence exists as a concrete, team-based goal relying on group rather than individual performance to realize its full value in the market. The study results demonstrate some incongruity between a desire for collaboration and the bias against non-surgical colleagues. Of note, surgeon representation at the inaugural AFS meeting was nearly three times greater than that of gastroenterologists. A recent article published in a gastrointestinal journal, co-authored by a gastroenterologist and a surgeon, discusses that though technological advancements in both specialties have begun to “erode the traditional turf” between these two parties, differences persist in protocols, which feed an underlying competitive spirit . Shared bias, ingrained by siloed physician training structures and decades of competition , may result in skepticism from gastroenterologists and threaten further recruitment to the cause. Consistent with theories discussed above, precipitous innovation in this disease space seems to motivate the self-described specialists to seek exclusive control over new procedures, thereby enhancing their prior claims to superior knowledge. Prior alimentary turf losses to gastroenterologists with the advent of the endoscope, the historical reputation of foregut surgery, and recent, rapid innovation converge to create a critical moment in the foregut subspecialty, in which surgeons and gastroenterologists can influence the outcome of this movement by choosing to collaborate and overcome professional differences. Strengths of the study include the duration and quality of interviews obtained, which resulted in detailed transcripts for interpretation. Limitations of the study included its cross-sectional design, which may have resulted in bias particular to the types of providers most willing to discuss evolution in the space. It was more difficult to elicit interviews from providers that have not chosen to join the emerging society to gain their perspective on the changes, though a few are represented in the data. The study is further limited by its US focus; there is currently an international discussion around specialization, as evidenced in the literature and by the recent inaugural meeting of the European Foregut Society Meeting in Vienna, Austria . A larger study collecting international data would reveal the potential global impact of these observations.
Self-identifying foregut surgeons collaborate across traditional competing specialties to establish professional dominance as sub-specialists in an increasingly focused US healthcare system. The manifested cross-disciplinary dialogue could advance quality improvement mechanisms and guideline alignment in this particular disease space nationally and internationally. Healthcare leaders pursuing the development of formal collaborative movements between these specialists, such as Esophageal Centers of Excellence or Heartburn Programs, should consider the social theory behind professional competition and historical outcomes of turf wars if they hope to achieve successful collaboration instead of further division between competing physicians.
S1 Appendix Interview guide. (DOCX) Click here for additional data file. S2 Appendix COREQ checklist. (PDF) Click here for additional data file.
|
Forensic Psychiatric Patients' Experiences of Personal Recovery | 7f93b2d7-5fb2-4bef-8179-f35c910803f3 | 11882176 | Forensic Medicine[mh] | In psychiatry in general, the recovery paradigm emphasizes patient-centered care and the uniqueness of each patient's personal recovery toward a meaningful life and well-being ( ; ). Recovery is defined as a personal process that encompasses changes to help the person enjoy a satisfying life and receive help from others (e.g., ; ). Recovery can be perceived in several different ways, for example, as clinical based, service based, and personal or client based ( ). However, it is common for forensic hospitals to strike a balance between patient empowerment and institutional control during the care process ( ; ; ). Forensic hospitals typically treat forensic patients, but in some countries, psychiatric patients who are dangerous and so severely mentally ill that they cannot be treated in general psychiatry may be treated in a forensic hospital ( ). In forensic hospitals, patients are detained in closed environments because of their mental illness and because they have offended or are at risk of offending ( ; ). Nurses need to balance between care and risk, and forensic patients need to deal with criminal responsibility, legal consequences, and self-determination ( ). The archetype of a forensic hospital traditionally serves as a place for “imposed” recovery, as it deprives patients of their freedom to choose their own path of recovery ( ). Lengthy treatment periods are typical for different kinds of patients in forensic hospitals ( ). In general, clinical, functional, social, and personal recovery can each be distinguished as a unique phenomenon. The concept of clinical recovery refers mostly to the mitigation of symptoms ( ) and adequate utilization of psychiatric care ( ; ). Functional recovery is related to improving performance in everyday activities, as assessed by both the patient and others ( ). Social recovery is connected to social inclusion ( ), which is especially challenging for forensic patients because of their criminal history, typical symptoms of long-term mental illness, long periods of hospitalization, and stigma associated with mental health disorders ( ). Personal recovery focuses on the patient's experiences of recovery as a journey toward a meaningful life and was therefore the concept chosen for this study ( ; ). The “CHIME” theoretical framework for recovery from mental illness was formulated over a decade ago. The acronym encompasses Connectedness, Hope and optimism about the future, Identity, Meaning in life, and Empowerment ( ). However, stated that the best results from CHIME can be obtained when it is applied to psychiatric care that meets certain contextual features; this need was the motivation for our study. We were interested in how the personal recovery of patients in forensic hospitals can be perceived when the care periods are long, the environment is highly restricted, and the patients are expected to be dangerous and are severely mentally ill ( ). In the latter part of the article, we suggest how nursing practice should consider the typical features of this patient population and care environment. Recently, the CHIME framework was expanded to the CHIME-S framework by to better fit the forensic context and highlight the importance of experiencing safety during forensic care. also studied forensic patients' recovery and provided an interpretation of the phenomenon. These researchers emphasized a personal sense of safety and linked this feature of care to experiences of a safe environment, facilities, and social interactions. They articulated this finding not as the core feature of recovery experienced by the patient but rather something that enables the recovery experience. Hope and social networks were presented as the second theme to express the importance of these factors during the recovery process. The third theme was work on identity, which describes how a patient's personal history (including both mental health disorder and criminal offenses) needs to be integrated into the recovering patient's life story for personal psychological development to take place. reported that forensic patients' perceptions of recovery and therapeutic measures are promoted through six themes: connectedness, sense of self, coming to terms with the past, freedom, hope, and health and intervention. The present article focuses on the concept of personal recovery experienced by forensic patients and aims to provide insight into the patient perspective. We considered the power imbalance between patients and staff in forensic psychiatry when formulating the methodology used and chose patients as the first source of information. Our methodological choice (concept analysis) indicates how we were interested in giving patients a voice, which is in line with the paradigm of recovery orientation. We also aimed at promoting nursing knowledge concerning care of forensic patients.
Concept Analysis A concept is the cognitive representation of a phenomenon that can be detected as it is ( , ) or by observing feelings, experiences, or other events connected to it ( ). Concept analysis can be used to examine a concept in the case that a consensus or clarification of the use and meaning of the chosen concept is needed ( ). This type of methodology is also beneficial for outlining how a certain concept is used in a new context or different field of science ( ; ; ). The use of a concept can modify the perceptions of the concept and understanding the phenomenon that it describes. In this sense, the concept and phenomenon that it describes are intricately intertwined ( ; ). The use of concepts varies in time ( , p. 184), and it affects our understanding and attitudes as well as behavior ( ) toward the phenomenon that the concept describes. Understanding what patients in forensic settings perceive as personal recovery promotes staff's capability of relating to the patient experience and mutual understanding during the care process. Evaluating the results of concept analysis assists theory development, understanding the elements of nursing, and the improvement of professional practice ( ; ; ). Concept analysis confronts the challenge of abstract definitions by distinguishing essential characteristics ( ) and constructing model cases ( ; ). The Wilson approach includes more phases than most other concept analysis methods, and some of the other phases differ from each other (e.g., ; ). work was intended to aid students to think more thoroughly about the features and traits of the phenomena that different concepts describe. The fact that Wilson did not intend to create a concept analysis method has been a source of criticism toward his work. Other methods have since been developed “to correct limitations” of Wilson's analysis process ( , p. 185). However, these methods have lost some of the nuances that made Wilson's analysis more profound ( ). The Wilson method links the concept to the context in a stronger way than, for example, method as it considers the situational factors and the environment (in Phase 8, Social Context, and Phase 9, Underlying Anxiety) and includes the practical results and results in language (Phases 10 and 11). In our study, the restrictive environment in forensic hospitals and forensic patients' history of mental illness as well as offending affect the recovery experiences of patients and cause underlying anxiety. point out that Wilson's method is dialogical, as the flow between different phases of analysis deepens the scope of results. In our analysis, we implemented this idea when we interpreted the meaning of different features from the perspective of the forensic patients. As such, the Wilsonian method of concept analysis ensured a deeper understanding of the phenomenon of personal recovery among forensic psychiatric patients. The different phases of the analysis performed in this study are typical of Wilson's method ( ) and are presented in more detail in Table . We later refer to these phases in the text as we describe the analysis process and report the results of our study. Data Collection The research question for our concept analysis can be defined as “What do the forensic psychiatric patients' experiences of recovery entail?” (Phase 1). We initially chose a preliminary definition of the concept (forensic psychiatric patients' experiences of personal recovery) and then collected data from peer-reviewed literature (Phase 1). The concept analysis aided us in defining the concept, and this definition could be compared with previous more general definitions of psychiatric recovery and definitions of forensic recovery. Searches were conducted in the CINAHL (EBSCO), PubMed (MEDLINE), and Social Science Premium Collection (ProQuest) databases. Search terms were as follows: in CINAHL, recovery OR rehabilitation OR healing AND forensic psychiatry OR forensic care OR high security; in PubMed, recovery AND forensic hospital; and in Social Science Premium Collection, forensic psychiatr* AND ti(recover*). Our search criteria for data in these databases were published in the English language, published in a peer-reviewed journal, original article, and the research having included a description of recovery from the perspective of a patient with severe mental illness in a forensic or high-security setting. We also included one book as a source of data because the contents focused on the chosen patient population and recovery, and we performed a manual search of the reference lists of identified relevant studies (see Figure ). We ensured all the references included in the data portrayed forensic hospital settings, patients with a history of offending or who had been labeled dangerous, and descriptions of recovery from the patient perspective. Data Analysis To increase the quality of the research, it was important to find a genuinely pure research question (concept) and avoid questions of value. During data collection, we needed to refocus our research question and simplify the concept we were analyzing (Phase 2). More specifically, we rejected definitions portraying what enables recovery or what recovery means from the perspective of staff or others. It is not expedient to list all possible features of the concept in this report. In our analysis, the essential features were found and grouped into themes and cases. We tolerated “hum noises” in the data, remembering the uniqueness of personal experiences, and some features were dismissed from the results because of their very low incidence. For example, in a book chapter in our data, the patient answered a question concerning their experience of recovery: “Right medication.” This unique feature can be understood rather as a helpful form of care than an essential feature of the personal recovery experience. We extracted the possible features of personal recovery of forensic patients from the original texts and gathered them into a list based on previous definitions by and . These definitions are more general descriptions of psychiatric recovery and were used to test whether our findings would fit the previous descriptions or differ from it. We constructed a phrase: “Recovery is 1) personal and unique , 2) a progressive journey with many stages and obstacles towards 3) a good life which is 4) more than absence of symptoms or overcoming the burden of disease and includes 5) citizenship and integration into society through 6) development of emotions and perceptions .” This proved to be an inaccurate way of organizing our findings, and we proceeded to create our own mode of categorization that would illuminate the forensic patients' experiences more accurately. We investigated the prevalence of different features identified in the chosen literature to define the typical nuances of forensic recovery experiences. We read through the list of features several times and, during the analysis process, found special features and unique meanings concerning forensic patients. Features resembling each other were grouped together until five themes entailed the typical nuances of personal recovery of patients in forensic care.
A concept is the cognitive representation of a phenomenon that can be detected as it is ( , ) or by observing feelings, experiences, or other events connected to it ( ). Concept analysis can be used to examine a concept in the case that a consensus or clarification of the use and meaning of the chosen concept is needed ( ). This type of methodology is also beneficial for outlining how a certain concept is used in a new context or different field of science ( ; ; ). The use of a concept can modify the perceptions of the concept and understanding the phenomenon that it describes. In this sense, the concept and phenomenon that it describes are intricately intertwined ( ; ). The use of concepts varies in time ( , p. 184), and it affects our understanding and attitudes as well as behavior ( ) toward the phenomenon that the concept describes. Understanding what patients in forensic settings perceive as personal recovery promotes staff's capability of relating to the patient experience and mutual understanding during the care process. Evaluating the results of concept analysis assists theory development, understanding the elements of nursing, and the improvement of professional practice ( ; ; ). Concept analysis confronts the challenge of abstract definitions by distinguishing essential characteristics ( ) and constructing model cases ( ; ). The Wilson approach includes more phases than most other concept analysis methods, and some of the other phases differ from each other (e.g., ; ). work was intended to aid students to think more thoroughly about the features and traits of the phenomena that different concepts describe. The fact that Wilson did not intend to create a concept analysis method has been a source of criticism toward his work. Other methods have since been developed “to correct limitations” of Wilson's analysis process ( , p. 185). However, these methods have lost some of the nuances that made Wilson's analysis more profound ( ). The Wilson method links the concept to the context in a stronger way than, for example, method as it considers the situational factors and the environment (in Phase 8, Social Context, and Phase 9, Underlying Anxiety) and includes the practical results and results in language (Phases 10 and 11). In our study, the restrictive environment in forensic hospitals and forensic patients' history of mental illness as well as offending affect the recovery experiences of patients and cause underlying anxiety. point out that Wilson's method is dialogical, as the flow between different phases of analysis deepens the scope of results. In our analysis, we implemented this idea when we interpreted the meaning of different features from the perspective of the forensic patients. As such, the Wilsonian method of concept analysis ensured a deeper understanding of the phenomenon of personal recovery among forensic psychiatric patients. The different phases of the analysis performed in this study are typical of Wilson's method ( ) and are presented in more detail in Table . We later refer to these phases in the text as we describe the analysis process and report the results of our study.
The research question for our concept analysis can be defined as “What do the forensic psychiatric patients' experiences of recovery entail?” (Phase 1). We initially chose a preliminary definition of the concept (forensic psychiatric patients' experiences of personal recovery) and then collected data from peer-reviewed literature (Phase 1). The concept analysis aided us in defining the concept, and this definition could be compared with previous more general definitions of psychiatric recovery and definitions of forensic recovery. Searches were conducted in the CINAHL (EBSCO), PubMed (MEDLINE), and Social Science Premium Collection (ProQuest) databases. Search terms were as follows: in CINAHL, recovery OR rehabilitation OR healing AND forensic psychiatry OR forensic care OR high security; in PubMed, recovery AND forensic hospital; and in Social Science Premium Collection, forensic psychiatr* AND ti(recover*). Our search criteria for data in these databases were published in the English language, published in a peer-reviewed journal, original article, and the research having included a description of recovery from the perspective of a patient with severe mental illness in a forensic or high-security setting. We also included one book as a source of data because the contents focused on the chosen patient population and recovery, and we performed a manual search of the reference lists of identified relevant studies (see Figure ). We ensured all the references included in the data portrayed forensic hospital settings, patients with a history of offending or who had been labeled dangerous, and descriptions of recovery from the patient perspective.
To increase the quality of the research, it was important to find a genuinely pure research question (concept) and avoid questions of value. During data collection, we needed to refocus our research question and simplify the concept we were analyzing (Phase 2). More specifically, we rejected definitions portraying what enables recovery or what recovery means from the perspective of staff or others. It is not expedient to list all possible features of the concept in this report. In our analysis, the essential features were found and grouped into themes and cases. We tolerated “hum noises” in the data, remembering the uniqueness of personal experiences, and some features were dismissed from the results because of their very low incidence. For example, in a book chapter in our data, the patient answered a question concerning their experience of recovery: “Right medication.” This unique feature can be understood rather as a helpful form of care than an essential feature of the personal recovery experience. We extracted the possible features of personal recovery of forensic patients from the original texts and gathered them into a list based on previous definitions by and . These definitions are more general descriptions of psychiatric recovery and were used to test whether our findings would fit the previous descriptions or differ from it. We constructed a phrase: “Recovery is 1) personal and unique , 2) a progressive journey with many stages and obstacles towards 3) a good life which is 4) more than absence of symptoms or overcoming the burden of disease and includes 5) citizenship and integration into society through 6) development of emotions and perceptions .” This proved to be an inaccurate way of organizing our findings, and we proceeded to create our own mode of categorization that would illuminate the forensic patients' experiences more accurately. We investigated the prevalence of different features identified in the chosen literature to define the typical nuances of forensic recovery experiences. We read through the list of features several times and, during the analysis process, found special features and unique meanings concerning forensic patients. Features resembling each other were grouped together until five themes entailed the typical nuances of personal recovery of patients in forensic care.
The collected data (see Table ) were categorized into five themes that describe the personal recovery experiences of forensic patients and emphasize the uniqueness of this group of psychiatric patients. Although we found similarities with previous descriptions of recovery in general ( ; ) and secure or forensic recovery ( ; ; ), we accentuate the different meanings of recovery for patients in forensic care. Our five overarching themes were as follows: personal development and autonomy, social inclusion and normality, redemption and overcoming, future orientation and hope, and advancing process. The collected data showed clear associations with these themes. Several original articles emphasized themes differently. The features in the data provided coherent examples of how patients experience their personal recovery process. It is noteworthy that some of our themes overlap, and certain features can be organized under two or more themes; this should not have a significant impact on the research because all of these features are a part of the concept under study. This overlapping serves as evidence that the personal recovery of the forensic patient is a complex phenomenon. The reasoning underlying the grouping of features into the five themes is depicted in Figure and in the following case examples. Personal Development and Autonomy All of the identified articles mentioned how a changing self-image and the development of new skills are a part of recovery. Examples of this included active coping ( ), utilizing strengths ( ), experiencing different emotions ( ), mastery over challenges ( ), feeling in control ( ), empowerment ( ; ; ), interests and dreams as aspirations ( ), making sense of past experiences as coming to terms with the past ( ), acceptance and working through denial ( ), autonomy ( ; ; ; ) and independence ( ), agency ( ; ), self-management and self-efficacy ( ), self-confidence and self-reflection ( ), a sense of achievement ( ; ), feeling better ( ; ), feeling good and important ( ), and being worthy ( ). Especially in the case of processing past life events, the offending history is a special feature for forensic patients. Empowerment and building autonomy and agency in a strictly closed environment are also special challenges for patients in forensic hospitals. Social Inclusion and Normality This theme primarily described integration into society after hospitalization, with the articles citing relevant activities such as social inclusion ( ), social bonding ( ), networking ( ), life as a citizen ( ; ; ), being accepted ( ), making a useful contribution ( ; ; , ), companionship as in close affectional relationships ( ), connecting with others ( ; ), being able to relate to people ( ; ; ; ; ; ), joining the work environment or enrolling in education ( , ), finding a home ( ), becoming a useful and contributing member of society ( ), internalizing social norms ( ), and becoming average ( ). This theme was discussed in 17 articles. Finding their own place in the society as full members can be especially challenging for forensic patients because of their long care in an institution. Normalcy can be perceived through this wish, to have the same possibilities as others who live without a history of mental illness and the role of a patient in a forensic or high-security hospital. Double stigma of mental illness and offending are present in forensic patients' lives. Redemption and Overcoming We found descriptions of overcoming the burden of disease and history of offending in 15 references. Redemption and overcoming were discussed in terms of criminal going straight and badness to redemption ( ), finding a new identity separate from the past ( ), helping instead of hurting ( ), hope for giving something back ( ), working on processing the crime ( ), understanding that an offense never leaves you ( ), and having an identity beyond being a patient or offender ( ). Although this theme was not the strongest, it clearly depicts special challenges for forensic recovery, the need for extra work on identity, and processing traumatic life events. Future Orientation and Hope In our sample, 11 references clarified how this theme is important to personal recovery. More specifically, the articles described how patients can gain future orientation and hope by pursuing a satisfying life ( ), feeling as though life is worth living ( ), not going back to previous ways ( ), letting go of the past ( ), looking into the future ( ), envisaging the future ( ), and perceiving hope as something to live for ( ). Orientation toward the future instead of clinging to the past experiences is a common theme for all psychiatric patients, but for forensic patients, not going back also means no more offending, no recidivism. Advancing Process The classic definition of recovery among psychiatric patients highlights that this is a continuous process. Our findings show that the recovery of forensic patients differs from the classic definition to some extent as only eight of the articles explicitly stated that recovery is a continuous, ongoing process. These articles commented that personal recovery involves building a life ( ), is a process ( ), can be considered as a journey ( ), is lifelong ( ) rather than an outcome one arrives at ( ), is nonlinear ( ), and encompasses one's own path ( ) or clearly moving on ( ). Perhaps the processual nature of recovery seems less important for patients in closed forensic institutions because of the typically long treatment periods. It has been reported that the length of stay at the forensic hospital can be unclear to the patients (e.g., ). During the long hospitalization, it can be difficult to start planning life outside the hospital and to see the inpatient period as a phase of the recovery process. This theme is closely linked to the previous one. method includes constructing cases to clarify the use of a concept. The cases (Wilson's Phases 3–6) created in this study are reported in the next paragraph. We only used invented cases (Phase 7). Social contextualization (Phase 8) and the question of underlying anxiety (Phase 9) were strongly present in our choice of concepts and phenomenon ( ). These contextual nuances were an important part of the analysis process. Social contextualization and the underlying anxiety are related to forensic patients' recovery context because the forensic hospital represents a high-security, closed institutional environment that involves limited possibilities for patients to choose social connections (e.g., ; ). The recovery process of forensic patients includes understanding their history as “offender patients” ( ) and managing a psychiatric condition (e.g., ), which can cause double stigmatization. Moreover, the care provided by staff involves a certain tension in balancing care and risk assessment ( ; ). We considered the social contextualization and underlying anxiety as important factors in the forensic context and regarded these during different phases of the analysis process and interpretation of the findings. This way of conducting the analysis is typical for Wilson's method. In this study, the cases were formulated as discussed below. The Model Case “I am moving forward toward my own goals, and I find it a promising journey. I am accompanied by my loved ones, and I am finding new roles and possibilities in society as perhaps almost anyone could. I acknowledge that previous crimes will influence my future, as does my psychiatric condition, but I have gained self-awareness, responsibility, and confidence. I do feel worthy, less impaired, and more capable.” This model case encompasses the essential features of personal recovery for forensic patients. The patient describes hope, self-determination, social networking, autonomy, overcoming adversity, finding new meaning in life, and trust in the future. The patient has found new capability of controlling their own life, conduct, interactions, and future. This statement acknowledges that the recovery process is a journey; moreover, instead of feeling hindered by their psychiatric illness and criminal history, the patient is finding new possibilities to look forward to. The Contrary Case “I feel limitations and restrictions and am somehow disconnected from the rest of the world. The symptoms have finally alleviated to some extent and the doctor should be happy. I do not know what I could hope for or be entitled to, either now or in the future. Perhaps I can return to my old life soon and forget about this whole mess caused by others.” In the contrary case, the patient does not feel in control of their life and has trouble being hopeful for new or better things. Furthermore, they have shifted the responsibility for their situation (a mess caused by others) to someone else, and the patient seems to lack motivation and purpose in recovering. As such, the essential features are not present in the contrary case ( ). The Related Case “I understand that what happened [the criminal offense] was because of my condition. I hope never to be and feel that way again. I am somewhat hopeless and lonely in this situation and do not have the necessary skills to proceed. Fortunately, I am getting support and assistance from my friends and the hospital staff. Maybe I can understand things better and find a new path in life.” In this case, the patient is interested in getting better but still doubtful about whether it will happen or how long it will take. This patient is connected to people, which creates hope. However, they are still lacking empowerment and strong motivation, but they are beginning to overcome the burden of their personal history. This example shows some of the elements of personal recovery but is not a pure case. It is a description of the personal development of a patient who is at the start of the recovery process ( ). The Borderline Case “I am getting some new ideas for the future. I am no longer as anxious as I was before, but sometimes I feel bitter and worry about how my life is going to turn out. This keeps me from experiencing true hope, and I do not quite feel worthy of happiness and a place in the outside world.” This patient is anticipating that something good and worthy will happen in the future but is still feeling insecure. The personal history is not yet integrated into this patient's life story and sense of self; as such, there has been no redemption or adopted responsibility. Although the symptoms have resolved, it remains unclear whether the psychiatric challenges have been accepted and understood. In this borderline case, the patient describes the features of recovery on a rather superficial level and experiences difficulties in self-expression ( ). The last phases of Wilson's concept analysis (Phases 10 and 11) include the practical results and clarifying the concept in the studied context, that is, “results in language.” Our results can help practitioners to better understand forensic psychiatric patients' experiences, which would benefit interventions aimed at designing appropriate care for the forensic environment. It can help the staff address important issues during the therapeutic process. Nevertheless, it is important to note that the analysis is not generalizable to all cases, as recovery has been described as a subjective and unique experience that changes over time ( ; ; ). In the best-case scenario, forensic psychiatric care can help a patient recover from their illness and mitigate the risk for recidivism by education about self-management approaches ( ).
All of the identified articles mentioned how a changing self-image and the development of new skills are a part of recovery. Examples of this included active coping ( ), utilizing strengths ( ), experiencing different emotions ( ), mastery over challenges ( ), feeling in control ( ), empowerment ( ; ; ), interests and dreams as aspirations ( ), making sense of past experiences as coming to terms with the past ( ), acceptance and working through denial ( ), autonomy ( ; ; ; ) and independence ( ), agency ( ; ), self-management and self-efficacy ( ), self-confidence and self-reflection ( ), a sense of achievement ( ; ), feeling better ( ; ), feeling good and important ( ), and being worthy ( ). Especially in the case of processing past life events, the offending history is a special feature for forensic patients. Empowerment and building autonomy and agency in a strictly closed environment are also special challenges for patients in forensic hospitals.
This theme primarily described integration into society after hospitalization, with the articles citing relevant activities such as social inclusion ( ), social bonding ( ), networking ( ), life as a citizen ( ; ; ), being accepted ( ), making a useful contribution ( ; ; , ), companionship as in close affectional relationships ( ), connecting with others ( ; ), being able to relate to people ( ; ; ; ; ; ), joining the work environment or enrolling in education ( , ), finding a home ( ), becoming a useful and contributing member of society ( ), internalizing social norms ( ), and becoming average ( ). This theme was discussed in 17 articles. Finding their own place in the society as full members can be especially challenging for forensic patients because of their long care in an institution. Normalcy can be perceived through this wish, to have the same possibilities as others who live without a history of mental illness and the role of a patient in a forensic or high-security hospital. Double stigma of mental illness and offending are present in forensic patients' lives.
We found descriptions of overcoming the burden of disease and history of offending in 15 references. Redemption and overcoming were discussed in terms of criminal going straight and badness to redemption ( ), finding a new identity separate from the past ( ), helping instead of hurting ( ), hope for giving something back ( ), working on processing the crime ( ), understanding that an offense never leaves you ( ), and having an identity beyond being a patient or offender ( ). Although this theme was not the strongest, it clearly depicts special challenges for forensic recovery, the need for extra work on identity, and processing traumatic life events.
In our sample, 11 references clarified how this theme is important to personal recovery. More specifically, the articles described how patients can gain future orientation and hope by pursuing a satisfying life ( ), feeling as though life is worth living ( ), not going back to previous ways ( ), letting go of the past ( ), looking into the future ( ), envisaging the future ( ), and perceiving hope as something to live for ( ). Orientation toward the future instead of clinging to the past experiences is a common theme for all psychiatric patients, but for forensic patients, not going back also means no more offending, no recidivism.
The classic definition of recovery among psychiatric patients highlights that this is a continuous process. Our findings show that the recovery of forensic patients differs from the classic definition to some extent as only eight of the articles explicitly stated that recovery is a continuous, ongoing process. These articles commented that personal recovery involves building a life ( ), is a process ( ), can be considered as a journey ( ), is lifelong ( ) rather than an outcome one arrives at ( ), is nonlinear ( ), and encompasses one's own path ( ) or clearly moving on ( ). Perhaps the processual nature of recovery seems less important for patients in closed forensic institutions because of the typically long treatment periods. It has been reported that the length of stay at the forensic hospital can be unclear to the patients (e.g., ). During the long hospitalization, it can be difficult to start planning life outside the hospital and to see the inpatient period as a phase of the recovery process. This theme is closely linked to the previous one. method includes constructing cases to clarify the use of a concept. The cases (Wilson's Phases 3–6) created in this study are reported in the next paragraph. We only used invented cases (Phase 7). Social contextualization (Phase 8) and the question of underlying anxiety (Phase 9) were strongly present in our choice of concepts and phenomenon ( ). These contextual nuances were an important part of the analysis process. Social contextualization and the underlying anxiety are related to forensic patients' recovery context because the forensic hospital represents a high-security, closed institutional environment that involves limited possibilities for patients to choose social connections (e.g., ; ). The recovery process of forensic patients includes understanding their history as “offender patients” ( ) and managing a psychiatric condition (e.g., ), which can cause double stigmatization. Moreover, the care provided by staff involves a certain tension in balancing care and risk assessment ( ; ). We considered the social contextualization and underlying anxiety as important factors in the forensic context and regarded these during different phases of the analysis process and interpretation of the findings. This way of conducting the analysis is typical for Wilson's method. In this study, the cases were formulated as discussed below. The Model Case “I am moving forward toward my own goals, and I find it a promising journey. I am accompanied by my loved ones, and I am finding new roles and possibilities in society as perhaps almost anyone could. I acknowledge that previous crimes will influence my future, as does my psychiatric condition, but I have gained self-awareness, responsibility, and confidence. I do feel worthy, less impaired, and more capable.” This model case encompasses the essential features of personal recovery for forensic patients. The patient describes hope, self-determination, social networking, autonomy, overcoming adversity, finding new meaning in life, and trust in the future. The patient has found new capability of controlling their own life, conduct, interactions, and future. This statement acknowledges that the recovery process is a journey; moreover, instead of feeling hindered by their psychiatric illness and criminal history, the patient is finding new possibilities to look forward to. The Contrary Case “I feel limitations and restrictions and am somehow disconnected from the rest of the world. The symptoms have finally alleviated to some extent and the doctor should be happy. I do not know what I could hope for or be entitled to, either now or in the future. Perhaps I can return to my old life soon and forget about this whole mess caused by others.” In the contrary case, the patient does not feel in control of their life and has trouble being hopeful for new or better things. Furthermore, they have shifted the responsibility for their situation (a mess caused by others) to someone else, and the patient seems to lack motivation and purpose in recovering. As such, the essential features are not present in the contrary case ( ). The Related Case “I understand that what happened [the criminal offense] was because of my condition. I hope never to be and feel that way again. I am somewhat hopeless and lonely in this situation and do not have the necessary skills to proceed. Fortunately, I am getting support and assistance from my friends and the hospital staff. Maybe I can understand things better and find a new path in life.” In this case, the patient is interested in getting better but still doubtful about whether it will happen or how long it will take. This patient is connected to people, which creates hope. However, they are still lacking empowerment and strong motivation, but they are beginning to overcome the burden of their personal history. This example shows some of the elements of personal recovery but is not a pure case. It is a description of the personal development of a patient who is at the start of the recovery process ( ). The Borderline Case “I am getting some new ideas for the future. I am no longer as anxious as I was before, but sometimes I feel bitter and worry about how my life is going to turn out. This keeps me from experiencing true hope, and I do not quite feel worthy of happiness and a place in the outside world.” This patient is anticipating that something good and worthy will happen in the future but is still feeling insecure. The personal history is not yet integrated into this patient's life story and sense of self; as such, there has been no redemption or adopted responsibility. Although the symptoms have resolved, it remains unclear whether the psychiatric challenges have been accepted and understood. In this borderline case, the patient describes the features of recovery on a rather superficial level and experiences difficulties in self-expression ( ). The last phases of Wilson's concept analysis (Phases 10 and 11) include the practical results and clarifying the concept in the studied context, that is, “results in language.” Our results can help practitioners to better understand forensic psychiatric patients' experiences, which would benefit interventions aimed at designing appropriate care for the forensic environment. It can help the staff address important issues during the therapeutic process. Nevertheless, it is important to note that the analysis is not generalizable to all cases, as recovery has been described as a subjective and unique experience that changes over time ( ; ; ). In the best-case scenario, forensic psychiatric care can help a patient recover from their illness and mitigate the risk for recidivism by education about self-management approaches ( ).
“I am moving forward toward my own goals, and I find it a promising journey. I am accompanied by my loved ones, and I am finding new roles and possibilities in society as perhaps almost anyone could. I acknowledge that previous crimes will influence my future, as does my psychiatric condition, but I have gained self-awareness, responsibility, and confidence. I do feel worthy, less impaired, and more capable.” This model case encompasses the essential features of personal recovery for forensic patients. The patient describes hope, self-determination, social networking, autonomy, overcoming adversity, finding new meaning in life, and trust in the future. The patient has found new capability of controlling their own life, conduct, interactions, and future. This statement acknowledges that the recovery process is a journey; moreover, instead of feeling hindered by their psychiatric illness and criminal history, the patient is finding new possibilities to look forward to.
“I feel limitations and restrictions and am somehow disconnected from the rest of the world. The symptoms have finally alleviated to some extent and the doctor should be happy. I do not know what I could hope for or be entitled to, either now or in the future. Perhaps I can return to my old life soon and forget about this whole mess caused by others.” In the contrary case, the patient does not feel in control of their life and has trouble being hopeful for new or better things. Furthermore, they have shifted the responsibility for their situation (a mess caused by others) to someone else, and the patient seems to lack motivation and purpose in recovering. As such, the essential features are not present in the contrary case ( ).
“I understand that what happened [the criminal offense] was because of my condition. I hope never to be and feel that way again. I am somewhat hopeless and lonely in this situation and do not have the necessary skills to proceed. Fortunately, I am getting support and assistance from my friends and the hospital staff. Maybe I can understand things better and find a new path in life.” In this case, the patient is interested in getting better but still doubtful about whether it will happen or how long it will take. This patient is connected to people, which creates hope. However, they are still lacking empowerment and strong motivation, but they are beginning to overcome the burden of their personal history. This example shows some of the elements of personal recovery but is not a pure case. It is a description of the personal development of a patient who is at the start of the recovery process ( ).
“I am getting some new ideas for the future. I am no longer as anxious as I was before, but sometimes I feel bitter and worry about how my life is going to turn out. This keeps me from experiencing true hope, and I do not quite feel worthy of happiness and a place in the outside world.” This patient is anticipating that something good and worthy will happen in the future but is still feeling insecure. The personal history is not yet integrated into this patient's life story and sense of self; as such, there has been no redemption or adopted responsibility. Although the symptoms have resolved, it remains unclear whether the psychiatric challenges have been accepted and understood. In this borderline case, the patient describes the features of recovery on a rather superficial level and experiences difficulties in self-expression ( ). The last phases of Wilson's concept analysis (Phases 10 and 11) include the practical results and clarifying the concept in the studied context, that is, “results in language.” Our results can help practitioners to better understand forensic psychiatric patients' experiences, which would benefit interventions aimed at designing appropriate care for the forensic environment. It can help the staff address important issues during the therapeutic process. Nevertheless, it is important to note that the analysis is not generalizable to all cases, as recovery has been described as a subjective and unique experience that changes over time ( ; ; ). In the best-case scenario, forensic psychiatric care can help a patient recover from their illness and mitigate the risk for recidivism by education about self-management approaches ( ).
Both the CHIME framework ( ) and the previous definitions of forensic recovery ( ; ; ) are applicable to the forensic context, yet our findings promote a deeper understanding of the concept of personal recovery from the patient's perspective. The theme personal development and autonomy resembles previous definitions of forensic recovery, that is, identity, meaning in life, and empowerment, as well as identity work and developing a sense of self ( ; ; ). Moreover, the definition used in this study categorizes meaning in life and empowerment within personal development and autonomy. As such, the definition links our findings to the recovery paradigm, which emphasizes how a meaningful life and well-being are connected to personal development ( ; ). Our theme of social inclusion and normality is closely related to connectedness ( ; ; ) and social networks ( ), and encompasses special features in the case of forensic patients. Our definition contains two distinct themes—” redemption and overcoming,” which was also considered important in the work by , and “advancing process.” The CHIME terms connectedness and hope and optimism about the future reflect, to some extent, the features of future orientation and hope and social inclusion and normality in our definition. Normalcy plays a significant role in the recovery of all psychiatric patients and is particularly important to forensic patients. Hope and looking forward to the future are commonly noted as central characteristics of recovery (e.g., ; ; ). In this way, previous knowledge of psychiatric recovery emphasizes this theme more than our findings from a population of forensic psychiatric patients. It was surprising that only 11 references of the identified 21 touched upon the theme of future orientation and hope, which is a central feature of the general recovery paradigm. Instead, several studies of forensic psychiatric patients included descriptions of long treatment periods and time seemingly “standing still.” This presenteeism can be a challenge, especially in forensic care ( ). Perhaps the processual nature of the forensic recovery process was rarely mentioned in the identified studies because of presenteeism. The duration of hospital care can seem infinite from the patient perspective ( ). This infiniteness is associated to the arbitrary length of stay conditional upon the patient's mental state and risk assessment, both of which are evaluated by the staff.
Our concept analysis included five overarching themes that contain the dimensions of recovery for forensic psychiatric patients. In comparison with previous definitions of personal recovery, our results stress that the recovery process of forensic psychiatric patients involves some special features. Each of the different dimensions of recovery (clinical, functional, social, and personal) was present in our findings and interacts in the patients' experiences in the clinical environment. The dimensions also overlap and change in time. Implications for Clinical Forensic Nursing Practice The results of this study provide several insights into the development of forensic nursing practice (Phase 10). First, supporting a patient's capacity to integrate their personal history into their life story and creating a sense of self should be critical elements in forensic nursing. Therapeutic work that focuses on the experience of the criminal offense, as well as its consequences for the patient and others, should be the primary form of support for forensic patients with a psychiatric illness. Second, inclusion and belonging can be promoted by providing possibilities for reorientation into society both during and after hospitalization, which is typically long-term. The boundaries of the hospital should be considered a risk for alienation, and protective activities should be organized through work with the patients' families and possibilities for engaging in professional activities and/or education. Moreover, patients should be afforded the opportunity to train for the skills required for common everyday activities outside the institution. Third, personal development and finding meaning in life should be a key part of the therapeutic relationship between nurses and forensic patients. It is important to state that the goal of treatment should not be a symptom-free life, but rather a life without the excess limitations caused by a psychiatric condition and criminal history. As such, supporting a patient's capacity to create new self-awareness, self-management skills, and self-efficacy will improve their sense of agency and add meaning to their life. This is critical to helping a patient get involved in social activities, which are a large part of reducing recidivism.
The results of this study provide several insights into the development of forensic nursing practice (Phase 10). First, supporting a patient's capacity to integrate their personal history into their life story and creating a sense of self should be critical elements in forensic nursing. Therapeutic work that focuses on the experience of the criminal offense, as well as its consequences for the patient and others, should be the primary form of support for forensic patients with a psychiatric illness. Second, inclusion and belonging can be promoted by providing possibilities for reorientation into society both during and after hospitalization, which is typically long-term. The boundaries of the hospital should be considered a risk for alienation, and protective activities should be organized through work with the patients' families and possibilities for engaging in professional activities and/or education. Moreover, patients should be afforded the opportunity to train for the skills required for common everyday activities outside the institution. Third, personal development and finding meaning in life should be a key part of the therapeutic relationship between nurses and forensic patients. It is important to state that the goal of treatment should not be a symptom-free life, but rather a life without the excess limitations caused by a psychiatric condition and criminal history. As such, supporting a patient's capacity to create new self-awareness, self-management skills, and self-efficacy will improve their sense of agency and add meaning to their life. This is critical to helping a patient get involved in social activities, which are a large part of reducing recidivism.
|
Development and validation of a machine learning model to predict hemostatic intervention in patients with acute upper gastrointestinal bleeding | 54433700-6692-426e-9e72-3eb5c3c95629 | 11934503 | Pathologic Processes[mh] | Acute upper gastrointestinal bleeding (UGIB) remains a prevalent and significant medical condition encountered in routine clinical practice, with an annual incidence of 80–150 per 100,000 population and mortality rates spanning from 2% to 15% [ – ]. Given a considerable spectrum of severity, multiple international guidelines recommended pre-endoscopic risk stratification to assess and triage patients according to severity. Esophagogastroduodenoscopy (EGD) should be performed within a suggested time frame of 12–24 h from the onset of presentation [ – ]. Endoscopy is the most effective tool for diagnosing UGIB, and endoscopic therapy is indicated for lesions with high-risk stigmata to control bleeding and prevent rebleeding. For risk stratification, various scoring systems have been developed during the past two decades, such as the pre-endoscopic Rockall score (RS), Glasgow-Blatchford score (GBS), and AIMS65 score [ – ]. These scores aim to classify patients into low-risk or high-risk groups, thereby guiding the subsequent treatment strategy. Patients classified as a high-risk group carry higher mortality and rebleeding rates and, therefore, require in-hospital management. The pre-endoscopic Rockall score and the AIMS65 score were designed to predict mortality, while the GBS aimed to predict the likelihood of in-hospital management, including endoscopy, transfusion support, or surgery. Studies comparing the performance of these scoring systems revealed that the GBS has the highest performance and very high sensitivity [ – ]. However, the specificity was still limited . Furthermore, these scoring systems did not focus mainly on predicting hemostatic intervention. The role of artificial intelligence (AI) in the medical profession is undergoing rapid expansion since machine learning can effectively handle large, complex, and heterogeneous datasets, extract correlations of parameters in detail, and accurately predict the outcome with experience [ – ]. As evidenced by numerous studies, the application of AI has yielded satisfying results in risk stratification, endoscopic findings, and mortality rate in UGIB [ – ]. In 2020, Seo et al. created a machine learning model to predict adverse events such as mortality, hypotension, and rebleeding in patients with initially stable nonvariceal bleeding. The model showed a higher ability to detect adverse events than conventional scores . The other two models that focus on predicting blood transfusion or mortality for UGIB in the intensive care unit exhibited impressive performance, with an area under receiver operating characteristic (AUROC) exceeding 0.80 . Moreover, Shung et al. developed an available online machine-learning model from multicenter patient data to stratify low-risk patients who can be treated in an outpatient setting . A multicenter study proved that the model performs better than the GBS, with great sensitivity and high specificity . From our perspective, predicting the need for endoscopic intervention in patients with UGIB is one of the key decisions in management flow, and it would benefit physicians in the resource-limited area. This prospect could optimize the selective referral of patients from the primary healthcare center to the endoscopic center. Currently, only a few machine learning models precisely predict the need for endoscopic intervention in UGIB . Then, we developed a simplified machine-learning model with structured data analysis for clinical decision-supporting systems to indicate the need for endoscopic intervention in patients with acute UGIB.
Patients Prospectively collected data from adult patients who presented acute overt UGIB at Siriraj Hospital, Bangkok, Thailand, from January 2011 to December 2020 were retrospectively reviewed. The UGIB management protocol in our hospital followed international guidelines [ – ]. However, the final treatment decision was based on the attending physician and the bleeding team, including the endoscopist, radiological interventionist, and surgeon, who were available on a 24–7 basis. An expert endoscopist or trainees under close supervision performed the endoscopy. Endoscopic findings and hemostatic intervention were recorded using an Endosmart program. Hemostatic interventions for ulcers with high-risk bleeding [Forrest classification Ia (spurting bleeding), Ib (oozing bleeding), and IIa (non-bleeding visible vessel)] include a single or combination of adrenaline injection, hemostatic clip, thermal hemostasis, or hemostatic powder. For adherent clots, an attempt was made to remove the clot and examine the character of the underlying lesion. The intervention will not be applied in low-risk ulcers, including clean base ulcers and hematin spots. For variceal bleeding, rubber band ligation or glue injection was usually used for hemostasis of active bleeding lesions; for nonbleeding esophageal varices with high-risk stigmata, prophylaxis band ligation was performed depending on the endoscopist’s decision. The study included patients aged 18 years and older with acute UGIB who underwent EGD with comprehensive documentation of endoscopic findings. The inclusion was done by retrieving the patients from the Siriraj GI endoscopic center database using the search term “upper GI bleeding or UGIB” from the indication for EGD. The exclusion criteria were patients with in-hospital onset of UGIB, onset of UGIB more than 72 hours before hospital visit, and missing data on baseline characteristics, bleeding presentation, laboratory findings, endoscopic result, or intervention. Eligible patients who met the defined criteria were analyzed and used to develop and validate the machine-learning model. Data collection and outcome The data from the chart and endoscopic record were extracted manually by well-trained GI fellows. They were divided into three categories: patient characteristic data, bleeding characteristic data, and laboratory data. All the data must be available before endoscopy. The baseline characteristics were extracted from the patient’s history, which was documented before the hospital visit. The bleeding presentation was noted by the first physician who encountered the patient, and laboratory tests were the initial test after the patient visited the hospital and before the GI consultation. Characteristic data of the patient included age, sex, co-morbidities documented in the medical file before the bleeding event, such as heart disease, stroke, chronic kidney disease, cirrhosis, and active malignancy, use of antithrombotic agents [conventional non-steroidal anti-inflammatory drugs (NSAIDs), cyclooxygenase-(COX) 2 inhibitors, aspirin, clopidogrel, and oral anticoagulants], previous episodes of UGIB and duration of bleeding before hospitalization documented by the first physician who encounters the patient. The definition of each comorbidity was described in Supplementary Material Appendix . Characteristics of bleeding composed of the clinical manifestation of bleeding, such as type of vomitus (red or coffee-ground emesis) and type of stool (melena, maroon stool, red stool), the presence of syncope, altered consciousness (defined by a decreased Glasgow coma score less than 13 or documented as the chief complaint to the hospital), systolic blood pressure (SBP), heart rate, and need for resuscitation (fluid therapy with the rate of fluid higher than 500 ml/h without vasopressor or vasopressor needed after optimal fluid therapy). Initial laboratory test data included hemoglobin, platelet count, blood urea nitrogen (BUN), creatinine, serum albumin, and international normalized ratio (INR). The BUN and creatinine were analyzed as BUN/creatinine ratio as it would diminish the falsely high BUN from renal causes such as pre-renal azotemia or chronic kidney disease. These parameters are used in previous scoring systems associated with hospital intervention and mortality [ – ]. The endpoint selected to develop the machine learning model was the requirement of endoscopic hemostatic intervention for variceal or non-variceal procedures, including epinephrine injection, hemostatic clip, thermal hemostasis, rubber band ligation, glue injection therapy, and hemostatic powder. All the data was cleaned and rechecked for exclusion criteria. For the data pre-processing step, we adopted the One-Hot Encoding method for this experiment because most machine learning algorithms are not capable of handling categorical data without encoding. Furthermore, this experiment also adopted the normalization technique, which is the z-score, as part of the data pre-processing step for machine learning. This technique was used to rescale the values of numeric columns in the dataset without distorting differences in the ranges of values. Model development For this study, we built and developed machine learning models in Python to predict the need for endoscopic intervention (Version 3.10, 64 bits). The data set of patients was randomly divided into two groups with an 80%–20% ratio as training and test sets. To reduce the risk of overfitting and ensure that the model can perform well in various samples, we applied a stratified 5-fold cross-validation as a validation technique. The training data were then fed to all models available in the model library using cross-validation to train and validate the models. These 15 supervised learning models included Linear Discriminant Analysis, Logistic Regression, Naïve Bayes, CatBoost Classifier, Extra Trees Classifier, Quadratic Discriminant Analysis, Random Forest Classifier, Gradient Boosting Classifier, Ada Boost Classifier, Light Gradient Boosting Machine, Extreme Gradient Boosting, K-Neighbors Classifier, Decision Tree Classifier, Dummy Classifier, and Lasso regression. They are available in a Python library called Pycaret (version 3.1.0). Since we adopted the stratified 5-fold cross-validation, the training data was randomly divided into five subsets or folds. The model was trained and evaluated five times, using a different fold as the validation set. Then, performance metrics from each fold were averaged to estimate the model’s performance, which was shown as the average area under the receiver operating characteristic (AUROC), accuracy, sensitivity, and specificity on the validation sets across five folds. Based on these comparison results, we determined the optimal model to predict the need for endoscopic intervention by selecting the model that could achieve the highest AUROC. Since default hyperparameters were implemented in all models, the best model was subsequently adjusted with a hyperparameter-tuning method in Pycaret to find the best prediction performance. In this experiment, we implemented a random grid search over a pre-defined grid search for hyperparameter tuning. In addition, we increased the number of iterations, ranging from 100 to 1000 at intervals of 100, to find the best performance. However, the same results were obtained, so the minimum number of iterations, which is 100, was chosen in this experiment. Then, the tuned model was internally validated with the test set to evaluate the performance and analyze the essential factors for the prediction. Finally, the model was deployed on the local host with the Python library, Streamlit. The outcome was shown as the need or the lack of a need for endoscopic intervention. The prediction probability was noted in the result as a percentage for the user to make decisions for further management. Statistical analysis Qualitative data were analyzed by frequency and percentage, while quantitative data were analyzed by mean and standard deviation. The difference in variables between the two groups was analyzed using Fischer’s exact test or the Mann–Whitney U test. The prediction performance was measured by the AUROC curve analysis, sensitivity, specificity, and accuracy. The negative predictive value (NPV) and positive predictive value (PPV) of the model were calculated. The AUROC value was predefined as follows: acceptable threshold (≥ 0.7), fair performance (≥ 0.7 but < 0.8), good performance (≥ 0.8 but < 0.9), and excellent performance (≥ 0.9). A comparison of the AUROC of the proposed model and the conventional score was performed using a paired permutation test. A P-value < 0.05 was considered statistical significance. All statistical analyses were performed using Python (Version 3.10, 64 bits). The sample size recruitment strategy was designed to include as many patients as possible to achieve a more efficient machine-learning model. This study followed the TRIPOD-AI reporting guideline and the ethical guidelines of the Declaration of Helsinki and was approved by the Siriraj Institutional Review Board (COA No. Si 1028/2021). The checklist of the TRIPOD-AI reporting guideline is provided in Supplementary Material Appendix . Since this was a retrospective analysis, informed consent was not obtained from the patients.
Prospectively collected data from adult patients who presented acute overt UGIB at Siriraj Hospital, Bangkok, Thailand, from January 2011 to December 2020 were retrospectively reviewed. The UGIB management protocol in our hospital followed international guidelines [ – ]. However, the final treatment decision was based on the attending physician and the bleeding team, including the endoscopist, radiological interventionist, and surgeon, who were available on a 24–7 basis. An expert endoscopist or trainees under close supervision performed the endoscopy. Endoscopic findings and hemostatic intervention were recorded using an Endosmart program. Hemostatic interventions for ulcers with high-risk bleeding [Forrest classification Ia (spurting bleeding), Ib (oozing bleeding), and IIa (non-bleeding visible vessel)] include a single or combination of adrenaline injection, hemostatic clip, thermal hemostasis, or hemostatic powder. For adherent clots, an attempt was made to remove the clot and examine the character of the underlying lesion. The intervention will not be applied in low-risk ulcers, including clean base ulcers and hematin spots. For variceal bleeding, rubber band ligation or glue injection was usually used for hemostasis of active bleeding lesions; for nonbleeding esophageal varices with high-risk stigmata, prophylaxis band ligation was performed depending on the endoscopist’s decision. The study included patients aged 18 years and older with acute UGIB who underwent EGD with comprehensive documentation of endoscopic findings. The inclusion was done by retrieving the patients from the Siriraj GI endoscopic center database using the search term “upper GI bleeding or UGIB” from the indication for EGD. The exclusion criteria were patients with in-hospital onset of UGIB, onset of UGIB more than 72 hours before hospital visit, and missing data on baseline characteristics, bleeding presentation, laboratory findings, endoscopic result, or intervention. Eligible patients who met the defined criteria were analyzed and used to develop and validate the machine-learning model.
The data from the chart and endoscopic record were extracted manually by well-trained GI fellows. They were divided into three categories: patient characteristic data, bleeding characteristic data, and laboratory data. All the data must be available before endoscopy. The baseline characteristics were extracted from the patient’s history, which was documented before the hospital visit. The bleeding presentation was noted by the first physician who encountered the patient, and laboratory tests were the initial test after the patient visited the hospital and before the GI consultation. Characteristic data of the patient included age, sex, co-morbidities documented in the medical file before the bleeding event, such as heart disease, stroke, chronic kidney disease, cirrhosis, and active malignancy, use of antithrombotic agents [conventional non-steroidal anti-inflammatory drugs (NSAIDs), cyclooxygenase-(COX) 2 inhibitors, aspirin, clopidogrel, and oral anticoagulants], previous episodes of UGIB and duration of bleeding before hospitalization documented by the first physician who encounters the patient. The definition of each comorbidity was described in Supplementary Material Appendix . Characteristics of bleeding composed of the clinical manifestation of bleeding, such as type of vomitus (red or coffee-ground emesis) and type of stool (melena, maroon stool, red stool), the presence of syncope, altered consciousness (defined by a decreased Glasgow coma score less than 13 or documented as the chief complaint to the hospital), systolic blood pressure (SBP), heart rate, and need for resuscitation (fluid therapy with the rate of fluid higher than 500 ml/h without vasopressor or vasopressor needed after optimal fluid therapy). Initial laboratory test data included hemoglobin, platelet count, blood urea nitrogen (BUN), creatinine, serum albumin, and international normalized ratio (INR). The BUN and creatinine were analyzed as BUN/creatinine ratio as it would diminish the falsely high BUN from renal causes such as pre-renal azotemia or chronic kidney disease. These parameters are used in previous scoring systems associated with hospital intervention and mortality [ – ]. The endpoint selected to develop the machine learning model was the requirement of endoscopic hemostatic intervention for variceal or non-variceal procedures, including epinephrine injection, hemostatic clip, thermal hemostasis, rubber band ligation, glue injection therapy, and hemostatic powder. All the data was cleaned and rechecked for exclusion criteria. For the data pre-processing step, we adopted the One-Hot Encoding method for this experiment because most machine learning algorithms are not capable of handling categorical data without encoding. Furthermore, this experiment also adopted the normalization technique, which is the z-score, as part of the data pre-processing step for machine learning. This technique was used to rescale the values of numeric columns in the dataset without distorting differences in the ranges of values.
For this study, we built and developed machine learning models in Python to predict the need for endoscopic intervention (Version 3.10, 64 bits). The data set of patients was randomly divided into two groups with an 80%–20% ratio as training and test sets. To reduce the risk of overfitting and ensure that the model can perform well in various samples, we applied a stratified 5-fold cross-validation as a validation technique. The training data were then fed to all models available in the model library using cross-validation to train and validate the models. These 15 supervised learning models included Linear Discriminant Analysis, Logistic Regression, Naïve Bayes, CatBoost Classifier, Extra Trees Classifier, Quadratic Discriminant Analysis, Random Forest Classifier, Gradient Boosting Classifier, Ada Boost Classifier, Light Gradient Boosting Machine, Extreme Gradient Boosting, K-Neighbors Classifier, Decision Tree Classifier, Dummy Classifier, and Lasso regression. They are available in a Python library called Pycaret (version 3.1.0). Since we adopted the stratified 5-fold cross-validation, the training data was randomly divided into five subsets or folds. The model was trained and evaluated five times, using a different fold as the validation set. Then, performance metrics from each fold were averaged to estimate the model’s performance, which was shown as the average area under the receiver operating characteristic (AUROC), accuracy, sensitivity, and specificity on the validation sets across five folds. Based on these comparison results, we determined the optimal model to predict the need for endoscopic intervention by selecting the model that could achieve the highest AUROC. Since default hyperparameters were implemented in all models, the best model was subsequently adjusted with a hyperparameter-tuning method in Pycaret to find the best prediction performance. In this experiment, we implemented a random grid search over a pre-defined grid search for hyperparameter tuning. In addition, we increased the number of iterations, ranging from 100 to 1000 at intervals of 100, to find the best performance. However, the same results were obtained, so the minimum number of iterations, which is 100, was chosen in this experiment. Then, the tuned model was internally validated with the test set to evaluate the performance and analyze the essential factors for the prediction. Finally, the model was deployed on the local host with the Python library, Streamlit. The outcome was shown as the need or the lack of a need for endoscopic intervention. The prediction probability was noted in the result as a percentage for the user to make decisions for further management.
Qualitative data were analyzed by frequency and percentage, while quantitative data were analyzed by mean and standard deviation. The difference in variables between the two groups was analyzed using Fischer’s exact test or the Mann–Whitney U test. The prediction performance was measured by the AUROC curve analysis, sensitivity, specificity, and accuracy. The negative predictive value (NPV) and positive predictive value (PPV) of the model were calculated. The AUROC value was predefined as follows: acceptable threshold (≥ 0.7), fair performance (≥ 0.7 but < 0.8), good performance (≥ 0.8 but < 0.9), and excellent performance (≥ 0.9). A comparison of the AUROC of the proposed model and the conventional score was performed using a paired permutation test. A P-value < 0.05 was considered statistical significance. All statistical analyses were performed using Python (Version 3.10, 64 bits). The sample size recruitment strategy was designed to include as many patients as possible to achieve a more efficient machine-learning model. This study followed the TRIPOD-AI reporting guideline and the ethical guidelines of the Declaration of Helsinki and was approved by the Siriraj Institutional Review Board (COA No. Si 1028/2021). The checklist of the TRIPOD-AI reporting guideline is provided in Supplementary Material Appendix . Since this was a retrospective analysis, informed consent was not obtained from the patients.
Patient characteristic The database of 2,201 patients with acute UGIB was reviewed. Among these, 635 patients with missing data, 170 patients with delayed hospitalization, and seven patients with in-hospital UGIB were excluded, resulting in 1,389 patients being eligible for model development. All patients underwent upper endoscopy within 120 h, and 615 (44.3%) of the cohorts received the endoscopic intervention; 293 variceal interventions, 336 nonvariceal interventions, and 14 patients received both variceal and nonvariceal interventions. The baseline characteristics, bleeding characteristics, and laboratory findings are presented in Table . For the total cohort, the mean age was 64.3 years, with a male predominance of 65%. The patients in the intervention group were younger, and male patients with coexisting cirrhosis and active malignancy were more prevalent. Furthermore, patients in the intervention group came to the hospital earlier. They required a higher rate of resuscitation, which was consistent with a significantly lower systolic blood pressure, a higher heart rate, more red emesis, a lower platelet number, a lower albumin level, and a higher INR level. For medication, the use of NSAIDs, COX-2 inhibitors, and clopidogrel was comparable between the two groups, but the nonintervention group consisted of higher aspirin and anticoagulant users. Model parameters After analyzing the data for the model development, some independent parameters were correlated with each other. For example, patients with a history of stroke or heart disease tend to use antiplatelet agents; creatinine levels could reflect chronic kidney disease; syncope and altered mental status can be evaluated as unstable vital signs. These categorical parameters, such as the history of stroke, heart disease, chronic kidney disease, syncope, and alteration of consciousness, were dropped out because of their probable co-linearity by logical assumption, as they displayed equivalent properties of the subjects and caused unstable coefficient estimates or overfitting models. We performed the Variance Inflation Factor (VIF) to evaluate the co-linearity of the numerical parameters. The result showed that hemoglobin, systolic blood pressure, heart rate, and albumin had high VIF. However, they were crucial parameters used in many previous prediction models [ – ]. Therefore, these parameters were retained in the models. Several types of antithrombotic drugs were grouped as the preliminary models showed similar precision between the grouped parameters and the distinct parameters of this medication. Finally, 18 parameters, including age, sex, presence of cirrhosis, active malignancy, use of antithrombotic drugs, previous history of UGIB, vomitus characters (red emesis or coffee-ground emesis), and stool characters (melena, maroon stool or red stool), duration of UGIB before hospitalization, resuscitation requirement, systolic blood pressure, heart rate, hemoglobin level, platelet number, serum albumin level, blood urea nitrogen level, creatinine level and INR level were used as input for the machine learning models. From 8 categorical data and 10 numerical data, all categorical features were transformed by the One-Hot Encoding method. Each categorical level becomes a separate feature in the dataset containing binary values, either 0 or 1. In doing so, the total of eighteen features was extended to twenty-one features. The blood urea nitrogen level and creatinine level will be computed as a ratio for machine learning analysis. As mentioned above, 80% of the cohort (1,111 patients) was used as a training set for machine learning models. The baseline characteristics of the training set and the test set are shown in Table . There were no significant differences in patient profile, bleeding presentation, laboratory results, and endoscopic hemostatic intervention between these two sets. Model performance According to the comparison results in Table , the linear discriminant analysis model demonstrated the highest AUROC, accuracy, and specificity. The AUROC of this model is 0.74, with a sensitivity of 57%, a specificity of 80%, and an accuracy of 70%. The result of the 5-fold cross-validation of the linear discriminant analysis model is shown in Supplementary Material Appendix . The model with the highest sensitivity of 63% was the Naïve Bayes model, but the AUROC was only 0.73. The linear discriminant analysis model was chosen for fine-tuning to develop the best performance, and its AUROC slightly increased to 0.75. For performance evaluation, the model was internally validated with the test set, which demonstrated an AUROC of 0.81, as shown in Fig. . The NPV and PPV of our model were calculated from the confusion matrix of the test set, and the results were 0.75 and 0.74, respectively, with a prevalence of UGIB of 0.44. The importance of the features was analyzed and shown in Fig. . Cirrhosis, red emesis, and the need for resuscitation were the three most important features of the model prediction of the need for endoscopic intervention. The probability threshold values to classify the patients into intervention groups or non-intervention groups are plotted in Fig. . The threshold of the model can be adjusted with the same AUROC result. At the default threshold of 0.5, our model has sensitivity and specificity of 74.5% and 81.4%, respectively. When the probability threshold value decreases, the model’s sensitivity increases, but the specificity decreases. For example, at the probability threshold of 0.17, the sensitivity could reach 99.2% with a specificity of 17.5%. After considering the accuracy in the figure, we found that the accuracy values were high when the probability threshold values ranged from 0.4 to 0.6, especially at values around 0.5. Therefore, this study used 0.5 as the threshold value to classify the patients into groups. The developed Linear Discriminant Analysis model, GBS, pre-endoscopic RS, and AIMS65 score were applied to the test set cohort, and the AUROC of each score was analyzed, as shown in Fig. . The results showed that our model was superior to conventional scoring systems in predicting the need for endoscopic intervention (AUC, developed model 0.81 [95% CI 0.76–0.87] vs GBS 0.55 [95% CI 0.48–0.61] p < 0.001, pre-endoscopic RS 0.60 [95% CI 0.53–0.67] p < 0.001, AIMS65 score 0.54 [95% CI 0.47–0.61] p < 0.001). After we achieved the optimal prediction model based on linear discriminant analysis, we implemented this model in a local web application using Streamlit (Python Library), as demonstrated in Fig. . The panel on the left side will be used for data input. After inserting all the data, the result and probability of prediction will be instantly shown on the right side. This finalized program can be used on a local host computer.
The database of 2,201 patients with acute UGIB was reviewed. Among these, 635 patients with missing data, 170 patients with delayed hospitalization, and seven patients with in-hospital UGIB were excluded, resulting in 1,389 patients being eligible for model development. All patients underwent upper endoscopy within 120 h, and 615 (44.3%) of the cohorts received the endoscopic intervention; 293 variceal interventions, 336 nonvariceal interventions, and 14 patients received both variceal and nonvariceal interventions. The baseline characteristics, bleeding characteristics, and laboratory findings are presented in Table . For the total cohort, the mean age was 64.3 years, with a male predominance of 65%. The patients in the intervention group were younger, and male patients with coexisting cirrhosis and active malignancy were more prevalent. Furthermore, patients in the intervention group came to the hospital earlier. They required a higher rate of resuscitation, which was consistent with a significantly lower systolic blood pressure, a higher heart rate, more red emesis, a lower platelet number, a lower albumin level, and a higher INR level. For medication, the use of NSAIDs, COX-2 inhibitors, and clopidogrel was comparable between the two groups, but the nonintervention group consisted of higher aspirin and anticoagulant users.
After analyzing the data for the model development, some independent parameters were correlated with each other. For example, patients with a history of stroke or heart disease tend to use antiplatelet agents; creatinine levels could reflect chronic kidney disease; syncope and altered mental status can be evaluated as unstable vital signs. These categorical parameters, such as the history of stroke, heart disease, chronic kidney disease, syncope, and alteration of consciousness, were dropped out because of their probable co-linearity by logical assumption, as they displayed equivalent properties of the subjects and caused unstable coefficient estimates or overfitting models. We performed the Variance Inflation Factor (VIF) to evaluate the co-linearity of the numerical parameters. The result showed that hemoglobin, systolic blood pressure, heart rate, and albumin had high VIF. However, they were crucial parameters used in many previous prediction models [ – ]. Therefore, these parameters were retained in the models. Several types of antithrombotic drugs were grouped as the preliminary models showed similar precision between the grouped parameters and the distinct parameters of this medication. Finally, 18 parameters, including age, sex, presence of cirrhosis, active malignancy, use of antithrombotic drugs, previous history of UGIB, vomitus characters (red emesis or coffee-ground emesis), and stool characters (melena, maroon stool or red stool), duration of UGIB before hospitalization, resuscitation requirement, systolic blood pressure, heart rate, hemoglobin level, platelet number, serum albumin level, blood urea nitrogen level, creatinine level and INR level were used as input for the machine learning models. From 8 categorical data and 10 numerical data, all categorical features were transformed by the One-Hot Encoding method. Each categorical level becomes a separate feature in the dataset containing binary values, either 0 or 1. In doing so, the total of eighteen features was extended to twenty-one features. The blood urea nitrogen level and creatinine level will be computed as a ratio for machine learning analysis. As mentioned above, 80% of the cohort (1,111 patients) was used as a training set for machine learning models. The baseline characteristics of the training set and the test set are shown in Table . There were no significant differences in patient profile, bleeding presentation, laboratory results, and endoscopic hemostatic intervention between these two sets.
According to the comparison results in Table , the linear discriminant analysis model demonstrated the highest AUROC, accuracy, and specificity. The AUROC of this model is 0.74, with a sensitivity of 57%, a specificity of 80%, and an accuracy of 70%. The result of the 5-fold cross-validation of the linear discriminant analysis model is shown in Supplementary Material Appendix . The model with the highest sensitivity of 63% was the Naïve Bayes model, but the AUROC was only 0.73. The linear discriminant analysis model was chosen for fine-tuning to develop the best performance, and its AUROC slightly increased to 0.75. For performance evaluation, the model was internally validated with the test set, which demonstrated an AUROC of 0.81, as shown in Fig. . The NPV and PPV of our model were calculated from the confusion matrix of the test set, and the results were 0.75 and 0.74, respectively, with a prevalence of UGIB of 0.44. The importance of the features was analyzed and shown in Fig. . Cirrhosis, red emesis, and the need for resuscitation were the three most important features of the model prediction of the need for endoscopic intervention. The probability threshold values to classify the patients into intervention groups or non-intervention groups are plotted in Fig. . The threshold of the model can be adjusted with the same AUROC result. At the default threshold of 0.5, our model has sensitivity and specificity of 74.5% and 81.4%, respectively. When the probability threshold value decreases, the model’s sensitivity increases, but the specificity decreases. For example, at the probability threshold of 0.17, the sensitivity could reach 99.2% with a specificity of 17.5%. After considering the accuracy in the figure, we found that the accuracy values were high when the probability threshold values ranged from 0.4 to 0.6, especially at values around 0.5. Therefore, this study used 0.5 as the threshold value to classify the patients into groups. The developed Linear Discriminant Analysis model, GBS, pre-endoscopic RS, and AIMS65 score were applied to the test set cohort, and the AUROC of each score was analyzed, as shown in Fig. . The results showed that our model was superior to conventional scoring systems in predicting the need for endoscopic intervention (AUC, developed model 0.81 [95% CI 0.76–0.87] vs GBS 0.55 [95% CI 0.48–0.61] p < 0.001, pre-endoscopic RS 0.60 [95% CI 0.53–0.67] p < 0.001, AIMS65 score 0.54 [95% CI 0.47–0.61] p < 0.001). After we achieved the optimal prediction model based on linear discriminant analysis, we implemented this model in a local web application using Streamlit (Python Library), as demonstrated in Fig. . The panel on the left side will be used for data input. After inserting all the data, the result and probability of prediction will be instantly shown on the right side. This finalized program can be used on a local host computer.
Our study successfully developed a machine learning model to predict the need for endoscopic hemostatic intervention in patients with acute UGIB. The model was operated by entering 18 simple parameters, including 6 demographic data, 6 bleeding characteristic data, and 6 initial laboratory data. As confirmed by a large validation cohort, the accuracy of this model ranged from fair to good, with an AUROC of 0.81, an accuracy of 70%, a sensitivity of 57%, and a specificity of 80%. The results revealed that the linear discriminant analysis outperformed other machine-learning algorithms. This might be because our data is in a relationship in such a way that it could utilize the advantages of linear discriminant analysis. To illustrate this point, linear discriminant analysis (LDA) maximizes the separation between classes and helps reduce the overlap between classes in the reduced-dimensional space. This makes it highly effective for classification. Additionally, it helps prevent overfitting by reducing dimensionality while maintaining class-related information. Prior studies have shown that the characteristics of active bleeding, such as red emesis and bloody gastric content, physical signs of hypovolemia, such as the presence of syncope, lower mean arterial pressure, and laboratory findings of low hemoglobin level, prolonged prothrombin time, and higher BUN level, were correlated with the performance of therapeutic intervention [ – ]. Combining those factors predicted endoscopic intervention with an AUROC comparable to all the parameters of GBS . Subsequently, several scores were developed to exclusively predict endoscopic intervention in UGIB with parameters less complex than GBS (Table ). The MAP(ASH) score was created in Spain and validated by a large cohort of the multiethnic population. The AUROC of the MAP(ASH) score was similar to the GBS but was significantly higher than the AIMS65 score . Two Japanese scores were described: Nagoya University and H3B2 scores . Both of which demonstrated an AUROC higher than GBS. The London Haemostat Score (LHS) also showed higher accuracy than GBS . Although these scoring systems provide similar or better accuracy to GBS with less complex parameters, their performance is still limited. An artificial model focusing on endoscopic intervention prediction has been developed to overcome the limited performance of scoring systems. Veisman et al. conducted a machine learning tool in Israel using 34 parameters to assess the possible correlation between baseline characteristics and endoscopic intervention in 883 patients . The Random Forest model was created with a sensitivity, specificity, and AUROC of 0.55, 0.71, and 0.68, respectively. The AUROC of the model is higher than those of GBS (0.54) and pre-endoscopic RS (0.56), which is similar to the result of our population. The analysis plot showed that syncope, cirrhosis, and erythromycin use are correlated with the risk of intervention. Compared to our model, the sensitivity, specificity, and AUROC are higher than those of Veisman et al., with fewer parameters being used. The better performance of our model could be because we included a higher number of training populations for model derivation, and we tried creating a variety of models and selected the best performance among them. However, the significant parameters for the prediction were in the exact correlation. Cirrhosis is a common risk feature in both models, as it carries various mechanisms that potentiate endoscopic intervention, including esophagogastric varices, thrombocytopenia, and coagulopathy. UGIB patients with cirrhosis generally have varices and usually need endoscopic intervention for both therapeutic and prophylaxis purposes. The presence of red emesis or syncope and the requirement for fluid resuscitation may reflect the severity of bleeding from the underlying high-risk stigmata lesion that leads to endoscopic intervention. For erythromycin, the prokinetic effect was discussed to improve endoscopic visualization, and then the endoscopic intervention can be performed consecutively. However, the portion of erythromycin used in the study was small. In our center, intravenous erythromycin is unavailable, so our model did not include pre-endoscopic prokinetic use as a parameter. In this study, the hemoglobin level was not an influential factor in predicting the endoscopic intervention. The explanation could be that it was only a one-time static parameter that did not represent the severity of bleeding. Many patients might have a baseline for chronic anemia from other medical conditions. From the previous model predicting endoscopic intervention by Veisman, et al. and Ito et al. , hemoglobin was not correlated with increased risk for endoscopic intervention as well. The changes in hemoglobin level from baseline or the first test should be more accurately related to bleeding severity. However, this information was not available in our population. Although numerous models were developed, the best model with great precision could not be achieved to predict the UGIB outcome. This could be explained by the hypothesis that UGIB is a complex disease with dynamic conditions, and the current risk scores are not dynamic . The clinical severity could change from the first hospital visit to the endoscopy time due to the etiology of bleeding, patient comorbidity, anticoagulation, and resuscitation process. Pre-endoscopic medication, such as proton pump inhibitors, can downstage the endoscopic stigmata and reduce the need for endoscopic intervention . A study comparing urgent endoscopy (<6 h) and early endoscopy (<24 h) in UGIB patients showed more high-risk stigmata lesions in the urgent group (66.4 vs. 47.8%). However, the mortality outcome was not different in both groups . Therefore, static parameters at one time may not accurately predict the need for hemostatic intervention in UGIB. The intervention performed during index EGD may not represent the culprit lesion of that bleeding episode, and different physicians may also decide differently on the same lesion in each situation. For example, a cirrhotic patient with melena from a peptic ulcer, which at the time of endoscopy showed a clean base ulcer and a column of large esophageal varices with high-risk stigmata, may undergo rubber band ligation for primary prophylaxis as recommended in the clinical algorithm . This could cause cirrhosis as the most crucial factor for model prediction. Furthermore, in real-life practice, patients who did not require hemostatic intervention still need endoscopic evaluation to diagnose the underlying etiology. Referral to an endoscopist is still mandatory but less urgent than a patient who needs hemostatic intervention. If the model were applied to the primary medical center, the patients with the initial diagnosis of UGIB would be evaluated and stabilized regularly. After inputting all parameters, the model will predict the endoscopic intervention probability of each patient. The physician would use this result to prioritize the urgency of an endoscopic intervention consultant or referral. Patients with low risk for intervention could be admitted for medical management and consulted for endoscopy later as elective or OPD cases. The proposed algorithm is shown in Fig. . For example, in our UGIB population, with a prevalence of 0.44 and NPV of 0.75, patients with predicted results as no need for intervention would have a 75% probability that they did not truly require urgent endoscopy. However, the physician should not miss the 25% risk. Other in-hospital evaluations, such as vital sign monitoring, serial hemoglobin level, and proper elevation of hemoglobin after blood transfusion, are still crucial for referral judgment. This protocol may reduce unnecessary urgent endoscopic procedures and optimize resource allocation in resource-limited settings. The strength of our model is that we use numerous databases of more than 1000 patients with UGIB, including both variceal and nonvariceal bleeding. We generated 15 models and compared them to select the most productive model for prediction. Our model provided accuracy superior to that of the GBS. However, some limitations were noted in our model. First, this study included 10 years of data collection. Over the decades, the development and changes in practice may affect physicians’ decisions over time. However, the resuscitation and endoscopic treatment recommendations have not changed significantly since 2010 . There are updates on restrictive blood transfusion strategies and new options for rescue therapeutic intervention, such as the over-the-scope clip and hemostatic spray, which do not affect the core concept of acute UGIB management. Second, about 30% of the patients with missing data were excluded, mainly due to missing the onset of symptoms and the INR. The characteristics of those missing data were compared to the eligible data and shown in Supplementary Material Appendix . Compared to the eligible population, the onset of symptoms and INR were not significantly different in both groups. There were significant differences in some parameters which might lead to bias. Also, some groups of patients were not included in the model computation, for example, the patients who did not undergo EGD due to low risk for GBS calculation, early death, or limitation of care. Moreover, some medical history not mentioned is considered to be absent such as unrecognized underlying disease, and untold over-the-counter NSAIDs or other medication. Third, certain critical factors that could affect the management decision were not included in the model, such as other anti-platelets in P2Y12 antagonists. They have been available in our center for the past few years. However, with high costs and reimbursement limitations, only a few patients received this medication. From our cohort, there were no patients prescribed these medications. The other crucial factors are time to endoscopy, pre-endoscopic medication, dynamic change in the bleeding characteristic or hemodynamic status after initial management, or the difference in hemoglobin level at present compared to baseline to assess the chronicity and severity of bleeding. In practice, these parameters are essential for the physician’s decision but require multiple data input steps and may not be practical in the primary setting. Further study, including multiple steps of data collection, would be helpful to maximize the model’s performance. Fourth, the CI95% for AUC, sensitivity, and specificity of the model were not presented as the results computed by Pycaret could not be further calculated to CI95%, but it was described in a similar pattern to the previous study . Lastly, the management of UGIB patients relies on physician judgment. In our center, there were no hospital-specific protocols for physicians in the emergency department. Different physicians, including endoscopists, may act differently when encountering similar situations. The threshold for resuscitation can be varied in some cases. For example, the dynamic changes in clinical and laboratory parameters could lead to further resuscitation. Even though the management of endoscopists for intervention is based on the current guidelines mentioned in the method, some interventions depend on the endoscopist’s judgment, such as prophylaxis EV ligation in cirrhotic patients with non-variceal bleeding. In real-life applications, regardless of the accuracy of the models, physicians must combine these prediction results with other dynamic factors to get the most appropriate management for the patients.
The prediction for endoscopic intervention in acute UGIB patients is complex and dynamic. In response to this challenge, our machine learning model, which used simple clinical parameters, performed fairly well in identifying UGIB patients who need endoscopic intervention. The practical implication of this study is that physicians in primary care units could prioritize patients who need a referral to endoscopic centers. Further development and external validation to identify more specific features will improve prediction performance.
Below is the link to the electronic supplementary material. Supplementary Material 1
|
DETECT: DEveloping and testing a model to identify preventive vision loss among older paTients in gEneral praCTice – protocol for a complex intervention in Denmark | afbeba35-f080-4a11-a287-66836055a4b1 | 10230986 | Family Medicine[mh] | It is estimated that 2.2 billion people have impaired vision, and of these, at least 1 billion people have a vision loss that could have been prevented or reduced by earlier detection or by access to treatment. The most common causes of moderate to severe visual impairment are uncorrected refractive errors, unoperated cataract, age-related macular degeneration (AMD), glaucoma and diabetic retinopathy. Also in affluent welfare states such as Denmark—constituting the setting of this study—visual impairment is a problem. The incidence of visual impairment increases with age, and due to the sociodemographical trajectory of an increasing elderly population, the prevalence of patients with visual impairment will increase. Timely access to healthcare has a major influence on the progression of eye conditions. The consequences of vision loss significantly affect the person’s quality of life, dependence and increases the risk of recurrent falls and fractures, which is a significant threat to mobility in old age. A YouGov poll showed that sight was by far the sensory function, people fear losing the most. Vision loss can result in worsened mental health, cognition and social functioning. Disease progression can be complicated by other chronic conditions and can complicate management of multimorbidity due to decreased self-care, ability to visit clinics and adherence to medication. Finally, vision loss can increase the risks of placement in nursing homes. Thus, visual impairment and loss has a great impact on the individual, their relatives, society in general and on the healthcare system. General practice and visual impairment General practitioners (GPs) handle preventive healthcare, diagnostics, treatment and care of chronic conditions as well as coordinate services from various healthcare professionals. In the Global North, the GPs handle the majority of all medical matters. A survey set in English general practice from 1998 concluded that eye problems, including undiagnosed glaucoma and AMD, were quite frequent among elderly patients consulting their GP. One study found that patients were more likely to have their eyes checked if their GP suggests it. In addition, an increased focus on eye health in at-risk populations in general practice is suggested to be more effective for early detection than broader screening programmes. However, a recent UK-based survey of GPs indicated that although up to 5% of the primary consultations were eye related, GPs ability to identify red flags was low. The literature points to a gap where, even though patients are in contact with their GP concerning symptoms related to their vision, an unidentified number of patients may suffer from unrecognised visual impairment that is not detected in general practice. Collaboration across healthcare professions Since vision problems are rarely detected in general practice, the need for research into how patients and health professionals collaborate to identify and manage visual impairment becomes a relevant matter. This is in line with a recent Cochrane review, which concludes that future research should look at optimised primary care-based vision screening interventions. Patients with visual impairment often have contacts with many different healthcare professionals. Therefore, it is important to incorporate collaboration across health professions and sectors in a GP intervention aimed at improving identification of patients with visual impairment. Optometrists constitute an occupational group who may be the first line of contact for some patients who experience visual changes. Optometrists will in most cases operate independently without a formal collaboration with other health professionals such as GPs. Given the optometrists contact with the patients and level of equipment for measuring vision and evaluating the eye, they pose a potentially important resource when collaboration across healthcare professions is rethought and are therefore important to include in the study. Their commercial agenda may influence their work and this will be evaluated in the collaboration. Aim and objectives In this study, we aim to develop a health intervention in a Danish general practice setting to improve the detection and care of visual impairment. The patient target group is middle-aged and older adults and their relatives, with GP’s constituting the primary professional target group. We define visual impairment broadly to be the patient experience of symptoms related to vision and findings identified by health professionals—such as reduced vision field—not yet experienced by the patient, but a serious threat to patient vision. Visual impairment is in this respect not connected to specific diagnoses, but as previously stated, we assume frequent eye diseases such as glaucoma and AMD will be well represented. The overall aim of DETECT is thus to: Develop an intervention in general practice aimed at identifying visual impairment among elderly patients with chronic conditions. Test the feasibility of the intervention model in general practice with a focus on ensuring improved patient support and education.
General practitioners (GPs) handle preventive healthcare, diagnostics, treatment and care of chronic conditions as well as coordinate services from various healthcare professionals. In the Global North, the GPs handle the majority of all medical matters. A survey set in English general practice from 1998 concluded that eye problems, including undiagnosed glaucoma and AMD, were quite frequent among elderly patients consulting their GP. One study found that patients were more likely to have their eyes checked if their GP suggests it. In addition, an increased focus on eye health in at-risk populations in general practice is suggested to be more effective for early detection than broader screening programmes. However, a recent UK-based survey of GPs indicated that although up to 5% of the primary consultations were eye related, GPs ability to identify red flags was low. The literature points to a gap where, even though patients are in contact with their GP concerning symptoms related to their vision, an unidentified number of patients may suffer from unrecognised visual impairment that is not detected in general practice.
Since vision problems are rarely detected in general practice, the need for research into how patients and health professionals collaborate to identify and manage visual impairment becomes a relevant matter. This is in line with a recent Cochrane review, which concludes that future research should look at optimised primary care-based vision screening interventions. Patients with visual impairment often have contacts with many different healthcare professionals. Therefore, it is important to incorporate collaboration across health professions and sectors in a GP intervention aimed at improving identification of patients with visual impairment. Optometrists constitute an occupational group who may be the first line of contact for some patients who experience visual changes. Optometrists will in most cases operate independently without a formal collaboration with other health professionals such as GPs. Given the optometrists contact with the patients and level of equipment for measuring vision and evaluating the eye, they pose a potentially important resource when collaboration across healthcare professions is rethought and are therefore important to include in the study. Their commercial agenda may influence their work and this will be evaluated in the collaboration.
In this study, we aim to develop a health intervention in a Danish general practice setting to improve the detection and care of visual impairment. The patient target group is middle-aged and older adults and their relatives, with GP’s constituting the primary professional target group. We define visual impairment broadly to be the patient experience of symptoms related to vision and findings identified by health professionals—such as reduced vision field—not yet experienced by the patient, but a serious threat to patient vision. Visual impairment is in this respect not connected to specific diagnoses, but as previously stated, we assume frequent eye diseases such as glaucoma and AMD will be well represented. The overall aim of DETECT is thus to: Develop an intervention in general practice aimed at identifying visual impairment among elderly patients with chronic conditions. Test the feasibility of the intervention model in general practice with a focus on ensuring improved patient support and education.
The study will be conducted in Denmark and is thus inscribed in a Scandinavian health system with universal access to healthcare. The general health status in Denmark is relatively high, and as far as vision is concerned, the incidence of legal blindness has decreased along with improved treatment options. The average life expectancy has increased over the last 70 years, which is positive, but it also entails a rise in age-related sight-threatening eye diseases, such as glaucoma and AMD. Despite the decreasing incidence of legal blindness due to AMD, many patients are diagnosed late with irreversible vision loss. It seems relevant to diagnose eye diseases earlier and optimise the coordination of care. It is difficult to provide an exact number of people in Denmark who live with visual impairment. A national survey of health, quality of life and morbidity from 2007 shows that 3.8% of the population over 60 reported difficulties in reading a newspaper text, while the Danish Eye Association estimate that 50.000 people in Denmark above 60 years are blind or visually impaired (total population 60+: 1.554.542 ). In Denmark, the GP is the patient’s primary entry point to the healthcare system, and the GP treats 90% of all medical cases. All Danes are assigned to a default general practice, and as many as 80% consult their GP at least annually, with an increased frequency among patients aged 50 or older. People with chronic conditions are offered an annual health check at their GP. A Danish survey from 2019 showed that 82.4% of men and 86.7% of women had their vision measured at their GP within the last 3 years. However, these figures are from 2010 to 2017. In 2017, regulations regarding driving licenses were changed, resulting in vision being measured only every 15 years. This may result in a lower frequency of vision acuity measurements at the GP today, but from the 2017 figures we can assume that the practice of performing vision measurements during the annual consultation for older adults is a well-known procedure. Ophthalmologists can diagnose, treat and carry out the necessary checks of, for example, glaucoma and atrophic AMD in the primary sector. If indicated, patients are referred to secondary care in the hospitals’ eye departments. Examples of referral indications may be neovascular AMD, proliferative diabetic retinopathy and medically uncontrollable glaucoma. At the hospitals, ophthalmologists work publicly funded and university hospital clinics have an obligation to do research within the field. Consultations with the GP or ophthalmologist are tax financed and without an out-of-pocket fee to patients. Hospital-based eye clinics are also free of charge, but the patient must be referred by a primary sector ophthalmologist for treatment. In cases of acute vision loss or pain, patients can be seen directly in the emergency room and referred from there to the on-call ophthalmologist in the hospital eye department. Anyone can book an appointment with the ophthalmologist in the primary sector without a referral, but due to a low number of ophthalmologists compared with the increasing demand, it is often difficult to book a consultation within a reasonable time frame and geographical distance. On the other hand, the GP must be available to all patients inscribed in his/her practice and have an in-depth knowledge of the patient’s general health and condition. The GPs, therefore, seem to be in an ideal situation to identify visual impairment and coordinate the management. It is estimated that around 2000 optometrists operate in Denmark (total population 5.8 million) and opticians shops can be found in most smaller cities, making it accessible even in rural areas to visit an optician shop. In most optician shops, it is free of charge to have vision tested. Many optometrists offer intraocular pressure measurements and fundus photographs as additional procedures for a fee. Optometrists are thus a professional stakeholder that we find interesting to explore further. Study phases We apply the Medical Research Council guidance on developing, testing and evaluating complex interventions. To operationalise the framework for complex interventions, we apply a temporal structure from the tradition of human-centred design to divide the project cycle into three main phases: (I) identify the problem, (II) develop the model and (III) test the feasibility of the model. Phase I+II focus on intervention development applying qualitative methods and phase III aims to first pilot test the intervention for feasibility and following implement it broader in a cohort study to measure effect. The intervention explores what patients, carers and professionals perceive as pivotal to improve regarding detection, navigating care and health services and support possibilities for people living with visual impairment. We are explicitly reflective on, how process and product are interwoven and to a very high extent dependent on the context it unfolds within. Following a structured and well-documented design process, we identify the changes made and insights produced (see ). Phase I: identify—identifying key issues to address in the health intervention The aim of phase I is to explore the problem we are addressing. We will perform a literature search on detection of eye diseases in general practice and conduct background interviews with a broad selection of relevant stakeholders to help us map the current practice in detecting and diagnosing eye conditions across sectors in the healthcare system (see ). An essential element in human-centred design processes is to create understanding and empathy for the end-user —in our case people living with visual impairment and their relatives. We aim to incorporate a wide spectrum of eye diagnoses and develop a model to identify and diagnose visual impairment that works from the onset of the patient’s early symptoms, such as stumbling over doorsteps, difficulties in reading, distorted vision or difficulties related to the transition from dark to light spaces and vice versa. The intervention aim of increased patient support also includes support of relatives, who carry a large part of the disease burden and may experience stress and depression due to their loved one’s visual impairment. Patient and public involvement in phase I Patients, relatives and professionals will be involved by providing their perspectives in various stages of the study and codesign core elements of the model constituting the intervention. The patient engagement process is informed by a thorough report produced by the Danish Center for Social Science Research on older adults with visual impairment and vision loss. The report underlines the need for increased knowledge on how visual impairment affects patients’ every day life in all aspects. The experiences, needs, preferences, and values of patients and relatives will thus be explored. See for an overview of the involvement of participants across the three project phases. In phase I, we perform semistructured interviews with patients and relatives, preferably in their home to gain an insight in their experiences in the context in which they occur. Here, we focus on the patient journey from the time patients experience symptoms leading to a diagnosis and handling life afterwards with a diagnosis. After the interview, patients and relatives are encouraged to contact the researcher if they would like her/him to participate in, for example, a visit to the ophthalmologist or if they have further input or concerns at a later stage. We will furthermore perform focus group interviews with older adults to investigate (1) the expectations to vision in old age and (2) which health professionals’ older adults identify as relevant when they experience vision changes. The focus group interviews supplement the interviews with patient and relative with a view to gain insight into the social norms and prominent attitudes towards visual impairment among older adults. The participants in the focus group interviews are asked to complete the validated Visual Function Questionnaire-25 to provide more individual knowledge about the participants own perception of their vision function. Through participant observations, we will generate knowledge on the everyday working environment and challenges that health professionals, private optometrists and communal workers navigate in concerning people with visual impairment. Phase II: develop—developing the intervention model In phase II, we operationalise the insights from phase I. This will be done through three consecutive content-developing workshops using the creative method of graphic facilitation. Graphic facilitation is well suited when elaborating on ideas and problem-solving processes because it allows for a transparent process open to multiple agendas. The method can be relevant for redistributing power and expertise in a codesign activity and the physical product of the three content workshops will act as design principles for the intervention. CTS facilitates the workshops and we invite a graphic facilitator to analogue draw and write inputs from the participants on a wall-to-wall paper during the workshops. The graphical recordings from the three workshops constitute a collective overview of core elements to include in the intervention model (see description of specifics on the workshops below and for details on participants). Choosing a visual method to engage participants could seem an unusual choice in a project focusing on visual impairment and vision loss. However, the participating patients are not blind. They live with a visual impairment, which poses a range of consequences and constraints in their every day life, but it does not prevent them from being able to participate in a graphical facilitated workshop. If needed, relevant aids will be provided—in example to enlarge the graphical recordings on a tablet Flow of the three workshops: The graphic facilitator and project researchers develop a template for the workshops. Participants for the workshops are recruited among participants in phase I. Workshop 1 In this workshop, the graphic facilitator engages patients and relatives to formulate a graphical recording on what is lacking in the identification, diagnosis and patient support concerning visual impairment. Workshop 2 The focus for this workshop is to include perspectives from relevant health professionals in the design phase. The health professionals are asked to identify possibilities and barriers of an intervention concerning vision impairment in general practice, including a discussion on key eye examinations and measures that would be feasible in a general practice setting as well as to identify the most relevant eye diseases. Workshop 3 Aims to synthesise the knowledge produced in the previous two workshops by formulating the specific activities, concrete consultation type and identify final intervention effect measurements. This also includes choosing the relevant guidelines for screening and diagnosing to apply in the intervention. Participants are GP’s and project researchers. Other stakeholders will be invited if relevant based on the insights from workshop 1+2. Phase III: feasibility—feasibility test in general practice Part 1: pilot test The intervention model will be first be validated and adjusted accordingly by patient representatives and ophthalmologists. The model will then be tested in a general practice setting to establish face validity and adjust according to the experiences. The model must be clinically relevant and feasible for implementation in a clinical practice. Data production continues in this phase through observational studies and interviews with GPs to disseminate the experiences with the model in general practice and whether the model can be part of improved collaboration between health professionals. Part 2: cohort study Based on findings from the pilot test, we expand the intervention by including 10–15 GP practices in the Capital and Zeeland Region, Denmark to participate in testing the GP’s possibilities and barriers to detect visual impairment and vision loss. According to previous literature, we need to include 1500–2000 patients in general practice aged 65 or older to identify 150–200 with visual impairment. The practices will receive the developed intervention model and recruit in up to 18 months. Patients 65+ who consult their GP as part of an annual consultation for a chronic condition will be informed about the study in the waiting room and asked to complete a questionnaire based on the validated Visual functioning Questionnaire-25. The questionnaire measures the dimensions of self-reported vision-targeted health status that are most important to individuals who have chronic eye diseases. A dedicated staff or GP will examine the vision according to the guidelines formulated in the model. If visual impairment is detected, the patient will undergo further examinations assessed by an ophthalmologist. The specifics of the examinations in general practice and at an ophthalmologist cannot be reported until the phases I+II have been completed, since these are to be developed in the codesign process. At present, we assume an ordinary vision test, visual fields and contrast vision as well as a function test could be included in the GP setting. We assume that measurement of the visual acuity, tonometry, macular and parapapillary optical coherence tomography (OCT) scans and fundus photographs could be part of the extended examinations at an ophthalmologist. An important outcome of phase II is thus detailed information on effect measurements, chosen guidelines and possibly a narrowed focus on specific eye diseases to address in the intervention. The results from the cohort study will function as a reference standard and allow us to study the prevalence of visual impairments as well as eye diseases and study predictor’s for visual impairment as well. Due to the Danish registers, it is possible to follow the cohort for a long period. This follow-up study is not part of the present project, but will be planned later. Analysis The analytical process will be carried out iteratively, which covers analytical steps taken between each of the three project phases to ensure appropriate adjustments in the design of the intervention model. All formal interviews from the study will be audio recorded and transcribed following a project guideline to ensure uniformity. The three phases will generate observations and informal talks, which will be documented through field notes and pictures. Details on the analytical steps are illustrated in .
We apply the Medical Research Council guidance on developing, testing and evaluating complex interventions. To operationalise the framework for complex interventions, we apply a temporal structure from the tradition of human-centred design to divide the project cycle into three main phases: (I) identify the problem, (II) develop the model and (III) test the feasibility of the model. Phase I+II focus on intervention development applying qualitative methods and phase III aims to first pilot test the intervention for feasibility and following implement it broader in a cohort study to measure effect. The intervention explores what patients, carers and professionals perceive as pivotal to improve regarding detection, navigating care and health services and support possibilities for people living with visual impairment. We are explicitly reflective on, how process and product are interwoven and to a very high extent dependent on the context it unfolds within. Following a structured and well-documented design process, we identify the changes made and insights produced (see ).
The aim of phase I is to explore the problem we are addressing. We will perform a literature search on detection of eye diseases in general practice and conduct background interviews with a broad selection of relevant stakeholders to help us map the current practice in detecting and diagnosing eye conditions across sectors in the healthcare system (see ). An essential element in human-centred design processes is to create understanding and empathy for the end-user —in our case people living with visual impairment and their relatives. We aim to incorporate a wide spectrum of eye diagnoses and develop a model to identify and diagnose visual impairment that works from the onset of the patient’s early symptoms, such as stumbling over doorsteps, difficulties in reading, distorted vision or difficulties related to the transition from dark to light spaces and vice versa. The intervention aim of increased patient support also includes support of relatives, who carry a large part of the disease burden and may experience stress and depression due to their loved one’s visual impairment.
Patients, relatives and professionals will be involved by providing their perspectives in various stages of the study and codesign core elements of the model constituting the intervention. The patient engagement process is informed by a thorough report produced by the Danish Center for Social Science Research on older adults with visual impairment and vision loss. The report underlines the need for increased knowledge on how visual impairment affects patients’ every day life in all aspects. The experiences, needs, preferences, and values of patients and relatives will thus be explored. See for an overview of the involvement of participants across the three project phases. In phase I, we perform semistructured interviews with patients and relatives, preferably in their home to gain an insight in their experiences in the context in which they occur. Here, we focus on the patient journey from the time patients experience symptoms leading to a diagnosis and handling life afterwards with a diagnosis. After the interview, patients and relatives are encouraged to contact the researcher if they would like her/him to participate in, for example, a visit to the ophthalmologist or if they have further input or concerns at a later stage. We will furthermore perform focus group interviews with older adults to investigate (1) the expectations to vision in old age and (2) which health professionals’ older adults identify as relevant when they experience vision changes. The focus group interviews supplement the interviews with patient and relative with a view to gain insight into the social norms and prominent attitudes towards visual impairment among older adults. The participants in the focus group interviews are asked to complete the validated Visual Function Questionnaire-25 to provide more individual knowledge about the participants own perception of their vision function. Through participant observations, we will generate knowledge on the everyday working environment and challenges that health professionals, private optometrists and communal workers navigate in concerning people with visual impairment.
In phase II, we operationalise the insights from phase I. This will be done through three consecutive content-developing workshops using the creative method of graphic facilitation. Graphic facilitation is well suited when elaborating on ideas and problem-solving processes because it allows for a transparent process open to multiple agendas. The method can be relevant for redistributing power and expertise in a codesign activity and the physical product of the three content workshops will act as design principles for the intervention. CTS facilitates the workshops and we invite a graphic facilitator to analogue draw and write inputs from the participants on a wall-to-wall paper during the workshops. The graphical recordings from the three workshops constitute a collective overview of core elements to include in the intervention model (see description of specifics on the workshops below and for details on participants). Choosing a visual method to engage participants could seem an unusual choice in a project focusing on visual impairment and vision loss. However, the participating patients are not blind. They live with a visual impairment, which poses a range of consequences and constraints in their every day life, but it does not prevent them from being able to participate in a graphical facilitated workshop. If needed, relevant aids will be provided—in example to enlarge the graphical recordings on a tablet Flow of the three workshops: The graphic facilitator and project researchers develop a template for the workshops. Participants for the workshops are recruited among participants in phase I. Workshop 1 In this workshop, the graphic facilitator engages patients and relatives to formulate a graphical recording on what is lacking in the identification, diagnosis and patient support concerning visual impairment. Workshop 2 The focus for this workshop is to include perspectives from relevant health professionals in the design phase. The health professionals are asked to identify possibilities and barriers of an intervention concerning vision impairment in general practice, including a discussion on key eye examinations and measures that would be feasible in a general practice setting as well as to identify the most relevant eye diseases. Workshop 3 Aims to synthesise the knowledge produced in the previous two workshops by formulating the specific activities, concrete consultation type and identify final intervention effect measurements. This also includes choosing the relevant guidelines for screening and diagnosing to apply in the intervention. Participants are GP’s and project researchers. Other stakeholders will be invited if relevant based on the insights from workshop 1+2.
In this workshop, the graphic facilitator engages patients and relatives to formulate a graphical recording on what is lacking in the identification, diagnosis and patient support concerning visual impairment.
The focus for this workshop is to include perspectives from relevant health professionals in the design phase. The health professionals are asked to identify possibilities and barriers of an intervention concerning vision impairment in general practice, including a discussion on key eye examinations and measures that would be feasible in a general practice setting as well as to identify the most relevant eye diseases.
Aims to synthesise the knowledge produced in the previous two workshops by formulating the specific activities, concrete consultation type and identify final intervention effect measurements. This also includes choosing the relevant guidelines for screening and diagnosing to apply in the intervention. Participants are GP’s and project researchers. Other stakeholders will be invited if relevant based on the insights from workshop 1+2.
Part 1: pilot test The intervention model will be first be validated and adjusted accordingly by patient representatives and ophthalmologists. The model will then be tested in a general practice setting to establish face validity and adjust according to the experiences. The model must be clinically relevant and feasible for implementation in a clinical practice. Data production continues in this phase through observational studies and interviews with GPs to disseminate the experiences with the model in general practice and whether the model can be part of improved collaboration between health professionals. Part 2: cohort study Based on findings from the pilot test, we expand the intervention by including 10–15 GP practices in the Capital and Zeeland Region, Denmark to participate in testing the GP’s possibilities and barriers to detect visual impairment and vision loss. According to previous literature, we need to include 1500–2000 patients in general practice aged 65 or older to identify 150–200 with visual impairment. The practices will receive the developed intervention model and recruit in up to 18 months. Patients 65+ who consult their GP as part of an annual consultation for a chronic condition will be informed about the study in the waiting room and asked to complete a questionnaire based on the validated Visual functioning Questionnaire-25. The questionnaire measures the dimensions of self-reported vision-targeted health status that are most important to individuals who have chronic eye diseases. A dedicated staff or GP will examine the vision according to the guidelines formulated in the model. If visual impairment is detected, the patient will undergo further examinations assessed by an ophthalmologist. The specifics of the examinations in general practice and at an ophthalmologist cannot be reported until the phases I+II have been completed, since these are to be developed in the codesign process. At present, we assume an ordinary vision test, visual fields and contrast vision as well as a function test could be included in the GP setting. We assume that measurement of the visual acuity, tonometry, macular and parapapillary optical coherence tomography (OCT) scans and fundus photographs could be part of the extended examinations at an ophthalmologist. An important outcome of phase II is thus detailed information on effect measurements, chosen guidelines and possibly a narrowed focus on specific eye diseases to address in the intervention. The results from the cohort study will function as a reference standard and allow us to study the prevalence of visual impairments as well as eye diseases and study predictor’s for visual impairment as well. Due to the Danish registers, it is possible to follow the cohort for a long period. This follow-up study is not part of the present project, but will be planned later.
The intervention model will be first be validated and adjusted accordingly by patient representatives and ophthalmologists. The model will then be tested in a general practice setting to establish face validity and adjust according to the experiences. The model must be clinically relevant and feasible for implementation in a clinical practice. Data production continues in this phase through observational studies and interviews with GPs to disseminate the experiences with the model in general practice and whether the model can be part of improved collaboration between health professionals.
Based on findings from the pilot test, we expand the intervention by including 10–15 GP practices in the Capital and Zeeland Region, Denmark to participate in testing the GP’s possibilities and barriers to detect visual impairment and vision loss. According to previous literature, we need to include 1500–2000 patients in general practice aged 65 or older to identify 150–200 with visual impairment. The practices will receive the developed intervention model and recruit in up to 18 months. Patients 65+ who consult their GP as part of an annual consultation for a chronic condition will be informed about the study in the waiting room and asked to complete a questionnaire based on the validated Visual functioning Questionnaire-25. The questionnaire measures the dimensions of self-reported vision-targeted health status that are most important to individuals who have chronic eye diseases. A dedicated staff or GP will examine the vision according to the guidelines formulated in the model. If visual impairment is detected, the patient will undergo further examinations assessed by an ophthalmologist. The specifics of the examinations in general practice and at an ophthalmologist cannot be reported until the phases I+II have been completed, since these are to be developed in the codesign process. At present, we assume an ordinary vision test, visual fields and contrast vision as well as a function test could be included in the GP setting. We assume that measurement of the visual acuity, tonometry, macular and parapapillary optical coherence tomography (OCT) scans and fundus photographs could be part of the extended examinations at an ophthalmologist. An important outcome of phase II is thus detailed information on effect measurements, chosen guidelines and possibly a narrowed focus on specific eye diseases to address in the intervention. The results from the cohort study will function as a reference standard and allow us to study the prevalence of visual impairments as well as eye diseases and study predictor’s for visual impairment as well. Due to the Danish registers, it is possible to follow the cohort for a long period. This follow-up study is not part of the present project, but will be planned later.
The analytical process will be carried out iteratively, which covers analytical steps taken between each of the three project phases to ensure appropriate adjustments in the design of the intervention model. All formal interviews from the study will be audio recorded and transcribed following a project guideline to ensure uniformity. The three phases will generate observations and informal talks, which will be documented through field notes and pictures. Details on the analytical steps are illustrated in .
Ethical issues will be a consideration at all levels of the study both when involving patients in the participatory design and during the cohort study. The study is registered in the records of research projects containing personal data at University of Copenhagen (J.nr: 514-0701/22-3000). It will be conducted according to the ethical standards of the Declaration of Helsinki and general data protection regulations (GDPR). Ethical approval was waived by the Danish National Ethical Research Committee because no biomaterial is included in the study. In data production, all participants will be asked to read and sign a consent form regarding their specific participation and kind of information, including how we will handle the information provided to us as, well as information on how to withdraw consent at a later stage. In the qualitative data production, we will produce photographical and graphical material, which requires further ethical reflections in terms of anonymisation. The cohort study involves a risk of overdiagnostic practice due to the tests and screening involved. Any potential harms, overdiagnosis, labelling effect and consequences of receiving the intervention will be scrutinised during the study. Age-related visual impairment diagnoses including glaucoma and AMD meet the requirements for screening formulated by WHO. During the analysis, both benefits and harms of the intervention will be investigated and presented as results. Results will be published in peer-reviewed journals, preferably open-access. Patients, relatives and health professionals are invited in as coauthors where relevant. We will present our results in relevant fora nationally and internationally (conferences, annual meetings, etc). In addition, we will organise a symposium directed at stakeholders from health and social care sector and employers. The participants from the three content-developing workshops in phase II will be invited to participate in the symposia and share their experiences of being part of the research process. For communication to lay persons, we will produce a podcast on sensory loss in old age focused on vision and participate in the yearly Danish democracy and community festival ‘Folkemødet’, which has a specific focus on communicating public health science. The proposed study is relevant for ensuring kind and empathic care with time to guide and comfort patients. This requires knowledge about how the patient experiences visual impairment as well as identification of the current challenges in the health services provided, which we aim to improve following the DETECT intervention. Specifically relevant in this study is the focus on general practice in relation to visual impairment, which is currently an understudied area. Implications Collectively, the output of intervention will help us understand, how to support and treat patients with impaired vision and to define an expedient role for general practice. In this respect, adding knowledge on the GP perspective will strengthen the feasibility of the intervention. The development of a codesigned intervention can have an important impact on the delivered quality in the diagnosis and management of patients with visual impairment in primary care.
Collectively, the output of intervention will help us understand, how to support and treat patients with impaired vision and to define an expedient role for general practice. In this respect, adding knowledge on the GP perspective will strengthen the feasibility of the intervention. The development of a codesigned intervention can have an important impact on the delivered quality in the diagnosis and management of patients with visual impairment in primary care.
Reviewer comments Author's
manuscript
|
Comparison of cone-beam computed tomography with photon-counting detector computed tomography for dental implant surgery | 51202ba9-9e23-4d3c-8dc5-b15bfc920c5d | 11906956 | Musculoskeletal System[mh] | In the event of tooth loss, several potential treatment options are available to address the resulting gap and its impact on multiple domains of oral health-related quality of life (OHRQoL). Implants have been demonstrated to offer the most favorable long-term outcomes compared to other restorative procedures, with a success rate exceeding 95% after ten years . Achieving a favorable surgical and prosthetic outcome in dental implant surgery requires the implementation of an individualized, multidisciplinary therapeutic strategy that is founded upon the findings of a comprehensive clinical and radiological assessment. Cone-beam computed tomography (CBCT) is frequently utilized as an adjunct to conventional two-dimensional radiography in surgical treatment planning. The exposure levels range from approximately 18 to 200 µSv per scan, depending on the indication, exposure settings, and region of interest . The integration of CBCT into routine clinical workflows is supported by its accuracy and reliability in cross-sectional structural analysis of bone density and volume [ – ]. Implant planning using CBCT enables precise implant positioning, prosthetically guided placement, and the prevention of damage to critical structures such as neurovascular structures or the maxillary sinus . Additionally, CBCT helps prevent implant failure caused by inadequate bone volume surrounding the implant or improper placement due to poor bone structure . The increasing use of CBCT scans, while providing valuable perioperative insights into surgically relevant parameters at implant sites, raises concerns about cumulative radiation exposure, particularly in genetically susceptible younger patients. Repeated imaging may elevate the risk of adverse health effects, potentially increasing the likelihood of developing thyroid cancer and meningiomas . In light of the proposed shift from the “as low as reasonably achievable” (ALARA) principle to the “as low as diagnostically acceptable” (ALADA) principle and its subsequent adaptation to “as low as diagnostically acceptable being indication-oriented and patient specific” (ALADAIP) principle , studies are currently examining low-dose CBCT protocols in the context of dental implant surgery . Preliminary findings suggest the feasibility of using low-dose CBCTs perioperatively compared to standard-dose CBCTs . However, there are limitations in the generalizability of these results, particularly regarding reliability and validity in the posterior mandible and the standardization of measurement sites . Photon-counting detectors (PCD) are the most recent development in computed tomography (CT) and employ semiconductors to directly convert incoming X-ray photons into electrical signals, providing superior spatial resolution, less electronic noise, enhanced contrast-to-noise ratio, and distinctive spectral features . PCD-CT addresses shortcomings of CBCT in dental implant imaging, such as low spatial resolution, deficiencies in soft-tissue contrast, and increased susceptibility to metal artifacts . In the ultra-high-resolution mode of PCD-CT, detector pixels measure 0.151 × 0.176 mm 2 at the isocenter, which translates to a maximum spatial in-plane image resolution of 0.11 mm and a maximum through-plane resolution of 0.16 mm . Thus, PCD-CT scans offer spatial resolutions of < 200 μm, comparable to or even superior to dental CBCT, and provide high-quality volumetric imaging with enhanced hard and soft tissue contrast and shorter scan times . The use of PCD-CT in dentomaxillofacial imaging has been investigated in only a few studies, yet the findings are promising, including accurate anatomical depiction, improved reduction of metal artifacts from dental implants , and improved radiation dose efficiency, with doses as low as a quarter of those used in CBCT . A feasibility study directly compared artifacts from a titanium implant using CBCT and PCD-CT, supporting PCD-CT’s superior artifact-reduction potential . To the best of the author’s knowledge, this is the first study to directly compare PCD-CT with CBCT in the same samples at equivalent radiation doses for dental implant imaging. The objective of this ex vivo study was to compare CBCT with PCD-CT at equivalent radiation doses, focusing on qualitative and quantitative parameters relevant to dental implant surgery. Given PCD-CT’s potential for superior artifact reduction, improved radiation dose efficiency, and high image quality at lower doses, this study investigates its ability to improve diagnostic accuracy, minimize implant planning errors, and thus support more precise clinical decision making.
Study design and ethics In this ex vivo comparative study, six pig mandibles were obtained from a local butcher’s shop in Zurich, Switzerland. Twelve dental implants were placed in the pig mandibles in a randomized order. Each mandible received two implants, with one implant placed in each quadrant between the canine and the first premolar. The procedure was conducted by an experienced senior physician (S.V., a board-certified oral surgeon with 11 years of experience). The study utilized implants from four commonly used brands in clinical practice: Dentsply Sirona ( Astra Tech OsseoSpeed ® EV 4.2 S, Mölndal, Sweden ), Nobel Biocare ( NobelActive ® TiUltra ™, Göteborg, Sweden ), Straumann ( Standard Plus SLActive ® , Basel, Switzerland ), and Thommen Medical ( SPI ® ELEMENT Implantat RC INICELL ® , Grenchen, Switzerland ). Due to ethical considerations and radiation safety reasons, conducting this study in a living organism was not feasible . Nevertheless, pig cadavers, which closely align with the human oral and maxillofacial system, are widely regarded as suitable animal models for orofacial research. Consequently, they were used as an alternative in this study . A formal declaration of non-responsibility was provided by the Office of Animal Welfare and 3R of the University of Zurich, confirming that all experiments comply with the Swiss federal guidelines for the use of animals in experimental research. This article’s reporting complies with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. Image acquisition CBCT data acquisition All mandibles were imaged using the Orthophos SL three-dimensional (3D) scanner (Dentsply-Sirona, Bensheim, Germany) following the manufacturer’s predefined standard and low-dose CBCT protocols. The mandibles were centrally positioned and aligned on a platform using the scanner’s positioning lights. To replicate in vivo conditions as accurately as possible, a cold pack (12 × 29 cm, GELLO Geltechnik GmbH, Ahaus, Germany) was placed at the center of each mandible to simulate the presence of soft tissue. The standard-dose imaging protocol was conducted with 85 kV, 13 mA, 4.4 s exposure time, 160 μm pixel size, and an 11 × 10 cm field of view, with an effective dose of 145 µSv. The low-dose imaging protocol used 85 kV, 13 mA, 2.2 s exposure time, 160 μm pixel size, and an 11 × 10 cm field of view, with an effective dose of 20 µSv. PCD-CT data acquisition Scans were acquired on a first-generation dual-source PCD-CT system (NAEOTOM Alpha; Siemens Healthineers AG, Forchheim, Germany) equipped with two cadmium telluride detectors using the ultra-high resolution mode with a detector collimation of 120 × 0.2 mm. The tube voltage was set to 140 kV, using tin pre-filtration, and the pitch factor was 0.85. Radiation doses of the PCD-CT scans were varied to match the doses of the CBCT scans, i.e., equivalent to the standard-dose protocol by adjusting the volume CT dose index (CTDI vol ) to 2.4 mGy, resulting in a dose-length-product (DLP) of 61 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 122 µSv (using a conversion factor of 0.002 mSv \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:\:$$\end{document} mGy − 1 \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm − 1 according to ), and equivalent to the low-dose protocol the by adjusting the CTDI vol to 0.4 mGy, resulting in DLP of 10 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 20 µSv. In addition, CTDI vol was adjusted to 7.0 mGy, resulting in a DLP of 180 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 360 µSv. Scans were reconstructed as polychromatic images (T3D) with a slice thickness of 0.2 mm and an increment of 0.2 mm. The sharp kernel Hr76 was applied, and the matrix size was 1024 × 1024 pixels. Image analysis CBCT and PCD-CT image data were evaluated on a 2-megapixel high-quality liquid-crystal display. A total of 30 scans, compromising 12 CBCT and 18 PCD-CT scans, were assessed by three readers with different levels of experience and specialization. Reader A (A.A.H.) is a resident in oral surgery with four years of experience; Reader B (N.A.H.) is an attending physician, board-certified in reconstructive dentistry, holding a Master of Advanced Studies in reconstructive and implant dentistry, with nine years of experience; and Reader C (S.V.) is a senior physician, board-certified in oral surgery, with 11 years of experience. Prior to the evaluation, a calibration session was conducted to standardize the assessment process. Each examiner received instructions from one of the principal investigators (A.A.H.). To address and eliminate any potential ambiguities, three randomly selected protocols were evaluated. To ensure objective and unbiased assessments, all readers performed the evaluations of the various scans, including protocol and modality, in randomized order. They were blinded to each other’s evaluations, the imaging protocols (low-dose, standard-dose, and high-dose), and the types of implants used. Qualitative measurements The objective analysis of the overall technical image quality and the burden of technical artifacts was graded using a modified 5-point Likert scale, as previously described by Dillinger et al. in the context of PCD-CT dental implant research : 5, excellent image quality with full diagnostic capabilities; 4, good image quality with sufficient diagnostic capabilities; 3, intermediate image quality with restricted diagnostic capabilities; 2, poor image quality with very limited diagnostic capabilities; and 1, indicated very poor image quality, allowing no diagnostic use. For technical artifact burden, the following scale was employed: 5, no artifacts; 4, minimal streaks; 3, intermediate streaks; 2, pronounced streaks; and 1, massive artifacts. The prevalence of hyperdense and hypodense artifacts caused by the placed dental implants was assessed using the following modified 5-Likert Scale according to Patzer et al. : 5, absent/almost absent, no metallic artifact; 4, minor, metallic artifacts are minimal and do not affect scan quality or diagnosis; 3, moderate, metallic artifacts are present and affect overall scan quality, but they do not interfere with the evaluation of adjacent anatomy or diagnosis; 2, pronounced, metallic artifacts are present and affect overall scan quality, and interfere with the evaluation of adjacent anatomy and diagnosis; 1, severe, metallic artifacts are significant, severely affect scan quality, obscure adjacent anatomy, and compromise diagnosis. Additionally, the diagnostic interpretability of soft and hard tissue was evaluated using the following Likert scale : 5, fully diagnostic; 4, minor artifacts with marginal impairment of diagnostic interpretability; 3, artifacts with impaired, mediocre diagnostic interpretability; 2, artifacts with significantly impaired diagnostic interpretability; 1, insufficient interpretability due to artifacts. Quantitative measurements Quantitative analysis of the CBCT and PCD-CT scans relevant to dental implant treatment planning in the posterior mandible was performed on a dedicated, commercially available software (OnDemand 3D, Cybermed, Seoul, Korea). Linear bone measurements in dental implant surgery were conducted according to the method proposed by Kaaber et al. . The image generation process was initiated using the standard-dose CBCT images, where the most appropriate reconstructions, both in the sagittal and coronal planes, were selected for the visualization of the implant site. Subsequently, the defined implant site was replicated using the low-dose CBCT and high, standard, and low PCD-CT images. To obtain measurements at the same site in the specified object, the investigators were instructed to take the measurement along the vertical and horizontal guiding lines. The software automatically sets the window width and window plane for the standard-dose CBCT protocol, thereby standardizing the viewing conditions of the image for all the protocols in each modality. The assessment included measuring bone height, defined as the vertical distance from the alveolar crest to the upper border of the mandibular canal, along the vertical guiding line, and bone width, defined as the horizontal distance three millimeters apical to the alveolar crest between the buccal and lingual cortical boundaries, measured along the horizontal guiding line . The measurements, which were taken in millimeters, were conducted by the same three readers. Accordingly, two implant sites were selected from each quadrant of the posterior mandible, resulting in four bone height and width measurements per mandible for each imaging protocol and reader. If the investigator was uncertain about performing the measurements due to the suboptimal image quality or the presence of artifacts, the case could be classified as unsuitable for measurement. To prevent recall bias, each measurement was repeated after a three-week interval. In the event of a discrepancy between the initial and subsequent measurements, the mean of the two values was calculated. Statistical analysis Descriptive statistics were used to analyze the qualitative data obtained, including overall technical image quality, the burden of technical artifacts, artifacts caused by the placed dental implants, and diagnostic interpretability of soft and hard tissue. The analysis involved calculations of means, standard deviations (SD), medians, minimums, maximums, and ranges. Additionally, inter-reader agreement for the assessed qualitative variables was determined and expressed as a percentage agreement or by analyzing the intraclass correlation coefficient (ICC) type 2:1 and the 95% confidence interval (CI) based on the agreement according to the 2-way random model. In accordance with the selected 95% CI, the ICC values and their agreement beyond chance were interpreted as follows: poor (< 0.5), moderate (0.5–0.75), good (0.75–0.9), and excellent (> 0.9) . Regarding quantitative parameters, the mean difference among readers and the inter-reader reliability of absolute length measurements were evaluated and expressed as ICCs. All statistical analyses were conducted with a significance level of α = 0.05. All statistical analyses were performed using IBM SPSS Statistics software (version 29.0, IBM Chicago, IL, USA).
In this ex vivo comparative study, six pig mandibles were obtained from a local butcher’s shop in Zurich, Switzerland. Twelve dental implants were placed in the pig mandibles in a randomized order. Each mandible received two implants, with one implant placed in each quadrant between the canine and the first premolar. The procedure was conducted by an experienced senior physician (S.V., a board-certified oral surgeon with 11 years of experience). The study utilized implants from four commonly used brands in clinical practice: Dentsply Sirona ( Astra Tech OsseoSpeed ® EV 4.2 S, Mölndal, Sweden ), Nobel Biocare ( NobelActive ® TiUltra ™, Göteborg, Sweden ), Straumann ( Standard Plus SLActive ® , Basel, Switzerland ), and Thommen Medical ( SPI ® ELEMENT Implantat RC INICELL ® , Grenchen, Switzerland ). Due to ethical considerations and radiation safety reasons, conducting this study in a living organism was not feasible . Nevertheless, pig cadavers, which closely align with the human oral and maxillofacial system, are widely regarded as suitable animal models for orofacial research. Consequently, they were used as an alternative in this study . A formal declaration of non-responsibility was provided by the Office of Animal Welfare and 3R of the University of Zurich, confirming that all experiments comply with the Swiss federal guidelines for the use of animals in experimental research. This article’s reporting complies with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines.
CBCT data acquisition All mandibles were imaged using the Orthophos SL three-dimensional (3D) scanner (Dentsply-Sirona, Bensheim, Germany) following the manufacturer’s predefined standard and low-dose CBCT protocols. The mandibles were centrally positioned and aligned on a platform using the scanner’s positioning lights. To replicate in vivo conditions as accurately as possible, a cold pack (12 × 29 cm, GELLO Geltechnik GmbH, Ahaus, Germany) was placed at the center of each mandible to simulate the presence of soft tissue. The standard-dose imaging protocol was conducted with 85 kV, 13 mA, 4.4 s exposure time, 160 μm pixel size, and an 11 × 10 cm field of view, with an effective dose of 145 µSv. The low-dose imaging protocol used 85 kV, 13 mA, 2.2 s exposure time, 160 μm pixel size, and an 11 × 10 cm field of view, with an effective dose of 20 µSv. PCD-CT data acquisition Scans were acquired on a first-generation dual-source PCD-CT system (NAEOTOM Alpha; Siemens Healthineers AG, Forchheim, Germany) equipped with two cadmium telluride detectors using the ultra-high resolution mode with a detector collimation of 120 × 0.2 mm. The tube voltage was set to 140 kV, using tin pre-filtration, and the pitch factor was 0.85. Radiation doses of the PCD-CT scans were varied to match the doses of the CBCT scans, i.e., equivalent to the standard-dose protocol by adjusting the volume CT dose index (CTDI vol ) to 2.4 mGy, resulting in a dose-length-product (DLP) of 61 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 122 µSv (using a conversion factor of 0.002 mSv \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:\:$$\end{document} mGy − 1 \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm − 1 according to ), and equivalent to the low-dose protocol the by adjusting the CTDI vol to 0.4 mGy, resulting in DLP of 10 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 20 µSv. In addition, CTDI vol was adjusted to 7.0 mGy, resulting in a DLP of 180 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 360 µSv. Scans were reconstructed as polychromatic images (T3D) with a slice thickness of 0.2 mm and an increment of 0.2 mm. The sharp kernel Hr76 was applied, and the matrix size was 1024 × 1024 pixels.
All mandibles were imaged using the Orthophos SL three-dimensional (3D) scanner (Dentsply-Sirona, Bensheim, Germany) following the manufacturer’s predefined standard and low-dose CBCT protocols. The mandibles were centrally positioned and aligned on a platform using the scanner’s positioning lights. To replicate in vivo conditions as accurately as possible, a cold pack (12 × 29 cm, GELLO Geltechnik GmbH, Ahaus, Germany) was placed at the center of each mandible to simulate the presence of soft tissue. The standard-dose imaging protocol was conducted with 85 kV, 13 mA, 4.4 s exposure time, 160 μm pixel size, and an 11 × 10 cm field of view, with an effective dose of 145 µSv. The low-dose imaging protocol used 85 kV, 13 mA, 2.2 s exposure time, 160 μm pixel size, and an 11 × 10 cm field of view, with an effective dose of 20 µSv.
Scans were acquired on a first-generation dual-source PCD-CT system (NAEOTOM Alpha; Siemens Healthineers AG, Forchheim, Germany) equipped with two cadmium telluride detectors using the ultra-high resolution mode with a detector collimation of 120 × 0.2 mm. The tube voltage was set to 140 kV, using tin pre-filtration, and the pitch factor was 0.85. Radiation doses of the PCD-CT scans were varied to match the doses of the CBCT scans, i.e., equivalent to the standard-dose protocol by adjusting the volume CT dose index (CTDI vol ) to 2.4 mGy, resulting in a dose-length-product (DLP) of 61 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 122 µSv (using a conversion factor of 0.002 mSv \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:\:$$\end{document} mGy − 1 \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm − 1 according to ), and equivalent to the low-dose protocol the by adjusting the CTDI vol to 0.4 mGy, resulting in DLP of 10 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 20 µSv. In addition, CTDI vol was adjusted to 7.0 mGy, resulting in a DLP of 180 mGy \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\bullet\:$$\end{document} cm and an effective dose of 360 µSv. Scans were reconstructed as polychromatic images (T3D) with a slice thickness of 0.2 mm and an increment of 0.2 mm. The sharp kernel Hr76 was applied, and the matrix size was 1024 × 1024 pixels.
CBCT and PCD-CT image data were evaluated on a 2-megapixel high-quality liquid-crystal display. A total of 30 scans, compromising 12 CBCT and 18 PCD-CT scans, were assessed by three readers with different levels of experience and specialization. Reader A (A.A.H.) is a resident in oral surgery with four years of experience; Reader B (N.A.H.) is an attending physician, board-certified in reconstructive dentistry, holding a Master of Advanced Studies in reconstructive and implant dentistry, with nine years of experience; and Reader C (S.V.) is a senior physician, board-certified in oral surgery, with 11 years of experience. Prior to the evaluation, a calibration session was conducted to standardize the assessment process. Each examiner received instructions from one of the principal investigators (A.A.H.). To address and eliminate any potential ambiguities, three randomly selected protocols were evaluated. To ensure objective and unbiased assessments, all readers performed the evaluations of the various scans, including protocol and modality, in randomized order. They were blinded to each other’s evaluations, the imaging protocols (low-dose, standard-dose, and high-dose), and the types of implants used.
The objective analysis of the overall technical image quality and the burden of technical artifacts was graded using a modified 5-point Likert scale, as previously described by Dillinger et al. in the context of PCD-CT dental implant research : 5, excellent image quality with full diagnostic capabilities; 4, good image quality with sufficient diagnostic capabilities; 3, intermediate image quality with restricted diagnostic capabilities; 2, poor image quality with very limited diagnostic capabilities; and 1, indicated very poor image quality, allowing no diagnostic use. For technical artifact burden, the following scale was employed: 5, no artifacts; 4, minimal streaks; 3, intermediate streaks; 2, pronounced streaks; and 1, massive artifacts. The prevalence of hyperdense and hypodense artifacts caused by the placed dental implants was assessed using the following modified 5-Likert Scale according to Patzer et al. : 5, absent/almost absent, no metallic artifact; 4, minor, metallic artifacts are minimal and do not affect scan quality or diagnosis; 3, moderate, metallic artifacts are present and affect overall scan quality, but they do not interfere with the evaluation of adjacent anatomy or diagnosis; 2, pronounced, metallic artifacts are present and affect overall scan quality, and interfere with the evaluation of adjacent anatomy and diagnosis; 1, severe, metallic artifacts are significant, severely affect scan quality, obscure adjacent anatomy, and compromise diagnosis. Additionally, the diagnostic interpretability of soft and hard tissue was evaluated using the following Likert scale : 5, fully diagnostic; 4, minor artifacts with marginal impairment of diagnostic interpretability; 3, artifacts with impaired, mediocre diagnostic interpretability; 2, artifacts with significantly impaired diagnostic interpretability; 1, insufficient interpretability due to artifacts.
Quantitative analysis of the CBCT and PCD-CT scans relevant to dental implant treatment planning in the posterior mandible was performed on a dedicated, commercially available software (OnDemand 3D, Cybermed, Seoul, Korea). Linear bone measurements in dental implant surgery were conducted according to the method proposed by Kaaber et al. . The image generation process was initiated using the standard-dose CBCT images, where the most appropriate reconstructions, both in the sagittal and coronal planes, were selected for the visualization of the implant site. Subsequently, the defined implant site was replicated using the low-dose CBCT and high, standard, and low PCD-CT images. To obtain measurements at the same site in the specified object, the investigators were instructed to take the measurement along the vertical and horizontal guiding lines. The software automatically sets the window width and window plane for the standard-dose CBCT protocol, thereby standardizing the viewing conditions of the image for all the protocols in each modality. The assessment included measuring bone height, defined as the vertical distance from the alveolar crest to the upper border of the mandibular canal, along the vertical guiding line, and bone width, defined as the horizontal distance three millimeters apical to the alveolar crest between the buccal and lingual cortical boundaries, measured along the horizontal guiding line . The measurements, which were taken in millimeters, were conducted by the same three readers. Accordingly, two implant sites were selected from each quadrant of the posterior mandible, resulting in four bone height and width measurements per mandible for each imaging protocol and reader. If the investigator was uncertain about performing the measurements due to the suboptimal image quality or the presence of artifacts, the case could be classified as unsuitable for measurement. To prevent recall bias, each measurement was repeated after a three-week interval. In the event of a discrepancy between the initial and subsequent measurements, the mean of the two values was calculated.
Descriptive statistics were used to analyze the qualitative data obtained, including overall technical image quality, the burden of technical artifacts, artifacts caused by the placed dental implants, and diagnostic interpretability of soft and hard tissue. The analysis involved calculations of means, standard deviations (SD), medians, minimums, maximums, and ranges. Additionally, inter-reader agreement for the assessed qualitative variables was determined and expressed as a percentage agreement or by analyzing the intraclass correlation coefficient (ICC) type 2:1 and the 95% confidence interval (CI) based on the agreement according to the 2-way random model. In accordance with the selected 95% CI, the ICC values and their agreement beyond chance were interpreted as follows: poor (< 0.5), moderate (0.5–0.75), good (0.75–0.9), and excellent (> 0.9) . Regarding quantitative parameters, the mean difference among readers and the inter-reader reliability of absolute length measurements were evaluated and expressed as ICCs. All statistical analyses were conducted with a significance level of α = 0.05. All statistical analyses were performed using IBM SPSS Statistics software (version 29.0, IBM Chicago, IL, USA).
Three readers qualitatively evaluated 30 DICOM data sets (12 CBCT, 18 PCD-CT), while each reader quantitatively assessed 24 implant planning sites in the posterior mandible per imaging protocol, resulting in a total of 120 implant sites per reader. Qualitative results The mean overall technical image quality ranged from good to excellent, demonstrating full diagnostic capabilities for high-dose PCD-CT (4.89 ± 0.27), standard-dose PCD-CT (4.50 ± 0.73), and standard-dose CBCT (4.33 ± 0.61). In contrast, low-dose protocols for both modalities showed intermediate image quality with restricted diagnostic capabilities (CBCT: 3.22 ± 0.35; PCD-CT: 3.17 ± 0.63). The artifact burden was most pronounced among all readers in the low-dose protocols for both imaging modalities. Notably, standard-dose PCD-CT (4.11 ± 0.27) exhibited lower artifact susceptibility compared to standard-dose CBCT (3.67 ± 0.52), with the high-dose PCD-CT protocol demonstrating the least artifact burden (4.89 ± 0.27). Hypodense and hyperdense artifacts related to the material properties of implants were either absent or minimal in high-dose PCD-CT scans (4.67 ± 0.52). Standard-dose PCD-CT scans demonstrated minor metallic artifacts (4.17 ± 0.82), not compromising image quality or diagnostic accuracy. In contrast, standard-dose CBCT exhibited moderate metallic artifacts (3.45 ± 0.59) that significantly reduced overall scan quality (Figs. and ). The diagnostic interpretability of both hard and soft tissues was generally rated as fully diagnostic, with minor artifacts causing only marginal impairment. Among the evaluated protocols, the high-dose PCD-CT achieved the highest scores for diagnostic interpretability (hard tissue: 4.83 ± 0.41; soft tissue: 4.44 ± 0.52). Standard-dose PCD-CT also outperformed the radiation dose-equivalent CBCT protocol for both hard tissue (4.56 ± 0.53 vs. 3.33 ± 0.52) and soft tissue (3.78 ± 0.76 vs. 2.44 ± 0.54) visualization (Fig. ). Results of all assessed qualitative parameters are presented in Table . The inter-reader agreement for all parameters ranged from moderate to good across all imaging modalities and protocols, with intraclass correlation coefficient (ICC) values reaching up to 0.89 (range: 0.70–0.89). The highest reliability was observed in the high-dose PCD-CT and the standard-dose PCD-CT and CBCT protocols, whereas the low-dose protocols yielded the lowest ICC values. All parameters demonstrated statistically significant agreement, with p -values < 0.05. Detailed inter-reader agreement for each parameter, assessed across the three readers within each modality and protocol, is presented in Table . Regarding manufacturer-specific differences in implant susceptibility to artifacts, high-dose PCD-CT consistently produced the fewest artifacts and the highest image quality across all implant manufacturers. Notably, no specific implant subtype was associated with a significantly higher incidence of artifacts (Table ). Quantitative results The quantitative analysis of linear bone measurements in the posterior mandible revealed that one case exhibited non-measurable height measurements, while all cases yielded successfully obtained width measurements. The mean differences among the three readers were minimal for both height and width measurements, demonstrating that both imaging modalities fulfill the clinical precision requirements for implant planning across different radiation levels, regardless of the reader’s experience or subspecialty. The inter-reader agreement for absolute length measurements was rated as good to excellent across all imaging protocols for both height and width measurements, with ICC ranges of 0.793 to 1.0 and 0.817 to 0.991 (all p < 0.001), respectively. A protocol-specific analysis revealed that both high-dose and standard-dose protocols demonstrated higher inter-reader agreement than low-dose protocols, regardless of the imaging modality. Notably, PCD-CT low-dose protocols demonstrated superior performance to CBCT protocols at equivalent radiation doses in most inter-reader comparisons, with the exception of one instance (Tables and ).
The mean overall technical image quality ranged from good to excellent, demonstrating full diagnostic capabilities for high-dose PCD-CT (4.89 ± 0.27), standard-dose PCD-CT (4.50 ± 0.73), and standard-dose CBCT (4.33 ± 0.61). In contrast, low-dose protocols for both modalities showed intermediate image quality with restricted diagnostic capabilities (CBCT: 3.22 ± 0.35; PCD-CT: 3.17 ± 0.63). The artifact burden was most pronounced among all readers in the low-dose protocols for both imaging modalities. Notably, standard-dose PCD-CT (4.11 ± 0.27) exhibited lower artifact susceptibility compared to standard-dose CBCT (3.67 ± 0.52), with the high-dose PCD-CT protocol demonstrating the least artifact burden (4.89 ± 0.27). Hypodense and hyperdense artifacts related to the material properties of implants were either absent or minimal in high-dose PCD-CT scans (4.67 ± 0.52). Standard-dose PCD-CT scans demonstrated minor metallic artifacts (4.17 ± 0.82), not compromising image quality or diagnostic accuracy. In contrast, standard-dose CBCT exhibited moderate metallic artifacts (3.45 ± 0.59) that significantly reduced overall scan quality (Figs. and ). The diagnostic interpretability of both hard and soft tissues was generally rated as fully diagnostic, with minor artifacts causing only marginal impairment. Among the evaluated protocols, the high-dose PCD-CT achieved the highest scores for diagnostic interpretability (hard tissue: 4.83 ± 0.41; soft tissue: 4.44 ± 0.52). Standard-dose PCD-CT also outperformed the radiation dose-equivalent CBCT protocol for both hard tissue (4.56 ± 0.53 vs. 3.33 ± 0.52) and soft tissue (3.78 ± 0.76 vs. 2.44 ± 0.54) visualization (Fig. ). Results of all assessed qualitative parameters are presented in Table . The inter-reader agreement for all parameters ranged from moderate to good across all imaging modalities and protocols, with intraclass correlation coefficient (ICC) values reaching up to 0.89 (range: 0.70–0.89). The highest reliability was observed in the high-dose PCD-CT and the standard-dose PCD-CT and CBCT protocols, whereas the low-dose protocols yielded the lowest ICC values. All parameters demonstrated statistically significant agreement, with p -values < 0.05. Detailed inter-reader agreement for each parameter, assessed across the three readers within each modality and protocol, is presented in Table . Regarding manufacturer-specific differences in implant susceptibility to artifacts, high-dose PCD-CT consistently produced the fewest artifacts and the highest image quality across all implant manufacturers. Notably, no specific implant subtype was associated with a significantly higher incidence of artifacts (Table ).
The quantitative analysis of linear bone measurements in the posterior mandible revealed that one case exhibited non-measurable height measurements, while all cases yielded successfully obtained width measurements. The mean differences among the three readers were minimal for both height and width measurements, demonstrating that both imaging modalities fulfill the clinical precision requirements for implant planning across different radiation levels, regardless of the reader’s experience or subspecialty. The inter-reader agreement for absolute length measurements was rated as good to excellent across all imaging protocols for both height and width measurements, with ICC ranges of 0.793 to 1.0 and 0.817 to 0.991 (all p < 0.001), respectively. A protocol-specific analysis revealed that both high-dose and standard-dose protocols demonstrated higher inter-reader agreement than low-dose protocols, regardless of the imaging modality. Notably, PCD-CT low-dose protocols demonstrated superior performance to CBCT protocols at equivalent radiation doses in most inter-reader comparisons, with the exception of one instance (Tables and ).
The objective of this ex vivo study was to evaluate the diagnostic interpretability of PCD-CT imaging protocols compared to clinically established CBCT protocols across varying equivalent radiation exposure levels, focusing on surgically relevant qualitative and quantitative parameters in dental implant surgery. Evaluating 30 reconstructions, high-dose PCD-CT demonstrated the best overall performance, offering superior image quality, minimal artifacts, and excellent diagnostic accuracy. At equivalent radiation exposure levels, PCD-CT performed superior to CBCT, while low-dose protocols of both modalities exhibited intermediate image quality and limited diagnostic capabilities. Both modalities achieved high inter-reader agreement and met clinical quantitative precision requirements regarding implant planning capabilities. However, PCD-CT consistently exhibited superior performance in reducing implant-induced metallic artifacts across all dose levels and implant manufacturers. These findings are in line with previous studies that highlight the advanced capabilities of PCD-CT for dental implant imaging . Dentomaxillofacial imaging constitutes the most frequently performed X-ray-based procedures globally, particularly in healthy individuals, accounting for up to 46% of all biomedical imaging procedures . Although the radiation doses associated with individual dentomaxillofacial examinations are relatively low compared to those used in the medical field, it is nevertheless essential to optimize this exposure in accordance with the “ALADAIP” principle . This is because even small doses of X-ray exposure can increase the likelihood of adverse health outcomes . This study’s findings suggest that low-dose protocols are feasible for diagnostic purposes in both imaging modalities, with PCD-CT offering superior image quality compared to CBCT . These findings align with prior research supporting low-dose CBCT in dental implant imaging and further highlight PCD-CT’s potential to reduce radiation exposure in clinical practice . Our results indicate that PCD-CT provides superior image quality and reduced susceptibility to technical artifacts compared to CBCT. This advantage was observed consistently across both standard and low-dose protocols, with robust inter-reader agreement across all protocols, ranging from moderate to good. These findings are consistent with those of previous studies, which reported significantly higher image quality and approximately 30% improved contrast-to-noise ratios for PCD-CT compared to CBCT . Moreover, a qualitative assessment of dentomaxillofacial structures based on Likert rating scales showed that, in a low-dose inter-modality comparison, PCD-CT reconstructions achieved superior depiction with higher ratings and inter-reader agreement (ICC > 0.6; p < 0.05) . These results highlight the enhanced diagnostic clarity provided by PCD-CT while maintaining diagnostic accuracy and minimizing radiation exposure, which is further supported by the findings of this study’s low-dose intermodality comparison. In this study, all PCD-CT scans were acquired in the ultra-high resolution mode with dose settings matching the CBCT scans to ensure comparability. The findings of this study indicate that PCD-CT scans reconstructed with a slice thickness of 0.2 mm, when using alongside established protocols, are a reliable and versatile imaging option for perioperative dental implant surgery. These approaches enabled detailed visualization of soft- and hard-tissue structures and reduced implant-related artifacts. In addition to overcoming several inherent limitations of CBCT, this technique could also reveal pathologies that would otherwise be obscured by metallic artifacts. This enhanced clarity may facilitate the early detection and diagnosis of these conditions, thus, potentially improving clinical outcomes especially in anatomically complex regions. Overall, this study highlights the clinical value of ultra-high resolution PCD-CT scans, offering a balanced combination of diagnostic interpretability and artifact reduction at low radiation dose. To the best of our knowledge, this is the first study to evaluate the quantitative assessment of bone height and width in PCD-CT, specifically in the context of dental implant placement. The findings of this study demonstrate that PCD-CT provides accurate and reliable measurements, with minimal differences compared to CBCT. Moreover, PCD-CT fulfills the clinical precision standards necessary for implant planning, regardless of the radiation dose applied or the reader’s experience and subspecialty. This allows for a precise radiological assessment of osseous parameters essential for the long-term stability and successful integration of the implant to be performed prior to surgical intervention . While this study design provides valuable insights, it is important to acknowledge the methodology’s inherent limitations. First, although pig mandibles are commonly used in dentomaxillofacial research, they do not fully replicate the anatomical conditions observed in human clinical settings. However, this limitation applies equally to both imaging modalities, ensuring the comparability of the acquired data. Second, the small sample size and the exclusive use of titanium as an implant material limit the generalizability of our findings. Further research with larger sample sizes is mandatory to investigate whether the observed differences between imaging modalities are applicable to a broader range of implant manufacturers, designs, materials, placement techniques, or anatomical variations in different regions of the mandible. Third, the higher cost, limited availability of PCD-CT scanners, and increased staffing and logistical requirements currently pose challenges to their implementation in dental practices. Moreover, validation in human subjects is essential to assess the clinical applicability of PCD-CT in dentomaxillofacial workflows, particularly in comparison with CBCT or other established radiological methods.
This comparative ex vivo study demonstrates the potential of PCD-CT to impact dental implant imaging significantly. At equivalent radiation doses, PCD-CT offers several advantages over conventional CBCT, including excellent spatial resolution and a significant reduction in implant-related artifacts at all radiation exposure levels. Moreover, PCD-CT demonstrates comparable reliability in quantitative measurements essential for implant planning. With ongoing advancements in cost-effectiveness, accessibility, and validation in clinical trials, PCD-CT has the potential to emerge as a promising perioperative diagnostic modality. From a clinical standpoint, low-dose PCD-CT protocols, maintaining both quantitative and qualitative diagnostic accuracy in treatment planning, can play a significant role in enhancing patient safety by reducing radiation exposure. This study further highlights a more personalized imaging approach that balances radiation dose and diagnostic performance according to the indication-specific requirements.
|
Revealing the structural microenvironment of high metastatic risk uveal melanomas following decellularisation | 57788ef5-4351-4f62-bb80-488235ae79c9 | 11538295 | Biochemistry[mh] | Uveal melanoma (UM) is a rare aggressive intraocular tumour that accounts for around 13% of all melanoma deaths . Although the primary tumour (pUM) is usually effectively treated, it metastasises in approximately 50% of patients, most commonly to the liver , . Metastatic UM (mUM) has limited treatment options and overall survival is short . The treatment of mUM is a significant challenge due to diffuse dissemination of UM cells in the liver, and a lack of effective systemic treatments, including immunotherapy, in most affected patients. The tumour microenvironment (TME) plays a critical role in tumour growth and the development of metastasis where the interaction between tumour cells and the associated stroma and cellular components, modulate tumour progression and patient prognosis , . We have previously demonstrated that the secretome of pUM with a high risk (HR) of metastatic spread is characterised by an upregulation of ECM proteins (collagens, fibronectin and laminin, and glycoproteins; aggrecan and thrombospondin) compared with low metastatic risk pUM . Bioinformatic analysis of the secretome data also identified hepatic fibrosis as one of the most differentially upregulated biological processes in HR pUM, suggesting that this is a prerequisite during tumour progression . Indeed, although there are variable hepatic growth patterns in mUM, many exhibit a fibrotic wall surrounding the metastatic tumour nodule , . This is often associated with a peri-tumoural distribution of lymphocytes, suggesting that this fibrotic wall may hinder their infiltration into the mUM (i.e., creating an “immune cold” tumour nodule) . In this context, the biochemical composition and three-dimensional (3D) structure of the extracellular matrix (ECM) plays a crucial functional and structural role in establishing permissive conditions for local invasion, metastatic spread and response to therapy . However, our overall understanding of ECM composition and how these processes influence tumour progression and tumour cell behaviour in pUM and mUM is limited. The human matrisome is a complex mixture of structural proteins, proteoglycans and glycoproteins that are important not only as cellular scaffolds but also in the provision of molecular cues that can modulate cellular function , . Recent evidence suggests that both stromal cells and cancer cells can significantly influence matrix structure and composition . Much pre-clinical cancer research is performed using two-dimensional (2D) cell culture systems that poorly recapitulate TME conditions. Although collagen and Matrigel™ can be used to provide a three-dimensional (3D) matrix in culture, the composition and biochemical properties of these may not be representative of the tumour ECM in vivo. Breakthroughs in the field of tissue engineering include the development of tissue decellularisation methods, which strip a tissue of cells to leave the complex composition of human ECM with preserved tissue architecture – . This has been used both to understand biochemical and biomechanical aspects as well as provide a biological scaffold for re-cellularisation, thus restoring the function of damaged or dysfunctional tissues. There has been some progress to use similar methods in the study of tumour progression, but these are currently limited, despite their potential to provide information that can be used to develop models that mimic the metastatic niche. Van Tienderen et al., developed an optimized decellularisation technique to characterize the ECM of hepatocellular carcinoma and cholangiocarcinoma, uncovering distinct malignancy-related ECM signatures that would likely be undetected in proteomic analysis of intact tumour material due to protein abundance and resolution . In addition, Xiong et al., demonstrated differences between the abilities of metastatic and non-metastatic breast cancer cells to colonize and grow in decellularised lung matrix . Our study herein outlines a novel protocol for the decellularisation of pUM tissue that retains its architecture. We examine the protein composition of the decellularised material and identify differences in the ECM signature between high and low-metastatic risk pUM.
Decellularisation of pUM tissue for isolation of extracellular matrix scaffold Primary UM tissue samples of known metastatic risk, with varied pigmentation and structure were decellularised with alternating hypo/hypertonic solutions followed by Triton X-100 incubation (Fig. a,b). Quantitative analysis of DNA content ( n = 5) confirmed successful decellularisation with an average decrease of 87.36% (Fig. c) compared with control tissue. In addition, histological analysis of sections of decellularised tissue demonstrated an absence of DAPI stained nuclei when compared with control material confirming successful removal of cellular content (Fig. d). The PAS and Gomori stains highlighted the positivity of the connective tissue and collagen fibres often surrounding the tubular ghost-like blood vessels and connective tissue loops, in the acellular tissue architecture (Fig. d). Proteomic characterisation of decellularised scaffold reveals ECM-associated proteins in HR pUM Proteomic analysis identified 488 proteins across all eight pUM samples based on 3 or more unique peptide identifications per protein and an FDR of < 1% (Supplementary Table ). Principal Component Analysis (PCA) demonstrated separation of HR-M3 (purple) from LR-D3 (blue) pUM based on protein expression (Fig. a). Differentially expressed proteins between HR-M3 and LR-D3 pUM groups with a log 2 fold change ≥ 2, P < 0.05 identified 93 proteins (Fig. b) and unsupervised hierarchical clustering separated these proteins into HR-M3 and LR-D3 groupings (Fig. c). Interestingly sample S228.13, which shows some similarities with HR-M3 UM (Fig. c) was noted in the pathology report to have a small focal extraocular melanoma extension at primary diagnosis, which can be associated with an increased risk of metastatic spread. GSEA of all 488 proteins revealed several Gene Ontology (GO) terms relating to extracellular encapsulating structure organisation, cell adhesion and integrin mediated signalling pathways, P-adjusted < 0.05 (Fig. d). Using the matrisome database 2.0 to filter the 488 proteins revealed 45 associated with ECM (Supplementary Table ); 31 core matrisome proteins and 14 matrisome-associated proteins (Fig. e). Of the 45 ECM related proteins identified, 34 (76%) were upregulated more than 1.5-fold in the HR-M3 pUM samples (Supplementary Table ) and 14 with statistical significance, P < 0.05 including NID1 , COL6A1 , COL4A2 (proteins in bold Fig. e). Comparative analysis reveals uniquely expressed ECM proteins To further evaluate the importance of the identified decellularised proteins, two additional in-house and publicly available pUM proteomic datasets - iTRAQ (whole cell) and secretome (secreted) - were interrogated in a combined analysis approach (Fig. a). The datasets differed in the number of proteins identified, ranging from decellularised, n = 488, secretome n = 758 and iTRAQ n = 3935, paralleling the source of each dataset - ECM scaffold to secreted proteins to whole cell. Using all three datasets in downstream analyses removes limitations of each proteomic measurement (with regards to complexity and protein abundance) and increases the power of any resulting biological predictions and pathways. For example, only glycoproteins, secreted factors and collagens were present in the iTRAQ dataset whilst ECM affiliated, ECM regulators and proteoglycans were additionally present in both secretome and decellularised datasets (Fig. b). All datasets were filtered using the matrisome 2.0 database and ECM associated proteins for decellularised, secretome (Supplementary Table ) and iTRAQ (Supplementary Table ) were reduced to 45, 115 and 141 proteins, respectively. The combination of the datasets identified 197 unique ECM-related proteins (Fig. b.i., Supplementary Table ). Thirty-one proteins were found in all three datasets including nidogens 1 and 2 ( NID1 , NID2 ), collagens 6A1, A2, A3 ( COL6A1 , COL6A2 , COL6A3 ), transforming growth factor beta 1 ( TGFB1 ), galectin 3 ( LGALS3 ) and tenascin-C ( TN-C ) (Supplementary Table ). Of the 197 proteins, 76 were upregulated in HR-M3 pUM with fold change ≥ 1.5, 64 of those with P < 0.05 significance (Fig. b.ii., Supplementary Table ). These 76 proteins were classified by the matrisome subdivisions and categories , and are shown for each dataset separately and combined (Fig. c). The seven proteins included in the overlap (FC ≥ 1.5, Fig. .b.ii) were collagen 6A3 ( COL6A3 ), laminin B2 ( LAMB2 ), laminin C1 ( LAMC1 ), alpha-2-macroglobulin (A2M), agrin ( AGRN ), cathepsin B ( CTSB ), and collagen 18A1 ( COL18A1 ). Over representation analysis confirms ECM-receptor involvement in KEGG pathways Over-representation analysis was used to identify biological functions and pathways enriched in the combined dataset of the 76 proteins upregulated in HR-M3 pUM samples. GO classifications identified 29 cellular components and 10 molecular functions which included ‘structural molecule activity’, and ‘collagen/laminin/integrin binding’ (Fig. d.i). Kyoto Encyclopaedia of Genes and Genomes (KEGG – ) revealed 16 pathways with ‘ECM-receptor interaction’ and ‘focal adhesion’ as the top two pathways. Other pathways included ‘PI3K-Akt signalling’, ‘apoptosis’, and ‘complement and coagulation cascade’ (Fig. d.ii). Validation of proteins identified as upregulated in HR-M3 pUM pUM IHC was performed on both control and decellularised formalin-fixed paraffin-embedded pUM tissue sections from three LR-D3 and three HR-M3 pUM samples for collagen 6A1, nidogen 1 and collagen 4. Representative images are shown in Fig. . In the control tissue, staining was observed around blood vessels in the normal choroid for all three proteins and in the retina for collagen 4 and nidogen 1, acting as an internal positive control (Supplementary Figure ). In the tumour regions of the LR-D3 and HR-M3 pUM controls, staining was observed for collagen 4 around a few blood vessels and in the HR-M3 pUM collagen 4 also highlighted looping structures where present. Collagen 6A1 was not detected in any of the LR-D3 pUM controls and was noted to be present around blood vessels and fibrous structures in one of the HR-M3 pUM controls. Nidogen 1 was detected surrounding blood vessels and looping structures where present, in all pUM controls analysed irrespective of metastatic risk, although the levels appeared lower in the LR-D3 pUM most likely due to a lower number of blood vessels and absence of looping structures as would be expected for these cases. Interestingly, in the decellularised tissue, protein expression was more visible for all three proteins in both LR-D3 and HR-M3 pUM samples (Fig. ). Negative controls showed no staining in the decellularised tissue (Supplementary Figure ). Although, collagen 6A1, nidogen 1 and collagen 4 staining across the samples varied, the HR-M3 pUM cases appeared to have stronger and more widespread staining than that present in the LR-D3 samples. Moreover, the patterns of staining in the decellularised material were consistent with ghost-like vascular structures , and the ECM scaffold as also shown with SEM in Fig. . Additional staining with CD34 in 2 decellularised cases (S203.21 LR-D3, S295.12 h-M3) confirmed the presence of vascular structures in concordance with PAS staining (Supplementary Figure ). mUM Tissue from three cases of non-decellularised hepatic metastatic UM were also stained with collagen 4, nidogen 1 and collagen 6A1 (Fig. ). The staining was more visible and more widespread than seen in the control pUM samples although staining patterns were similar, highlighting the vascular structures, connective tissue loops and collagen-rich ECM lattice. Scaffold architecture analysis of decellularised tissue by SEM SEM was employed to provide high resolution, 3D images of the structure of pUM samples, both control and decellularised, allowing more in depth topographical, morphological, and compositional detail to be seen (Fig. ). Both control LR-D3 and HR-M3 pUM tissues revealed densely packed cellular structures with cell shape and size easily observed at 2000x magnification. Epithelioid and spindle cell morphology was seen at the highest magnifications and matched the pathological characteristics. All cell types appeared interconnected and in close contact and it was difficult to distinguish any clear fibre structures within control tissues. SEM of the decellularised tissue revealed the effective removal of cellular matter from all tissues and clear fibrillar structures could be seen (Fig. ). In terms of structure, decellularisation of HR-M3 pUM cases revealed ECM that appeared more compact and less porous (Fig. b) than that observed in LR-D3 pUM (Fig. a), possibly indicating remodelling of the interstitial matrix.
Primary UM tissue samples of known metastatic risk, with varied pigmentation and structure were decellularised with alternating hypo/hypertonic solutions followed by Triton X-100 incubation (Fig. a,b). Quantitative analysis of DNA content ( n = 5) confirmed successful decellularisation with an average decrease of 87.36% (Fig. c) compared with control tissue. In addition, histological analysis of sections of decellularised tissue demonstrated an absence of DAPI stained nuclei when compared with control material confirming successful removal of cellular content (Fig. d). The PAS and Gomori stains highlighted the positivity of the connective tissue and collagen fibres often surrounding the tubular ghost-like blood vessels and connective tissue loops, in the acellular tissue architecture (Fig. d).
Proteomic analysis identified 488 proteins across all eight pUM samples based on 3 or more unique peptide identifications per protein and an FDR of < 1% (Supplementary Table ). Principal Component Analysis (PCA) demonstrated separation of HR-M3 (purple) from LR-D3 (blue) pUM based on protein expression (Fig. a). Differentially expressed proteins between HR-M3 and LR-D3 pUM groups with a log 2 fold change ≥ 2, P < 0.05 identified 93 proteins (Fig. b) and unsupervised hierarchical clustering separated these proteins into HR-M3 and LR-D3 groupings (Fig. c). Interestingly sample S228.13, which shows some similarities with HR-M3 UM (Fig. c) was noted in the pathology report to have a small focal extraocular melanoma extension at primary diagnosis, which can be associated with an increased risk of metastatic spread. GSEA of all 488 proteins revealed several Gene Ontology (GO) terms relating to extracellular encapsulating structure organisation, cell adhesion and integrin mediated signalling pathways, P-adjusted < 0.05 (Fig. d). Using the matrisome database 2.0 to filter the 488 proteins revealed 45 associated with ECM (Supplementary Table ); 31 core matrisome proteins and 14 matrisome-associated proteins (Fig. e). Of the 45 ECM related proteins identified, 34 (76%) were upregulated more than 1.5-fold in the HR-M3 pUM samples (Supplementary Table ) and 14 with statistical significance, P < 0.05 including NID1 , COL6A1 , COL4A2 (proteins in bold Fig. e).
To further evaluate the importance of the identified decellularised proteins, two additional in-house and publicly available pUM proteomic datasets - iTRAQ (whole cell) and secretome (secreted) - were interrogated in a combined analysis approach (Fig. a). The datasets differed in the number of proteins identified, ranging from decellularised, n = 488, secretome n = 758 and iTRAQ n = 3935, paralleling the source of each dataset - ECM scaffold to secreted proteins to whole cell. Using all three datasets in downstream analyses removes limitations of each proteomic measurement (with regards to complexity and protein abundance) and increases the power of any resulting biological predictions and pathways. For example, only glycoproteins, secreted factors and collagens were present in the iTRAQ dataset whilst ECM affiliated, ECM regulators and proteoglycans were additionally present in both secretome and decellularised datasets (Fig. b). All datasets were filtered using the matrisome 2.0 database and ECM associated proteins for decellularised, secretome (Supplementary Table ) and iTRAQ (Supplementary Table ) were reduced to 45, 115 and 141 proteins, respectively. The combination of the datasets identified 197 unique ECM-related proteins (Fig. b.i., Supplementary Table ). Thirty-one proteins were found in all three datasets including nidogens 1 and 2 ( NID1 , NID2 ), collagens 6A1, A2, A3 ( COL6A1 , COL6A2 , COL6A3 ), transforming growth factor beta 1 ( TGFB1 ), galectin 3 ( LGALS3 ) and tenascin-C ( TN-C ) (Supplementary Table ). Of the 197 proteins, 76 were upregulated in HR-M3 pUM with fold change ≥ 1.5, 64 of those with P < 0.05 significance (Fig. b.ii., Supplementary Table ). These 76 proteins were classified by the matrisome subdivisions and categories , and are shown for each dataset separately and combined (Fig. c). The seven proteins included in the overlap (FC ≥ 1.5, Fig. .b.ii) were collagen 6A3 ( COL6A3 ), laminin B2 ( LAMB2 ), laminin C1 ( LAMC1 ), alpha-2-macroglobulin (A2M), agrin ( AGRN ), cathepsin B ( CTSB ), and collagen 18A1 ( COL18A1 ).
Over-representation analysis was used to identify biological functions and pathways enriched in the combined dataset of the 76 proteins upregulated in HR-M3 pUM samples. GO classifications identified 29 cellular components and 10 molecular functions which included ‘structural molecule activity’, and ‘collagen/laminin/integrin binding’ (Fig. d.i). Kyoto Encyclopaedia of Genes and Genomes (KEGG – ) revealed 16 pathways with ‘ECM-receptor interaction’ and ‘focal adhesion’ as the top two pathways. Other pathways included ‘PI3K-Akt signalling’, ‘apoptosis’, and ‘complement and coagulation cascade’ (Fig. d.ii).
pUM IHC was performed on both control and decellularised formalin-fixed paraffin-embedded pUM tissue sections from three LR-D3 and three HR-M3 pUM samples for collagen 6A1, nidogen 1 and collagen 4. Representative images are shown in Fig. . In the control tissue, staining was observed around blood vessels in the normal choroid for all three proteins and in the retina for collagen 4 and nidogen 1, acting as an internal positive control (Supplementary Figure ). In the tumour regions of the LR-D3 and HR-M3 pUM controls, staining was observed for collagen 4 around a few blood vessels and in the HR-M3 pUM collagen 4 also highlighted looping structures where present. Collagen 6A1 was not detected in any of the LR-D3 pUM controls and was noted to be present around blood vessels and fibrous structures in one of the HR-M3 pUM controls. Nidogen 1 was detected surrounding blood vessels and looping structures where present, in all pUM controls analysed irrespective of metastatic risk, although the levels appeared lower in the LR-D3 pUM most likely due to a lower number of blood vessels and absence of looping structures as would be expected for these cases. Interestingly, in the decellularised tissue, protein expression was more visible for all three proteins in both LR-D3 and HR-M3 pUM samples (Fig. ). Negative controls showed no staining in the decellularised tissue (Supplementary Figure ). Although, collagen 6A1, nidogen 1 and collagen 4 staining across the samples varied, the HR-M3 pUM cases appeared to have stronger and more widespread staining than that present in the LR-D3 samples. Moreover, the patterns of staining in the decellularised material were consistent with ghost-like vascular structures , and the ECM scaffold as also shown with SEM in Fig. . Additional staining with CD34 in 2 decellularised cases (S203.21 LR-D3, S295.12 h-M3) confirmed the presence of vascular structures in concordance with PAS staining (Supplementary Figure ). mUM Tissue from three cases of non-decellularised hepatic metastatic UM were also stained with collagen 4, nidogen 1 and collagen 6A1 (Fig. ). The staining was more visible and more widespread than seen in the control pUM samples although staining patterns were similar, highlighting the vascular structures, connective tissue loops and collagen-rich ECM lattice.
IHC was performed on both control and decellularised formalin-fixed paraffin-embedded pUM tissue sections from three LR-D3 and three HR-M3 pUM samples for collagen 6A1, nidogen 1 and collagen 4. Representative images are shown in Fig. . In the control tissue, staining was observed around blood vessels in the normal choroid for all three proteins and in the retina for collagen 4 and nidogen 1, acting as an internal positive control (Supplementary Figure ). In the tumour regions of the LR-D3 and HR-M3 pUM controls, staining was observed for collagen 4 around a few blood vessels and in the HR-M3 pUM collagen 4 also highlighted looping structures where present. Collagen 6A1 was not detected in any of the LR-D3 pUM controls and was noted to be present around blood vessels and fibrous structures in one of the HR-M3 pUM controls. Nidogen 1 was detected surrounding blood vessels and looping structures where present, in all pUM controls analysed irrespective of metastatic risk, although the levels appeared lower in the LR-D3 pUM most likely due to a lower number of blood vessels and absence of looping structures as would be expected for these cases. Interestingly, in the decellularised tissue, protein expression was more visible for all three proteins in both LR-D3 and HR-M3 pUM samples (Fig. ). Negative controls showed no staining in the decellularised tissue (Supplementary Figure ). Although, collagen 6A1, nidogen 1 and collagen 4 staining across the samples varied, the HR-M3 pUM cases appeared to have stronger and more widespread staining than that present in the LR-D3 samples. Moreover, the patterns of staining in the decellularised material were consistent with ghost-like vascular structures , and the ECM scaffold as also shown with SEM in Fig. . Additional staining with CD34 in 2 decellularised cases (S203.21 LR-D3, S295.12 h-M3) confirmed the presence of vascular structures in concordance with PAS staining (Supplementary Figure ).
Tissue from three cases of non-decellularised hepatic metastatic UM were also stained with collagen 4, nidogen 1 and collagen 6A1 (Fig. ). The staining was more visible and more widespread than seen in the control pUM samples although staining patterns were similar, highlighting the vascular structures, connective tissue loops and collagen-rich ECM lattice.
SEM was employed to provide high resolution, 3D images of the structure of pUM samples, both control and decellularised, allowing more in depth topographical, morphological, and compositional detail to be seen (Fig. ). Both control LR-D3 and HR-M3 pUM tissues revealed densely packed cellular structures with cell shape and size easily observed at 2000x magnification. Epithelioid and spindle cell morphology was seen at the highest magnifications and matched the pathological characteristics. All cell types appeared interconnected and in close contact and it was difficult to distinguish any clear fibre structures within control tissues. SEM of the decellularised tissue revealed the effective removal of cellular matter from all tissues and clear fibrillar structures could be seen (Fig. ). In terms of structure, decellularisation of HR-M3 pUM cases revealed ECM that appeared more compact and less porous (Fig. b) than that observed in LR-D3 pUM (Fig. a), possibly indicating remodelling of the interstitial matrix.
Our study represents the first comparative analysis of the ECM-specific proteome in pUM of differing metastatic risk. We have: (1) identified a matrisome signature of core ECM and ECM-associated proteins upregulated in HR-M3 pUM when compared with LR-D3 pUM that may play a functional role in UM progression, which could be targeted and/or serve as biomarkers; and (2) developed a method for the decellularisation of pUM tissue that retains much of the biophysical scaffold that may serve as a platform for re-cellularisation and/or inform the creation of bioengineered matrices. In particular, whole tumour tissue decellularisation methods generated a global ECM secreted by both tumour and stromal cells, which removes the limitations of scaffolds created in vitro by or based solely on stromal cells, such as fibroblasts. The number of articles describing the importance of the ECM in tumours and their progression has more than doubled in the last 10 years, although most works have been undertaken on large transcriptomic datasets , , or in mouse models – . Relatively few studies have been undertaken in human tumour tissue and, indeed, this is the first study to use proteomic analysis of HR-M3 pUM and LR-D3 pUM to identify a matrisome profile associated with metastatic risk. In addition, histological, immunohistochemical, and ultrastructural evaluations of decellularised pUM confirmed the preservation of native tissue architecture and major ECM components including laminin, collagens, and glycosaminoglycans (GAG), such as heparin sulfate proteoglycan 2; GAG preservation is crucial to maintain the biological activity of the decellularised scaffolds . Perturbed collagen architectures are often observed surrounding tumours likely due to cellular remodelling of the ECM associated with tumour stiffness. Lysyl oxidases catalyse cross-linking reactions between collagen and elastin to regulate ECM formation, development, maturation, remodelling and thus stiffness . Although ECM stiffness was not examined in this study, LOXL2 and LOXL4 were upregulated in HR-M3 pUM. Kaluz et al., analysed The Cancer Genome Atlas (TCGA) data and reported Col6A1 and Col6A2 as the most abundantly expressed collagen genes in pUM , which is consistent with the abundance of collagens 6A1, A2 and A3 proteins in HR-M3 pUM and the high expression of this collagen in mUM in our study. Similarly, collagen 6 was detected by Daniels et al., in choroidal melanoma by IHC and not in the normal choroid, reflecting the remodelling potential of ECM by tumour cells . Of note, Li et al. who examined the 33 TCGA cancers analysed in detail, reported collagens 6A1, A2 and A3 to be upregulated in several malignancies associated with a TGFβ prominent immune profile and a poor outcome . Their study also reported collagens 6A1, A2 and A3 as potential treatment targets, due to their associations with chemotherapeutic sensitivity/resistance . This could provide a unique opportunity for drug targeting in UM based on collagen 6 expression, exploring the use of antifibrotic drugs. Older UM studies implicated collagen 6 in ECM remodelling and the authors hypothesized that collagen 6 and hyaluron were precursors of vascular networks, e.g., closed ‘vascular connective tissue loops’ that are often seen in poor prognosis UM . Indeed, these closed connective tissue loops were present in all HR-M3 samples used in our study for proteomic analysis; however, the small sample size examined does not allow any further correlations with these features in this study. Kaluz et al., also reported upregulation of the hypoxia regulated genes P4HA1 and P4HA2 in UM patients with metastatic disease, consistent with upregulation of P4HA1 protein in HR-M3 pUM in our study; these genes in turn regulate collagen maturation and deposition. Other collagens shown to be upregulated in the HR-M3 pUM datasets have also been associated with poor prognoses in several tumour types. Collagen 4 is secreted by pancreatic cancer cells, with circulating and stromal collagen 4 associated with poor survival . Type 4 collagens form a supramolecular structure involved in the basement membrane that influences adhesion and migration of epithelial cells . In our study, collagen 4A2 was upregulated in HR-M3 pUM and a pan-collagen 4 antibody identified collagen fibres in both HR-M3 pUM and mUM tissues. Collagen 4 anchors the tumour cells, and a study in UM showed increased cell adhesion with a collagen 4 matrix and invasive UM cells , potentially stimulating growth and migration . Metastases occur in approximately 50% of UM patients via the bloodstream predominantly to the liver . In a prior study, ECM networks were identified in collagen I gel matrices remodelled by isolated pUM cells; and implicated the interaction of laminins and metalloproteinases (MMPs) in the degradation and formation of these networks . Within our datasets collagen 1A1 was upregulated in HR-M3 pUM as was MMP1 and ADAM10. ADAM10 has previously been reported as highly expressed in UM cell lines with gene silencing resulting in reduced invasiveness of 92.1 μm cells . Consistent with analyses in other tumour types, we identified Tenascin-C in high abundance and upregulated in HR-M3 pUM. Tenascin-C has widespread protein distribution in embryonic tissues, where it is found surrounding motile cells , whilst in adult tissues it has a more restricted distribution associated with inflammation and stem cell niches , perhaps linking this ECM protein with a more dedifferentiated phenotype of HR-M3 pUM. Recent studies using a pancreatic neuroendocrine mouse model showed knocking out Tenascin-C decreased angiogenesis thereby suggesting this molecule as an angiogenic modulator , consistent with increased vascular loops in HR-M3 pUM. Tenascin-C has recently been proposed as a prognostic biomarker of increased UM mortality . Several agents either targeting Tenascin-C directly or combination therapies with Tenascin-C antibodies are in clinical trials in other cancers . Metastatic UM is refractory to most immune checkpoint inhibitors and other immunotherapy modalities . Evidence suggests that soluble galectins released in the TME bind specific glycoproteins and glycolipids exposed on the plasma membrane of tumour infiltrating lymphocytes (TIL) modulating their function thus contributing to tumour immune escape . We previously reported upregulation of galectin 3 in hepatic mUM , and both galectin 3 and galectin 7 were identified as part of the HR-M3 upregulated matrisome in this study suggesting the relevance of galectin targeting therapies in mUM. Several other factors detected as upregulated in HR-M3 pUM in this study have previously been associated with tumour progression and metastasis in UM, including growth differentiating factor 15 ( GDF-15 ), thrombospondin 2 ( THBS2 ), and cysteine-rich angiogenic inducer 61 ( CYR61 ) – . Our SEM studies of pUM decellularised ECM are novel; SEM revealed compacted tumour cells in non-decellularised control tissue with few ECM fibres, whilst intact fibres forming complex networks are more demonstrable in the decellularised tissue, both for HR-M3 and LR-D3 tissue. The current study provides proteomic and structural characterisation of the ECM of pUM; however, with metastatic disease remaining the main challenge in UM, characterisation of mUM is vital to better understand the role of ECM in metastases. This study has successfully established a ‘proof of concept’ methodology for effective decellularisation of pUM with the aim of utilising this technique in metastatic UM tissue, which is a limited resource, due to the rare nature of this disease and the small number of liver resections performed per year. Despite this, expression of Collagen 4, Collagen 6A1 and Nidogen 1 proteins by IHC in mUM tissue did show expression levels resembling pUM HR-M3 tissue, and suggested ECM morphology similarities with pUM. Previous studies have also reported morphological and ECM similarities between primary and metastatic UM , , with the latter manipulating the liver parenchyma to create distinct hepatic growth patterns that support its growth , , , . Such ‘mirror-like’ reconstruction of tumour morphology and supporting matrisome has been seen in other malignancies, e.g. metastatic colon cancer and breast cancer , .
Our study provides a novel approach to studying the ECM protein composition of pUM and offers unique insights into metastatic risk-specific protein profiles. These data will be linked to mUM ECM protein composition and stiffness in future studies, to help define proteins involved in guiding tumour cells to the hepatic premetastatic niche. By combining this with the bioinformatic analyses and GO and KEGG pathways already revealed in this study, which highlight ECM receptor interaction and PI3K/AKT signalling, it is now crucial to dissect the molecular mechanisms regulated by these proteins. Critically, the generation of 3D biomaterial scaffolds resulting from these analyses will provide new insights into cellular behaviour, and importantly provide a platform to test novel therapeutics targeting the ECM.
Clinical samples This study conformed to the principles of the Declaration of Helsinki, and all procedures and methods relating to the human tissue used were approved by the Health Research Authority under the REC Ref 11/NW/0568. All samples and pseudo-anonymized data including clinical, histopathological (including nuclear BAP1 (nBAP1) protein expression) and genetic information (chromosome 3 and 8q copy number) were provided by the Ocular Oncology Biobank (REC ref 21/NW/0139). All patients had provided informed consent for the use of their samples and data in research. All samples were snap frozen following isolation and stored long-term at -80 o C or were available as formalin fixed paraffin embedded tissue. DNA quantification was undertaken in initial protocol optimisation on five anonymized primary UM (pUM) samples not included in any downstream processes to confirm acellular material. Proteomic samples Four pUM samples classified using chromosome 3 status as high metastatic risk monosomy 3 (HR-M3) and four classified as low metastatic risk disomy 3 (LR-D3), where the patients had died of metastasis within 5 years of follow-up or were still alive respectively, were selected for decellularisation and proteomic analysis. Immunohistochemistry (IHC) and scanning electron microscopy (SEM) samples Six pUM samples of known metastatic risk were also selected for decellularisation, IHC and SEM analysis. Patient data for pUM samples used in proteomic-, IHC- and SEM analyses are detailed in Table . mUM IHC samples mUM tissue samples ( n = 3) were used for IHC staining with selected antibodies. Decellularisation procedure pUM samples selected had sufficient tissue available to yield two samples between 15 and 35 mg in weight, to provide both ‘control’ (non-decellularised) and ‘decellularised’ specimens. Tissue samples were decellularised in individual wells of a 12-well plate over a 72-hour period. In brief, samples were incubated in 1 ml hypotonic solution (10mM Tris-HCl, 5mM EDTA, pH8.0) for 24 h on a shaking platform at room temperature (RT). The solution was then removed and the tissue washed for 15 min in 1 ml phosphate buffered saline (PBS) before the addition of 1 ml of hypertonic solution (50 mM Tris-HCl, 0.5 M NaCl, 10 mM EDTA, pH8.0) for a further 48 h at RT. Tissue samples were then washed for 15 min in 2 ml PBS before incubation in 0.5% TX100 in dH 2 O for 3 h and a final incubation in dH 2 O for 1 h. Following decellularisation, the tissue pieces were either snap frozen and stored at -80 °C prior to proteomic analyses, fixed in 10% neutral buffered formalin (NBF) for processing and paraffin embedding, fixed in glutaraldehyde/paraformaldehyde (PFA) solution for SEM, or underwent DNA extraction. DNA quantification The efficiency of the tissue decellularisation process was initially confirmed by assessing DNA content across five pUM samples. DNA extraction was performed with the Qiagen Blood and Tissue DNA extraction kit according to manufacturer’s instructions with an overnight extended proteinase K incubation. DNA was quantified using the NanoDrop spectrophotometer (ThermoFisher, UK). Data are mean ± SD, compared using unpaired parametric student’s t-test, P < 0.05 considered significant. Proteomic sample preparation Lysis and protein estimation Each sample was weighed and nine times the weight in volume of complete EDTA free protease inhibitor cocktail in 25 mM Ammonium Bicarbonate ( AmBic) was added (e.g., 10 µg of sample, 90 µL of Lysis buffer). Protein concentration was estimated for each sample using a Bradford assay. In-solution digestion 50 µg total protein from each sample was denatured in 1% Rapigest™ solution heated to 80˚C for 10 min. Cysteine reduction was performed with dithiothreitol in 25 mM Ambic incubated at 60˚C for 10 min. Subsequent alkylation was performed by adding iodoacetamide in 25 mM Ambic and incubating for 30 min in the dark. Digestion was performed with 0.2 µg/µL trypsin (50:1 ratio of sample: trypsin) at 37˚C for 16 h on an orbital shaker. Rapigest™ inactivation was achieved by adding trifluoroacetic acid, ensuring a pH of 2 or less. Finally, samples were incubated at 37˚C for 45 min before spinning at 13,000 x g for 15 min at 7 ˚C. LC-MS analysis Samples were analysed with an Ultimate 3000 RSLC™ nano-system (Thermo Scientific) coupled to a Q Exactive Quadrupole-Orbitrap™ mass spectrometer (Thermo Scientific). The data-dependent program used for data acquisition consisted of a 70,000-resolution full-scan MS scan in the orbitrap. All samples were analysed in random order. Data analysis Data were searched using Proteome Discoverer (v 2.4) and the Mascot search engine (v 2.8) against the UniProt database of human reviewed proteins. Fixed cysteine carbamidomethylation and variable modification of methionine oxidation were specified and limited to 1 missed cleavage. Data was processed using Progenesis. Label free quantitation was performed on the top 3 unique and razor peptides. Proteins identified with less than 3 unique/razor peptides and greater than 1% FDR were filtered out. Bioinformatic analyses Bioinformatic analyses were performed with R 4.3.2 (cran.r-project.org) on the decellularised dataset generated above and two additional independent proteomic datasets, including secretome protein data from pUM specimens ( n = 10 h; n = 4 LR) and isobaric tag for relative and absolute quantification (iTRAQ) labelled pUM protein data ( n = 53 h; n = 47 LR) . Proteins in each dataset were interrogated using the following criteria: decellularised dataset - ≥3 unique peptides with a 1% False Discovery Rate (FDR); secretome dataset - ≥3 unique peptides with a 1% FDR; and iTRAQ dataset – ≥2 unique peptides relative to a pooled non-involved choroid control from UM eyes. For each dataset proteins were converted to the gene name and gene entrez ID numbers were obtained using the maplds function from the org.Hs.eg.db R package. Initial analysis was performed on the decellularised dataset al.one using gene set enrichment analysis (GSEA) R package “clusterProfiler” version 4.10.0 function gseGO. Subsequent combined dataset analyses used proteins that were only present in the publicly available Homo sapiens Matrisome database 2.0 (matrisomedb.org accessed date: 11 December 2023 ). Differentially expressed proteins upregulated in HR-M3 pUM defined by a fold change (FC) ≥ 1.5 for both decellularised and secretome datasets, and ≥ 1 std dev from the mean HR: LR linear ratio for the iTRAQ dataset, were taken forward. Over Representation Analysis (ORA) was performed using the R package “clusterProfiler” version 4.10.0 . GO analysis was performed using the function EnrichGO with background gene list set to the matrisome database 2.0 as above. KEGG analysis was performed using EnrichKEGG function with background set to ‘ Homo sapiens’ . Histology and immunohistochemistry Sections of non-decellularised ‘control’ and decellularised pUM tissue were stained with; Haematoxylin and Eosin (H&E) to highlight tissue architecture, Periodic Acid Schiff (PAS) to highlight the basement membrane, Gomori trichrome to detect collagen, and 4′,6-diamidino-2-phenylindole (DAPI) to identify cell nuclei. Immunohistochemical detection of key ECM proteins identified following bioinformatic analyses was undertaken in sections of control and decellularised pUM and mUM tissue using the Leica Bond RXm and Bond Polymer Refine detection kit according to the manufacturer’s instructions. Goat antibodies required removal of the post primary step and addition of a rabbit anti goat IgG as a linker for the horse radish peroxidase enzyme. Antibodies and conditions for antigen retrieval are provided in Table . Scanning electron microscopy (SEM) Matched control and decellularised samples were fixed with glutaraldehyde/paraformaldehyde (PFA) fixative solution (0.5 ml 2.5% glutaraldehyde, 1.25 ml 4% PFA, 0.5 ml 1 x phosphate buffered solution (PBS), 2.75 ml dH 2 O) overnight at 4 o C. Samples were washed with PBS x3, placing on a test tube rotator for 3 min in-between washes. Samples were placed in osmium tetroxide (OsO 4 ) stain (2% working solution, filtered) overnight. Samples were then washed 5x in dH 2 O for 5 min on the rotator and dehydrated using graded ethanol (33-, 50-, 70-, 90-, 100%) for 10 min again on the rotator. A second and final 100% ethanol wash step was performed, and the sample cooled to approximately 5 o C using critical point drying to replace EtOH with liquid CO 2 . Samples were then heated until pressure reached 90 Bar, upon which the pressure was allowed to release slowly overnight. Inert samples were placed on carbon stubs and coated with 5 µn gold plating and viewed on FEG-SEM using TM4000 software.
This study conformed to the principles of the Declaration of Helsinki, and all procedures and methods relating to the human tissue used were approved by the Health Research Authority under the REC Ref 11/NW/0568. All samples and pseudo-anonymized data including clinical, histopathological (including nuclear BAP1 (nBAP1) protein expression) and genetic information (chromosome 3 and 8q copy number) were provided by the Ocular Oncology Biobank (REC ref 21/NW/0139). All patients had provided informed consent for the use of their samples and data in research. All samples were snap frozen following isolation and stored long-term at -80 o C or were available as formalin fixed paraffin embedded tissue. DNA quantification was undertaken in initial protocol optimisation on five anonymized primary UM (pUM) samples not included in any downstream processes to confirm acellular material. Proteomic samples Four pUM samples classified using chromosome 3 status as high metastatic risk monosomy 3 (HR-M3) and four classified as low metastatic risk disomy 3 (LR-D3), where the patients had died of metastasis within 5 years of follow-up or were still alive respectively, were selected for decellularisation and proteomic analysis. Immunohistochemistry (IHC) and scanning electron microscopy (SEM) samples Six pUM samples of known metastatic risk were also selected for decellularisation, IHC and SEM analysis. Patient data for pUM samples used in proteomic-, IHC- and SEM analyses are detailed in Table . mUM IHC samples mUM tissue samples ( n = 3) were used for IHC staining with selected antibodies.
Four pUM samples classified using chromosome 3 status as high metastatic risk monosomy 3 (HR-M3) and four classified as low metastatic risk disomy 3 (LR-D3), where the patients had died of metastasis within 5 years of follow-up or were still alive respectively, were selected for decellularisation and proteomic analysis.
Six pUM samples of known metastatic risk were also selected for decellularisation, IHC and SEM analysis. Patient data for pUM samples used in proteomic-, IHC- and SEM analyses are detailed in Table .
mUM tissue samples ( n = 3) were used for IHC staining with selected antibodies.
pUM samples selected had sufficient tissue available to yield two samples between 15 and 35 mg in weight, to provide both ‘control’ (non-decellularised) and ‘decellularised’ specimens. Tissue samples were decellularised in individual wells of a 12-well plate over a 72-hour period. In brief, samples were incubated in 1 ml hypotonic solution (10mM Tris-HCl, 5mM EDTA, pH8.0) for 24 h on a shaking platform at room temperature (RT). The solution was then removed and the tissue washed for 15 min in 1 ml phosphate buffered saline (PBS) before the addition of 1 ml of hypertonic solution (50 mM Tris-HCl, 0.5 M NaCl, 10 mM EDTA, pH8.0) for a further 48 h at RT. Tissue samples were then washed for 15 min in 2 ml PBS before incubation in 0.5% TX100 in dH 2 O for 3 h and a final incubation in dH 2 O for 1 h. Following decellularisation, the tissue pieces were either snap frozen and stored at -80 °C prior to proteomic analyses, fixed in 10% neutral buffered formalin (NBF) for processing and paraffin embedding, fixed in glutaraldehyde/paraformaldehyde (PFA) solution for SEM, or underwent DNA extraction.
The efficiency of the tissue decellularisation process was initially confirmed by assessing DNA content across five pUM samples. DNA extraction was performed with the Qiagen Blood and Tissue DNA extraction kit according to manufacturer’s instructions with an overnight extended proteinase K incubation. DNA was quantified using the NanoDrop spectrophotometer (ThermoFisher, UK). Data are mean ± SD, compared using unpaired parametric student’s t-test, P < 0.05 considered significant.
Lysis and protein estimation Each sample was weighed and nine times the weight in volume of complete EDTA free protease inhibitor cocktail in 25 mM Ammonium Bicarbonate ( AmBic) was added (e.g., 10 µg of sample, 90 µL of Lysis buffer). Protein concentration was estimated for each sample using a Bradford assay. In-solution digestion 50 µg total protein from each sample was denatured in 1% Rapigest™ solution heated to 80˚C for 10 min. Cysteine reduction was performed with dithiothreitol in 25 mM Ambic incubated at 60˚C for 10 min. Subsequent alkylation was performed by adding iodoacetamide in 25 mM Ambic and incubating for 30 min in the dark. Digestion was performed with 0.2 µg/µL trypsin (50:1 ratio of sample: trypsin) at 37˚C for 16 h on an orbital shaker. Rapigest™ inactivation was achieved by adding trifluoroacetic acid, ensuring a pH of 2 or less. Finally, samples were incubated at 37˚C for 45 min before spinning at 13,000 x g for 15 min at 7 ˚C. LC-MS analysis Samples were analysed with an Ultimate 3000 RSLC™ nano-system (Thermo Scientific) coupled to a Q Exactive Quadrupole-Orbitrap™ mass spectrometer (Thermo Scientific). The data-dependent program used for data acquisition consisted of a 70,000-resolution full-scan MS scan in the orbitrap. All samples were analysed in random order. Data analysis Data were searched using Proteome Discoverer (v 2.4) and the Mascot search engine (v 2.8) against the UniProt database of human reviewed proteins. Fixed cysteine carbamidomethylation and variable modification of methionine oxidation were specified and limited to 1 missed cleavage. Data was processed using Progenesis. Label free quantitation was performed on the top 3 unique and razor peptides. Proteins identified with less than 3 unique/razor peptides and greater than 1% FDR were filtered out.
Each sample was weighed and nine times the weight in volume of complete EDTA free protease inhibitor cocktail in 25 mM Ammonium Bicarbonate ( AmBic) was added (e.g., 10 µg of sample, 90 µL of Lysis buffer). Protein concentration was estimated for each sample using a Bradford assay.
50 µg total protein from each sample was denatured in 1% Rapigest™ solution heated to 80˚C for 10 min. Cysteine reduction was performed with dithiothreitol in 25 mM Ambic incubated at 60˚C for 10 min. Subsequent alkylation was performed by adding iodoacetamide in 25 mM Ambic and incubating for 30 min in the dark. Digestion was performed with 0.2 µg/µL trypsin (50:1 ratio of sample: trypsin) at 37˚C for 16 h on an orbital shaker. Rapigest™ inactivation was achieved by adding trifluoroacetic acid, ensuring a pH of 2 or less. Finally, samples were incubated at 37˚C for 45 min before spinning at 13,000 x g for 15 min at 7 ˚C.
Samples were analysed with an Ultimate 3000 RSLC™ nano-system (Thermo Scientific) coupled to a Q Exactive Quadrupole-Orbitrap™ mass spectrometer (Thermo Scientific). The data-dependent program used for data acquisition consisted of a 70,000-resolution full-scan MS scan in the orbitrap. All samples were analysed in random order.
Data were searched using Proteome Discoverer (v 2.4) and the Mascot search engine (v 2.8) against the UniProt database of human reviewed proteins. Fixed cysteine carbamidomethylation and variable modification of methionine oxidation were specified and limited to 1 missed cleavage. Data was processed using Progenesis. Label free quantitation was performed on the top 3 unique and razor peptides. Proteins identified with less than 3 unique/razor peptides and greater than 1% FDR were filtered out.
Bioinformatic analyses were performed with R 4.3.2 (cran.r-project.org) on the decellularised dataset generated above and two additional independent proteomic datasets, including secretome protein data from pUM specimens ( n = 10 h; n = 4 LR) and isobaric tag for relative and absolute quantification (iTRAQ) labelled pUM protein data ( n = 53 h; n = 47 LR) . Proteins in each dataset were interrogated using the following criteria: decellularised dataset - ≥3 unique peptides with a 1% False Discovery Rate (FDR); secretome dataset - ≥3 unique peptides with a 1% FDR; and iTRAQ dataset – ≥2 unique peptides relative to a pooled non-involved choroid control from UM eyes. For each dataset proteins were converted to the gene name and gene entrez ID numbers were obtained using the maplds function from the org.Hs.eg.db R package. Initial analysis was performed on the decellularised dataset al.one using gene set enrichment analysis (GSEA) R package “clusterProfiler” version 4.10.0 function gseGO. Subsequent combined dataset analyses used proteins that were only present in the publicly available Homo sapiens Matrisome database 2.0 (matrisomedb.org accessed date: 11 December 2023 ). Differentially expressed proteins upregulated in HR-M3 pUM defined by a fold change (FC) ≥ 1.5 for both decellularised and secretome datasets, and ≥ 1 std dev from the mean HR: LR linear ratio for the iTRAQ dataset, were taken forward. Over Representation Analysis (ORA) was performed using the R package “clusterProfiler” version 4.10.0 . GO analysis was performed using the function EnrichGO with background gene list set to the matrisome database 2.0 as above. KEGG analysis was performed using EnrichKEGG function with background set to ‘ Homo sapiens’ .
Sections of non-decellularised ‘control’ and decellularised pUM tissue were stained with; Haematoxylin and Eosin (H&E) to highlight tissue architecture, Periodic Acid Schiff (PAS) to highlight the basement membrane, Gomori trichrome to detect collagen, and 4′,6-diamidino-2-phenylindole (DAPI) to identify cell nuclei. Immunohistochemical detection of key ECM proteins identified following bioinformatic analyses was undertaken in sections of control and decellularised pUM and mUM tissue using the Leica Bond RXm and Bond Polymer Refine detection kit according to the manufacturer’s instructions. Goat antibodies required removal of the post primary step and addition of a rabbit anti goat IgG as a linker for the horse radish peroxidase enzyme. Antibodies and conditions for antigen retrieval are provided in Table .
Matched control and decellularised samples were fixed with glutaraldehyde/paraformaldehyde (PFA) fixative solution (0.5 ml 2.5% glutaraldehyde, 1.25 ml 4% PFA, 0.5 ml 1 x phosphate buffered solution (PBS), 2.75 ml dH 2 O) overnight at 4 o C. Samples were washed with PBS x3, placing on a test tube rotator for 3 min in-between washes. Samples were placed in osmium tetroxide (OsO 4 ) stain (2% working solution, filtered) overnight. Samples were then washed 5x in dH 2 O for 5 min on the rotator and dehydrated using graded ethanol (33-, 50-, 70-, 90-, 100%) for 10 min again on the rotator. A second and final 100% ethanol wash step was performed, and the sample cooled to approximately 5 o C using critical point drying to replace EtOH with liquid CO 2 . Samples were then heated until pressure reached 90 Bar, upon which the pressure was allowed to release slowly overnight. Inert samples were placed on carbon stubs and coated with 5 µn gold plating and viewed on FEG-SEM using TM4000 software.
Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Supplementary Material 6 Supplementary Material 7 Supplementary Material 8
|
Understanding weight management experiences from patient perspectives: qualitative exploration in general practice | 50728995-830a-4844-b936-4a7844312ca4 | 9926650 | Family Medicine[mh] | Obesity affects over 650 million people across the world and leads to further physical, psychosocial and cultural health issues . Obesity and its related comorbidities reportedly cost over US$990bn globally , which is unsustainable and threatens to bankrupt national health systems, including the UK National Health Service (NHS) . The predominant clinical view of obesity is that it presents a significant health risk. However, many other perspectives of obesity exist, including obesity not being classified as a health risk as well as obesity being the preferred ‘ideal’ in some cultures and therefore not warranting clinical ‘intervention’ or ‘treatment’. However, obesity is a stigmatized health concern in western contexts with discrimination reportedly experienced in all levels of life further contributing to a reduced quality of life . The World Health Organisation defines obesity as preventable and reversable through effective weight management strategies . The national health systems in the UK, Canada, Australia, America and New Zealand (NZ) position primary care as suitable for obesity healthcare. Guidelines recommend routine identification and treatment of obesity in primary care to reduce obesity rates . The Body Mass Index (BMI) is (while arguably a flawed tool used to measure obesity levels in primary care, with a healthy weight range classed as 18.5-24.9, overweight between 25 and 29.9, and obese over 30 . However, BMI is reported to be under-recorded and weight loss interventions under-referred in general practice. Weight management options are available through general practice, privately, through commercial avenues, or internet-based sources . In NZ, primary care offers weight management advice via national guidelines . Secondary care referral options for a clinician include dietitian consultations, weight management medication, hospital weight management clinic and bariatric surgery . In so saying, there are limited publicly funded spaces in these programmes, and options such as bariatric surgery, low calorie diet plans or exercise establishment memberships, are increasingly being offered to patients who can self-fund or pay for private health insurance . However, many people at risk of developing obesity live in high-deprivation and are financially unable to access this care , which can contribute to increasing the health inequity gap . Achieving weight management has been argued to be simply an issue of balancing an ‘energy in versus energy out’ eq. . However, evidence indicates it is more complex, due to a myriad of additional contributing factors, including the obesogenic environment, psychological factors, sociocultural norms, adverse or traumatic life events, colonisation impacts (for indigenous populations) and social determinants of health . While acknowledging the significant role modern obesogenic environments and an individual’s choice to engage with weight management plays, one of the most effective ways to achieve weight management is through a combination of diet, exercise, and behavioural change conducted in culturally appropriate ways . This combination and balance of factors needs to be calibrated to the individual for suitability as no one diet suits all. Despite this literature, obesity rates continue to rise worldwide, including the UK, suggesting there are barriers as current weight management interventions in general practice are ineffective. NZ has high obesity rates with 34% of adults classed as obese . There is significant health inequity experienced by the Indigenous Māori population in NZ with 51% obesity rate, as well as Pacific Island populations in NZ at 71% . While effective management of weight is complicated and influenced by many compounding factors both within and outside the scope of general practice, Māori also face additional challenges when engagaing with public health systems such as experiences of hostility, alienation, racism and trying to navigate a health system that does not align with a Maori health worldview reported . Yet, there is limited literature that focus on the experiences of weight management in general practice from the patient perspective . NZ populations at high risk of developing obesity include rural communities, Pacific and Indigenous Māori populations, and those living in high-deprivation areas who experience inequities . The aim of this study was to explore the patient perspectives of their weight management experiences in general practice to identify barriers and ways to improve health outcomes. It is hoped that this study will suggest new ways to offer weight management strategies within general practice and the community.
Participants Participants were recruited through rural general practices.. Participant criteria included: aged > 25 years, currently or recently resided in a rural Waikato location, not on any weight influencing medication, and identified as having had some experience with weight management in general practice context. While acknowledging the subjective nature of obesity and the definition of ‘obesity’ being socio-culturally influenced , for the purposes of this research the clinical measurement of obesity was used to demarcate weight and identify participants who were eligible for this study (BMI of > 30) . Data collection Rural general practices and Māori healthcare providers across the Waikato region were contacted via phone and email and invited to participate. They were asked to identify and pass on the female researchers (KN) details (or gain consent to be contacted by the researcher) to any of their patients they saw that fit the criteria for this study. Seven participants were recruited through this avenue. Due to the potential that some patients who fit the criteria of the study may not have visited their general practice recently, a snowballing strategy was utilised. All recruited participants were invited to share the researchers details to those in their community they knew who might want to take part. Ten participants were recruited from this method across the Waikato region. Purposeful sampling was conducted towards the end to recruit males, as only one out of the first 14 participants was male. Three males were recruited, however one male participant was excluded during the interview weight gain was found to be influenced by medication which was an exclusion criterion. A total of 16 participants were recruited and the demographic details are in Table . Information sheets and consent forms were given to all participants, rapport was built with the participants, the reasons behind why the study was being conducted, as well as any questions or concerns answered before consent was signed and interviews commenced. Interviews were held in person or via Skype at a time and place preferred by the participant . A Māori cultural advisor and GP was included and contributed by guiding the research process, including processes such as karakia (Māori prayer), whakawhanaungatanga (process of establishing relationships), koha use (gift and gratitude for participant), and contributing to interpretation of Māori narratives in a western health context. While the interviewer identifies as non-indigenous, she has lived experience with significant weight management, and extensive experience in qualitative interviews and analysis, including awareness of the limitations of her own experiences when collecting or interpreting indigenous narratives and actively sought guidance throughout the study. These factors contributed to reducing power imbalances. No participants wanted copies of transcripts. Ethical approval was granted by the University of Waikato Human Research Ethics Committee reference HREC2020#38. Interviews were semi-structured and guided by a set of questions to ensure that all participants were asked the same open-ended / exploratory questions, and to ensure participants had agency to share their story in their own words. The open-ended interview questions included: ‘could you please tell me about your experience with weight management?’ and ‘could you please tell me about your experience with any barriers to weight related health engagements?’ All participants were encouraged to speak about their experience for as long as they wanted to. Interviews were audio recorded for later transcription, notes were taken by interviewer, participants were thanked and compensated for their time with a $30 voucher. Analysis All interview data was transcribed verbatim and analysed using reflexive thematic analysis . Each transcript was printed out, read and re-read by the researcher for immersion in the data. In the right-hand margin of each transcript, sections of conversation were analysed and labelled with no pre-defined categories, enabling the concepts that were significant to the participants’ experiences to be identified and highlighted. These ideas and interpretive notes were used for the codes of this study and were checked by second researcher. Each transcript was analysed in turn, and then re-analysed for any missing codes. The Māori cultural advisor read and ensured that appropriate interpretation occurred for all participants identifying as Māori. Once all codes were listed, any redundant or double-up codes were removed. KN, RK, and LB were all involved with analysis and formulation of themes. Whilst Braun and Clarke highlight that the ability to achieve data saturation is situated and subjective , this analysis found no new themes when revisiting the transcripts and reflecting on codes already identified.
Participants were recruited through rural general practices.. Participant criteria included: aged > 25 years, currently or recently resided in a rural Waikato location, not on any weight influencing medication, and identified as having had some experience with weight management in general practice context. While acknowledging the subjective nature of obesity and the definition of ‘obesity’ being socio-culturally influenced , for the purposes of this research the clinical measurement of obesity was used to demarcate weight and identify participants who were eligible for this study (BMI of > 30) .
Rural general practices and Māori healthcare providers across the Waikato region were contacted via phone and email and invited to participate. They were asked to identify and pass on the female researchers (KN) details (or gain consent to be contacted by the researcher) to any of their patients they saw that fit the criteria for this study. Seven participants were recruited through this avenue. Due to the potential that some patients who fit the criteria of the study may not have visited their general practice recently, a snowballing strategy was utilised. All recruited participants were invited to share the researchers details to those in their community they knew who might want to take part. Ten participants were recruited from this method across the Waikato region. Purposeful sampling was conducted towards the end to recruit males, as only one out of the first 14 participants was male. Three males were recruited, however one male participant was excluded during the interview weight gain was found to be influenced by medication which was an exclusion criterion. A total of 16 participants were recruited and the demographic details are in Table . Information sheets and consent forms were given to all participants, rapport was built with the participants, the reasons behind why the study was being conducted, as well as any questions or concerns answered before consent was signed and interviews commenced. Interviews were held in person or via Skype at a time and place preferred by the participant . A Māori cultural advisor and GP was included and contributed by guiding the research process, including processes such as karakia (Māori prayer), whakawhanaungatanga (process of establishing relationships), koha use (gift and gratitude for participant), and contributing to interpretation of Māori narratives in a western health context. While the interviewer identifies as non-indigenous, she has lived experience with significant weight management, and extensive experience in qualitative interviews and analysis, including awareness of the limitations of her own experiences when collecting or interpreting indigenous narratives and actively sought guidance throughout the study. These factors contributed to reducing power imbalances. No participants wanted copies of transcripts. Ethical approval was granted by the University of Waikato Human Research Ethics Committee reference HREC2020#38. Interviews were semi-structured and guided by a set of questions to ensure that all participants were asked the same open-ended / exploratory questions, and to ensure participants had agency to share their story in their own words. The open-ended interview questions included: ‘could you please tell me about your experience with weight management?’ and ‘could you please tell me about your experience with any barriers to weight related health engagements?’ All participants were encouraged to speak about their experience for as long as they wanted to. Interviews were audio recorded for later transcription, notes were taken by interviewer, participants were thanked and compensated for their time with a $30 voucher.
All interview data was transcribed verbatim and analysed using reflexive thematic analysis . Each transcript was printed out, read and re-read by the researcher for immersion in the data. In the right-hand margin of each transcript, sections of conversation were analysed and labelled with no pre-defined categories, enabling the concepts that were significant to the participants’ experiences to be identified and highlighted. These ideas and interpretive notes were used for the codes of this study and were checked by second researcher. Each transcript was analysed in turn, and then re-analysed for any missing codes. The Māori cultural advisor read and ensured that appropriate interpretation occurred for all participants identifying as Māori. Once all codes were listed, any redundant or double-up codes were removed. KN, RK, and LB were all involved with analysis and formulation of themes. Whilst Braun and Clarke highlight that the ability to achieve data saturation is situated and subjective , this analysis found no new themes when revisiting the transcripts and reflecting on codes already identified.
All interviews lasted up to 60 minutes. Six initial themes were formed from the coding lists and after reflection of the transcripts, four overarching themes were identified: Inconsistent Information, Significance of Holistic Factors, Obesity Centre Need, and Education. Inconsistent information Inconsistent information around food dietary advice was expressed as significant in the weight management process. One woman reported “ Knowing how many calories to eat is what I struggle with” (Participant 04). Despite accessing multiple health ‘sources’ and ‘professionals’, the actual calorie deficit amount for her weight management journey was still a mystery – making her weight goals unachievable before even starting her diet plan: “What actually is it? You put it in my fitness pal [app], it’s 2500 [calories]. You do a [gym] body scan, it’s 2700. I went to [commercial nutritionist] it was 2200. There’s like 200-300-500 calorie difference- it’s a whole meal!” (Participant 04) Popular diets such as Ketogenic (low carbohydrate diet ) provoked tension for some. One participant described concern about going on a Keto diet saying that it is “actually bad for you” (Participant 09) after being advised to try it. Another participant declared that the concept of only eating fats to lose fat went against his ‘general’ understanding of weight loss whereby “ It’s kind of like the opposite of everything you learn of good nutrition” (Participant 17). Accessing quality and reputable information became a “ struggle” (Participant 04) as participants described many self-advertised weight related health ‘professionals’ (outside general practice and commonly in the commercial sector) as unqualified to give accurate dietary advice. Participants expressed that the ability to rely on dietary information became unachievable as there was “ so much misinformation available to everyone” (Participant 09). Confusion around what to believe caused further tension: “There is contradictory information out there” (Participant 09) The consistent misleading or confusing information was expressed with feelings of helplessness and powerlessness to achieve their weight goals. As summarised by one participant: “Where- what do you trust?” (Participant 09) Visiting their general practice for dietary advice was not actioned by all participants. Some participants highlighted that they did not think their GP would be a place for this type of health advice: “Going to the GP would be like a last resort” (Participant 11) And sometimes actively avoided: “I don’t think I’ve ever gone to a GP [solely for weight management advice]- but I don’t think I would, because I don’t think it would benefit me. My perspective of it is I feel like all they would say is ‘eat better and go to the gym’ And that’s what I’ve been currently trying to do” (Participant 04) Experiences with weight management options through a GP varied. Medication was “extremely expensive” , made one participant “violently ill” (Participant 14) and others had heard “ traumatizing things about the side effects” (Participant 11) of particular medication. GPs were approached for bariatric surgery as one participant described, “ I had to be GP referred to go privately” (Participant 14) . Commercial weight management programmes were viewed with scepticism as one woman reported: “There is all these different companies that are just trying to make money and like, [commercial business] they’re all businesses, they’re all trying to make money. Like, yes they are trying to ‘help’ people, but they’re also a business that’s trying to make money ” (Participant 11) Advertising of weight management through ‘X week challenges’ from commercial gyms implied ‘expertise’. As one participant highlighted this presumed ‘expertise’ ended up being generalised nutritional advice and she got “ really nothing out of it” (Participant 04). Her failure to reap any rewards from this advice generated further disappointment, frustration and depression: “How is it that I followed the nutrition plan and worked out for like four or five days a week and I lost 800 grams?! I was just so heartbroken. I was like -what’s the point? I’m trying so hard and it’s just not working. So then I could that kind of sent me back on a downward spiral” (Participant 04) Further confusion and tension surrounded the definition of a business operating as a weight management ‘expert’ as there was little transparency in terms of qualifications. Trying to identify who was a reliable information resource among all the available sources was difficult for many. One participant highlighted their frustration at wanting to find a reliable weight management professional: “I said to [doctor] I’ve been to a nutritionist, and it didn’t really fit me what that nutritionist has given me. I don’t know enough information about a difference between a nutritionist and a dietitian, do you think it would be better for me to go to a dietitian, like I’m happy to pay to go I just don’t know the differences easily. Or do I try a different nutritionist? Like, I want to get my food right!” (Participant 04) Significance of holistic factors For those who had achieved their weight loss goals, or who had achieved some weight loss in the past, a healthy mind set was crucial. Prior to losing weight, understanding why she ate was important for her success and adherence to her choice of calorie deficit plan: “I had to learn the association of what I did when I was depressed or feeling down, you know, I ate.” (Participant 01) Recognising personal relationships with food and eating behaviour were vital for any dietary changes to take place. Emotional connections to food, emotional eating, or using food to feel ‘good’ were identified as reasons for weight gain in some participants’ journeys: “I have changed my entire mental health, mental shift and food association with mental health so I don’t need chocolate to make me feel good anymore.” (Participant 01) “I think people’s mental health has a direct impact on weight loss. And when you’re depressed, you just eat crap. You eat crap, because you feel like crap and you think you’re crap” (Participant 14) Psychological aspects to weight management were also recognised as contributors to eating behaviour: “Part of the problem for me is my depression and anxiety. When they play up I tend not to pay as much attention to what I’m eating and not eating and things like that” (Participant 09) Participants reported the need for a ‘holistic’ view of weight management that incorporated many aspects to weight management and “ not just my diet ” (Participant 08) as it would “ just be a better way” (Participant 09) . One participant indicated: “[I need to] have my complete entire well being checked out- my mind, my spirituality, my environment” (Participant 08) Whilst another highlighted that balancing both physical and holistic aspects to weight management was key for effectiveness: “[It’s] very holistic, but also very scientific. This is why you do what you do. And this is why your body is reacting the way it’s reacting” (Participant 14) Feelings of failure were significant to further psychological harm with one participant reporting the whole experience being “ really disheartening” (Participant 02). Another participant described being “stuck in a cycle” (Participant 04) of failed diets and that: “It makes me feel like shit, to be fair, because I feel like I’m doing something wrong” (Participant 04) Obesity health centre Participants expressed a desire for a service that could meet their weight management needs. The concept of a “ health centre rather than a medical centre” (Participant 07) or “ weight care centres” (Participant 06) was reported as a desired ‘place to go’ for these participants for weight management needs. Weight management centres were positioned as a service that could provide reputable and reliable information as well as access to qualified health professionals who could help these participants. One woman described that having “more access to information” (Participant 14) was crucial, while another participant highlighted: “There needs to be somewhere where there is clear information from the government or actually from the medical professionals, saying, ‘This is what you can do to be better’” (Participant 09) Participants reported a significant desire and expectation that health professionals are proficient in the complexities of weight management: “Someone who is qualified and done research and knows what they’re talking about, and had experience with this, people, situations, so they know not every [diet] works for the same people” (Participant 04) “I want to be able to have access to a practitioner that understands the multi-dimensional layers to obesity” (Participant 08) Expectations on a single health professional to provide all the needs for weight management were low due to the variances of weight management needs. Difficulties with trying to deal with a health issue that is “ not just black and white” (Participant 02) with a GP only having “10 minutes to make that assessment” (Participant 16) was highlighted as an issue that needs addressing. One participant expressed: “It’s probably really hard to find someone like that [to cover complexities]. But if one person can’t do it, get a team, you know?” (Participant 02) Education Whether participants had achieved their desired weight, or were still on the weight management journey, all participants positioned education about healthy living as important. The change in societal norms was described in many forms. One participant highlighted disgust that advertising and processed foods companies are using discourse such as “organic sweetening agent e105a” as a way of “hiding what [sugar level] is in” (Participant 03) their food products. Education around processed food labels was positioned as vital to one participant’s success at weight management: “Anything with a square on it explaining what’s in it, to me that’s a warning sign” (Participant 03) Education in schools was positioned as crucial to save the next generation from suffering from obesity. Teaching them how to cook food that “ could actually fuel you and taste good” (Participant 14) and the need for teaching to be about “healthy kai (food)” (Participant 06) was important. As one participant expressed, the youth are “the victim of the sugar” (Participant 03). Awareness that the weight management “ wasn’t a diet- it was a lifestyle” (Participant 01) was crucial for long-term effectiveness. As one participant indicated: “Teaching about healthy food choices in teaching about healthy, what healthy bodies actually are is important” (Participant 09)
Inconsistent information around food dietary advice was expressed as significant in the weight management process. One woman reported “ Knowing how many calories to eat is what I struggle with” (Participant 04). Despite accessing multiple health ‘sources’ and ‘professionals’, the actual calorie deficit amount for her weight management journey was still a mystery – making her weight goals unachievable before even starting her diet plan: “What actually is it? You put it in my fitness pal [app], it’s 2500 [calories]. You do a [gym] body scan, it’s 2700. I went to [commercial nutritionist] it was 2200. There’s like 200-300-500 calorie difference- it’s a whole meal!” (Participant 04) Popular diets such as Ketogenic (low carbohydrate diet ) provoked tension for some. One participant described concern about going on a Keto diet saying that it is “actually bad for you” (Participant 09) after being advised to try it. Another participant declared that the concept of only eating fats to lose fat went against his ‘general’ understanding of weight loss whereby “ It’s kind of like the opposite of everything you learn of good nutrition” (Participant 17). Accessing quality and reputable information became a “ struggle” (Participant 04) as participants described many self-advertised weight related health ‘professionals’ (outside general practice and commonly in the commercial sector) as unqualified to give accurate dietary advice. Participants expressed that the ability to rely on dietary information became unachievable as there was “ so much misinformation available to everyone” (Participant 09). Confusion around what to believe caused further tension: “There is contradictory information out there” (Participant 09) The consistent misleading or confusing information was expressed with feelings of helplessness and powerlessness to achieve their weight goals. As summarised by one participant: “Where- what do you trust?” (Participant 09) Visiting their general practice for dietary advice was not actioned by all participants. Some participants highlighted that they did not think their GP would be a place for this type of health advice: “Going to the GP would be like a last resort” (Participant 11) And sometimes actively avoided: “I don’t think I’ve ever gone to a GP [solely for weight management advice]- but I don’t think I would, because I don’t think it would benefit me. My perspective of it is I feel like all they would say is ‘eat better and go to the gym’ And that’s what I’ve been currently trying to do” (Participant 04) Experiences with weight management options through a GP varied. Medication was “extremely expensive” , made one participant “violently ill” (Participant 14) and others had heard “ traumatizing things about the side effects” (Participant 11) of particular medication. GPs were approached for bariatric surgery as one participant described, “ I had to be GP referred to go privately” (Participant 14) . Commercial weight management programmes were viewed with scepticism as one woman reported: “There is all these different companies that are just trying to make money and like, [commercial business] they’re all businesses, they’re all trying to make money. Like, yes they are trying to ‘help’ people, but they’re also a business that’s trying to make money ” (Participant 11) Advertising of weight management through ‘X week challenges’ from commercial gyms implied ‘expertise’. As one participant highlighted this presumed ‘expertise’ ended up being generalised nutritional advice and she got “ really nothing out of it” (Participant 04). Her failure to reap any rewards from this advice generated further disappointment, frustration and depression: “How is it that I followed the nutrition plan and worked out for like four or five days a week and I lost 800 grams?! I was just so heartbroken. I was like -what’s the point? I’m trying so hard and it’s just not working. So then I could that kind of sent me back on a downward spiral” (Participant 04) Further confusion and tension surrounded the definition of a business operating as a weight management ‘expert’ as there was little transparency in terms of qualifications. Trying to identify who was a reliable information resource among all the available sources was difficult for many. One participant highlighted their frustration at wanting to find a reliable weight management professional: “I said to [doctor] I’ve been to a nutritionist, and it didn’t really fit me what that nutritionist has given me. I don’t know enough information about a difference between a nutritionist and a dietitian, do you think it would be better for me to go to a dietitian, like I’m happy to pay to go I just don’t know the differences easily. Or do I try a different nutritionist? Like, I want to get my food right!” (Participant 04)
For those who had achieved their weight loss goals, or who had achieved some weight loss in the past, a healthy mind set was crucial. Prior to losing weight, understanding why she ate was important for her success and adherence to her choice of calorie deficit plan: “I had to learn the association of what I did when I was depressed or feeling down, you know, I ate.” (Participant 01) Recognising personal relationships with food and eating behaviour were vital for any dietary changes to take place. Emotional connections to food, emotional eating, or using food to feel ‘good’ were identified as reasons for weight gain in some participants’ journeys: “I have changed my entire mental health, mental shift and food association with mental health so I don’t need chocolate to make me feel good anymore.” (Participant 01) “I think people’s mental health has a direct impact on weight loss. And when you’re depressed, you just eat crap. You eat crap, because you feel like crap and you think you’re crap” (Participant 14) Psychological aspects to weight management were also recognised as contributors to eating behaviour: “Part of the problem for me is my depression and anxiety. When they play up I tend not to pay as much attention to what I’m eating and not eating and things like that” (Participant 09) Participants reported the need for a ‘holistic’ view of weight management that incorporated many aspects to weight management and “ not just my diet ” (Participant 08) as it would “ just be a better way” (Participant 09) . One participant indicated: “[I need to] have my complete entire well being checked out- my mind, my spirituality, my environment” (Participant 08) Whilst another highlighted that balancing both physical and holistic aspects to weight management was key for effectiveness: “[It’s] very holistic, but also very scientific. This is why you do what you do. And this is why your body is reacting the way it’s reacting” (Participant 14) Feelings of failure were significant to further psychological harm with one participant reporting the whole experience being “ really disheartening” (Participant 02). Another participant described being “stuck in a cycle” (Participant 04) of failed diets and that: “It makes me feel like shit, to be fair, because I feel like I’m doing something wrong” (Participant 04)
Participants expressed a desire for a service that could meet their weight management needs. The concept of a “ health centre rather than a medical centre” (Participant 07) or “ weight care centres” (Participant 06) was reported as a desired ‘place to go’ for these participants for weight management needs. Weight management centres were positioned as a service that could provide reputable and reliable information as well as access to qualified health professionals who could help these participants. One woman described that having “more access to information” (Participant 14) was crucial, while another participant highlighted: “There needs to be somewhere where there is clear information from the government or actually from the medical professionals, saying, ‘This is what you can do to be better’” (Participant 09) Participants reported a significant desire and expectation that health professionals are proficient in the complexities of weight management: “Someone who is qualified and done research and knows what they’re talking about, and had experience with this, people, situations, so they know not every [diet] works for the same people” (Participant 04) “I want to be able to have access to a practitioner that understands the multi-dimensional layers to obesity” (Participant 08) Expectations on a single health professional to provide all the needs for weight management were low due to the variances of weight management needs. Difficulties with trying to deal with a health issue that is “ not just black and white” (Participant 02) with a GP only having “10 minutes to make that assessment” (Participant 16) was highlighted as an issue that needs addressing. One participant expressed: “It’s probably really hard to find someone like that [to cover complexities]. But if one person can’t do it, get a team, you know?” (Participant 02)
Whether participants had achieved their desired weight, or were still on the weight management journey, all participants positioned education about healthy living as important. The change in societal norms was described in many forms. One participant highlighted disgust that advertising and processed foods companies are using discourse such as “organic sweetening agent e105a” as a way of “hiding what [sugar level] is in” (Participant 03) their food products. Education around processed food labels was positioned as vital to one participant’s success at weight management: “Anything with a square on it explaining what’s in it, to me that’s a warning sign” (Participant 03) Education in schools was positioned as crucial to save the next generation from suffering from obesity. Teaching them how to cook food that “ could actually fuel you and taste good” (Participant 14) and the need for teaching to be about “healthy kai (food)” (Participant 06) was important. As one participant expressed, the youth are “the victim of the sugar” (Participant 03). Awareness that the weight management “ wasn’t a diet- it was a lifestyle” (Participant 01) was crucial for long-term effectiveness. As one participant indicated: “Teaching about healthy food choices in teaching about healthy, what healthy bodies actually are is important” (Participant 09)
Summary This study demonstrated many aspects to the patient experience of weight management including not only the need for a suitable calorie deficit dietary plan, but also addressing holistic aspects to their health such as psychological or cultural related experiences with weight. Expressions of confusion, frustration and deception around weight management advice and commercial sources of ‘help’ were found to be pervasive. Patients reported wanting education from ‘trust-worthy’ qualified professionals who could meet their wider health needs, a feat in which a GP could not achieve in their small 10-minute consultation. Surprisingly, minimal discourse linked weight management to general practice or interventions and some explicitly highlighted they would not consider visiting their GP for weight advice. Strengths and limitations As with any qualitative study, the unconscious bias from researchers can influence design and analysis. Recognising the potential for bias, this study was designed and analysed by a team of academic, general practitioner, and lived obesity experience researchers which actively included processes of cultural awareness and reflexivity throughout the research entirety. While qualitative findings cannot be generalised, this research provides novel insights to the experience of weight management from the patient perspective, which is imperative to understand if any future weight management interventions are to be effective. While the sample size was small and rurally based, it is relevant to all people attempting weight management. The research achieved saturation in the interviews with themes consistent across narratives and no new themes emerging. However, it is acknowledged that whilst the experiences and themes from both Māori and non-Māori participants were similar throughout this study, using an indigenous health worldview lens would likely elicit a wider range of themes and understandings for Māori participants.. Comparison with existing literature An unexpected finding was the lack of discourse around weight management experiences in general practice, despite this being the context for this research. Many patients positioned general practice as unsuitable to deliver effective weight management healthcare, a perspective that contradicts the national health policy and clinical guidelines in the UK, America, Australia, Canada and NZ . When general practice was talked about, it was positioned with negative clinical options (such as medication and bariatric surgery), and an overall inability to provide the obesity management patients desired. For example, addressing the holistic needs, including spiritual and cultural factors, to weight management and lifestyle habits was positioned as unsuitable for the time-poor consultation with a GP. Further, some patients specifically stated they would not even engage with their GP for weight management as it was viewed as ‘unhelpful’, which supports one UK study where patients did not see the GP or NHS as appropriate for this healthcare . It is little wonder that obesity and obesity comorbidity rates are increasing in the UK, and worldwide, given that not only do GPs experience many barriers to effective obesity healthcare delivery in their practice but their patients potentially do not come to them for this healthcare in the first place. Instead, many patients who chose to engage with weight management did so through non-general practice avenues such as fad diets or commercial companies. However, significant dis-trust, confusion, and feelings of deception were associated with these options. Commercial companies selling ‘personalised’ programmes for weight loss results that premised on very little ‘science’ were commonly reported throughout these narratives. With obesity stigma and the ‘thin ideal’ (a body image concept that is promoted to be aspired to) being pervasive in Western culture it is unsurprising that commercial endeavours such as private companies and food marketing tactics would be used to exploit those who are ostracized and vulnerable. One UK study explored patients experiences of a GP (and therefore ‘reliable’) referral to commercial weight loss programmes was welcomed as patients viewed weight as more lifestyle issues requiring a non-medical solution. However, participants in this study highlighted that their commercial weight management programmes failed to meet their comprehensive needs, and only addressed one layer of the complex weight management experience (either food, exercise or behaviour change) which contradicts the national guidelines and effective weight management literature in the UK and NZ . Issues around neo-liberal capitalist behaviours were also noted by participants whereby products consistently acted in ways that ‘hide’ sugar content and using language to imply they are qualified to give specialist advice (nutritionist versus dietitian for example), further deceiving the individual seeking help. Whilst some participants understood the economical concepts of weight loss programmes, the ‘service’ or ‘product’ they paid for did not meet their expectations despite being advertised as ‘effective’. This generated more confusion about where to go for help, what to ‘believe’ anymore or who the ‘experts’ actually are for all patients. Clear information about nutrition and exercise was desired by these patients supporting previous findings . However, this study found the information or ‘education’ sought after transcended the ‘reductionist’ nutritional or lifestyle weight management advice of previous findings and included factors such as how to navigate this current obesogenic climate and avoid consumer ‘traps’. Surprisingly, patients called for the establishment of an obesity healthcare centre. This ‘one-stop obesity shop’ was positioned to provide holistic obesity services that could extend beyond a GP (in)capability and not have a financial interest in repeat business that commercial avenues were viewed to have. Facilitating access or providing care for the myriad of factors that are recognised to contribute to obesity including culturally appropriate services for indigenous populations was stressed as crucial for successful weight management. Previous literature has also indicated that trauma and adverse life events can contribute to weight , indicating that obesity healthcare could benefit from including psychological services such as counselling as a way to improve some patients’ relationships with food and extend beyond programmes that only include dietary manipulation and exercise increase. In addition, this centre could mitigate the confusion and ‘dis-trust’ experienced by patients through employing regulated health professionals, or ‘actual experts’ that could offer reliable ‘trust-worthy’ weight advice. While the capacity for general practice to provide obesity healthcare has been questioned in previous urban literature with many barriers identified , this study extends this need for a specialised obesity referral service to indigenous and rural areas who experience significant health inequities. Implications for clinical practice This study found the patients perspective did not fully align with the national position that general practice is ‘best suited’ for effective obesity healthcare. Future research should investigate the percentage of patients utilising general practice for weight management as these efforts could be mis-placed. Further, an appraisal focused on the suitability of general practice to provide obesity healthcare is strongly recommended, as this was found to be questionable and potentially, hindering obesity reduction efforts before attempts are even made. In addition, research into the feasibility of an obesity centre establishment is recommended as this could reduce the strain on general practice and provide patients with comprehensive, culturally appropriate healthcare. Many participants felt that their ‘holistic’ obesity related health needs were not met in their general practice and desired access to a helpful referral pathway which was positioned as a ‘trustworthy’ source of information through their primary care clinician. Potentially, an effective primary care health service for obesity could be one that supports a specialised secondary service that can meet the ‘holistic’ health needs of patients. Previous literature has indicated that primary care is a valuable system that can contribute to better health outcomes and equity . Investigation into the development of obesity health services and how the division of work between primary and secondary care should be explored for efficacy purposes in the future. Public health education on obesity management urgently needs updating to include wider aspects to weight management besides calorie manipulation. Education needs to include factors within the reach of the individual, such as the ability to comprehend food labels, understanding biomedical responses to lifestyle factors, cultural influences on food consumption, and an awareness of personal psychosocial behavioural connections with food. However, the wider political climate also needs to be understood, regulated and held accountable for the factors that directly influence the individual’s ability to engage with a healthy lifestyle.
This study demonstrated many aspects to the patient experience of weight management including not only the need for a suitable calorie deficit dietary plan, but also addressing holistic aspects to their health such as psychological or cultural related experiences with weight. Expressions of confusion, frustration and deception around weight management advice and commercial sources of ‘help’ were found to be pervasive. Patients reported wanting education from ‘trust-worthy’ qualified professionals who could meet their wider health needs, a feat in which a GP could not achieve in their small 10-minute consultation. Surprisingly, minimal discourse linked weight management to general practice or interventions and some explicitly highlighted they would not consider visiting their GP for weight advice.
As with any qualitative study, the unconscious bias from researchers can influence design and analysis. Recognising the potential for bias, this study was designed and analysed by a team of academic, general practitioner, and lived obesity experience researchers which actively included processes of cultural awareness and reflexivity throughout the research entirety. While qualitative findings cannot be generalised, this research provides novel insights to the experience of weight management from the patient perspective, which is imperative to understand if any future weight management interventions are to be effective. While the sample size was small and rurally based, it is relevant to all people attempting weight management. The research achieved saturation in the interviews with themes consistent across narratives and no new themes emerging. However, it is acknowledged that whilst the experiences and themes from both Māori and non-Māori participants were similar throughout this study, using an indigenous health worldview lens would likely elicit a wider range of themes and understandings for Māori participants..
An unexpected finding was the lack of discourse around weight management experiences in general practice, despite this being the context for this research. Many patients positioned general practice as unsuitable to deliver effective weight management healthcare, a perspective that contradicts the national health policy and clinical guidelines in the UK, America, Australia, Canada and NZ . When general practice was talked about, it was positioned with negative clinical options (such as medication and bariatric surgery), and an overall inability to provide the obesity management patients desired. For example, addressing the holistic needs, including spiritual and cultural factors, to weight management and lifestyle habits was positioned as unsuitable for the time-poor consultation with a GP. Further, some patients specifically stated they would not even engage with their GP for weight management as it was viewed as ‘unhelpful’, which supports one UK study where patients did not see the GP or NHS as appropriate for this healthcare . It is little wonder that obesity and obesity comorbidity rates are increasing in the UK, and worldwide, given that not only do GPs experience many barriers to effective obesity healthcare delivery in their practice but their patients potentially do not come to them for this healthcare in the first place. Instead, many patients who chose to engage with weight management did so through non-general practice avenues such as fad diets or commercial companies. However, significant dis-trust, confusion, and feelings of deception were associated with these options. Commercial companies selling ‘personalised’ programmes for weight loss results that premised on very little ‘science’ were commonly reported throughout these narratives. With obesity stigma and the ‘thin ideal’ (a body image concept that is promoted to be aspired to) being pervasive in Western culture it is unsurprising that commercial endeavours such as private companies and food marketing tactics would be used to exploit those who are ostracized and vulnerable. One UK study explored patients experiences of a GP (and therefore ‘reliable’) referral to commercial weight loss programmes was welcomed as patients viewed weight as more lifestyle issues requiring a non-medical solution. However, participants in this study highlighted that their commercial weight management programmes failed to meet their comprehensive needs, and only addressed one layer of the complex weight management experience (either food, exercise or behaviour change) which contradicts the national guidelines and effective weight management literature in the UK and NZ . Issues around neo-liberal capitalist behaviours were also noted by participants whereby products consistently acted in ways that ‘hide’ sugar content and using language to imply they are qualified to give specialist advice (nutritionist versus dietitian for example), further deceiving the individual seeking help. Whilst some participants understood the economical concepts of weight loss programmes, the ‘service’ or ‘product’ they paid for did not meet their expectations despite being advertised as ‘effective’. This generated more confusion about where to go for help, what to ‘believe’ anymore or who the ‘experts’ actually are for all patients. Clear information about nutrition and exercise was desired by these patients supporting previous findings . However, this study found the information or ‘education’ sought after transcended the ‘reductionist’ nutritional or lifestyle weight management advice of previous findings and included factors such as how to navigate this current obesogenic climate and avoid consumer ‘traps’. Surprisingly, patients called for the establishment of an obesity healthcare centre. This ‘one-stop obesity shop’ was positioned to provide holistic obesity services that could extend beyond a GP (in)capability and not have a financial interest in repeat business that commercial avenues were viewed to have. Facilitating access or providing care for the myriad of factors that are recognised to contribute to obesity including culturally appropriate services for indigenous populations was stressed as crucial for successful weight management. Previous literature has also indicated that trauma and adverse life events can contribute to weight , indicating that obesity healthcare could benefit from including psychological services such as counselling as a way to improve some patients’ relationships with food and extend beyond programmes that only include dietary manipulation and exercise increase. In addition, this centre could mitigate the confusion and ‘dis-trust’ experienced by patients through employing regulated health professionals, or ‘actual experts’ that could offer reliable ‘trust-worthy’ weight advice. While the capacity for general practice to provide obesity healthcare has been questioned in previous urban literature with many barriers identified , this study extends this need for a specialised obesity referral service to indigenous and rural areas who experience significant health inequities.
This study found the patients perspective did not fully align with the national position that general practice is ‘best suited’ for effective obesity healthcare. Future research should investigate the percentage of patients utilising general practice for weight management as these efforts could be mis-placed. Further, an appraisal focused on the suitability of general practice to provide obesity healthcare is strongly recommended, as this was found to be questionable and potentially, hindering obesity reduction efforts before attempts are even made. In addition, research into the feasibility of an obesity centre establishment is recommended as this could reduce the strain on general practice and provide patients with comprehensive, culturally appropriate healthcare. Many participants felt that their ‘holistic’ obesity related health needs were not met in their general practice and desired access to a helpful referral pathway which was positioned as a ‘trustworthy’ source of information through their primary care clinician. Potentially, an effective primary care health service for obesity could be one that supports a specialised secondary service that can meet the ‘holistic’ health needs of patients. Previous literature has indicated that primary care is a valuable system that can contribute to better health outcomes and equity . Investigation into the development of obesity health services and how the division of work between primary and secondary care should be explored for efficacy purposes in the future. Public health education on obesity management urgently needs updating to include wider aspects to weight management besides calorie manipulation. Education needs to include factors within the reach of the individual, such as the ability to comprehend food labels, understanding biomedical responses to lifestyle factors, cultural influences on food consumption, and an awareness of personal psychosocial behavioural connections with food. However, the wider political climate also needs to be understood, regulated and held accountable for the factors that directly influence the individual’s ability to engage with a healthy lifestyle.
|
Increasing access to psychological services within pediatric rheumatology care | bfa891b1-473a-4930-b862-e755111a9cc1 | 10234679 | Internal Medicine[mh] | A QI team was assembled and consisted of a psychologist, rheumatologists, a rheumatology fellow, and quality improvement specialist. This QI team identified key drivers and interventions aimed to increase access to psychological services for youth with rheumatic disease. The Plan-Do-Study Act method of quality improvement was applied. Authors followed the Standards for Quality Improvement Reporting Excellence (SQUIRE 2.0) Guidelines in the creation of this manuscript. Authors received ethics board approval and exemption from the institutional review board (IRB # STUDY00002401). Written informed consent was waived based on the IRB determination that this project is not research involving human subjects. Data was collected for a 6-month baseline period (April through September 2017) and 4-year intervention period (October 2017 through November 2021). Data from baseline and intervention periods were obtained by extraction from the electronic health record. A psychology charge within an encounter was used as a proxy for psychology contact with a patient. Referrals to psychology were provided by rheumatology physicians and fellows, or the rheumatology psychologist. Referrals addressed common presenting problems that affect youth with rheumatic disease, including adjustment to chronic illness and treatment demands, medical adherence, and needle phobia, as well as management of acute and/or chronic pain, anxiety, and depression, and promotion of healthy lifestyle behaviors. Psychological services specifically targeted presenting problems that affect the patient; however, if caregiver or sibling needs, and/or family dynamics were directly impacting the patient, the psychologist provided either direct intervention or recommendations to address these family concerns. Psychology referrals were routinely reviewed by an intake coordinator and rheumatology psychologist to determine their appropriateness. Evidence-based screening tools, such as the PHQ-9, were used to inform referral decisions along with patient and family interview and provider observations. Key drivers for this project were identified as reducing mental health stigma, increasing awareness of mental health issues within pediatric rheumatology, reducing the demand and waitlist for psychological services, and increasing identification of mental health needs. See Fig. for key driver diagram. Interventions were performed over a 4-year period to increase access to psychological services among youth with rheumatic disease and to promote integration of psychological services into pediatric rheumatology care at a children’s hospital. The first intervention included a staffing change (i.e., the existing psychologist supporting the rheumatology service was replaced by a new psychologist) and implementation of a multidisciplinary clinic for patients with jSLE, which included psychology. As part of this multidisciplinary clinic, patients with jSLE were offered psychological services in addition to their rheumatology and nephrology care, and support from social work and pharmacy services. The second intervention involved initiation of annual depression screening with the Patient Health Questionnaire-9 (PHQ-9) to patients with jSLE as a standard of care. The third intervention included implementation of a multidisciplinary clinic for patients with joint hypermobility and Ehlers-Danlos syndrome. As part of this intervention, psychology closely followed these patients and intervened to support management of chronic pain, anxiety and depression, as well as other behavioral challenges. Youth with jSLE and hypermobility were selected as the first patient groups to receive integrated psychological services based on the demonstrated behavioral needs within these subpopulations, as well as the authors’ desire to implement quality improvement interventions on a smaller sample to ensure feasibility and troubleshoot any challenges associated with our process. For the fourth intervention, an integrated clinic was created in order to provide youth with any diagnosis of rheumatic disease with access to joint psychology and rheumatology appointments. For the fifth and final intervention, psychology only appointments were made available within rheumatology clinic in order to allow timely access to psychological services within the medical clinic setting and reduce barriers associated with seeking outpatient psychological services in a behavioral health clinic. Statistical process control was employed to monitor data throughout the course of this quality improvement project. We applied the American Society for Quality criteria to adjust the center line and control limits, evaluated each of our outcomes monthly, and modified interventions as needed. Individual measurements control charts (I-Charts) were performed to track improvement over time and t-tests were performed to assess for statistical significance. A run chart was performed to track the percentage of patients with jSLE who demonstrated a decreased or stable score on a standardized measure of depression (the PHQ-9) over time.
The sample was primarily female and Caucasian, with an average age of 13. See Table for demographic characteristics of the current sample. The I-chart featured in Fig. highlights the number of patients with rheumatic disease who were seen by a psychologist over the intervention period. There were two statistically significant center line shift in the number of patients seen by psychology over time with applied stepwise interventions. Specifically, patients seen by a psychologist increased by 3,173% from a baseline average of 1.8 to 59.9 patients seen per month from April 2017 to November 2021 ( p < 0.03). Figure depicts the number of referrals to psychology for patients with rheumatic disease being followed for outpatient rheumatology care. There was a statistically significant center line shift in the number of patients referred to psychology over time. Specifically, psychology referrals increased by 48% from a baseline average of 9.85 to 14.58 referrals per month over the intervention period from March 2017 to November 2021 ( p < 0.01). Common reasons for psychology referral within this patient population included: needle fears and phobia, pain management, and treatment of anxiety and depression. Seasonal shifts in referral frequency were observed in line with seasonal variations in stress level and school demands for patients and families, as well as pandemic conditions. Additionally, Fig. demonstrates a trend, in which, the majority of patients with jSLE who received psychological services during rheumatology care maintained a reduced or stable PHQ-9 score across annual screenings for depression. Patient satisfaction data was collected as part of the current study, which revealed an average rating of 4.6 on a 0–5 scale. Qualitative review of responses showed that patients and families perceived a benefit from receipt of psychological services within the rheumatology clinic setting. Patients and families specifically indicated feeling comfortable meeting with the psychologist, having increased access to mental health treatment as a function of our programming, and that psychology involvement helped them to better manage their rheumatic diseases. Furthermore, patients and families reflected upon their increased comfort level upon receiving psychological care within their rheumatology clinic and expressed appreciation for being able to forgo a separate behavioral health or psychological evaluation outside of the medical setting. Based on qualitative feedback from the physician team, psychology involvement in rheumatology care allowed physicians to spend reduced time focusing on the behavioral factors that impact disease management and thereby, increased provider efficiency.
Youth with rheumatic disease are at increased risk for mental health disorders, though availability of mental health care remains limited in this population. We found that youth with rheumatic disease received increased access to mental health treatment when psychological services were integrated and embedded within their routine rheumatology care. Referrals to psychology also increased significantly over the intervention period, suggesting that psychology integration within a medical clinic can increase identification of psychosocial and behavioral needs among patients and families. Additionally, the majority of youth with jSLE demonstrated reduced or stable depression scores over time when receiving psychological treatment as a component of their medical care. Results of this project suggest that psychology integration into rheumatology care remains feasible, and increases access to mental health treatment and identification of psychological needs in this at-risk population. Of note, there were no established models to follow for integration of psychological services into pediatric rheumatology care at the time of this project’s implementation. This integrated model represents the first of its kind to include psychology service embedment throughout all aspects of pediatric rheumatology care. Extant literature supports the merging medical and psychology specialties into an integrated plan of care to improve mental and physical health outcomes in youth with chronic illness . In line with a biopsychosocial approach to care, we have found substantial value added by integration of psychological services into routine rheumatology care. Psychology integration into medical care has been shown to reduce mental health stigma and barriers to care . This was observed within our population where access to psychological services proved invaluable in normalizing discussion of mental health concerns, and promoting patient and family’s buy-in and follow through with mental health evaluation and treatment. It remains important to note that our patient population represented a large catchment area, including a significant percentage of patients from rural areas and two neighboring states with poor access to mental health services. As such, embedment of psychological treatment into medical care allowed for increased access to behavioral health care. Patient satisfaction data revealed an overwhelmingly positive response to our integrated model, with patients and their families expressing satisfaction and appreciation of psychology’s integration into rheumatology clinic, and reflecting upon their increased access to mental health treatment and improved disease management. This project was completed over the course of 4 years, in which psychology was gradually integrated into pediatric rheumatology care. Our QI team has learned the value of approaching integration with a stepwise progression to ensure that team members are aware of mental health needs among youth with rheumatic disease, as well as the contribution that psychological services lend to a patient’s plan of care. Psychological services were first provided to youth with rheumatic disease within a separate behavioral health clinic and targeted common reasons for referral, such as anxiety/needle fears, pain management, and depression. Clinical screening was later introduced to identify patients with mental health or behavioral needs who were then seen by psychology across behavioral health clinic or multidisciplinary clinic settings. Psychological services were strategically integrated into multidisciplinary clinics within rheumatology, particularly among populations with high needs (e.g., those with jSLE or joint hypermobility), and eventually, expanded to patients with wide-ranging diagnoses within the rheumatology clinic setting. Primary interventions within this quality improvement initiative included implementation of mental health-focused education among providers and staff, mental health screening, and multidisciplinary medical clinics with psychology involvement. Psychological services were made available on a preventative basis for those with a new diagnosis or within certain high-risk groups, as well as when patients were presenting with acute problems that were in need of treatment. There are barriers to psychology integration into rheumatology care, including limited access to or funding for a psychologist or mental health professional. Should resources be available to obtain a psychologist, other barriers may include team cohesion and awareness of mental health impact on physical functioning and psychology’s role, as well as the psychologist’s ability to remain productive with a sufficient referral base. There were also time demands associated with psychology integration into rheumatology care. Integrating psychological services into rheumatology care increased visit duration by approximately 30 to 60 min, though patients and families consented to extend their visit, and clinical flow and templates were adjusted accordingly. Patients were scheduled within a lengthier time slot when behavioral needs were anticipated to ensure sufficient time for evaluation and treatment of rheumatic disease, as well as provision of psychological services. Alternatively, when behavioral needs were discovered spontaneously during the visit, psychological intervention directly followed medical management to limit disruptions to the clinic flow and the physician’s schedule. Based on qualitative feedback, psychology’s involvement allowed providers to spend reduced time focusing on the behavioral factors that impact disease management and thereby, increased provider efficiency. Future steps within our program include broadening of our mental health screening to include assessment of anxiety and suicide risk, ensuring access to screening across rheumatic diseases, as well as increasing the scope of psychology integration into rheumatology clinic and outcome measurement in the form of provider satisfaction with integrated services. Additionally, future studies of interest include exploration of the impact of psychology referral and involvement on rheumatic disease scales, such as patient global score. Psychology integration into pediatric rheumatology care allowed for exponential growth in access to mental health support, at a time where children and adolescents are experiencing heightened levels of emotional distress, especially those facing additional risk factors of chronic physical illness and pain. Through integration of psychological services into rheumatology care, receipt of mental health treatment was normalized and de-stigmatized within our patient population. Psychosocial and geographical barriers for patients were also minimized by embedding psychological services into established medical care. Physicians also gained an appreciation for the contribution of psychological services in their practice upon observing improved outcomes for their patients when psychology was involved, as well as the reduction in the demands on rheumatologists when behavioral targets were instead addressed by a psychologist. This study suggests the feasibility of psychology integration into pediatric rheumatology care, and highlights the capacity for psychology integration to increase access to mental health and behavioral support among patients and families affected by pediatric rheumatic disease.
Youth with rheumatic disease received increased access to mental health treatment when psychological services were integrated and embedded within pediatric rheumatology care. Referrals to psychology also increased significantly over the intervention period, suggesting that psychology integration within a rheumatology clinic can increase identification of psychosocial and behavioral needs among youth and families. Results of this project suggest that psychology integration into rheumatology care remains feasible, and increases access to mental health treatment and identification of psychological needs in this at-risk population. While psychological services have been increasingly integrated into pediatric primary care and other specialty care settings , no current models or research exist for the integration of psychological services into pediatric rheumatology. This manuscript describes an integrated model of pediatric rheumatology care and highlights novel interventions that may be employed to embed psychological support within a rheumatology clinic and increase access to psychological services for youth with pediatric rheumatic disease.
|
Discrimination of Anti-Donor Response in Allogeneic Transplantation Using an Alloreactive T-Cell Detection Assay | 65ccd66d-1858-41ff-a8a7-4c934961aa8f | 11810569 | Surgical Procedures, Operative[mh] | T cells play pivotal roles in orchestrating immune responses after solid organ transplantation . Through their unique T-cell receptors (TCRs), these cells recognize antigens presented on the peptide-major histocompatibility complex (pMHC) on antigen-presenting cells (APCs) . After transplantation, alloreactive T cells can enhance and mediate immune responses, resulting in organ damage and memory formation . Donor-reactive T cells, which are quantitatively rare, reflect the anti-donor immune status, which may elucidate the hidden mechanisms underlying complex interactions in T-cell activation and regulation during the immune response . Next-generation sequencing is a robust tool for comprehensive and high-throughput TCR profiling and facilitates the determination of the entire T-cell repertoire profile and tracing of antigen-specific T cells . Although the MHC multimer is also an excellent marker for detecting antigen-specific T-cell clones in the total pool , it is challenging to identify alloreactive T cells in the clinical context owing to alloantigen diversity and variability . Mixed lymphocyte reaction (MLR) is a classical and reliable method for estimating T-cell response in allogeneic transplantation and is useful for detecting clones against heterogenous allo-antigens. Previously, a novel comprehensive alloreactive T-cell detection (cATD) assay was developed using the MLR platform with activating markers (CD137 and CD154) . In the present study, we aimed to investigate the relevance of alloreactive T cells via a direct pathway detected using this assay in a transplantation model. Specifically, we monitored alloreactive T cells in a mouse skin transplant model to clarify the importance of detected alloreactive T cells for rejection. In addition, we investigated whether this method could be useful to estimate the immune tolerance status. Flow Cytometry The following antibodies were used: anti-AF700-CD8a (53-6.7), anti-APC-CD154 (MR1), anti-APCCy7-CD8a (53-6.7), anti-PE-CD137 (17B5), anti-PE-CD4 (GK1.5), anti-PerCPcy5.5-CD3 (17A2), anti-BV421-CD62L (MEL-14), anti-BV421-granzyme B (GZMB; QA18A28), anti-BV605-CD4 (RM4-5), anti-BV711-CD44 (IM7), and anti-BV711-interferon gamma (IFN-γ; XMG1.2), purchased from BioLegend (San Diego, CA, United States). Anti-APCCy7-CD19 (1D3) and anti-PE Cy7-FoxP3 (FJK-16s) were purchased from BD Biosciences (San Jose, CA, United States). Nonspecific FcγR binding of labeled monoclonal antibodies (mAbs) was blocked using anti-mouse CD16/32 (2.4G2; BD Pharmingen, Hamburg, Germany). Dead cells were excluded from analysis using the forward Zombie Aqua Fixable Viability Kit (BioLegend), the Zombie NIR Fixable Viability Kit (BioLegend), or 7-aminoactinomycin D (7-AAD; BD Biosciences) staining. For intracellular staining, cells were fixed and permeabilized using the FoxP3/Transcription Factor Staining Buffer Set (BD Biosciences), according to the manufacturer’s instructions. To assess cytokine production, the cells were stimulated using monensin (BD Biosciences) in a culture medium at 37°C in a 5% CO 2 incubator for 4 h prior to staining. The data were collected using LSRFortessa X-20, FACS Canto II, or FACS Celesta (BD Biosciences) and were analyzed using FlowJo v. 10 (Tree Star, Ashland, OR, United States). Mice C57BL/6 (H-2Db), BALB/c (H-2Dd), and C3H/HeJ (H-2Dk) mice were purchased from CLEA (Osaka, Japan) and maintained in a pathogen-free animal facility of Hiroshima University, Hiroshima, Japan. Female mouse were used at an age of 10–12 weeks. When indicated, the mice were euthanized through cervical dislocation after isoflurane inhalation. All efforts were made to minimize animal suffering . This study was performed in strict accordance with the “Guide for the Care and Use of Laboratory Animals” prepared by the Institute of Laboratory Animal Resources and published by the National Institutes of Health. All mice received humane care in compliance with the Principles of Laboratory Animal Care formulated by the National Society for Medical. The experimental protocol was approved by the Ethics Review Committee for Animal Experimentation of the Graduate School of Biomedical Sciences, Hiroshima University (Permit Number: A23-17). A part of this work was performed at the Research Facilities for Laboratory Animal Science, Natural Science Center for Basic Research and Development (N-BARD), Hiroshima University. Skin Transplantation Full-thickness skin grafts were transplanted onto the left lateral dorsum of a recipient. Briefly, donor skin tissues were removed from the tails and trimmed into 10 mm × 10 mm strips. Recipient mice were anesthetized using intraperitoneal injection of xylazine (5 mg/kg body weight) and ketamine (100 mg/kg body weight). Skin tissues of the same size were removed from the recipients’ backs and replaced with donor grafts. The skin grafts were covered with bandages for 5 days, and graft survival was evaluated through daily visual inspection. Rejection was defined as destruction of >95% of the skin transplant . An MHC full-mismatch BALB/c into C57BL/6 combination was employed as a rejection model. A BALB/c into C57BL/6 or C3H/HeJ combination previously reported as a tolerance induction model treated with CTLA-4 IgG (abatacept, 200 μg; Bristol-Myers Squibb, Braine-l’Alleud, Belgium) on days 0, 2, 4, and 6, and anti-CD154 antibody (MR1, 250 μg; BioLegend, San Diego, CA, United States) on days 0, 2, and 4 was used for monitoring peripheral tolerance induction. cATD Assay We prepared mononuclear cell suspensions of BALB/c mouse spleens and purified the B cells via positive selection using CD19 MicroBeads (Miltenyi Biotec, San Diego, CA, United States) in an autoMACS Pro Separator (Miltenyi Biotec), according to the manufacturer’s instructions . The purity of the sorted cells was consistently >95%. Using a cocktail of recombinant mouse CD40L multimer (100 ng/mL; AdipoGen, San Diego, CA, United States) and recombinant mouse IL-4 (10 ng/mL; R&D Systems, Minneapolis, MN, United States), activated B cells were generated by culturing 0.2 × 10 6 cells/mL at 37°C under 5% CO 2 for 24 h. All cell cultures were performed in complete medium [RPMI 1640 medium (Nacalai Tesque, Kyoto, Japan) supplemented with 5% fetal bovine serum (SERANA, Pessin, Germany), 100 mM sodium pyruvate (Thermo Fisher Scientific, Waltham, MA), 100 U/mL penicillin–streptomycin (Thermo Fisher Scientific), 1% HEPES buffer (Thermo Fisher Scientific), and 50 µM 2-ME] in a 48-well flat-bottom plate. Using activated B cells as stimulators, MLR culture was performed, after which alloreactive T cells were identified. Prior to culturing, the stimulators were irradiated with 40 Gy. Responder T cells were purified from recipient splenocytes via negative selection, using a Pan T-Cell isolation kit (Miltenyi Biotec) in the autoMACS Pro Separator (Miltenyi Biotec), according to the manufacturer’s instructions. The purity of the sorted cells was consistently >95%. Responders and stimulators were co-cultured at a 1:1 ratio (10 6 cells each) in 96-well U-bottom plates, with 200 µL complete medium containing APC-conjugated anti-CD154-labeled mAbs (MR1, 1 μL; BioLegend) for 18 h. Protein transport inhibitor (monensin, 2 μL; BD Biosciences) was added to the culture medium for the last 4 h of incubation. Alloreactive CD4 + and CD8 + T cells were identified as CD3 + CD4 + CD154 + and CD3 + CD8 + CD137 + responders, respectively. We collected at least 100,000 counts during flow cytometry acquisition for detecting 0.1% population to keep the coefficient of validation up to 10%. Proliferation Assay Recipient splenocytes were labeled with 5 µM carboxy fluorescein succinimidyl ester (CFSE; Molecular Probes) for 5 min prior to culturing. The activated B-cell stimulators were prepared as described in cATD Assay . Responders and stimulators were co-cultured at a 1:1 ratio (2 × 10 5 cells each) for 4 days in 96-well U-bottom plates with 200 µL medium. Attenuation of CFSE fluorescence intensity was evaluated as proliferating activity gated on CD4 + and CD8 + T cells. Mitotic index (MI) was calculated as previously described . Statistical Analysis Statistical analyses were performed using JMP 16 (SAS Institute, Cary, NC, United States). Chi-square or Fisher’s exact test was used to compare categorical variables, and Student’s t -test or Mann–Whitney U-test was used for continuous variables. Comparisons between groups were made using the one-way analysis of variance (ANOVA), and significant differences were examined using Tukey–Kramer’s multiple-comparison post-hoc test. Differences with p < 0.05 were considered statistically significant. The following antibodies were used: anti-AF700-CD8a (53-6.7), anti-APC-CD154 (MR1), anti-APCCy7-CD8a (53-6.7), anti-PE-CD137 (17B5), anti-PE-CD4 (GK1.5), anti-PerCPcy5.5-CD3 (17A2), anti-BV421-CD62L (MEL-14), anti-BV421-granzyme B (GZMB; QA18A28), anti-BV605-CD4 (RM4-5), anti-BV711-CD44 (IM7), and anti-BV711-interferon gamma (IFN-γ; XMG1.2), purchased from BioLegend (San Diego, CA, United States). Anti-APCCy7-CD19 (1D3) and anti-PE Cy7-FoxP3 (FJK-16s) were purchased from BD Biosciences (San Jose, CA, United States). Nonspecific FcγR binding of labeled monoclonal antibodies (mAbs) was blocked using anti-mouse CD16/32 (2.4G2; BD Pharmingen, Hamburg, Germany). Dead cells were excluded from analysis using the forward Zombie Aqua Fixable Viability Kit (BioLegend), the Zombie NIR Fixable Viability Kit (BioLegend), or 7-aminoactinomycin D (7-AAD; BD Biosciences) staining. For intracellular staining, cells were fixed and permeabilized using the FoxP3/Transcription Factor Staining Buffer Set (BD Biosciences), according to the manufacturer’s instructions. To assess cytokine production, the cells were stimulated using monensin (BD Biosciences) in a culture medium at 37°C in a 5% CO 2 incubator for 4 h prior to staining. The data were collected using LSRFortessa X-20, FACS Canto II, or FACS Celesta (BD Biosciences) and were analyzed using FlowJo v. 10 (Tree Star, Ashland, OR, United States). C57BL/6 (H-2Db), BALB/c (H-2Dd), and C3H/HeJ (H-2Dk) mice were purchased from CLEA (Osaka, Japan) and maintained in a pathogen-free animal facility of Hiroshima University, Hiroshima, Japan. Female mouse were used at an age of 10–12 weeks. When indicated, the mice were euthanized through cervical dislocation after isoflurane inhalation. All efforts were made to minimize animal suffering . This study was performed in strict accordance with the “Guide for the Care and Use of Laboratory Animals” prepared by the Institute of Laboratory Animal Resources and published by the National Institutes of Health. All mice received humane care in compliance with the Principles of Laboratory Animal Care formulated by the National Society for Medical. The experimental protocol was approved by the Ethics Review Committee for Animal Experimentation of the Graduate School of Biomedical Sciences, Hiroshima University (Permit Number: A23-17). A part of this work was performed at the Research Facilities for Laboratory Animal Science, Natural Science Center for Basic Research and Development (N-BARD), Hiroshima University. Full-thickness skin grafts were transplanted onto the left lateral dorsum of a recipient. Briefly, donor skin tissues were removed from the tails and trimmed into 10 mm × 10 mm strips. Recipient mice were anesthetized using intraperitoneal injection of xylazine (5 mg/kg body weight) and ketamine (100 mg/kg body weight). Skin tissues of the same size were removed from the recipients’ backs and replaced with donor grafts. The skin grafts were covered with bandages for 5 days, and graft survival was evaluated through daily visual inspection. Rejection was defined as destruction of >95% of the skin transplant . An MHC full-mismatch BALB/c into C57BL/6 combination was employed as a rejection model. A BALB/c into C57BL/6 or C3H/HeJ combination previously reported as a tolerance induction model treated with CTLA-4 IgG (abatacept, 200 μg; Bristol-Myers Squibb, Braine-l’Alleud, Belgium) on days 0, 2, 4, and 6, and anti-CD154 antibody (MR1, 250 μg; BioLegend, San Diego, CA, United States) on days 0, 2, and 4 was used for monitoring peripheral tolerance induction. We prepared mononuclear cell suspensions of BALB/c mouse spleens and purified the B cells via positive selection using CD19 MicroBeads (Miltenyi Biotec, San Diego, CA, United States) in an autoMACS Pro Separator (Miltenyi Biotec), according to the manufacturer’s instructions . The purity of the sorted cells was consistently >95%. Using a cocktail of recombinant mouse CD40L multimer (100 ng/mL; AdipoGen, San Diego, CA, United States) and recombinant mouse IL-4 (10 ng/mL; R&D Systems, Minneapolis, MN, United States), activated B cells were generated by culturing 0.2 × 10 6 cells/mL at 37°C under 5% CO 2 for 24 h. All cell cultures were performed in complete medium [RPMI 1640 medium (Nacalai Tesque, Kyoto, Japan) supplemented with 5% fetal bovine serum (SERANA, Pessin, Germany), 100 mM sodium pyruvate (Thermo Fisher Scientific, Waltham, MA), 100 U/mL penicillin–streptomycin (Thermo Fisher Scientific), 1% HEPES buffer (Thermo Fisher Scientific), and 50 µM 2-ME] in a 48-well flat-bottom plate. Using activated B cells as stimulators, MLR culture was performed, after which alloreactive T cells were identified. Prior to culturing, the stimulators were irradiated with 40 Gy. Responder T cells were purified from recipient splenocytes via negative selection, using a Pan T-Cell isolation kit (Miltenyi Biotec) in the autoMACS Pro Separator (Miltenyi Biotec), according to the manufacturer’s instructions. The purity of the sorted cells was consistently >95%. Responders and stimulators were co-cultured at a 1:1 ratio (10 6 cells each) in 96-well U-bottom plates, with 200 µL complete medium containing APC-conjugated anti-CD154-labeled mAbs (MR1, 1 μL; BioLegend) for 18 h. Protein transport inhibitor (monensin, 2 μL; BD Biosciences) was added to the culture medium for the last 4 h of incubation. Alloreactive CD4 + and CD8 + T cells were identified as CD3 + CD4 + CD154 + and CD3 + CD8 + CD137 + responders, respectively. We collected at least 100,000 counts during flow cytometry acquisition for detecting 0.1% population to keep the coefficient of validation up to 10%. Recipient splenocytes were labeled with 5 µM carboxy fluorescein succinimidyl ester (CFSE; Molecular Probes) for 5 min prior to culturing. The activated B-cell stimulators were prepared as described in cATD Assay . Responders and stimulators were co-cultured at a 1:1 ratio (2 × 10 5 cells each) for 4 days in 96-well U-bottom plates with 200 µL medium. Attenuation of CFSE fluorescence intensity was evaluated as proliferating activity gated on CD4 + and CD8 + T cells. Mitotic index (MI) was calculated as previously described . Statistical analyses were performed using JMP 16 (SAS Institute, Cary, NC, United States). Chi-square or Fisher’s exact test was used to compare categorical variables, and Student’s t -test or Mann–Whitney U-test was used for continuous variables. Comparisons between groups were made using the one-way analysis of variance (ANOVA), and significant differences were examined using Tukey–Kramer’s multiple-comparison post-hoc test. Differences with p < 0.05 were considered statistically significant. cATD Assay Detected Sensitization Leading to Acute Rejection in the Mouse Skin Transplantation Model Skin allografts were rejected from 7 to 15 days in the full MHC mismatched rejection model (BALB/c into C57BL/6) (MST 11 days, ). We did not observe a sensitized reaction in peripheral CD4 + and CD8 + T cells at 3 days after transplantation, as determined using a proliferation assay (syngeneic vs. rejection model, median MI; CD4 + 0.18 vs. 0.08, p = 0.53 (upper) and CD8 + 0.20 vs. 0.10, p = 0.80 (lower), ). Seven days after transplantation, we observed a higher proliferation response of both CD4 + /CD8 + T cells in the rejection model than in the syngeneic model (median MI; CD4 + 0.18 vs. 0.46, p < 0.05 (upper) and CD8 + 0.57 vs. 1.28, p < 0.05 (lower), ). The cATD assay revealed a sensitized immune response after skin transplantation at the same time point as the proliferation assay, showing a higher proportion of donor-reactive CD4 + CD154 + /CD8 + CD137 + T cells than that in the syngeneic model at 7 days after transplantation (syngeneic vs. rejection model, CD4 + CD154 + in total CD4 + ; 1.0% vs. 1.4%, p < 0.05 (upper) and CD8 + CD137 + in total CD8 + ; 0.27% vs. 0.53%, p < 0.05 (lower), ). Donor-reactive T cells identified in this assay showed an increase in proportion and enhancement in function under antigen-specific stimulation in recipients sensitized with BALB/c mouse graft . The multiparametric flowcytometric analyses demonstrated a unique functionality of donor-reactive CD8 + T cells in the rejection model; for instance, the production ability of the crucial effectors, GZMB and IFN-γ, was specifically enhanced in donor-reactive CD8 + T cells in the rejection model at 7 days after transplantation (syngeneic vs. rejection model, % positive for GZMB 16.1% vs. 54.7%, p < 0.05 (upper), and IFN-γ 1.82% vs. 8.18%, p < 0.05 (lower), ). As a proof of sensitization, effector memory T (TEM; CD44 + CD62L − ) cells were enriched in the donor-reactive population after transplantation (syngeneic vs. rejection model, median % TEM in CD4 + CD154 + 20.2% vs. 32.9%, p < 0.05 (upper), and CD8 + CD137 + 9.2% vs. 19.5%, p < 0.05 (lower), ). Quantitative and Qualitative Analyses of Donor-Reactive T Cells for Monitoring Tolerance Induction in the Treated Mouse Skin Transplantation Model Permanent engraftment was observed in C3H/HeJ recipients of the full MHC mismatched BALB/c graft treated with CTLA-4 IgG and anti-CD154 antibody (treated tolerance (TT) model, ≥30-day survival was recorded in 16/19 animals, 84.2%), whereas all C57BL/6 recipients with the same immunosuppression eventually experienced allograft rejection within 20 days (treated rejection (TR) model, ). We investigated the immunological status at 7 and 30 days after transplantation, that is, before and after rejection, respectively. The proportion of FOXP3 + Tregs in CD4 + T cells was comparable, despite tolerance induction, between 7 and 30 days after transplantation . The proliferation assay conducted at 7 days after transplantation showed a significant reduction in response to immunosuppression in both C3H/HeJ and C57BL/6 recipients compared with that in an untreated rejection (UR) model. However, the conventional proliferation readout results did not show the differential immune response at 7 days after transplantation between C3H/HeJ and C57BL/6 recipients, despite a different outcome (TT model vs. TR model, median MI; CD4 + 0.41 vs. 0.21, p = 0.44 and CD8 + 0.15 vs. 0.19, p = 0.70, respectively, ). The cATD assay revealed that the proportion of CD8 + donor-reactive T cells in the TT model was lower than that in the TR model at 7 days after transplantation (TT model vs. TR model, median % donor-reactive CD8 + ; 0.18% vs. 0.35%, p < 0.05, ). The GZMB- and IFN-γ-producing capacity of the CD8 + donor-reactive T cells was comparatively low in the three groups . Regardless of the final outcome, models with immunosuppression exhibited impaired memory formation in donor-reactive T cells . At 30 days after transplantation, the proliferation assay showed a lower MI of CD8 + T cells for the response of the TT model than that for the response of both UR and TR models (TT model vs. UR and TR models, median CD8 + MI: 0.18 vs. 0.93 and 0.66, p < 0.05, respectively, ). The cATD assay performed at 30 days after transplantation revealed that donor-reactive T cells were detectable in the TT model, similar to those in the UR and TR models (UR vs. TT vs. TR, %CD4 + CD154 + in total CD4 + was 2.01%, 1.77%, and 2.35%, %CD8 + CD137 + in total CD8 + was 1.00%, 0.7%, and 0.81%, respectively, ). As expected, the functionality of donor-reactive CD8 + T cells in the TT model was lower than that in the UR model (UR vs. TT model, % positive in donor-reactive CD8 + T cells, GZMB; 31.9% vs. 11.2%, p < 0.05, and IFN-γ; 33.9% vs. 3.04%, p < 0.05, respectively, ). However, there were no differences in functionality and memory formation between the TT and TR models . Skin allografts were rejected from 7 to 15 days in the full MHC mismatched rejection model (BALB/c into C57BL/6) (MST 11 days, ). We did not observe a sensitized reaction in peripheral CD4 + and CD8 + T cells at 3 days after transplantation, as determined using a proliferation assay (syngeneic vs. rejection model, median MI; CD4 + 0.18 vs. 0.08, p = 0.53 (upper) and CD8 + 0.20 vs. 0.10, p = 0.80 (lower), ). Seven days after transplantation, we observed a higher proliferation response of both CD4 + /CD8 + T cells in the rejection model than in the syngeneic model (median MI; CD4 + 0.18 vs. 0.46, p < 0.05 (upper) and CD8 + 0.57 vs. 1.28, p < 0.05 (lower), ). The cATD assay revealed a sensitized immune response after skin transplantation at the same time point as the proliferation assay, showing a higher proportion of donor-reactive CD4 + CD154 + /CD8 + CD137 + T cells than that in the syngeneic model at 7 days after transplantation (syngeneic vs. rejection model, CD4 + CD154 + in total CD4 + ; 1.0% vs. 1.4%, p < 0.05 (upper) and CD8 + CD137 + in total CD8 + ; 0.27% vs. 0.53%, p < 0.05 (lower), ). Donor-reactive T cells identified in this assay showed an increase in proportion and enhancement in function under antigen-specific stimulation in recipients sensitized with BALB/c mouse graft . The multiparametric flowcytometric analyses demonstrated a unique functionality of donor-reactive CD8 + T cells in the rejection model; for instance, the production ability of the crucial effectors, GZMB and IFN-γ, was specifically enhanced in donor-reactive CD8 + T cells in the rejection model at 7 days after transplantation (syngeneic vs. rejection model, % positive for GZMB 16.1% vs. 54.7%, p < 0.05 (upper), and IFN-γ 1.82% vs. 8.18%, p < 0.05 (lower), ). As a proof of sensitization, effector memory T (TEM; CD44 + CD62L − ) cells were enriched in the donor-reactive population after transplantation (syngeneic vs. rejection model, median % TEM in CD4 + CD154 + 20.2% vs. 32.9%, p < 0.05 (upper), and CD8 + CD137 + 9.2% vs. 19.5%, p < 0.05 (lower), ). Permanent engraftment was observed in C3H/HeJ recipients of the full MHC mismatched BALB/c graft treated with CTLA-4 IgG and anti-CD154 antibody (treated tolerance (TT) model, ≥30-day survival was recorded in 16/19 animals, 84.2%), whereas all C57BL/6 recipients with the same immunosuppression eventually experienced allograft rejection within 20 days (treated rejection (TR) model, ). We investigated the immunological status at 7 and 30 days after transplantation, that is, before and after rejection, respectively. The proportion of FOXP3 + Tregs in CD4 + T cells was comparable, despite tolerance induction, between 7 and 30 days after transplantation . The proliferation assay conducted at 7 days after transplantation showed a significant reduction in response to immunosuppression in both C3H/HeJ and C57BL/6 recipients compared with that in an untreated rejection (UR) model. However, the conventional proliferation readout results did not show the differential immune response at 7 days after transplantation between C3H/HeJ and C57BL/6 recipients, despite a different outcome (TT model vs. TR model, median MI; CD4 + 0.41 vs. 0.21, p = 0.44 and CD8 + 0.15 vs. 0.19, p = 0.70, respectively, ). The cATD assay revealed that the proportion of CD8 + donor-reactive T cells in the TT model was lower than that in the TR model at 7 days after transplantation (TT model vs. TR model, median % donor-reactive CD8 + ; 0.18% vs. 0.35%, p < 0.05, ). The GZMB- and IFN-γ-producing capacity of the CD8 + donor-reactive T cells was comparatively low in the three groups . Regardless of the final outcome, models with immunosuppression exhibited impaired memory formation in donor-reactive T cells . At 30 days after transplantation, the proliferation assay showed a lower MI of CD8 + T cells for the response of the TT model than that for the response of both UR and TR models (TT model vs. UR and TR models, median CD8 + MI: 0.18 vs. 0.93 and 0.66, p < 0.05, respectively, ). The cATD assay performed at 30 days after transplantation revealed that donor-reactive T cells were detectable in the TT model, similar to those in the UR and TR models (UR vs. TT vs. TR, %CD4 + CD154 + in total CD4 + was 2.01%, 1.77%, and 2.35%, %CD8 + CD137 + in total CD8 + was 1.00%, 0.7%, and 0.81%, respectively, ). As expected, the functionality of donor-reactive CD8 + T cells in the TT model was lower than that in the UR model (UR vs. TT model, % positive in donor-reactive CD8 + T cells, GZMB; 31.9% vs. 11.2%, p < 0.05, and IFN-γ; 33.9% vs. 3.04%, p < 0.05, respectively, ). However, there were no differences in functionality and memory formation between the TT and TR models . Allogeneic reactive T cells play a pivotal role in the process of promoting or conversely regulating rejection in allogeneic solid organ transplantation . Understanding the characteristics and behavior of alloreactive T cells is vital for assessing the immune response after allogeneic transplantation . MLR is a classical but practical method to assess allo-response. The precursor frequency of alloreactive T cells has been reported to be 1%–10% under various assay conditions and readouts in both murine and human T-cell repertoires . Proliferation, which requires a culture period of 4–5 days, has been widely used as an accessible readout to visualize and quantify the responsiveness of alloreactive T cells using MLR. However, with advancements in flow cytometry technology, it has become feasible to perform multiparametric evaluations of rare populations of less than 1%. This finding suggests the possibility of assessing these infrequent alloreactive T cells without the need for proliferation. In line with this prospect, a previous study demonstrated that the cATD assay, using activated allogeneic B-cell stimulators and very early activation markers, enables the detection of alloreactive T cells with high precision in a short-term culture system . In the present study, we validated the utility of the cATD assay for rapid evaluation of donor-reactive T cells in an in vivo transplantation model. The usefulness of CD154 and CD137 for detecting antigen-specific CD4 + and CD8 + T cells as rapid-activating molecules has been demonstrated using viral peptides and toxins, respectively . CD154 is preferentially expressed on effector CD4 + T cells and memory CD8 + T cells . Although CD137 expression can be induced on CD4 + T cells, the combination of CD137 + CD154 − expression after allo-stimulation has been reported to delineate activated FOXP3 + regulatory T cells that exhibit a specific suppressive capacity against corresponding allo-stimulation . Single-cell TCR analysis has revealed that CD137 expression on CD8 + T cells after allogeneic stimulation is a marker for oligoclonal expanded alloreactive T cells during acute cellular rejection (ACR) after lung transplantation . Moreover, alloreactive CD154 expression on CD8 + memory T cells has been reported to be associated with acute rejection after pediatric liver, intestine, and kidney transplantation . Although CD154 could be used as a candidate for predicting rejection by analyzing memory CD8 + T cells, CD137 can be used as a marker to detect a variety of CD8 + T-cell subsets including a substantial portion of naïve populations . Consistent with the results of the previous study, we observed a considerable proportion of a naïve phenotype in donor-reactive CD8 + T cells using CD137 detection. CD137 alloreactive CD8 + T cells showed greater functional molecule expression than those detected by CD154 in our rejection model mice . In clinical settings, the cATD assay enables repeated monitoring of circulating alloreactive T cells. The significance of alloreactive T-cell clones in circulation as the pathological effector of rejection after transplantation may be controversial. A recent TCR repertoire analysis using next-generation sequencing revealed that expanded circulating T-cell clones during ACR were observed in the circulation before ACR after lung , liver , and kidney transplantation . Furthermore, expanded clones in circulation have been reported to overlap with infiltrated T-cell clones in the liver and kidney allografts . An interesting case report of malignant melanoma treated with an immune checkpoint inhibitor after kidney transplantation indicated that the alloreactive T-cell cluster in renal biopsy identified through single-cell RNA sequencing overlapped with circulating clones, which were identified both before and after rejection of the allograft . According to these observations, we believe that circulating alloreactive T cells reflect immune responses after solid organ transplantation. In the current era where organ transplantation is a standard therapy for patients with organ failure, a standard approach to monitor harmful alloimmune responses is lacking . A previous study reported the usefulness of quantified proliferation in MLR to diagnose immunological rejection . The proliferation and cATD assays assess different time points and readouts, suggesting that they can identify different T-cell populations. During the proliferation assay, in vitro culture of T cells is performed over several days to amplify them and obtain T cells of various developmental stages. On the contrary, the cATD assay detects the population that responds rapidly in MLR initiated through overnight culturing, which may indicate a highly primed status and is directly linked to impending rejection. As this assay assesses alloreactivity through a direct pathway, missing the component through indirect pathways could be a limitation when monitoring long-term allo-response after transplantation. However, we believe that its relevance to in vivo acute rejection models makes it a useful tool for immune monitoring. We observed different outcomes and immunological findings in tolerance induction between C3H/HeJ (TT) and C57BL/6 (TR) recipients. C3H/HeJ mice express a dysfunctional toll-like receptor 4, which reduces macrophage and B-cell proliferation and antigen-presenting capabilities, possibly leading to different immune responses and outcomes . Interestingly, the cATD assay showed quantitatively different priming status of donor-reactive CD8 + T cells between the TT and TR models before rejection. After rejection when the rejected graft was lost, the cATD assay did not show differential findings between the TT and TR models; however, the proliferation assay reliably showed sensitization potential in the TR model, based on the results obtained 30 days after transplantation. These findings may be attributed to the feature of alloreactive T cells detected using the cATD assay. This study has some limitations. Notably, the immunological response in skin transplantation is potentially different from that in organ transplantation. Investigation of other organ transplant models and clinical samples could further validate the relevance of the findings of the present study across diverse transplantation settings. However, the cATD assay, which enables real-time and repeatable detection of donor-reactive effectors, might be clinically relevant in diagnosing harmful allo-responses directly linked to the region responsible for rejection. Future research should compare the TCR repertoire of reactive T cells at rejection or upon achieving tolerance between proliferation and cATD assays to obtain differential immunological information. Multifaceted evaluation through the cATD assay facilitates the investigation of superior functional molecules and biomarkers for monitoring clinical conditions such as tolerance status. Additionally, it enables the retrieval of rare live alloreactive T-cell populations for downstream investigation via fluorescence-activated cell sorting and provides valuable information for further studies in the field of translational research. In conclusion, the cATD assay using CD154 and CD137 as alloreactive markers effectively distinguished immune responses in in vivo mouse transplantation models, highlighting its potential to facilitate prompt quantitative and qualitative estimation of alloreactive T cells after allogeneic transplantation. |
Health Literacy in Africa—A Scoping Review of Scientific Publications | 894583eb-c45c-4b75-b0a8-acf75601668a | 11594271 | Health Literacy[mh] | A key driver for sustainable development is a future in which every African can enjoy a life of better health and well-being. Africa is currently the world’s second-largest and second-most-populous continent, with the fastest-growing population globally. Its population of 1.4 billion has an average age of 18.8 years, making it the youngest of any continent . Sub-Saharan Africa faces considerable educational and financial challenges, with the highest illiteracy rate in the world and a large proportion of people living in extreme poverty. These social determinants of health, combined with the diverse environmental and political conditions, as well as the great diversity of ethnicities, cultures, languages, and historical developments, present both richness and challenges for each African country including for the health status and health systems. Moreover, African countries face a double burden of persistently high infectious diseases and increasingly prevalent chronic diseases . Nonetheless, African nations have also made remarkable strides in enhancing the health outcomes of their populations in recent decades . The Africa health landscape is rapidly evolving, with better control of communicable diseases and rising prevalence of non-communicable diseases, specifically diabetes and cancer. It is crucial to continue concerted action in Africa to curb the spread of infectious diseases, while also facilitating appropriate management of chronic conditions and general disease prevention and health promotion to support healthier populations . According to the Nairobi Declaration on Health Promotion , investment in health literacy is key in this process. 1.1. Health Literacy The concept of health literacy has received considerable attention in the past decades . As the knowledge and competencies necessary to use health information and services , promoting health literacy has become an incremental component of many global and national health plans, e.g., the Shanghai Declaration on promoting health in the 2030 Agenda for Sustainable Development or the national health literacy action plans of China , Germany , Scotland , and the United States , as an enabler of sustainable development, vital for good health and well-being, disease prevention, and achieving universal health coverage . Health literacy is considered a relational, modifiable determinant of health that focuses on the skills of individuals and communities and the health literacy responsiveness of service providers and systems to maintain and promote health and well-being . It can be defined as a multidimensional concept that includes the knowledge, motivation, and competencies for people to access, understand, appraise, and apply information to form judgments and make decisions concerning health care, disease prevention, and health promotion in everyday life to maintain and improve quality of life throughout the course of their life . Importantly, it also emphasizes how health providers, organizations, and settings enable people to cope with health challenges . 1.2. Health Literacy in Africa Although, several reports have sketched the international progress of health literacy as a growing global movement , they only included few records from African countries and did not provide a comprehensive overview of health literacy endeavors across the African continent. Thus, to date, there are limited insights about how health literacy is being researched, developed and implemented in the diverse African societies. Nonetheless, to improve health literacy in Africa, it is crucial to understand and address the advancements based on the distinctive African contexts . 1.3. Study Aim To bridge the gap, this study aimed to shed light on the scope and scale of health literacy development in Africa by providing a first general overview of the state of the art. Therefore, the objectives included a scoping review to analyze scientific publications on health literacy in Africa with regards to (1) research approaches, (2) historical trends, (3) geographical origins, (4) target groups and settings, and (5) thematic analysis of specified health literacy aspects. The insights gained may help to inform further progress and prioritize future actions for capacity building, policy development, and informing practices on health literacy for all in Africa.
The concept of health literacy has received considerable attention in the past decades . As the knowledge and competencies necessary to use health information and services , promoting health literacy has become an incremental component of many global and national health plans, e.g., the Shanghai Declaration on promoting health in the 2030 Agenda for Sustainable Development or the national health literacy action plans of China , Germany , Scotland , and the United States , as an enabler of sustainable development, vital for good health and well-being, disease prevention, and achieving universal health coverage . Health literacy is considered a relational, modifiable determinant of health that focuses on the skills of individuals and communities and the health literacy responsiveness of service providers and systems to maintain and promote health and well-being . It can be defined as a multidimensional concept that includes the knowledge, motivation, and competencies for people to access, understand, appraise, and apply information to form judgments and make decisions concerning health care, disease prevention, and health promotion in everyday life to maintain and improve quality of life throughout the course of their life . Importantly, it also emphasizes how health providers, organizations, and settings enable people to cope with health challenges .
Although, several reports have sketched the international progress of health literacy as a growing global movement , they only included few records from African countries and did not provide a comprehensive overview of health literacy endeavors across the African continent. Thus, to date, there are limited insights about how health literacy is being researched, developed and implemented in the diverse African societies. Nonetheless, to improve health literacy in Africa, it is crucial to understand and address the advancements based on the distinctive African contexts .
To bridge the gap, this study aimed to shed light on the scope and scale of health literacy development in Africa by providing a first general overview of the state of the art. Therefore, the objectives included a scoping review to analyze scientific publications on health literacy in Africa with regards to (1) research approaches, (2) historical trends, (3) geographical origins, (4) target groups and settings, and (5) thematic analysis of specified health literacy aspects. The insights gained may help to inform further progress and prioritize future actions for capacity building, policy development, and informing practices on health literacy for all in Africa.
The study design followed the Arksey and O’Malley steps for scoping reviews and adhered to the PRISMA Guidelines for Scoping Reviews, which allow for systematic and transparent data identification, selection, synthesis, and appraisal of research-related literature . Identification of relevant studies: The search strategy employed a comprehensive and systematic approach to identify scientific evidence from publications that could potentially inform trends in health literacy development in Africa. Thus, the systematic literature search was conducted in six online scientific databases: PubMed, PsycINFO, Cochrane Library, ERIC, African Journals Online, and African Index Medicus. The search string used the PCC approach covering participants, content, and context as suggested by the Johanna Briggs Institute . Thus, the search terms entailed the concept “health literacy” OR “littératie en santé” OR “compétence en matière de santé”, combined with the Boolean operator AND with the search terms for context and population “Afric*” OR “Afriq*” OR all of the names of the 54 African countries of the African Union and their associated adjectives, but NOT “African American*”. An initial search using the Portuguese term ‘literacia em saude’ did not provide any results regarding African publications and was therefore not included in the search string. The inclusion of Afrikaans, Swahili, and other languages spoken and published in Africa apart from English, French, and Portuguese was out of the scope of this study. Eventually all the six databases generated abstracts in English, thus no further analysis with regards to publication language was made. The truncation symbol was used to increase the search’s sensitivity and include additional terms, such as African countries or the African continent. Please refer to for the search strategy. The search was not restricted to a specific time period. The study was registered at OSF https://doi.org/10.17605/OSF.IO/2QPXU (accessed on 11 September 2024). Selection of the studies: The literature search and data collection took place in July 2022 (11 July 2022). Three researchers screened the identified records separately. Any discrepancies regarding the inclusion of studies were resolved through discussion. Articles were included which were published in English or French language and utilized the term ‘health literacy’ in the title or abstract and had full-text available online (either through open access or university access). The exclusion criteria comprised studies conducted in African regions with only limited or no international recognition, such as Somaliland and Western Sahara. Additionally, studies conducted outside of Africa, including those focusing on African Americans in the United States of America or African migrants in Europe, were excluded. Initially, the authors intended to employ forward and ancestry citation searching and manual searching of journals, yet due to the vast number of records identified and the completely voluntarily nature of this research, this proved to be beyond the scope of this research project. To ensure that no relevant studies were overlooked, the authors consulted members of the African Health Literacy Network. As critical appraisal of individual sources of evidence is not required in scoping reviews, this was not conducted. Charting data and collating and reporting results: The lead authors KS, SH, and VK reviewed the research literature based on the pre-defined inclusion criteria. They coded the data in Microsoft Excel using a jointly developed coding scheme that reflected the research objectives such as author, year, title, country, type of manuscript, and specific study characteristics such as research methodology, target population, setting, and research theme. The data for each article was extracted by one researcher and doublechecked by two researchers. Collaboratively, they analyzed and synthesized the findings corresponding to the study objectives. Firstly, the authors extracted data based on the original wording and terminologies in the records. Secondly, the data was analyzed, main categories were developed inductively for each research objective, and the overall number of records per category was counted. We performed a qualitative content analysis of each objective to identify occurring health literacy themes and sub-themes which are all presented in aggregated forms in this article due to vast number of records found. Key informants from the African Health Literacy Network provided thorough feedback on the research design, process, and findings and contributed as co-authors to the discussion of the findings and recommendations.
The findings of the scoping review of scientific publications with regards to health literacy in Africa are presented below. 3.1. Characteristics of the Studies The initial search yielded a total of 876 records after removal of duplicates. The co-authoring members of the African Health Literacy Network suggested various studies, but these were already included in the initial set of 876 records. The titles and abstracts were screened for eligibility. As illustrated in the PRISMA flowchart , the screening process yielded 487 records, of which 137 were excluded due to inaccessibility of the full text and 13 due to unavailability of official results or inappropriateness for the purpose of the study. Ultimately, 336 records were included in the review of health literacy development in Africa. The complete list of articles can be found in . The scoping review and synthesis according to research methods, countries, populations, settings, and health literacy factors revealed new insights on the scope and scale of health literacy development in Africa. 3.2. Research Approaches Applied Out of the 336 articles, 22 (6.5%) were scoping literature reviews, 291 (86.6%) were original research articles, and 23 (6.8%) were opinions, perspectives, or commentaries published in scientific journals or academic books. Among the original research articles, 163 publications presented quantitative methods, mostly using cross-sectional designs to assess health literacy or disease-related health literacy in patients. Other articles assessed, for example, the effectiveness of controlled trials and interventions. One hundred and five articles used qualitative approaches, with the majority using semi-structured interviews or focus group discussions to investigate barriers to health care or appropriate health behaviors. Finally, 19 publications presented mixed- or multi-method approaches, such as developing or translating health literacy measures or assessing perceptions and health behaviors related to a specific health concern, and 4 described clinical trials. 3.3. Publication History of African Health Literacy Research Scientific publications on health literacy in Africa have increased exponentially since the beginning of the twenty-first century, as shown in . The first mention of health literacy in Africa was in Kickbusch’s seminal paper “Addressing the health and education divide” , which provided several examples. The first specific study on health literacy was published in 2005, focusing on HIV-related health literacy in Zambia. Since then, the number of annual publications steadily increased, reaching 60 publications in 2021 and likely more in the pipeline the following years. 3.4. Geographical Distribution The geographical analysis revealed that health literacy publications were available from 38 of the 54 African Union countries . Most studies were conducted in South Africa ( n = 84), followed by Nigeria ( n = 49), Ghana ( n = 30), Ethiopia ( n = 28), and Uganda ( n = 27). Many countries had only one or a few studies on health literacy, such as Burundi ( n = 1) or Sudan ( n = 1). Additionally, 31 publications concentrated on various regions such as West Africa ( n = 2), North Africa ( n = 1), Southern Africa ( n = 2), and East Africa ( n = 1). Ten publications specifically targeted sub-Saharan Africa, while one study focused on rural Africa. Thirteen publications discussed Africa in general without specifying a country or distinct region. The publications based on general and regional accounts are not displayed in , which only presents the country-specific studies. For a detailed summary of the health literacy content available in each country, see . 3.5. Analysis of Target Groups and Settings The analysis of target groups distinguished between population-based and professional-stakeholder-related studies . The population-based studies focused on the public or subgroups of the public, while the studies focusing on the professional perspective encompassed actors in the workforce (health professionals and other relevant stakeholders), organizations, systems, or the policy arena. does not include the 24 publications where it was not possible to specify a target group in their research, which was the case in some commentaries or policy papers. The largest proportion of stakeholders among all studies were patients ( n = 80). Other subgroups included community dwellers ( n = 33), adults ( n = 31), students ( n = 31), women ( n = 23), adolescents ( n = 23), caregivers ( n = 18), and mothers ( n = 15). Finally, high-risk groups, such as migrants, prisoners, the elderly, and indigenous populations, represented only 2% ( n = 6), and several single studies focused on men ( n = 1), consumers ( n = 1), employees ( n = 1), or several target groups ( n = 2), which were grouped under “other” ( n = 5). Target groups associated with the healthcare workforce form the largest proportion ( n = 39). Other professionals include professions outside the health sector ( n = 14). A limited number of publications focused on organizations ( n = 4) and NGOs or policy stakeholders ( n = 3). According to the WHO , a setting for health refers to a place or social context where individuals engage in daily activities, and where environmental, organizational, and personal factors interact to influence health and well-being. The findings revealed that the most common settings studies were the community and the health care settings, followed by educational settings such as secondary school or high school. As shown in , fewer studies were designated in digital/online settings, workplaces, or rural/urban settings. 3.6. Thematic Analysis of Health Literacy in Africa The thematic analysis identified four general trends covering 14 themes based on the types of health literacy categorized in the data, as illustrated in . The trends and themes are summarized in and briefly described in the text with keywords and number of papers identified for each theme. It is beyond the scope of this article to provide a more detailed description based on all the records. Instead, for the sake of transparency, the general data set is provided in . Firstly, there was a disease-oriented focus concentrating on mental health, communicable diseases, and non-communicable diseases. Moreover, maternal health was a priority focus, given the high levels of child morbidity and mortality in many African countries . Mental health literacy included 72 records related to a wide variety of themes including body image, COVID-19 and tuberculosis, suicide and depression literacy, obsessive-compulsive disorder (OCD) literacy, mental health literacy in general, mental health care and screening, mental disorders and illnesses, for instance schizophrenia, mental health governance, school mental health, socio-cultural factors, stigma, and promotion of well-being. The publications presented and discussed facets such as knowledge and capacity building, explanatory models, attitudes, impact of media, prevention and treatment, management, and policy. Communicable diseases included 55 records describing health literacy in a wide range of diseases such as HIV/AIDS, COVID-19, tuberculosis, malaria, Ebola, onchocerciasis, cholera, hepatitis, schistosomiasis, and foodborne diseases. Noncommunicable diseases (NCD) entailed 59 records describing NCD literacy in general and in relation to specific conditions such as aphasia, back pain, cancer, chronic obstructive pulmonary disease (COPD), cardiovascular diseases (CVD), epilepsy, gastrointestinal diseases, diabetes, hypertension, rheumatic heart disease, podoconiosis, systemic lupus erythematosus, and stroke. Maternal health literacy covered 26 records related to family planning, community care, healthy pregnancy, antenatal and postnatal care, neonatal jaundice, mortality and infant survival practices, and childcare by parents and other caregivers. Secondly, a trend focused on organizational and systemic aspects of health literacy in the domains of health systems, health care, prevention, and health promotion. 5. Health system entailed 21 records regarding health service literacy, health system literacy, and nursing literacy. 6. Healthcare included 11 records to health system issues such as treatment and medication literacy. Some specific features were identified, such as autopsy literacy, hemodialysis literacy, obstetric literacy, organ donation literacy, palliative care literacy, and genomic literacy. 7. Prevention covered 24 records associated with reproductive health literacy, HIV/AIDS literacy and sexual health, and vaccine literacy related to COVID-19 and HPV. The theme also included alcohol and drug literacy as well as oral health literacy, hygiene, and antimicrobial resistance literacy. In addition, occupational health literacy and health literacy in relation to child labor were included. 8. Health promotion included 17 records focusing broadly on health literacy-related health promotion strategies such as sexual health literacy, nutrition literacy, disability literacy, and self-care literacy, as well as environmental health literacy. Thirdly, there was a clear focus on communication and information , particularly in relation to digital health . 9. Information and communication health literacy was based on 19 records and focused on awareness, seeking and accessing information, and improving communication between, e.g., patients and healthcare providers. 10. Digital health literacy concerned seven studies and discussions on the use of eHealth literacy, mHealth literacy, social media, and other forms of technology and innovation. Lastly, there was a conceptual focus on concepts and cultural perspectives as well as outcomes and general measurement in relation to health literacy. 11. Conceptual perspectives referred to seven reflections on concepts and approaches related to health literacy, for instance, functional health literacy and the life course perspective. 12. Cultural perspectives included one study that highlighted the importance of cross-cultural understanding and collaboration. 13. Outcomes of health literacy were based on 13 records highlighting outcomes such as health behavior and practices, health status, participation and empowerment, quality of life, sustainability, and equity. 14. Measurement included 13 records emphasizing the wide range of tools, methods, and approaches used to assess the prevalence of health literacy or as an outcome of interventions. Tools included, for example, REALM-R, HLS-EU, HLQ, and the Mental Health Literacy Survey.
The initial search yielded a total of 876 records after removal of duplicates. The co-authoring members of the African Health Literacy Network suggested various studies, but these were already included in the initial set of 876 records. The titles and abstracts were screened for eligibility. As illustrated in the PRISMA flowchart , the screening process yielded 487 records, of which 137 were excluded due to inaccessibility of the full text and 13 due to unavailability of official results or inappropriateness for the purpose of the study. Ultimately, 336 records were included in the review of health literacy development in Africa. The complete list of articles can be found in . The scoping review and synthesis according to research methods, countries, populations, settings, and health literacy factors revealed new insights on the scope and scale of health literacy development in Africa.
Out of the 336 articles, 22 (6.5%) were scoping literature reviews, 291 (86.6%) were original research articles, and 23 (6.8%) were opinions, perspectives, or commentaries published in scientific journals or academic books. Among the original research articles, 163 publications presented quantitative methods, mostly using cross-sectional designs to assess health literacy or disease-related health literacy in patients. Other articles assessed, for example, the effectiveness of controlled trials and interventions. One hundred and five articles used qualitative approaches, with the majority using semi-structured interviews or focus group discussions to investigate barriers to health care or appropriate health behaviors. Finally, 19 publications presented mixed- or multi-method approaches, such as developing or translating health literacy measures or assessing perceptions and health behaviors related to a specific health concern, and 4 described clinical trials.
Scientific publications on health literacy in Africa have increased exponentially since the beginning of the twenty-first century, as shown in . The first mention of health literacy in Africa was in Kickbusch’s seminal paper “Addressing the health and education divide” , which provided several examples. The first specific study on health literacy was published in 2005, focusing on HIV-related health literacy in Zambia. Since then, the number of annual publications steadily increased, reaching 60 publications in 2021 and likely more in the pipeline the following years.
The geographical analysis revealed that health literacy publications were available from 38 of the 54 African Union countries . Most studies were conducted in South Africa ( n = 84), followed by Nigeria ( n = 49), Ghana ( n = 30), Ethiopia ( n = 28), and Uganda ( n = 27). Many countries had only one or a few studies on health literacy, such as Burundi ( n = 1) or Sudan ( n = 1). Additionally, 31 publications concentrated on various regions such as West Africa ( n = 2), North Africa ( n = 1), Southern Africa ( n = 2), and East Africa ( n = 1). Ten publications specifically targeted sub-Saharan Africa, while one study focused on rural Africa. Thirteen publications discussed Africa in general without specifying a country or distinct region. The publications based on general and regional accounts are not displayed in , which only presents the country-specific studies. For a detailed summary of the health literacy content available in each country, see .
The analysis of target groups distinguished between population-based and professional-stakeholder-related studies . The population-based studies focused on the public or subgroups of the public, while the studies focusing on the professional perspective encompassed actors in the workforce (health professionals and other relevant stakeholders), organizations, systems, or the policy arena. does not include the 24 publications where it was not possible to specify a target group in their research, which was the case in some commentaries or policy papers. The largest proportion of stakeholders among all studies were patients ( n = 80). Other subgroups included community dwellers ( n = 33), adults ( n = 31), students ( n = 31), women ( n = 23), adolescents ( n = 23), caregivers ( n = 18), and mothers ( n = 15). Finally, high-risk groups, such as migrants, prisoners, the elderly, and indigenous populations, represented only 2% ( n = 6), and several single studies focused on men ( n = 1), consumers ( n = 1), employees ( n = 1), or several target groups ( n = 2), which were grouped under “other” ( n = 5). Target groups associated with the healthcare workforce form the largest proportion ( n = 39). Other professionals include professions outside the health sector ( n = 14). A limited number of publications focused on organizations ( n = 4) and NGOs or policy stakeholders ( n = 3). According to the WHO , a setting for health refers to a place or social context where individuals engage in daily activities, and where environmental, organizational, and personal factors interact to influence health and well-being. The findings revealed that the most common settings studies were the community and the health care settings, followed by educational settings such as secondary school or high school. As shown in , fewer studies were designated in digital/online settings, workplaces, or rural/urban settings.
The thematic analysis identified four general trends covering 14 themes based on the types of health literacy categorized in the data, as illustrated in . The trends and themes are summarized in and briefly described in the text with keywords and number of papers identified for each theme. It is beyond the scope of this article to provide a more detailed description based on all the records. Instead, for the sake of transparency, the general data set is provided in . Firstly, there was a disease-oriented focus concentrating on mental health, communicable diseases, and non-communicable diseases. Moreover, maternal health was a priority focus, given the high levels of child morbidity and mortality in many African countries . Mental health literacy included 72 records related to a wide variety of themes including body image, COVID-19 and tuberculosis, suicide and depression literacy, obsessive-compulsive disorder (OCD) literacy, mental health literacy in general, mental health care and screening, mental disorders and illnesses, for instance schizophrenia, mental health governance, school mental health, socio-cultural factors, stigma, and promotion of well-being. The publications presented and discussed facets such as knowledge and capacity building, explanatory models, attitudes, impact of media, prevention and treatment, management, and policy. Communicable diseases included 55 records describing health literacy in a wide range of diseases such as HIV/AIDS, COVID-19, tuberculosis, malaria, Ebola, onchocerciasis, cholera, hepatitis, schistosomiasis, and foodborne diseases. Noncommunicable diseases (NCD) entailed 59 records describing NCD literacy in general and in relation to specific conditions such as aphasia, back pain, cancer, chronic obstructive pulmonary disease (COPD), cardiovascular diseases (CVD), epilepsy, gastrointestinal diseases, diabetes, hypertension, rheumatic heart disease, podoconiosis, systemic lupus erythematosus, and stroke. Maternal health literacy covered 26 records related to family planning, community care, healthy pregnancy, antenatal and postnatal care, neonatal jaundice, mortality and infant survival practices, and childcare by parents and other caregivers. Secondly, a trend focused on organizational and systemic aspects of health literacy in the domains of health systems, health care, prevention, and health promotion. 5. Health system entailed 21 records regarding health service literacy, health system literacy, and nursing literacy. 6. Healthcare included 11 records to health system issues such as treatment and medication literacy. Some specific features were identified, such as autopsy literacy, hemodialysis literacy, obstetric literacy, organ donation literacy, palliative care literacy, and genomic literacy. 7. Prevention covered 24 records associated with reproductive health literacy, HIV/AIDS literacy and sexual health, and vaccine literacy related to COVID-19 and HPV. The theme also included alcohol and drug literacy as well as oral health literacy, hygiene, and antimicrobial resistance literacy. In addition, occupational health literacy and health literacy in relation to child labor were included. 8. Health promotion included 17 records focusing broadly on health literacy-related health promotion strategies such as sexual health literacy, nutrition literacy, disability literacy, and self-care literacy, as well as environmental health literacy. Thirdly, there was a clear focus on communication and information , particularly in relation to digital health . 9. Information and communication health literacy was based on 19 records and focused on awareness, seeking and accessing information, and improving communication between, e.g., patients and healthcare providers. 10. Digital health literacy concerned seven studies and discussions on the use of eHealth literacy, mHealth literacy, social media, and other forms of technology and innovation. Lastly, there was a conceptual focus on concepts and cultural perspectives as well as outcomes and general measurement in relation to health literacy. 11. Conceptual perspectives referred to seven reflections on concepts and approaches related to health literacy, for instance, functional health literacy and the life course perspective. 12. Cultural perspectives included one study that highlighted the importance of cross-cultural understanding and collaboration. 13. Outcomes of health literacy were based on 13 records highlighting outcomes such as health behavior and practices, health status, participation and empowerment, quality of life, sustainability, and equity. 14. Measurement included 13 records emphasizing the wide range of tools, methods, and approaches used to assess the prevalence of health literacy or as an outcome of interventions. Tools included, for example, REALM-R, HLS-EU, HLQ, and the Mental Health Literacy Survey.
This comprehensive scientific review provides a novel overview of the scope and scale of health literacy developments in Africa. By incorporating global and African-specific scientific databases, the literature search was rigorously conducted and provided ample evidence that health literacy is an emerging feature of interest on the African continent. The large number of 876 publications covering health literacy, of which 336 publications were included, demonstrates its importance for research, policy, and practice in Africa. According to the study, the global community invested in health literacy, health promotion, and global health has a unique opportunity to learn successful approaches, particularly in adapting questionnaires and interventions to linguistic and cultural contexts from the African community. Over half of the research made use of quantitative approaches illustrating validation and measurement studies are common in Africa like in other parts of the world . While no in-depth analysis of definitions was conducted, the general review of literature revealed that the use of health literacy definitions referred to the most widely used in research . Although health literacy in Africa was briefly explored since 2001, a major shift in research development can be detected since 2017 where the research publications significantly increased. This trend is similar to the global trend, albeit on a smaller scale . Across the African Union, South Africa, Nigeria, Ethiopia, Ghana, and Uganda lead the research development on health literacy. It is worth noting that these countries consist of comparatively large populations in Africa. Moreover, all of these countries utilize English as one of the official languages, a legacy of the colonial era when they were under British rule. The question of whether establishment of universities in these countries in the 20th century by Britain, along with special scholarship programs (such as the Commonwealth Scholarship), or other programs that initiated and intensified long-standing collaboration between universities in these countries and those in the Global North played a role in faster adoption of new developments in global health cannot be satisfactorily answered based on the available data. Interestingly, most publications originated from English-speaking countries, with limited contributions from French- or Portuguese-speaking countries. The research and publication opportunities within and between countries may be affected by a lack of funding or language barriers for non-native English speakers in relation to publishing . The analysis of target groups indicated a greater emphasis on individual health literacy rather than on the organizational health literacy perspectives. This finding is consistent with global developments. This might be grounded in the historical evolvement of the concept which primarily focused on individual health literacy before embracing organizational health literacy responsiveness . In contrast to health literacy research in other world regions, the settings analysis highlighted the interesting feature that health literacy is promoted more extensively in communities and settings outside the health sector. Studies in Europe and North America often focus on clinical settings . The findings can be partly explained by the health challenges prevalent in certain areas of Africa, where access to health services is limited, universal health care is scarce, and health system literacy is often low . Consequently, interventions located in the community might be more appropriate and many African countries are more familiar with it, as they resemble the many interventions aimed at preventing infectious diseases such as malaria or HIV . The thematic analysis emphasized the specific health challenges facing Africa. Firstly, with regard to communicable diseases, health literacy was widely associated with communicable diseases, predominantly in the African region, such as Ebola, HIV/AIDS, and malaria . Moreover, the emphasis on maternal health is due to the heightened risk of morbidity and mortality within childhood (under-5 mortality) . The study suggests that mental health literacy remains a challenge due to insufficient knowledge and awareness, stigma, and prejudice, which is consistent with findings from other research studies . The widespread focus on non-communicable diseases such as cancer, diabetes, and cardiovascular diseases reflects the projected increase in global burden . Secondly, health information and communication remain an important avenue for raising awareness, especially through digital health literacy such as mHealth literacy, which appears to be a promising strategy given the increased access to mobile coverage . These findings align with global trends and policies . Thirdly, a cross-cutting topic was sexual and reproductive health literacy related to family planning, which remains a challenge in the world’s youngest population . Fourthly, the topical analysis highlighted efforts to bend the curves of limited health literacy regarding health systems and organizations, health care, disease prevention, and health promotion to improve outcomes. This is a universal challenge that is prevalent around the globe . Finally, the discussion of conceptual approaches, measurements, and ways to address cultural change was dominated by the academic perspectives, in line with international research developments . While this scoping review aimed to provide a general overview of the scope and scale of health literacy developments in Africa, the findings show that many more lessons can be learned through a more in-depth analysis of the themes and sub-themes in the future. 4.1. Strengths and Limitations Despite the rigorous study design, some limitations remained. Data search and collection focused on countries officially recognized as members of the African Union, which excluded countries and territories such as Somaliland and Western Sahara. These choices were made for administrative reasons but may have introduced bias into the data set. Moreover, the main purpose of this study was to provide a general overview of articles on health literacy trends in Africa, not to challenge and critique geographical boundaries. This study used official national borders, such as those used by the African Union, to categorize different geographic regions, recognizing that these borders do not always reflect the reality of ethnic groups that are found on either side of national borders. Given the vast diversity of ethnic groups across the African continent, including their linguistic, cultural, geographical, and often socio-economic particularities, it is crucial to account for these factors when designing future studies. The data collection was based on six international and Africa-focused databases. While this may not be an exhaustive list of all health literacy publications, the Africa-specific databases clearly yielded studies that would not have been found elsewhere. Scanning the gray literature was beyond the scope of the study; however, to be inclusive, all types of resources identified in the databases were included based on the inclusion criteria, not just the research publications. An initial test search revealed no results for Portuguese publications; hence, the study was based on a search of key terms in English and French, which are also the only or primary languages used as interfaces for these databases. A considerable number of these databases offer abstracts in English (or French) even in instances where the original article is available in a different language. In accordance with the aforementioned strategy, the authors were thus able to guarantee that all records with an abstract in one of these languages were identified, irrespective of the language employed in the full text. Following the screening of both the abstract and the full text, only records in English remained. It is possible that the application of this search strategy may have resulted in the overlooking of records in other major languages spoken in Africa, such as Portuguese, Swahili, or Afrikaans, which have not been indexed with an English or French abstract in the databases consulted. The aim of this review was analysis of publications using the term ‘health literacy’, so other studies that indirectly address ‘health literacy’ without mentioning the concept were not included. This review presents the general results of a broad trend analysis of health literacy publications in Africa. An in-depth meta-analysis of research articles or policy recommendations was beyond the scope of this review, although the data were available. Further research and publications are warranted to fill these gaps. 4.2. Implications for the Future Reviewing the developments of health literacy in Africa sparks food for thought. First and foremost, the study revealed that health literacy is on the agenda in Africa and that progress is expanding. Improving health literacy in Africa requires a multifaceted approach involving policy integration, education system enhancement, community engagement, technological innovation, health system improvements, research, and partnerships. It is paramount to create more visibility regarding the need for health literacy and good practice interventions and their importance in improving health outcomes and the quality of health services. By addressing these areas, African countries can empower their populations to make informed health decisions, ultimately improving health outcomes and reducing health disparities across the continent. The study highlighted the importance of contextualizing health literacy with respect to African society, culture, and traditions . A co-author highlighted that when searching Google and consulting the scientific discourse, health literacy is often portrayed, e.g., with an image of a person sitting in front of a computer navigating for health information or discussing and negotiating treatment options with a physician. These situations may not be appropriate in many African settings, where large portions of the population are rural dwellers, have limited access to digitalization, and often low functional literacy. Thus, the strategies, tools, and approaches for health literacy should be tailored to local opportunities and needs, for example, by translating and validating tools for health literacy assessments, using social mobilization, community dialogue, and engaging local religious and community leaders to provide information about health and available health services. Drawing on local wisdom and strategies of co-design and participation is recommended to define and conceptualize health literacy for each unique African context . This can be performed in collaboration with health literacy practitioners in the specific community, district, region, or country with the aim of making it a priority agenda for professionals and policymakers through advocacy using multiple platforms. Health literacy initiatives can tackle social norms through proven interventions, including stakeholder engagement and community dialogue. Harmful practices adversely affect the groups most at risk. Given the degree of vulnerability among the general population, efforts to engage the most vulnerable are often deprioritized to maximize scarce resources. In Africa, where health challenges are significant and resources are often limited, enhancing health literacy can play a transformative role. Similarly to trends in other world regions, health literacy should be a core component of national health policies and strategies. Governments should recognize it as a critical determinant of health and integrate it into treatment pathways and health promotion and disease prevention programs. This includes developing resources and providing ongoing professional development opportunities for health professionals, teachers, and community workers and leaders alike. As this study shows, it is important to collaborate with community leaders, religious figures, and traditional healers to promote health literacy. These trusted figures can help disseminate health information and encourage positive health behaviors. Use of health literacy champions may pave the way for better implementation of health literacy . Local and international investment and networking are needed for further research, better interventions, and policy development to advance health literacy in Africa. For example, health literacy strategies can be included in the health sector development plans of each African country. Investments in more health literacy interventions can benefit both populations and systems specifically if cultural sensitivity is considered when developing and implementing the interventions. The African Health Literacy Network is a crucial player in strengthening collaboration and further dissemination of the health literacy agenda to build the capacity of people, professionals, and societies in the African region. It was launched in 2017 and has grown to cover membership from 25 countries. It hosted its first pan-African conference in August 2023. Establish mechanisms to monitor and evaluate the effectiveness of health literacy programs in Africa. This includes conducting research to identify best practices and areas for improvement. The lessons learned from this study revealed that measurement and evaluation of health literacy takes place and can be amplified more progressively. Collecting data on health literacy levels across different regions and demographic groups to inform policy decisions and tailor interventions accordingly will improve the quality of implementation. As an example, the WHO is supporting countries like Liberia and St. Tome and Principe to measure population health literacy to inform and develop the national health literacy agendas (contact the corresponding author for more information).
Despite the rigorous study design, some limitations remained. Data search and collection focused on countries officially recognized as members of the African Union, which excluded countries and territories such as Somaliland and Western Sahara. These choices were made for administrative reasons but may have introduced bias into the data set. Moreover, the main purpose of this study was to provide a general overview of articles on health literacy trends in Africa, not to challenge and critique geographical boundaries. This study used official national borders, such as those used by the African Union, to categorize different geographic regions, recognizing that these borders do not always reflect the reality of ethnic groups that are found on either side of national borders. Given the vast diversity of ethnic groups across the African continent, including their linguistic, cultural, geographical, and often socio-economic particularities, it is crucial to account for these factors when designing future studies. The data collection was based on six international and Africa-focused databases. While this may not be an exhaustive list of all health literacy publications, the Africa-specific databases clearly yielded studies that would not have been found elsewhere. Scanning the gray literature was beyond the scope of the study; however, to be inclusive, all types of resources identified in the databases were included based on the inclusion criteria, not just the research publications. An initial test search revealed no results for Portuguese publications; hence, the study was based on a search of key terms in English and French, which are also the only or primary languages used as interfaces for these databases. A considerable number of these databases offer abstracts in English (or French) even in instances where the original article is available in a different language. In accordance with the aforementioned strategy, the authors were thus able to guarantee that all records with an abstract in one of these languages were identified, irrespective of the language employed in the full text. Following the screening of both the abstract and the full text, only records in English remained. It is possible that the application of this search strategy may have resulted in the overlooking of records in other major languages spoken in Africa, such as Portuguese, Swahili, or Afrikaans, which have not been indexed with an English or French abstract in the databases consulted. The aim of this review was analysis of publications using the term ‘health literacy’, so other studies that indirectly address ‘health literacy’ without mentioning the concept were not included. This review presents the general results of a broad trend analysis of health literacy publications in Africa. An in-depth meta-analysis of research articles or policy recommendations was beyond the scope of this review, although the data were available. Further research and publications are warranted to fill these gaps.
Reviewing the developments of health literacy in Africa sparks food for thought. First and foremost, the study revealed that health literacy is on the agenda in Africa and that progress is expanding. Improving health literacy in Africa requires a multifaceted approach involving policy integration, education system enhancement, community engagement, technological innovation, health system improvements, research, and partnerships. It is paramount to create more visibility regarding the need for health literacy and good practice interventions and their importance in improving health outcomes and the quality of health services. By addressing these areas, African countries can empower their populations to make informed health decisions, ultimately improving health outcomes and reducing health disparities across the continent. The study highlighted the importance of contextualizing health literacy with respect to African society, culture, and traditions . A co-author highlighted that when searching Google and consulting the scientific discourse, health literacy is often portrayed, e.g., with an image of a person sitting in front of a computer navigating for health information or discussing and negotiating treatment options with a physician. These situations may not be appropriate in many African settings, where large portions of the population are rural dwellers, have limited access to digitalization, and often low functional literacy. Thus, the strategies, tools, and approaches for health literacy should be tailored to local opportunities and needs, for example, by translating and validating tools for health literacy assessments, using social mobilization, community dialogue, and engaging local religious and community leaders to provide information about health and available health services. Drawing on local wisdom and strategies of co-design and participation is recommended to define and conceptualize health literacy for each unique African context . This can be performed in collaboration with health literacy practitioners in the specific community, district, region, or country with the aim of making it a priority agenda for professionals and policymakers through advocacy using multiple platforms. Health literacy initiatives can tackle social norms through proven interventions, including stakeholder engagement and community dialogue. Harmful practices adversely affect the groups most at risk. Given the degree of vulnerability among the general population, efforts to engage the most vulnerable are often deprioritized to maximize scarce resources. In Africa, where health challenges are significant and resources are often limited, enhancing health literacy can play a transformative role. Similarly to trends in other world regions, health literacy should be a core component of national health policies and strategies. Governments should recognize it as a critical determinant of health and integrate it into treatment pathways and health promotion and disease prevention programs. This includes developing resources and providing ongoing professional development opportunities for health professionals, teachers, and community workers and leaders alike. As this study shows, it is important to collaborate with community leaders, religious figures, and traditional healers to promote health literacy. These trusted figures can help disseminate health information and encourage positive health behaviors. Use of health literacy champions may pave the way for better implementation of health literacy . Local and international investment and networking are needed for further research, better interventions, and policy development to advance health literacy in Africa. For example, health literacy strategies can be included in the health sector development plans of each African country. Investments in more health literacy interventions can benefit both populations and systems specifically if cultural sensitivity is considered when developing and implementing the interventions. The African Health Literacy Network is a crucial player in strengthening collaboration and further dissemination of the health literacy agenda to build the capacity of people, professionals, and societies in the African region. It was launched in 2017 and has grown to cover membership from 25 countries. It hosted its first pan-African conference in August 2023. Establish mechanisms to monitor and evaluate the effectiveness of health literacy programs in Africa. This includes conducting research to identify best practices and areas for improvement. The lessons learned from this study revealed that measurement and evaluation of health literacy takes place and can be amplified more progressively. Collecting data on health literacy levels across different regions and demographic groups to inform policy decisions and tailor interventions accordingly will improve the quality of implementation. As an example, the WHO is supporting countries like Liberia and St. Tome and Principe to measure population health literacy to inform and develop the national health literacy agendas (contact the corresponding author for more information).
In conclusion, this review of health literacy trends in Africa underscores the call to action to develop the field of health literacy in ways that are responsive to the needs and concerns of African countries and contexts, rather than copying what has already been performed elsewhere without much adaptation. As public health services are currently being strengthened and expanded in many African countries , the valuable lessons learned from this study can be used as a reference point and incorporated from the outset, particularly with regard to building systemic health literacy capacity and incorporating health literacy practices as part of community development. Building on the widespread legacy of health education and community development in the African region, new approaches applied through a health literacy lens can help to target populations in vulnerable situations and make services more fit for purpose. Proposed solutions include co-designed, targeted programs and interventions for specific populations, as well as capacity building of the workforce and investment in education policies. Particular attention should be paid to ensure cultural appropriateness and adaptation to the needs and perceptions of the populations, especially women and high-risk groups, to leave no one behind. Strengthening health literacy in Africa is essential to building a future where every African can enjoy a life of better health and well-being. Importantly, this scoping review highlights that and how health literacy development is on the rise and is promisingly being recognized as a positive key driver of change across the African continent.
|
Impact of the COVID-19 Pandemic on the Wellbeing of International Oncology and Hematology Fellows at the Princess Margaret Cancer Center (PMCC) | d2dc9ae4-f963-4c07-bac9-d98d657b097b | 9452400 | Internal Medicine[mh] | Clinical fellowship programs have been widely implemented to provide opportunities for physicians to obtain advanced and/or specialized training. The Princess Margaret Cancer Centre (PMCC) in Toronto, Ontario, Canada offers a number of clinical and research fellowships in hematology and oncology that attracts physicians from all around the world. Although important from a career standpoint, moving from one country to another can be very difficult for a number of reasons. These include, but are not limited to, learning a new healthcare system, new language, different cultural customs and traditions and importantly separation from family, friends and social supports. In March 2020, the COVID-19 pandemic was declared. This had the potential to significantly exacerbate existing challenges and introduce new challenges for international fellows. One of the most important measures implemented to control the COVID-19 pandemic, was social distancing. Although very effective, it has had a significant impact on the mental health and well-being of society as a whole. This impact may be even more profound for international fellows adjusting to a new country and at the same time being concerned about the health and wellbeing of their loved ones back in their home countries. The primary aim of this fellow-led project was to identify and describe first hand experiences and key challenges faced by international fellows at the PMCC during the COVID-19 pandemic. We were also interested to understand how fellows were coping and to explore potential strategies that did or could have improved the overall fellowship experience during the pandemic. By sharing this data, we hope to provide guidance and tools that may help to improve mental health and well-being during the current pandemic, and importantly help us to be better equipped and prepared in the future.
This survey study was deemed exempt by the Institutional Review Board of the University Health Network (UHN) as it fell within the scope of minimal risk research and participation was completely voluntary. Sample and data collection The survey consisted of 60 questions, divided into 4 key parts including demographics, wellbeing assessment, fellowship specific questions (personal and professional), and coping strategies (see supplementary appendix). For the wellbeing assessment, the validated Short Warwick Edinburgh Wellbeing Scale (SWEMWBS) was used, after approval by the scale's creator. This scale uses 7 of the 14 statements present on the original scale which relate more to functioning than feeling. For the assessment of coping mechanisms, we used the Brief COPE scale, a validated self-report questionnaire consisting of 28 items to identify different coping mechanisms (supplementary appendix). In addition, there were open-ended questions about strategies that could be implemented by the fellowship programs to help better support fellows during the pandemic. Each of the 52 fellows, defined as physicians who have completed medical training and residency and are now in a subspecialty training program, was sent a link to an online survey by SurveyMonkey on July 6, 2020, with 5 reminders for the non-responders sent until August 10, 2020. Completion of the survey was voluntary and all responses were anonymous. Data interpretation The scoring by the SWEMWBS was done by summing the scores obtained for each of the items, to give a final raw score that was transformed into metric scores using a conversion table provided by the SWEMWBS developers. The final scores ranged from 7 to 35, reflecting the following: ≤17 probable depression, 18–20 possible depression, 21–27 average mental wellbeing and ≥28 high mental wellbeing. The Brief COPE scale, proposed by Carver in 1997 , is a 28 item self-report questionnaire designed to measure effective and ineffective ways to cope with a stressful life event. Coping was defined broadly as an effort to minimize distress associated with negative life experiences . The questions are divided into 14 subscales: self-distraction, denial, substance use, self-blame, behavioral disengagement, venting, emotional support, use of informational support, active coping, positive reframing, planning, humor, acceptance and religion. The first 6 subscales are considered as maladaptive coping approaches, with scores in the range of 12 to 48, and the other 8 subscales are described as adaptive approaches with scores in the range of 16 to 64.
The survey consisted of 60 questions, divided into 4 key parts including demographics, wellbeing assessment, fellowship specific questions (personal and professional), and coping strategies (see supplementary appendix). For the wellbeing assessment, the validated Short Warwick Edinburgh Wellbeing Scale (SWEMWBS) was used, after approval by the scale's creator. This scale uses 7 of the 14 statements present on the original scale which relate more to functioning than feeling. For the assessment of coping mechanisms, we used the Brief COPE scale, a validated self-report questionnaire consisting of 28 items to identify different coping mechanisms (supplementary appendix). In addition, there were open-ended questions about strategies that could be implemented by the fellowship programs to help better support fellows during the pandemic. Each of the 52 fellows, defined as physicians who have completed medical training and residency and are now in a subspecialty training program, was sent a link to an online survey by SurveyMonkey on July 6, 2020, with 5 reminders for the non-responders sent until August 10, 2020. Completion of the survey was voluntary and all responses were anonymous.
The scoring by the SWEMWBS was done by summing the scores obtained for each of the items, to give a final raw score that was transformed into metric scores using a conversion table provided by the SWEMWBS developers. The final scores ranged from 7 to 35, reflecting the following: ≤17 probable depression, 18–20 possible depression, 21–27 average mental wellbeing and ≥28 high mental wellbeing. The Brief COPE scale, proposed by Carver in 1997 , is a 28 item self-report questionnaire designed to measure effective and ineffective ways to cope with a stressful life event. Coping was defined broadly as an effort to minimize distress associated with negative life experiences . The questions are divided into 14 subscales: self-distraction, denial, substance use, self-blame, behavioral disengagement, venting, emotional support, use of informational support, active coping, positive reframing, planning, humor, acceptance and religion. The first 6 subscales are considered as maladaptive coping approaches, with scores in the range of 12 to 48, and the other 8 subscales are described as adaptive approaches with scores in the range of 16 to 64.
The survey was sent to all 52 fellows working at the PMCC on July 6, 2020, and reminder e-mails were sent on July 13, July 20, July 27, August 3 and August 10, 2020, resulting in a total of 24 (46.1%) respondents. Demographics Most of the respondents were men (54%), between the ages of 31–35 (48%); 65% were married and 49% had children. Overall, 48% started their program in 2019, 21/24 fellows were international with 48% from Asia ( ). There were 3 fellows from other Canadian provinces who also completed the survey. They were included in the analysis because we believe the challenges, they faced during the pandemic may be similar to international fellows, since the COVID-19 restrictions in Canada and imposed by the institution limited travel even between Canadian provinces. Wellbeing scale The final mean value, after using the conversion table, among all responders was 21.0. For married participants the mean score was 21.4; for those with children it was 21.0; and for single fellows the mean score was the lowest, at 19.6 ( ). Fellowship specific questions- personal When asked how often their wellbeing had been affected during the pandemic, 62.5% responded either “often” or “all of the time” and when asked about wellbeing before the pandemic, the majority responded that their wellbeing was not or rarely affected (58%). The majority of responders admitted feeling guilty some of the time for not being with their family (46%), and for not helping their country (42%) during the pandemic. Regarding financial status, 29% said they were often or always concerned about money. When inquired about stress in their relationship 26% of the fellows described it as being an issue “often” or “all of the time.” Among the fellows who were parents, 36% admitted to frequently being concerned about the lack of childcare. Fellowship specific questions-professional Half of the participants agreed that the weekly schedule of their fellowship program had changed considerably. However, most believed that there were little or no changes to the on-call periods during the pandemic (79%). Overall, 42% also believed their clinical work had suffered major changes, while 50% thought the same about their research projects. Most were satisfied with the way their fellowship had been conducted during the pandemic. Half of the survey participants admitted experiencing lack of energy and fatigue. Sleep disorders and loss of interest and pleasure in daily activities were described by 38%, 29% described weight alterations, and 20% reported agitation. When asked if any of these symptoms were present before the pandemic, 46% responded affirmatively. Due to the pandemic, personal events had to be postponed or cancelled by almost 80% of the fellows. Based on the PMCC directive to work from home where possible, the majority (66%) of fellows started working from home at least one day a week, however, 34% continued to come to the office and/or hospital every day. For 5 (21%) fellows, the total duration of the fellowship program was extended due to the pandemic. Coping mechanisms The most common coping mechanisms described was video conferencing with family and friends, focusing on research projects, cooking more frequently, doing physical exercise and planning weekly routines in advance. Yoga, meditation, reading, streaming television shows and movies were also described. Overall, 13% of the fellows admitted to smoking/drinking more often as another way of coping during the pandemic ( ). Regarding the Brief-COPE scale, we observed a mean score of 21.9 (ranging from 13 to 38) for the maladaptive coping strategies, the most relevant one being self-distraction, and a mean score of 39.2 (28–52) for the adaptive ones, mainly represented by planning, positive reframing and acceptance strategies ( ).
Most of the respondents were men (54%), between the ages of 31–35 (48%); 65% were married and 49% had children. Overall, 48% started their program in 2019, 21/24 fellows were international with 48% from Asia ( ). There were 3 fellows from other Canadian provinces who also completed the survey. They were included in the analysis because we believe the challenges, they faced during the pandemic may be similar to international fellows, since the COVID-19 restrictions in Canada and imposed by the institution limited travel even between Canadian provinces.
The final mean value, after using the conversion table, among all responders was 21.0. For married participants the mean score was 21.4; for those with children it was 21.0; and for single fellows the mean score was the lowest, at 19.6 ( ).
When asked how often their wellbeing had been affected during the pandemic, 62.5% responded either “often” or “all of the time” and when asked about wellbeing before the pandemic, the majority responded that their wellbeing was not or rarely affected (58%). The majority of responders admitted feeling guilty some of the time for not being with their family (46%), and for not helping their country (42%) during the pandemic. Regarding financial status, 29% said they were often or always concerned about money. When inquired about stress in their relationship 26% of the fellows described it as being an issue “often” or “all of the time.” Among the fellows who were parents, 36% admitted to frequently being concerned about the lack of childcare.
Half of the participants agreed that the weekly schedule of their fellowship program had changed considerably. However, most believed that there were little or no changes to the on-call periods during the pandemic (79%). Overall, 42% also believed their clinical work had suffered major changes, while 50% thought the same about their research projects. Most were satisfied with the way their fellowship had been conducted during the pandemic. Half of the survey participants admitted experiencing lack of energy and fatigue. Sleep disorders and loss of interest and pleasure in daily activities were described by 38%, 29% described weight alterations, and 20% reported agitation. When asked if any of these symptoms were present before the pandemic, 46% responded affirmatively. Due to the pandemic, personal events had to be postponed or cancelled by almost 80% of the fellows. Based on the PMCC directive to work from home where possible, the majority (66%) of fellows started working from home at least one day a week, however, 34% continued to come to the office and/or hospital every day. For 5 (21%) fellows, the total duration of the fellowship program was extended due to the pandemic.
The most common coping mechanisms described was video conferencing with family and friends, focusing on research projects, cooking more frequently, doing physical exercise and planning weekly routines in advance. Yoga, meditation, reading, streaming television shows and movies were also described. Overall, 13% of the fellows admitted to smoking/drinking more often as another way of coping during the pandemic ( ). Regarding the Brief-COPE scale, we observed a mean score of 21.9 (ranging from 13 to 38) for the maladaptive coping strategies, the most relevant one being self-distraction, and a mean score of 39.2 (28–52) for the adaptive ones, mainly represented by planning, positive reframing and acceptance strategies ( ).
In this study led by our fellows, we were able to assess the wellbeing and coping strategies of predominantly international hematology and oncology fellows at one of the largest cancer centers in Canada during the first wave of the COVID-19 pandemic. This has particular relevance as it sheds further light on the important topic of physician wellbeing. Most of the fellows in the study stated that their wellbeing was frequently affected during the pandemic. This is probably due to a combination of factors including feeling guilty for not being physically close to their families and not being able to provide a service in their own country. Similar findings were also reported in a survey of international radiation oncology fellows . Other factors affecting well-being included concerns about financial status, relationship stress, and lack of childcare. Although not specifically addressed in our survey, significant changes in the work environment, social isolation, covering COVID triage clinics, and constant fear of catching COVID (before the availability of vaccines) also likely led in part to decreased wellbeing. These concerns echo those expressed by other health care providers surveyed during the pandemic . In our survey the observed mean score of SWEWS was 21.0, which represents the lower end of the average mental wellbeing scores which range from 21 to 27, and suggests a possible negative emotional impact caused by the pandemic, even though we do not have a pre-pandemic baseline assessment. A similar study conducted among 17 radiation-oncology trainees during the pandemic, demonstrated that burnout features in the domains of disengagement and exhaustion were present in 71% and 64%, respectively. But, importantly they also described evidence of resilience in 47%, as they felt that positive alterations brought about by the pandemic outnumbered the negative ones. In addition, good institutional leadership along with the rapid implementation of virtual care helped them to feel more energized . Still regarding the SWEWS, we detected a slight difference in the score between married (21.6) and single fellows (19.6). Although the difference is small, it generates a hypothesis that possibly having a partner during the pandemic may have had a protective effect, but this would need to be validated in a larger sample. Interestingly, this result is in keeping with a study on wellbeing among married and unmarried individuals that reported the association of marriage with overall perception of higher degree of wellbeing . Although participants felt that their fellowship had changed during the pandemic, most agreed that the hospital and university were able to guarantee a safe working environment both mentally and physically. Clear, quick and regular communications from hospital management about strategies to limit spread of COVID-19, COVID-19 protocols, and the number of positive cases admitted at the hospitals in the network, sent fellows a positive message that the situation was being managed as best as possible, and improved the sense of safety. Other important strategies, contributing positively to wellbeing, included the fact that personal protective equipment was widely available and training was provided to all fellows on how to manage patients with positive or suspected COVID infections. Importantly, this training was tailored to every possible scenario from the clinic to emergency situations such as a protected code blue. This guidance was particularly appreciated by the fellows, as this was all happening at a time of great uncertainty surrounding COVID-19. With regards to professional questions, almost half of the fellows agreed that their clinical work had suffered major changes, which was expected given the changes in the delivery of clinical care. Most in-person visits were converted to phone calls, but interestingly, more than half stated that the changes were not significant. This is an optimistic view of the situation which is possibly linked to their sense of purpose and feeling valued. In a study conducted by Pilar et al , radiation oncology trainees reported that good leadership that encouraged culture change, more time to develop research projects and rapid implementation of virtual care were seen as silver-linings of the pandemic . Results obtained through the Brief-COPE scale showed that in general the fellows have been using more adaptive mechanisms (approach coping) as opposed to mal-adaptive ones (avoidant coping), as previously described. As expected, approach coping is associated with better responses to adversity and is more effective at dealing with anxiety. A Saudi Arabian study from 2018 by Alosaimi et al assessing stress among 582 physicians, which used the same scale to evaluate coping mechanisms has also demonstrated a predominance of approach coping strategies. Even prior to the pandemic, the crisis of physician burnout has been well recognized and is a cause for concern not only among more experienced physicians but also among residents and even medical students , . In a study from 1991 assessing burnout syndrome amongst 598 oncologists, almost 60% had experienced burnout . In a survey of 1700 medical oncologists working in a community practice in the US, Allegra et al showed that nearly 62% had symptoms of burnout . More recently, data from the American Society of Clinical Oncology reported that 45% of their members have reported experiencing burnout-related symptoms , . For some physicians, wellbeing was already affected prior to the pandemic and may have been further compromised by the pandemic . Oncologists may be particularly vulnerable to COVID-19-related distress. This is not only because of their concern for their own health but also due to their patient's vulnerability to infection, the need to possibly suspend or delay anti-cancer treatment, and the inherent risks of decreasing chances of cure, and the expected increase in mortality associated with cancer, attributed to treatment delays , . A recent study by Lu et al revealed that during the pandemic, medical staff have experienced a higher incidence of fear, anxiety and depression when compared to administrative staff. Professionals at higher risk were working in the departments of respirology, emergency medicine, intensive care and infectious diseases . On the other hand, it has also been argued that the pandemic has brought an increased sense of altruism and restored some of the elements of autonomy, competency and relatedness, intimately associated with intrinsic motivation, which could reduce the risk of burnout . In a study from Wuhan comparing burnout frequency among oncology healthcare workers on the frontline and those in the usual wards during COVID-19, a lower frequency of burnout was seen in frontline staff. This could be explained by the fact that frontline workers may have a better sense of control over the situation and that more attention and recognition was paid to those working directly with symptomatic patients. Workers in the usual wards may not have had the same sense of control and may not have experienced the same degree of gratitude . Oncology and hematology fellows would be included in the second group, as most of them are not directly involved in the frontline care of COVID patients. Therefore, based on the premise previously stated, they could be facing a higher risk of burnout. The study has several limitations. These include the limited sample size, and the fact that only half of the fellows responded to the questionnaire, causing a possible selection bias. Moreover, this survey was conducted in a single cancer center, which does not have its own emergency room, and where the number of COVID 19 cases were low since the beginning of the pandemic. More robust and multicentric studies would help us to better understand the real impact of the pandemic on the wellbeing of international fellows. To this end, a follow-up study, among international fellows across the wider Department of Medicine, at the University of Toronto is being planned.
This survey, conducted early in the ongoing COVID-19 pandemic among international fellows, demonstrated lower wellbeing scores than the estimates for the general population. Understanding the specific challenges and coping mechanisms of international fellows may help institutions develop better targeted strategies to promote their overall wellbeing, professional development and high-quality patient care during unprecedented times such as the COVId-19 pandemic.
Carlos Stecca, No Relationships to Disclose. Di Maria Jiang, Consulting or Advisory Role – Bayer. Marie Alt, No Relationships to Disclose. Mary Elliott, No Relationships to Disclose. Nazanin Fallah-Rad, No Relationships to Disclose. Glaucia Michelis, No Relationships to Disclose. Srikala S. Sridhar, Consulting or Advisory Role - Astellas Pharma (Inst); AstraZeneca (Inst); Bayer (Inst); Bristol-Myers Squibb (Inst); Immunomedics (Inst); Janssen (Inst); Merck (Inst); Pfizer (Inst); Roche/Genentech (Inst); Sanofi (Inst)
|
Immunohistochemistry in postmortem diagnosis of acute cerebral hypoxia and ischemia | dc3833b3-e0f0-4f72-855d-472161301e30 | 8238305 | Anatomy[mh] | Introduction Discovery of evidence of acute brain ischemia or hypoxia and its differentiation from agonal hypoxia represents a task of particular interest but extremely difficult in forensic neuropathology. [ – ] Generally, more than 50% of forensic autopsies indicate evidence of brain induced functional arrest of the organ system, which can be the result of a hypoxic/ischemic brain event. The latter can be caused by multiple conditions such as traumatic or chemical events, respiratory and cardiac arrest, asphyxiation or obstruction of the cerebral or cervical vessels. At present, there are still no specific neurological signs in case of acute fatal hypoxia (which may occur, for example, in strangulation or drowning). [ , , ] Ischemia and hypoxia have often been considered of a similar nature and define the cell's inability to receive and use oxygen. However, ischemia is characterized by the reduction/absence of cerebral blood flow resulting in irreversible neuronal destruction while hypoxia foresees a lack of oxygen in the blood and indicates an increase in cerebral blood flow with a reversible alteration of brain functions. Specifically, in the event of the obstruction of the areas (aspiration, asthma, etc.) or absence of ambient oxygen (as in drowning), cerebral circulation continues but an increase in plasma carbon dioxide will occur (with a secondary dilation of the arteries) as well as a reduction in the partial pressure of oxygen (pO 2 ). The obstruction of the cerebral vessels or of the neck (as in the case of hanging or strangulation) induces both a reduction/arrest of the intracranial circulation as well as an alteration of the values of gases in the blood and the pH associated with acidosis and increase in lactates. As a result of these processes, a dysfunction of the sodium-potassium pump and cytotoxic edema occurs. In the case of reperfusion, the cytotoxic edema will be associated with vasogenic edema within 10 to 20 minutes. When analyzing ischemic-hypoxic brain damage, the susceptibility of specific brain areas (watershed area) must be considered in relation to vascular anatomy and in relation to the different types of neurons. In this sense, the fissure between the first and second turns of the frontal lobe, the CA1 region of the hippocampus, the Purkinje cells of the cerebellum and the pale globe represent the areas that are most susceptible to hypoxia and samples should be taken. Some neurobiological changes occur in neurons after hypoxic/ischemic damage but histological neuronal findings appear in the brain after several hours. On the other hand, neuronal necrosis and neuron loss are detectable after a long period of survival. [ – ] If death occurs within a few hours, it is extremely complicated (if not impossible) to observe specific macroscopic changes. Even from a microscopic point of view, the very acute ischemic/hypoxic area of the brain is difficult to trace. Ischemic neurons initially reveal collapsed, shrunken and pyknotic nuclei and intensely eosinophilic cytoplasm (in hematoxylin-eosin staining), and the Nissl substance appears dispersed and finely granular. Subsequently, when chromatin has degraded, the nuclei become more eosinophilic and appear to merge with the surrounding cytoplasm. Neutrophils can be mildly vacuolized (nonspecific relief) or normal. These changes (the first to occur in case of necrosis) are histologically visible only after a survival time of at least 4 to 6 hours or even 8 to 12 hours. [ , – ] In many cases, however, death occurs within hours or even minutes of the initial acute event. If death occurs after several minutes (as in cases of drowning or hypoxia) or within just a few hours, it would not be possible to identify ischemic/hypoxic brain damage through a conventional macroscopic and histological examination. After the initial ischemic/hypoxic damage, a series of structural alterations of proteins of the brain begins to take place. In this regard, several immunohistochemical markers have been investigated with the objective of identifying acute brain damage. Currently, a routine method that could resolve this problem has not been identified. On the basis of a critical analysis of the literature on the use of immunohistochemistry in hypoxic-ischemic injury, this review aims to analyze and summarize all the principal markers examined to date, regarding acute ischemic/hypoxic brain damage.
Materials and methods All the main scientific studies regarding the immunohistochemical evaluation of acute cerebral hypoxia/ischemia were examined. Specifically, we used the search engines PUMED ( https://www.ncbi.nlm.nih.gov/pmc/ ) and Scopus ( https://www.scopus.com ) to research the keywords “immunohistochemical markers,” “acute cerebral ischemia,” “ischemic or hypoxic brain damage,” and “acute cerebral hypoxia”. All major papers published in the English language in the last 25 years were also considered. The results of the search were screened on the basis of the titles and abstracts of the papers. We excluded papers that did not fully relate with the subject under examination: those that did not relate with the immunohistochemical diagnosis of acute hypoxic/ischemic brain injury. Only studies that considered an acute ischemic/hypoxic brain damage were considered and included in the review. Studies where the period of survival was greater than 12 to 24 hours were not taken into consideration. Articles deemed relevant to the issue under investigation were read and analyzed in their entirety. Furthermore, studies with mainly forensic purposes were mainly examined. We conducted a critical analysis of all the scientific papers selected, examining all the markers utilized and the immunohistochemical response.
Results The search yielded over 35 scientific papers deemed suitable for analysis. We identified the following markers that had been examined (in previous studies) for the purpose of diagnosing acute cerebral hypoxia and ischemia (Table ). The principal markers were: Tau protein: Tau protein plays an important role in the assembly and stabilization of microtubules. In addition, this protein is involved in the mechanisms of signal transduction, interaction with the cytoskeleton actin, neuritis outgrowth, and stabilization during brain development. An alteration of the Tau protein has been found in neurodegenerative diseases. Other studies correlate the aggregates of this protein with acute brain damage such as ischemia. Salama et al studied Tau expression in the animal models of acute hypoxia-ischemia. Tau aggregates were significantly greater in hypoxic-ischemic models than in controls. According to the authors, this protein could be useful in the forensic investigation of asphyxiated death. However, the study has several limitations: the limited number of cases studied, a long period of survival (1 day), the unlikelihood of confirming whether this result is a direct effect of hypoxia or an epiphenomenon linked to secondary injury cascade. S-100: S-100 protein is a binding-calcium protein with a subunit of A and B. [ – ] The S100B subunit is highly specific for ependymocytes, oligodendrocytes and astrocytes in the central nervous system. On the other hand, the S100A subunit is present in skeletal muscles, lungs, liver, kidneys, pancreas, and heart. [ – ] S100B has been clinically studied as a serum marker of brain damage. [ , – ] Li et al studied the immunohistochemical expression of the S100 protein in the cerebral cortex in forensic autopsy cases. A lower expression was seen in cases of asphyxia due to neck compression (strangulation and hanging) and drowning (than in other groups). The decrease in the number of S100-positive astrocytes was more evident in the cerebral cortex. No significant change in oligodendrocyte positivity was revealed. According to the authors, these results are suggestive of astrocytic diffuse damage (especially in the cerebral cortex) due to cerebral hypoxia and/or ischemia. Calbindin-D28K (CaBP-D28k): this protein is part of the EF-hand calcium-binding protein family and is expressed in the cytoplasm of neurons in many brain regions, including the Purkinje-cells of the cerebellum (PCs). Bartschat et al analyzed the cerebellar expression of CaBP-D28k in forensic autopsy cases of acute hypoxia such as drowning or asphyxia. In this study, the immunohistochemical analysis revealed a significant reduction in the expression of CaBP-D28k in cases of acute cerebral hypoxia compared to the control groups (polytrauma, heart failure). According to the authors, the discovery of a reduction in the concentration of calbindin-D28k could support the diagnosis of acute hypoxia. HIF-1 alpha: Hypoxia-inducible factor controls the expression of the genes involved in the hypoxic response. [ , – ] HIF-1 alpha is almost absent in normoxia and promotes the expression of vascular endothelial growth factor (VEGF) in order to maintain homeostasis in hypoxic conditions. In 1 study the cerebellar expression of HIF-1 alpha was evaluated in forensic autopsy cases of acute hypoxia such as drowning or asphyxia. As a control group, cases of polytrauma and heart failure were selected. HIF-1a staining revealed weak positive immunostaining in all cases (including in control cases). Therefore, according to the authors, this marker is not deemed useful in the diagnosis of cerebral hypoxia. VEGF: the protein plays an important role in angiogenesis and vascular permeability. [ – ] VEGF expression is promoted by HIF, stimulating neovascularization in the hypoxic/ischemic brain area. [ , – ] Authors analyzed the cerebellar expression of VEGF in forensic autopsy cases of acute hypoxia such as drowning. Cases of polytrauma and heart failure were selected as a control group. According to the authors, the immunoreaction of VEGF was consistently negative. The comparison between the groups did not indicate any change in immunoreactivity in cerebellar PCs due to hypoxic events. The study proposed by Bartschat et al did not demonstrate that VEGF is induced during early responses to brain ischemia/hypoxia. Cyclooxygenase-2 (Cox-2): it is an enzyme involved in the metabolization of arachidonic acid into prostanoids. The expression of Cox-2 is present in several brain areas (especially in the hippocampus and cerebral cortex) in normal conditions [5,19d3] and can be considerably induced under certain stimuli such as ischemia. In an experimental study, Sanz et al studied the brain expression of Cox-2 in mice 6, 12, and 24 hours after the occlusion of the middle cerebral artery (MCA). According to this study, an expression of Cox-2 was detected 6 to 24 hours from ischemic insult. At 6 hours this marker was identified within the MCA territory. At 24 hours the expression was restricted to the perifocal cortical area which indicated a high level of immunoreactivity. At 6 hours, Cox-2 was identified mainly in layer II of the ipsilateral cortex and rarely in striated neurons. At 24 hours, the immunoreactivity of Cox-2 was detected in the ipsilateral peripheral areas (layer II in the cingulate frontal cortex) surrounding the ischemic area. C-fos: it is a proto-oncogene involved in numerous cellular functions. The transient activation of c-fos follows a cortical brain injury of various nature. In a previous study the expression of c-fos in mice was analyzed after 6, 12, and 24 hours from occlusion of the middle cerebral artery (MCA). According to this study, the immunoreactivity of this marker was observed after 6 hours primarily in the superficial layers of the cortex within the MCA territory. Within 24 hours, the expression of c-fos was seldom identified within the MCA territory but was very distinct in the ipsilateral undamaged cortex, primarily in the superficial layers of the cingulate frontal cortex. According to a further study, the expression of c-fos had increased in and around the ischemic lesions (autopsy case studies were limited to only 6 cases). Shock Heat shock protein 70 (HSP70): it plays a central role in cellular repair and adaptation to stress. [ – ] HSP70 protects cells from a host of stresses, including heat, hypoxia and oxidative stress. Kitamura analyzed the expression of HSP70 in autopsy cases of hypoxic/ischemic brain damage. According to this study, the immunohistochemical expression of HSP70 was found in the hippocampal regions CA2, CA3, and CA4 principally in the event of long-term survival after severe toxic or ischemic injury. In another study, Sanz et al researched the expression of HSP70 in mice after MCA occlusion. According to this study, 6 hours after the occurrence of ischemia, an immunoreactivity of HSP70 was observed after 6 hours (survival time). The intensity of the Hsp70 expression increased from 6 to 24 hours after the ischemic insult. Also at 24 hours, a strong expression of Hsp70 was identified in the neurons surrounding the ischemic area (penumbra like-zone). A further neuropathological study on forensic cases revealed an expression of HSP70 in areas CA2, CA3, and CA4 primarily in long survival time after severe hypoxic/ischemic damage. Microtubule-associated protein 2 (MAP2): microtubule-associated proteins are the largest group of cytoskeleton proteins and play an important role in neuronal morphogenesis. MAP2 is the most abundant MAP family protein in the brain. This protein increases the elongation of the microtubules and reduces the rapid shortening frequency even if simply modifying but not prohibiting their dynamic behavior. MAP2 is considered a very early marker of ischemic neuronal damage, as demonstrated by the loss of neuronal MAP2 immunoreactivity from the results of experimental studies. [ , – ] In essence, studies in rats and gerbils have revealed results of an early loss of MAP2 immunoreactivity following ischemia, but there have been a limited number of studies on the human brain, [ , , ] and none of these studies refer to a broader number of cases comprised of different causes of ischemic/hypoxic brain death. Kuhn et al studied the immunohistochemical expression of MAP2 in a forensic autopsy case of hypoxia/ischemia (including cases of drowning and hanging). In the present study, cases of the hypoxia-ischemia group revealed a reduction in MAP2 immunostaining in hippocampal areas CA2 – CA4 and in cortical layers II-VI compared to controls. The most vulnerable regions were the hippocampal area CA4 and the cortical layers III –V. According to some studies, even 10 minutes of anoxia can induce a reduction of MAP2 immunoreactivity in the hippocampus. Monoclonal antibody to neurofilament protein: it is an antibody directed against nonphosphorylated neurofilaments. It tags dendrites and the cell body of a subtype of pyramidal neurons. Leifer et al conducted an immunohistochemical study on autopsy cases of cerebral hypoxia/ischemia. According to the authors, the expression of SMI32 was predominantly reduced in cases of acute ischemia even when the Nissl stain revealed only slight pycnosis. The staining was not present in the areas of necrosis. However, this study had experienced limitations: a very limited number of cases studied and the presence of cases with neurodegenerative pathologies. Albumin: Løberg et al conducted a study of experimental animal samples (together with several autopsy cases) in order to understand if an uptake of plasma proteins occurs in damaged neurons after ischemic/hypoxic damage. Anterior brain ischemia was induced in rats by carotid clamping and hypotension for 15 minutes, followed by recirculation for 6 hours, 24 hours, 48 hours, and 5 days. According to the authors, blood-brain barrier rupture with mild albumin extravasation was revealed 6 hours after ischemic/hypoxic damage in the lateral reticular nucleus of the thalamus, in the dorsolateral striatum and watershed area of the cerebral cortex. Previously albumin expression after ischemic/hypoxic damage has been studied by Maeda et al in experimental animal samples. According to the authors and their results from optical microscopy, there was no reaction to albumin for the first 12 hours after unilateral occlusion of the common carotid artery for 10 minutes as well as reperfusion. At 12 hours, the reaction was weak and limited in the CA1 subiculum region. More significant results were obtained by the same authors through the use of electron microscopy. However, it should be considered that the expression immunohistochemistry of albumin can occur as a postmortem phenomenon, therefore its objective application in the forensic field be extremely limited and not considered very useful. For this reason, Løberg et al proposed to test fibrinogen for autopsy cases of hypoxic brain damage. Glial fibrillary acid protein (GFAP) and Vimentin: The GFAP is the main intermediate filament protein in mature astrocytes. Vimentin is part of the family of intermediate filament proteins and plays a key role in controlling microglia activation and neurotoxicity during cerebral ischemia. In the rat hippocampus after ischemia, vimentin and GFAP positive astrocytes appeared solely in the CA1 region, indicating neuronal necrosis, while GFAP positive and vimentin negative cells were observed not only in the CA1 region but in the CA3 region as well, which indicated neuronal vitality. Based on these results, vimentin could be considered a useful marker for neuronal necrosis (but not in very acute hypoxic damage). According to Loefer et al, a reduction in the expression of GFAP was identified in very acute lesions of a series of autopsy specimens while the lack of reactivity of this marker had increased in damage after a few days. This study, however, has considerable limitations for it was conducted on a very limited series of cases, and a part of them experienced neurodegenerative diseases as well. GFAP and vimentin have also been studied with autoptic cases. According to Kitamura, the proliferation of GFAP-positive and vimentin-positive cells (astrocytes and microglia) was primarily observed after hypoxic/ischemic damage with prolonged survival time. In cases that have a previous history of hypoxic damage, the author indicated a proliferation of GFAP-positive and vimentin-negative astrocytes in the CA3 and CA4 regions of the hippocampus. However, and according to this study, the immunohistochemical evaluation of GFAP cannot distinguish hypoxic/ischemic damage from postmortem alterations. In conclusion, GFAP and vimentin play a vital role if the survival time is prolonged for; they are linked to the reaction of glial cells to hypoxic/ischemic damage. Therefore, the use of these markers is considered of no use in the case of very recent hypoxic damage.
Discussion In the field of forensic neuropathology, evidence of a hypoxic/ischemic brain injury is particularly important. Even if the brain is the target organ of this type of damage, at present, there are no specific neuropathological (macroscopic and histological) findings of hypoxic damage (such as drowning, hanging, carbon monoxide poisoning) or acute ischemic. [ , , , ] According to the literature, the mechanism of death in hypoxia/ischemia is too rapid to determine the presence of vital morphological changes. Postmortem changes can alter the cerebral parenchyma rendering the assessment of brain tissue even more difficult. In the postmortem diagnosis of ischemia/brain hypoxia, immunohistochemistry could help and overcome the limitations of conventional histology. Therefore, in-depth knowledge of cellular reactions triggered by neuronal hypoxic damage is particularly important. Ischemia/hypoxia induces severe stress on nerve cells which leads to the activation of immediate early genes (such as c-fos) and their coding for thermal shock proteins (such as HSP70). [ – ] While acute ischemic neuronal injury indicates axon sparing and selective neuronal injury (due to the release of large quantities of glutamate into the extracellular space), late neuronal death is associated with anti-apoptotic growth factors and reduced expression of microtubule-associated proteins and tubulin. Immunohistochemistry studies of acute hypoxic/ischemic brain damage have often been based on this knowledge. In this perspective, the review we have proposed summarizes and considers all scientific studies relating to the immunohistochemical diagnosis of acute cerebral hypoxia/ischemia in the last 25 years or so. In every original paper analyzed, an immunohistochemical study was performed in the brain areas most susceptible to ischemic damage, specifically in the hippocampus, cerebellum and watershed area. By far, microtubule-associated protein 2 (MAP2) is the most researched biomarker and has provided the best results. In fact, one of the studies in the forensic field has revealed a reduction of MAP2 expression in cases of cerebral hypoxia (such as hanging and drowning) in which the agonic time (of survival) was a few minutes. The results of this study seem very encouraging and so the marker could be very useful in the detection of very recent cerebral hypoxia. In our opinion, further scientific confirmation studies on autopsy specimens would be needed before MAP2 is utilized in common forensic practice. The results obtained by the study by Bartschat et al also seem very interesting. The authors, in fact, discovered a significant reduction in the cerebellar expression of calbindin-D28k in a Medico-Legal case of acute cerebral hypoxia (such as drowning or generic asphyxia). The survival time of the study group was only a few minutes; therefore, the marker could also be useful in supporting the diagnosis of acute hypoxia. This study can be considered an important pioneering analysis however additional confirmatory studies would be required. With regards to a large part of the other markers analyzed (such as HIF-1 alpha, HSP70, vimentin, VEGF and GFAP) the results were very poor and their use in the detection of acute cerebral hypoxia/ischemia is not possible. A problem that emerged from the original papers studied was that the markers were often tested on a small number of autopsy cases and further confirmation studies were carried out in a limited number. As a result, the studied markers have attracted just a theoretical interest (in some cases, furthering the pathological mechanism triggered by hypoxic/ischemic brain damage) but they do not have a significant application in forensic practice. In studies on animal samples, the experimental model was inconclusive and could not be compared to any specific autopsy case. For example, during the preparation of tissue samples, the occlusion times of cerebral arteries were long (approximately 1 hour), while in human beings the damage from hypoxic phenomena related to asphyxia (in drowning, for example) continues for only several minutes. Therefore, the sample obtained through animal studies could have a different or enhanced immunohistochemical expression compared to autopsy cases. Therefore, in the absence of other confirmatory studies on autopsy specimens, these experimental studies performed on animals (although they have particular theoretical and even clinical interests) cannot serve a forensic purpose. Another particular problem in the forensic field is postmortem alteration due to autolysis and putrefaction. In this regard, the quality of immunohistochemical staining always depends on the duration of autolysis, namely, on the period of the moment of death of an individual and the fixation of the samples. Autolysis and putrefaction cause tissue alteration with the degeneration of protein structures. These processes depend on various factors of both the corpse (type of death, physical constitution) and environmental factors (temperature, humidity, ventilation). [ , , ] It is not always possible to perform an autopsy immediately after death for the discovery of a corpse can occur sometime after death and/or due to the necessity of authorization from judicial authorities to perform the autopsy. For example, due to the movements of a corpse in water, its discovery could occur within a few days of death by drowning, therefore, severely limiting an immunohistochemical investigation. However, scientific studies in the literature often analyze cases with a very limited postmortem interval, and the immunohistochemical expression of markers in autolytic or initially putrefied samples is not evaluated. Therefore, the attempt to fully understand the resistance of the markers to post-lethal alterations is difficult when it comes to positive scientific significance. To sum up, despite many promising studies, at present, it is difficult to identify reliable and confirmed biomarkers from multiple studies in order to support a postmortem diagnosis of acute cerebral hypoxia/ischemia. Without a doubt, MAP2 is the most researched marker in the literature and the results obtained have proven to be quite useful however, only a few select studies have used human brain samples. Moreover, there are significant limitations in the studies analyzed, mainly in relation to the limited number of autopsy cases studied and the alterations due to autolytic and putrefactive phenomena. However, evidence of cerebral hypoxia is often important in judicial autopsies, especially in cases of violent mechanical asphyxia. The results of this review could also be useful for judicial court. In conclusion, for the time being, proof of cerebral hypoxia is particularly difficult if the survival time is very short which usually occurs in asphyxia. In these cases, it is necessary to accurately follow the indications suggested in the literature, [ , , – ] and always consider every brain area that is most susceptible to hypoxic damage. Only a few select immunohistochemistry studies have been performed to provide support in the detection of acute cerebral ischemia/hypoxia. The results from several of the biomarkers seem promising however further confirmatory studies are strongly recommended when it comes to their application in common forensic practice.
Conceptualization: Rosario Barranco, Francesco Ventura. Data curation: Rosario Barranco, Alessandro Bonsignore. Formal analysis: Alessandro Bonsignore. Methodology: Rosario Barranco. Supervision: Francesco Ventura. Validation: Francesco Ventura.
|
Perceptual inference employs intrinsic alpha frequency to resolve perceptual ambiguity | 2d2ed64c-3550-4648-833c-e172232b11cc | 6433295 | Physiology[mh] | The brain is increasingly being understood as engaged in probabilistic unconscious perceptual inference to actively predict and explain observed sensory inputs, which helps to resolve ambiguity in sensory signals [ – ]. Therefore, our perception of the world is not simply based on our sensory inputs. Instead, what we perceive is heavily altered by contextual information [ – ] and expectations [ – ]. Besides context and expectation, intrinsic neural oscillatory signatures and organization status of the brain dramatically modulate the perceptual outcome of ambiguous stimuli [ – ]. However, it remains unclear how perceptual inference employs intrinsic brain activity to bias the perception of ambiguous sensory information towards predicted percepts with the progress of time. The process of perceptual inference can be well illustrated by the phenomenon of bistable apparent motion in the Ternus display , in which subjective perception spontaneously alternates between spatial and temporal grouping interpretations of a constant ambiguous dynamic visual scene ( ). The human brain adopts two major strategies of perceptual grouping to achieve perceptual coherence along the spatial and temporal dimension, despite the ever-changing visual inputs and the resulting fragmentary nature of the retinal image across space and time [ – ]. Spatially, grouping local visual elements into a holistic percept allows us to perceive scenes and objects as a whole rather than a meaningless collection of unconnected parts [ – ]. Temporally, successive discrete visual events unfolding in time could be grouped based on temporal proximity to perceive the stability of object identity and location [ , , ]. In the classical Ternus display ( ), two horizontally spaced disks appear at shifted locations in two successive frames. Depending on the interframe interval (IFI), observers typically report two distinct percepts . Temporal grouping is explicitly dominant at short IFIs: the central overlapping disks between the two frames are temporally integrated as one single disk, and the visual persistence of the central overlapping disk makes the lateral disk in frame 1 appear to jump across the central disk, i.e., element motion (EM) ( , upper panel; ). On the other hand, spatial grouping is explicitly dominant at long IFIs: the two disks within each frame are spatially grouped and perceived as moving together as a group, i.e., group motion (GM) ( , lower panel; ). Most critically, when the IFI reaches a certain psychophysical threshold, the Ternus display becomes ambiguous/bistable: the report of EM (temporal grouping) versus GM (spatial grouping) percepts randomly fluctuates on a trial-by-trial base, resulting in comparable proportions of GM and EM percepts ( ) . Interestingly, the typical transition IFI threshold between temporal and spatial grouping in the Ternus display occurs around a time window of approximately 100 ms (see ), which corresponds to the average cycle of occipital alpha-band oscillations, with peak frequencies ranging between 8 and 13 Hz (i.e., 70- to 120-ms cycle). The alpha oscillations, as one of the most predominant oscillations in the visual system, are considered as one underlying mechanism of perceptual cycles by gating the transient temporal windows of perception [ – ]. Accordingly, accumulating evidence shows that the phase of ongoing alpha oscillations reflects cyclic shifts of neuronal excitability [ – ] and predicts not only behavioral performance [ , – ] but also a variety of subsequent neural signals related to stimulus processing [ , , ]. Besides the phasic effects, the peak frequency of alpha-band oscillations predicts reaction times (RTs) and variations in temporal resolution of perception [ , – ]. The phasic and frequency effect of alpha oscillations lead to the long-standing hypothesis that the alpha cycle provides the discrete temporal window of perceptual grouping: whether two stimuli are integrated into a single percept or segregated into separate events depends on whether they fall in the same cycle of the alpha oscillation . In terms of the Ternus paradigm, if the two frames fall in the same alpha cycle, they will be temporally integrated, resulting in the EM percepts; if the two frames fall in different alpha cycles, they will be segregated temporally, and spatial grouping will take place separately in the two frames, resulting in the GM percepts. Especially when the sensory inputs become ambiguous at the transition IFI threshold ( ), we hypothesize that the intrinsic prestimulus alpha frequency (PAF) plays a critical role in determining whether the two frames are grouped over time or not, which accordingly affects the codependent spatial grouping process. Specifically speaking, lower PAFs (i.e., longer prestimulus alpha cycles, , upper panel) allow the two consecutively presented frames to fall in the same alpha cycle, resulting in the EM percepts, while higher PAFs (i.e., shorter prestimulus alpha cycles, , lower panel) allow the two frames to fall in different alpha cycles, resulting in the GM percepts. We thus predict that the PAF can affect the outcome of bistable perceptual grouping in the Ternus paradigm, with higher PAFs preceding the bistable GM than EM percepts. Furthermore, we hypothesize that perceptual inference employs the intrinsic PAFs to predict the perceptual outcome in the bistable Ternus display. Specifically, the brain generates predictions towards the GM percepts according to higher PAFs since the higher PAFs make it more possible for the two frames to fall in different alpha cycles. On the other hand, the brain generates predictions towards the EM percepts according to lower PAFs since the lower PAFs make it more possible for the two frames to fall within the same alpha cycle. Under the framework of perceptual inference, combining the specific prediction and forthcoming inputs, perceptual inference towards one specific percept will be made [ – ]. The perceptual inference is efficient if it is consistent with the subsequently perceived percept but inefficient if inconsistent. We thus predict that the efficiency of the perceptual inference may bias neural representations of the perceived percepts with the progress of time. In particular, the efficient perceptual inference may induce the corresponding representation pattern underlying the predicted percepts even before the actual presentation of the stimuli. Alternatively, if perception was based solely on sensory inputs, one would assume that neural representations underlying the integrated percepts are induced only after the actual presentation of the stimuli. To distinguish between the above hypotheses, we adopted electroencephalography (EEG) in healthy adults and intracranial recordings in epileptic patients and used multivariate decoding techniques on the EEG data to further probe the representational content of neural signals in a time-resolved manner.
Behavioral performance In the EEG ( n = 17), intracranial ( n = 4), and functional magnetic resonance imaging (fMRI) ( n = 18) experiments, participants were asked to report the perceived EM versus GM percepts after viewing 1) the explicit EM stimuli with the short IFI, 2) the explicit GM stimuli with the long IFI, and 3) the bistable stimuli with the transition IFI threshold ( ). The transition IFI threshold, at which equal proportions of EM and GM trials were reported, was determined individually for each participant prior to the main experiment ( ; see ). The individual IFI threshold for each participant, estimated by a psychometric function fitted to the participant’s responses at each of the seven IFIs (see details in , Figs and ), was shown in for the EEG and fMRI experiments, respectively. A two-sample t test showed no significant difference between the two experiments in terms of the group mean IFI threshold, t (33) < 1. In the two explicit conditions, the mean accuracy rates in the explicit EM and explicit GM condition were comparable and both above 85% in both the EEG, t (16) < 1, and the fMRI experiment, t (17) < 1 ( ). The accuracy rate of the explicit trials was taken as an indicator of whether a participant could indeed clearly discriminate between the two different types of percept. Our data indicated that the participants could clearly distinguish the two explicitly different percepts at the short versus long IFIs. RTs, however, were significantly slower in the explicit EM than the explicit GM condition in both the fMRI experiment ( t (17) = 3.06, p < 0.01) and the EEG experiment ( t (16) = 3.73, p < 0.005) ( ). In the bistable condition, there was no significant difference between the bistable EM and the bistable GM conditions in terms of both the rates of choice (EEG experiment: t (16) < 1, fMRI experiment: t (17) < 1) ( ) and the RTs (EEG experiment: t (16) < 1, fMRI experiment: t (17) < 1) ( ). In addition, behavioral performance of the four epileptic patients with depth electrodes showed similar patterns as the healthy participants (see ). Neurophysiology results PAF predicted the outcome of bistable perceptual grouping The EEG data showed a clear peak in the alpha-band amplitude (8–13 Hz) ( ) and a posterior scalp distribution of alpha amplitude during the prestimulus period (–800 to 0 ms relative to the presentation of the first frame) for all the participants ( ). These results guided further analysis by confining the frequency of interest to 8–13 Hz and the region of interest to the posterior electrodes (Oz, O1, O2, POz, PO1, PO2, PO3, PO4). To understand the relationship between the alpha frequency and the perceived percepts of the bistable perceptual grouping, we first calculated the between-subject correlation between the individual transition IFI thresholds and the individual prestimulus peak alpha frequency. The individual peak alpha frequency was calculated based on the maximal prestimulus posterior alpha amplitude of each participant . Subsequently, the Pearson correlation between the individual alpha frequency and the individual transition IFI threshold was calculated. The two measures were significantly correlated ( n = 17, r = –0.7187, p = 0.0012) ( ). The significant between-subject correlation suggests that the faster the alpha oscillations (i.e., higher alpha frequencies and shorter alpha cycles) in an individual, the shorter the IFI required for the two frames to fall in different alpha cycles, and thus it takes shorter IFI for the later GM percepts at the longer IFIs to take dominance over the earlier EM percepts at the shorter IFIs. On the other hand, the slower the alpha oscillations (i.e., lower alpha frequencies and longer alpha cycles) in an individual, the longer the IFI required for the two frames to fall in different alpha cycles, and thus it takes longer IFIs for the transition from the earlier EM percepts to the later GM percepts. In addition, under the hypothesis that the alpha frequency gates the time window of temporal integration, one may expect that an increase in the alpha cycle length will increase the threshold IFI by the same amount. We accordingly tested whether the slope of the fitted line between the individual alpha cycle length and the individual transition IFI threshold ( ), which represents an increased transition threshold IFI per increment of the alpha cycle, was significantly different from the hypothetical slope of 1. The results showed that the slope of the linear regression does not significantly differ from 1, t < 1, indicating that an increase in the alpha cycle was neither significantly higher nor lower than an increase in the transition threshold of the IFI. Moreover, if, as we predicted, slower alpha cycles lead to higher possibilities that the two frames will fall in the same alpha cycle (i.e., temporal grouping, , upper panel) while faster alpha cycles lead to higher possibilities that the two frames will fall in different alpha cycles (i.e., spatial grouping, , lower panel), trial-by-trial variance in the PAF within each subject should predict the outcome of the bistable perceptual grouping, with higher PAFs on the bistable GM (spatial grouping) than the bistable EM (temporal grouping) trials. To test the above prediction, we analyzed time-resolved changes in the prestimulus derivative of the phase angle time series (see ), which corresponds to the instantaneous frequency of a signal within a band-limited range . For each subject, we calculated and compared the instantaneous alpha frequency in the prestimulus window (–800 to 0 ms relative to the presentation of the first frame, and no poststimulus signal was included in the analysis) for the bistable GM and the bistable EM percept trials, respectively. Consistent with our predictions, the results showed that the PAF was significantly higher in the bistable GM than bistable EM trials, from about –550 to –210 ms relative to the presentation of the first frame, cluster-based correction ( p < 0.05) ( ). To further test the consistency of the above PAF effect and its precise anatomical origins, we collected intracranial data from four epileptic patients with depth electrodes. For each patient, according to the present frequency of interest at 8–13 Hz, we computed the alpha amplitude for each and every contact and then selected the first 10 contacts with the strongest alpha amplitude. The anatomical locations of these selected contacts were mostly located in the occipital and parietal regions ( ), which was consistent with the posterior scalp distribution of alpha in the EEG experiment ( ). We subsequently performed further analysis on the within-subject trial-by-trial variance of instantaneous PAF in the selected contacts. The instantaneous PAF of the bistable EM and GM trials was further computed for each of the 10 contacts with the strongest alpha power, using similar methods as those for the EEG data analysis. The results showed that most of the selected contacts exhibited the trend of higher PAF in the bistable GM than bistable EM trials ( ). For all four patients, in the group mean of the 10 contacts with maximal alpha amplitude, the PAF was significantly higher for the bistable GM percepts than the bistable EM percepts using cluster-based permutation test ( , patient 1: from about –500 to –110 ms relative to the presentation of frame 1; , patient 2: from about –470 to –180 ms; , patient 3: from about –600 to –250 ms; and , patient 4: from about –410 to –300 ms, cluster-based correction, p < 0.05). PAF biased poststimulus neural representation by inducing preactivation of the subsequently reported bistable percepts We subsequently investigated how the PAF biases the neural representations of the spatially versus temporally integrated percepts. The working hypothesis ( ) is that lower PAFs will result in efficient perceptual inference on the EM percepts and inefficient perceptual inference on the GM percepts. On the other hand, higher PAFs will result in efficient perceptual inference on the GM percepts and inefficient perceptual inference on the EM percepts. Accordingly, for the bistable GM trials with higher PAFs and the bistable EM trials with lower PAFs, we expected to observe biased neural representations of the predicted percepts in the neural signals not only after but also before the actual stimulus onset. To test the above hypothesis, we employed the temporal generalization method, a time-resolved decoding approach, to characterize how neural representations are dynamically transformed with the progress of time . The classifiers used in the decoding analysis rely on boundaries through the high-dimensional activation space that maximally separate patterns of neural activity underlying different percepts (i.e., bistable EM versus bistable GM). Classifier performance should be better if the two representations in the activation space are clearly separated ( ). For the EEG data, all the bistable trials were first sorted according to the PAF and then half split into the high PAF and the low PAF sessions (see details in ). Subsequently, the bistable EM and GM trials in the high PAF session were selected as the high PAF EM and GM trials, respectively ( ). Similarly, the bistable EM and GM trials in the low PAF session were selected as the low PAF EM and GM trials, respectively ( ). To exclude the potential confounds caused by the different number of trials, the trial number in each of the above four types of trials was matched (see details in ). According to our hypothesis, the high PAF GM trials and the low PAF EM trials were designated as the efficient inference condition, while the low PAF GM trials and the high PAF EM trials were designated as the inefficient inference condition ( ). For each condition, we calculated the temporal generalization matrix, which contained the decoding performance between the bistable EM and bistable GM trials over time (quantified by the area under the receiver operator characteristic, i.e., AUC, using the leave-one-out cross-validation method). If the process of perceptual integration simply depends on sensory inputs, perceptual grouping between the two frames should happen only after the actual presentation of the second frame. We thus time locked the temporal generalization matrix to the presentation of the second frame to investigate how neural representations of the spatially versus temporally integrated percepts were encoded with the progress of time relative to the presentation of frame 2. For the efficient inference condition, neural representations of the bistable EM versus GM percepts could be successfully discriminated both before and after the onset of the second frame ( , from –280 ms to 400 ms relative to the second frame, cluster-based correction, p < 0.05). For the inefficient inference condition, however, the two types of percept could be successfully discriminated only after the onset of the second frame ( , from 100 ms to 400 ms after the second frame, cluster-based correction, p < 0.05). By directly comparing the decoding performance between the efficient versus inefficient inference condition, we found significantly better decoding performance in the efficient than inefficient inference condition, not only after but also before the onset of the second frame (a significant cluster from training time –150 to 100 ms and decoding time –280 to 160 ms, cluster-based correction, p < 0.05) ( ). In particular, the neural signals 0–100 ms after the onset of frame 2 could be significantly better generalized to the neural signals 0–280 ms before the onset of frame 2 in the efficient than inefficient inference condition and vice versa (0–140 ms before frame 2 onset being generalized to 0–160 ms after frame 2). For demonstration purposes, these generalized signals were further illustrated, for example, when the decoder was trained 0–50 ms after frame 2 ( ): in the efficient inference condition, neural signals from 0 to 220 ms before frame 2 were similar to those evoked by the actual onset of frame 2. Please note here, the 0–50 ms poststimulus training time window in was just one representative time window taken from and was not the only significant time window. Instead, the significant effects extend from 0 to 100 ms after the actual presentation of frame 2 ( ). Taken together, the above decoding results suggested that the efficient perceptual inference, based on the intrinsic PAFs, improved the readout of poststimulus temporally versus spatially integrated representations by preactivating the percept-like signals even before the actual onset of frame 2. Reduced prestimulus alpha power for the bistable EM percept In addition to the alpha frequency effect, we further tested whether the prestimulus alpha power could potentially influence the bistable perception. We calculated the prestimulus occipital alpha power, relative to the onset of the first frame, for the bistable EM and bistable GM trials, respectively (see also ). Significantly lower prestimulus alpha power was found in the bistable EM than bistable GM trials, from –580 to –220 ms relative to the onset of frame 1 ( ). fMRI data: Enhanced prestimulus activity and network dynamics in the frontoparietal network for the bistable EM percept For the fMRI data, we first identified the specific neural substrates underlying the bistable spatially versus temporally integrated percepts by directly contrasting the bistable EM and GM trials. Compared to the bistable GM trials, the bistable EM trials induced stronger positive activations in the left intraparietal sulcus (IPS) ( , upper panel, red; ), and stronger deactivations in the medial prefrontal cortex (MPFC) of the default-mode network (DMN) ( , upper panel, blue; ). As shown in the mean parameter estimates extracted from the activated clusters ( , lower panel), neural activity increased in the left IPS and decreased in the MPFC, specifically in the bistable EM trials. To further localize the specific neural regions in which prestimulus neural activity predicts the outcome of bistable perceptual grouping, we compared neural activity in the trials prior to the bistable EM and GM trials (see ). Left inferior parietal cortex and bilateral inferior frontal gyrus (IFG) showed significantly enhanced prestimulus neural activity of the bistable EM trials, compared to the bistable GM trials ( , upper panel). For example, the extracted mean parameter estimates in bilateral IFG showed that prestimulus neural activity was higher for the bistable EM than the bistable GM trials but was comparable between the explicit EM and the explicit GM trials ( , lower panel; ). The left inferior parietal cortex showed similar patterns. No significant activations were revealed in the reverse contrast, i.e., bistable GM > bistable EM. These results thus suggested that enhanced neural activity in the frontoparietal network prior to the presentation of the bistable stimuli predicted the bistable EM percepts. In addition to the height of prestimulus neural activity, prestimulus network dynamics in the frontoparietal attention network may play a critical role in predicting the outcome of bistable perceptual grouping as well. To address this, we compared the patterns of functional connectivity among the frontoparietal regions prior to the presentation of the bistable EM versus GM trials. Since the left IPS exhibited specific selectivity towards the bistable EM percepts during both the pre- ( ) and the poststimulus ( ) period, we used the left IPS as the seed region to perform the network analysis, focusing on the prestimulus period. Psychophysiological interaction (PPI) analysis treated prestimulus activity in the left IPS as the physiological factor and the contrast between the two bistable percepts (bistable EM versus bistable GM) as the psychological factor. In this way, we aimed to calculate how prestimulus changes in the functional connectivity of the left IPS predict the subsequent bistable EM versus GM percepts. The results showed that prestimulus functional connectivity between the left IPS and the frontal regions was significantly enhanced for the bistable EM trials compared to the bistable GM trials ( and ). No significant activations were found in the reverse contrast, i.e., bistable GM > bistable EM. Therefore, the enhanced dynamics in the frontoparietal network prior to the presentation of the bistable stimuli predicted the subsequent subjectively perceived EM percepts.
In the EEG ( n = 17), intracranial ( n = 4), and functional magnetic resonance imaging (fMRI) ( n = 18) experiments, participants were asked to report the perceived EM versus GM percepts after viewing 1) the explicit EM stimuli with the short IFI, 2) the explicit GM stimuli with the long IFI, and 3) the bistable stimuli with the transition IFI threshold ( ). The transition IFI threshold, at which equal proportions of EM and GM trials were reported, was determined individually for each participant prior to the main experiment ( ; see ). The individual IFI threshold for each participant, estimated by a psychometric function fitted to the participant’s responses at each of the seven IFIs (see details in , Figs and ), was shown in for the EEG and fMRI experiments, respectively. A two-sample t test showed no significant difference between the two experiments in terms of the group mean IFI threshold, t (33) < 1. In the two explicit conditions, the mean accuracy rates in the explicit EM and explicit GM condition were comparable and both above 85% in both the EEG, t (16) < 1, and the fMRI experiment, t (17) < 1 ( ). The accuracy rate of the explicit trials was taken as an indicator of whether a participant could indeed clearly discriminate between the two different types of percept. Our data indicated that the participants could clearly distinguish the two explicitly different percepts at the short versus long IFIs. RTs, however, were significantly slower in the explicit EM than the explicit GM condition in both the fMRI experiment ( t (17) = 3.06, p < 0.01) and the EEG experiment ( t (16) = 3.73, p < 0.005) ( ). In the bistable condition, there was no significant difference between the bistable EM and the bistable GM conditions in terms of both the rates of choice (EEG experiment: t (16) < 1, fMRI experiment: t (17) < 1) ( ) and the RTs (EEG experiment: t (16) < 1, fMRI experiment: t (17) < 1) ( ). In addition, behavioral performance of the four epileptic patients with depth electrodes showed similar patterns as the healthy participants (see ).
PAF predicted the outcome of bistable perceptual grouping The EEG data showed a clear peak in the alpha-band amplitude (8–13 Hz) ( ) and a posterior scalp distribution of alpha amplitude during the prestimulus period (–800 to 0 ms relative to the presentation of the first frame) for all the participants ( ). These results guided further analysis by confining the frequency of interest to 8–13 Hz and the region of interest to the posterior electrodes (Oz, O1, O2, POz, PO1, PO2, PO3, PO4). To understand the relationship between the alpha frequency and the perceived percepts of the bistable perceptual grouping, we first calculated the between-subject correlation between the individual transition IFI thresholds and the individual prestimulus peak alpha frequency. The individual peak alpha frequency was calculated based on the maximal prestimulus posterior alpha amplitude of each participant . Subsequently, the Pearson correlation between the individual alpha frequency and the individual transition IFI threshold was calculated. The two measures were significantly correlated ( n = 17, r = –0.7187, p = 0.0012) ( ). The significant between-subject correlation suggests that the faster the alpha oscillations (i.e., higher alpha frequencies and shorter alpha cycles) in an individual, the shorter the IFI required for the two frames to fall in different alpha cycles, and thus it takes shorter IFI for the later GM percepts at the longer IFIs to take dominance over the earlier EM percepts at the shorter IFIs. On the other hand, the slower the alpha oscillations (i.e., lower alpha frequencies and longer alpha cycles) in an individual, the longer the IFI required for the two frames to fall in different alpha cycles, and thus it takes longer IFIs for the transition from the earlier EM percepts to the later GM percepts. In addition, under the hypothesis that the alpha frequency gates the time window of temporal integration, one may expect that an increase in the alpha cycle length will increase the threshold IFI by the same amount. We accordingly tested whether the slope of the fitted line between the individual alpha cycle length and the individual transition IFI threshold ( ), which represents an increased transition threshold IFI per increment of the alpha cycle, was significantly different from the hypothetical slope of 1. The results showed that the slope of the linear regression does not significantly differ from 1, t < 1, indicating that an increase in the alpha cycle was neither significantly higher nor lower than an increase in the transition threshold of the IFI. Moreover, if, as we predicted, slower alpha cycles lead to higher possibilities that the two frames will fall in the same alpha cycle (i.e., temporal grouping, , upper panel) while faster alpha cycles lead to higher possibilities that the two frames will fall in different alpha cycles (i.e., spatial grouping, , lower panel), trial-by-trial variance in the PAF within each subject should predict the outcome of the bistable perceptual grouping, with higher PAFs on the bistable GM (spatial grouping) than the bistable EM (temporal grouping) trials. To test the above prediction, we analyzed time-resolved changes in the prestimulus derivative of the phase angle time series (see ), which corresponds to the instantaneous frequency of a signal within a band-limited range . For each subject, we calculated and compared the instantaneous alpha frequency in the prestimulus window (–800 to 0 ms relative to the presentation of the first frame, and no poststimulus signal was included in the analysis) for the bistable GM and the bistable EM percept trials, respectively. Consistent with our predictions, the results showed that the PAF was significantly higher in the bistable GM than bistable EM trials, from about –550 to –210 ms relative to the presentation of the first frame, cluster-based correction ( p < 0.05) ( ). To further test the consistency of the above PAF effect and its precise anatomical origins, we collected intracranial data from four epileptic patients with depth electrodes. For each patient, according to the present frequency of interest at 8–13 Hz, we computed the alpha amplitude for each and every contact and then selected the first 10 contacts with the strongest alpha amplitude. The anatomical locations of these selected contacts were mostly located in the occipital and parietal regions ( ), which was consistent with the posterior scalp distribution of alpha in the EEG experiment ( ). We subsequently performed further analysis on the within-subject trial-by-trial variance of instantaneous PAF in the selected contacts. The instantaneous PAF of the bistable EM and GM trials was further computed for each of the 10 contacts with the strongest alpha power, using similar methods as those for the EEG data analysis. The results showed that most of the selected contacts exhibited the trend of higher PAF in the bistable GM than bistable EM trials ( ). For all four patients, in the group mean of the 10 contacts with maximal alpha amplitude, the PAF was significantly higher for the bistable GM percepts than the bistable EM percepts using cluster-based permutation test ( , patient 1: from about –500 to –110 ms relative to the presentation of frame 1; , patient 2: from about –470 to –180 ms; , patient 3: from about –600 to –250 ms; and , patient 4: from about –410 to –300 ms, cluster-based correction, p < 0.05). PAF biased poststimulus neural representation by inducing preactivation of the subsequently reported bistable percepts We subsequently investigated how the PAF biases the neural representations of the spatially versus temporally integrated percepts. The working hypothesis ( ) is that lower PAFs will result in efficient perceptual inference on the EM percepts and inefficient perceptual inference on the GM percepts. On the other hand, higher PAFs will result in efficient perceptual inference on the GM percepts and inefficient perceptual inference on the EM percepts. Accordingly, for the bistable GM trials with higher PAFs and the bistable EM trials with lower PAFs, we expected to observe biased neural representations of the predicted percepts in the neural signals not only after but also before the actual stimulus onset. To test the above hypothesis, we employed the temporal generalization method, a time-resolved decoding approach, to characterize how neural representations are dynamically transformed with the progress of time . The classifiers used in the decoding analysis rely on boundaries through the high-dimensional activation space that maximally separate patterns of neural activity underlying different percepts (i.e., bistable EM versus bistable GM). Classifier performance should be better if the two representations in the activation space are clearly separated ( ). For the EEG data, all the bistable trials were first sorted according to the PAF and then half split into the high PAF and the low PAF sessions (see details in ). Subsequently, the bistable EM and GM trials in the high PAF session were selected as the high PAF EM and GM trials, respectively ( ). Similarly, the bistable EM and GM trials in the low PAF session were selected as the low PAF EM and GM trials, respectively ( ). To exclude the potential confounds caused by the different number of trials, the trial number in each of the above four types of trials was matched (see details in ). According to our hypothesis, the high PAF GM trials and the low PAF EM trials were designated as the efficient inference condition, while the low PAF GM trials and the high PAF EM trials were designated as the inefficient inference condition ( ). For each condition, we calculated the temporal generalization matrix, which contained the decoding performance between the bistable EM and bistable GM trials over time (quantified by the area under the receiver operator characteristic, i.e., AUC, using the leave-one-out cross-validation method). If the process of perceptual integration simply depends on sensory inputs, perceptual grouping between the two frames should happen only after the actual presentation of the second frame. We thus time locked the temporal generalization matrix to the presentation of the second frame to investigate how neural representations of the spatially versus temporally integrated percepts were encoded with the progress of time relative to the presentation of frame 2. For the efficient inference condition, neural representations of the bistable EM versus GM percepts could be successfully discriminated both before and after the onset of the second frame ( , from –280 ms to 400 ms relative to the second frame, cluster-based correction, p < 0.05). For the inefficient inference condition, however, the two types of percept could be successfully discriminated only after the onset of the second frame ( , from 100 ms to 400 ms after the second frame, cluster-based correction, p < 0.05). By directly comparing the decoding performance between the efficient versus inefficient inference condition, we found significantly better decoding performance in the efficient than inefficient inference condition, not only after but also before the onset of the second frame (a significant cluster from training time –150 to 100 ms and decoding time –280 to 160 ms, cluster-based correction, p < 0.05) ( ). In particular, the neural signals 0–100 ms after the onset of frame 2 could be significantly better generalized to the neural signals 0–280 ms before the onset of frame 2 in the efficient than inefficient inference condition and vice versa (0–140 ms before frame 2 onset being generalized to 0–160 ms after frame 2). For demonstration purposes, these generalized signals were further illustrated, for example, when the decoder was trained 0–50 ms after frame 2 ( ): in the efficient inference condition, neural signals from 0 to 220 ms before frame 2 were similar to those evoked by the actual onset of frame 2. Please note here, the 0–50 ms poststimulus training time window in was just one representative time window taken from and was not the only significant time window. Instead, the significant effects extend from 0 to 100 ms after the actual presentation of frame 2 ( ). Taken together, the above decoding results suggested that the efficient perceptual inference, based on the intrinsic PAFs, improved the readout of poststimulus temporally versus spatially integrated representations by preactivating the percept-like signals even before the actual onset of frame 2. Reduced prestimulus alpha power for the bistable EM percept In addition to the alpha frequency effect, we further tested whether the prestimulus alpha power could potentially influence the bistable perception. We calculated the prestimulus occipital alpha power, relative to the onset of the first frame, for the bistable EM and bistable GM trials, respectively (see also ). Significantly lower prestimulus alpha power was found in the bistable EM than bistable GM trials, from –580 to –220 ms relative to the onset of frame 1 ( ). fMRI data: Enhanced prestimulus activity and network dynamics in the frontoparietal network for the bistable EM percept For the fMRI data, we first identified the specific neural substrates underlying the bistable spatially versus temporally integrated percepts by directly contrasting the bistable EM and GM trials. Compared to the bistable GM trials, the bistable EM trials induced stronger positive activations in the left intraparietal sulcus (IPS) ( , upper panel, red; ), and stronger deactivations in the medial prefrontal cortex (MPFC) of the default-mode network (DMN) ( , upper panel, blue; ). As shown in the mean parameter estimates extracted from the activated clusters ( , lower panel), neural activity increased in the left IPS and decreased in the MPFC, specifically in the bistable EM trials. To further localize the specific neural regions in which prestimulus neural activity predicts the outcome of bistable perceptual grouping, we compared neural activity in the trials prior to the bistable EM and GM trials (see ). Left inferior parietal cortex and bilateral inferior frontal gyrus (IFG) showed significantly enhanced prestimulus neural activity of the bistable EM trials, compared to the bistable GM trials ( , upper panel). For example, the extracted mean parameter estimates in bilateral IFG showed that prestimulus neural activity was higher for the bistable EM than the bistable GM trials but was comparable between the explicit EM and the explicit GM trials ( , lower panel; ). The left inferior parietal cortex showed similar patterns. No significant activations were revealed in the reverse contrast, i.e., bistable GM > bistable EM. These results thus suggested that enhanced neural activity in the frontoparietal network prior to the presentation of the bistable stimuli predicted the bistable EM percepts. In addition to the height of prestimulus neural activity, prestimulus network dynamics in the frontoparietal attention network may play a critical role in predicting the outcome of bistable perceptual grouping as well. To address this, we compared the patterns of functional connectivity among the frontoparietal regions prior to the presentation of the bistable EM versus GM trials. Since the left IPS exhibited specific selectivity towards the bistable EM percepts during both the pre- ( ) and the poststimulus ( ) period, we used the left IPS as the seed region to perform the network analysis, focusing on the prestimulus period. Psychophysiological interaction (PPI) analysis treated prestimulus activity in the left IPS as the physiological factor and the contrast between the two bistable percepts (bistable EM versus bistable GM) as the psychological factor. In this way, we aimed to calculate how prestimulus changes in the functional connectivity of the left IPS predict the subsequent bistable EM versus GM percepts. The results showed that prestimulus functional connectivity between the left IPS and the frontal regions was significantly enhanced for the bistable EM trials compared to the bistable GM trials ( and ). No significant activations were found in the reverse contrast, i.e., bistable GM > bistable EM. Therefore, the enhanced dynamics in the frontoparietal network prior to the presentation of the bistable stimuli predicted the subsequent subjectively perceived EM percepts.
The EEG data showed a clear peak in the alpha-band amplitude (8–13 Hz) ( ) and a posterior scalp distribution of alpha amplitude during the prestimulus period (–800 to 0 ms relative to the presentation of the first frame) for all the participants ( ). These results guided further analysis by confining the frequency of interest to 8–13 Hz and the region of interest to the posterior electrodes (Oz, O1, O2, POz, PO1, PO2, PO3, PO4). To understand the relationship between the alpha frequency and the perceived percepts of the bistable perceptual grouping, we first calculated the between-subject correlation between the individual transition IFI thresholds and the individual prestimulus peak alpha frequency. The individual peak alpha frequency was calculated based on the maximal prestimulus posterior alpha amplitude of each participant . Subsequently, the Pearson correlation between the individual alpha frequency and the individual transition IFI threshold was calculated. The two measures were significantly correlated ( n = 17, r = –0.7187, p = 0.0012) ( ). The significant between-subject correlation suggests that the faster the alpha oscillations (i.e., higher alpha frequencies and shorter alpha cycles) in an individual, the shorter the IFI required for the two frames to fall in different alpha cycles, and thus it takes shorter IFI for the later GM percepts at the longer IFIs to take dominance over the earlier EM percepts at the shorter IFIs. On the other hand, the slower the alpha oscillations (i.e., lower alpha frequencies and longer alpha cycles) in an individual, the longer the IFI required for the two frames to fall in different alpha cycles, and thus it takes longer IFIs for the transition from the earlier EM percepts to the later GM percepts. In addition, under the hypothesis that the alpha frequency gates the time window of temporal integration, one may expect that an increase in the alpha cycle length will increase the threshold IFI by the same amount. We accordingly tested whether the slope of the fitted line between the individual alpha cycle length and the individual transition IFI threshold ( ), which represents an increased transition threshold IFI per increment of the alpha cycle, was significantly different from the hypothetical slope of 1. The results showed that the slope of the linear regression does not significantly differ from 1, t < 1, indicating that an increase in the alpha cycle was neither significantly higher nor lower than an increase in the transition threshold of the IFI. Moreover, if, as we predicted, slower alpha cycles lead to higher possibilities that the two frames will fall in the same alpha cycle (i.e., temporal grouping, , upper panel) while faster alpha cycles lead to higher possibilities that the two frames will fall in different alpha cycles (i.e., spatial grouping, , lower panel), trial-by-trial variance in the PAF within each subject should predict the outcome of the bistable perceptual grouping, with higher PAFs on the bistable GM (spatial grouping) than the bistable EM (temporal grouping) trials. To test the above prediction, we analyzed time-resolved changes in the prestimulus derivative of the phase angle time series (see ), which corresponds to the instantaneous frequency of a signal within a band-limited range . For each subject, we calculated and compared the instantaneous alpha frequency in the prestimulus window (–800 to 0 ms relative to the presentation of the first frame, and no poststimulus signal was included in the analysis) for the bistable GM and the bistable EM percept trials, respectively. Consistent with our predictions, the results showed that the PAF was significantly higher in the bistable GM than bistable EM trials, from about –550 to –210 ms relative to the presentation of the first frame, cluster-based correction ( p < 0.05) ( ). To further test the consistency of the above PAF effect and its precise anatomical origins, we collected intracranial data from four epileptic patients with depth electrodes. For each patient, according to the present frequency of interest at 8–13 Hz, we computed the alpha amplitude for each and every contact and then selected the first 10 contacts with the strongest alpha amplitude. The anatomical locations of these selected contacts were mostly located in the occipital and parietal regions ( ), which was consistent with the posterior scalp distribution of alpha in the EEG experiment ( ). We subsequently performed further analysis on the within-subject trial-by-trial variance of instantaneous PAF in the selected contacts. The instantaneous PAF of the bistable EM and GM trials was further computed for each of the 10 contacts with the strongest alpha power, using similar methods as those for the EEG data analysis. The results showed that most of the selected contacts exhibited the trend of higher PAF in the bistable GM than bistable EM trials ( ). For all four patients, in the group mean of the 10 contacts with maximal alpha amplitude, the PAF was significantly higher for the bistable GM percepts than the bistable EM percepts using cluster-based permutation test ( , patient 1: from about –500 to –110 ms relative to the presentation of frame 1; , patient 2: from about –470 to –180 ms; , patient 3: from about –600 to –250 ms; and , patient 4: from about –410 to –300 ms, cluster-based correction, p < 0.05).
We subsequently investigated how the PAF biases the neural representations of the spatially versus temporally integrated percepts. The working hypothesis ( ) is that lower PAFs will result in efficient perceptual inference on the EM percepts and inefficient perceptual inference on the GM percepts. On the other hand, higher PAFs will result in efficient perceptual inference on the GM percepts and inefficient perceptual inference on the EM percepts. Accordingly, for the bistable GM trials with higher PAFs and the bistable EM trials with lower PAFs, we expected to observe biased neural representations of the predicted percepts in the neural signals not only after but also before the actual stimulus onset. To test the above hypothesis, we employed the temporal generalization method, a time-resolved decoding approach, to characterize how neural representations are dynamically transformed with the progress of time . The classifiers used in the decoding analysis rely on boundaries through the high-dimensional activation space that maximally separate patterns of neural activity underlying different percepts (i.e., bistable EM versus bistable GM). Classifier performance should be better if the two representations in the activation space are clearly separated ( ). For the EEG data, all the bistable trials were first sorted according to the PAF and then half split into the high PAF and the low PAF sessions (see details in ). Subsequently, the bistable EM and GM trials in the high PAF session were selected as the high PAF EM and GM trials, respectively ( ). Similarly, the bistable EM and GM trials in the low PAF session were selected as the low PAF EM and GM trials, respectively ( ). To exclude the potential confounds caused by the different number of trials, the trial number in each of the above four types of trials was matched (see details in ). According to our hypothesis, the high PAF GM trials and the low PAF EM trials were designated as the efficient inference condition, while the low PAF GM trials and the high PAF EM trials were designated as the inefficient inference condition ( ). For each condition, we calculated the temporal generalization matrix, which contained the decoding performance between the bistable EM and bistable GM trials over time (quantified by the area under the receiver operator characteristic, i.e., AUC, using the leave-one-out cross-validation method). If the process of perceptual integration simply depends on sensory inputs, perceptual grouping between the two frames should happen only after the actual presentation of the second frame. We thus time locked the temporal generalization matrix to the presentation of the second frame to investigate how neural representations of the spatially versus temporally integrated percepts were encoded with the progress of time relative to the presentation of frame 2. For the efficient inference condition, neural representations of the bistable EM versus GM percepts could be successfully discriminated both before and after the onset of the second frame ( , from –280 ms to 400 ms relative to the second frame, cluster-based correction, p < 0.05). For the inefficient inference condition, however, the two types of percept could be successfully discriminated only after the onset of the second frame ( , from 100 ms to 400 ms after the second frame, cluster-based correction, p < 0.05). By directly comparing the decoding performance between the efficient versus inefficient inference condition, we found significantly better decoding performance in the efficient than inefficient inference condition, not only after but also before the onset of the second frame (a significant cluster from training time –150 to 100 ms and decoding time –280 to 160 ms, cluster-based correction, p < 0.05) ( ). In particular, the neural signals 0–100 ms after the onset of frame 2 could be significantly better generalized to the neural signals 0–280 ms before the onset of frame 2 in the efficient than inefficient inference condition and vice versa (0–140 ms before frame 2 onset being generalized to 0–160 ms after frame 2). For demonstration purposes, these generalized signals were further illustrated, for example, when the decoder was trained 0–50 ms after frame 2 ( ): in the efficient inference condition, neural signals from 0 to 220 ms before frame 2 were similar to those evoked by the actual onset of frame 2. Please note here, the 0–50 ms poststimulus training time window in was just one representative time window taken from and was not the only significant time window. Instead, the significant effects extend from 0 to 100 ms after the actual presentation of frame 2 ( ). Taken together, the above decoding results suggested that the efficient perceptual inference, based on the intrinsic PAFs, improved the readout of poststimulus temporally versus spatially integrated representations by preactivating the percept-like signals even before the actual onset of frame 2.
In addition to the alpha frequency effect, we further tested whether the prestimulus alpha power could potentially influence the bistable perception. We calculated the prestimulus occipital alpha power, relative to the onset of the first frame, for the bistable EM and bistable GM trials, respectively (see also ). Significantly lower prestimulus alpha power was found in the bistable EM than bistable GM trials, from –580 to –220 ms relative to the onset of frame 1 ( ).
For the fMRI data, we first identified the specific neural substrates underlying the bistable spatially versus temporally integrated percepts by directly contrasting the bistable EM and GM trials. Compared to the bistable GM trials, the bistable EM trials induced stronger positive activations in the left intraparietal sulcus (IPS) ( , upper panel, red; ), and stronger deactivations in the medial prefrontal cortex (MPFC) of the default-mode network (DMN) ( , upper panel, blue; ). As shown in the mean parameter estimates extracted from the activated clusters ( , lower panel), neural activity increased in the left IPS and decreased in the MPFC, specifically in the bistable EM trials. To further localize the specific neural regions in which prestimulus neural activity predicts the outcome of bistable perceptual grouping, we compared neural activity in the trials prior to the bistable EM and GM trials (see ). Left inferior parietal cortex and bilateral inferior frontal gyrus (IFG) showed significantly enhanced prestimulus neural activity of the bistable EM trials, compared to the bistable GM trials ( , upper panel). For example, the extracted mean parameter estimates in bilateral IFG showed that prestimulus neural activity was higher for the bistable EM than the bistable GM trials but was comparable between the explicit EM and the explicit GM trials ( , lower panel; ). The left inferior parietal cortex showed similar patterns. No significant activations were revealed in the reverse contrast, i.e., bistable GM > bistable EM. These results thus suggested that enhanced neural activity in the frontoparietal network prior to the presentation of the bistable stimuli predicted the bistable EM percepts. In addition to the height of prestimulus neural activity, prestimulus network dynamics in the frontoparietal attention network may play a critical role in predicting the outcome of bistable perceptual grouping as well. To address this, we compared the patterns of functional connectivity among the frontoparietal regions prior to the presentation of the bistable EM versus GM trials. Since the left IPS exhibited specific selectivity towards the bistable EM percepts during both the pre- ( ) and the poststimulus ( ) period, we used the left IPS as the seed region to perform the network analysis, focusing on the prestimulus period. Psychophysiological interaction (PPI) analysis treated prestimulus activity in the left IPS as the physiological factor and the contrast between the two bistable percepts (bistable EM versus bistable GM) as the psychological factor. In this way, we aimed to calculate how prestimulus changes in the functional connectivity of the left IPS predict the subsequent bistable EM versus GM percepts. The results showed that prestimulus functional connectivity between the left IPS and the frontal regions was significantly enhanced for the bistable EM trials compared to the bistable GM trials ( and ). No significant activations were found in the reverse contrast, i.e., bistable GM > bistable EM. Therefore, the enhanced dynamics in the frontoparietal network prior to the presentation of the bistable stimuli predicted the subsequent subjectively perceived EM percepts.
By using EEG, intracranial recordings, and fMRI, we investigated how the frequency of alpha-band oscillations acts as the critical neural dynamics that accommodate the temporal and spatial grouping during ambiguous perception in the Ternus paradigm and, more importantly, how the brain makes predictions, based on intrinsic alpha frequency, to resolve perceptual ambiguity. At the behavioral level, comparable task performance/judgment difficulty was revealed between the bistable temporal and spatial grouping condition ( ). Therefore, any neuronal difference between the two bistable conditions cannot be attributed to differences in judgment difficulty. At the neural level, both within and between subjects, peak prestimulus frequency of alpha oscillations in the occipitoparietal regions predicted the bistable temporal versus spatial grouping (Figs and ). Moreover, efficient perceptual inference, based on spontaneous variance in the intrinsic PAFs, induced a representation of the subsequently reported bistable percept in the neural signals before the actual appearance of the second frame, indicating a preactivation of the subjectively perceived bistable percepts ( ). Based on the above observations, we propose that the alpha frequency gates the time window for perceptual grouping and perceptual inference based on intrinsic alpha frequency biased poststimulus neural representations by inducing preactivation of the predicted percepts. In addition, the reduced prestimulus alpha power ( ), together with enhanced prestimulus blood-oxygen-level–dependent (BOLD) activity and network dynamics in the frontoparietal network ( ), predicted bistable EM rather than GM percepts. It has been proposed that perception is discrete and cyclic in a manner of perceptual cycles [ , , , , ]. Accumulating recent evidence showed that perceptual performance depends on the frequency of the critical rhythm at around the onset time of stimuli [ , , ]. A higher frequency of the brain oscillations should be equivalent to a faster frame rate of discrete perception and vice versa. Accordingly, lower alpha frequency was reported to be associated with poorer temporal resolution , as if the slower frame rate of perception made two successive flashes more likely to fall within the same frame and thus be perceived as one . In the present Ternus paradigm, the intrinsic alpha frequency determines whether the two frames are integrated over time (i.e., EM) or not (i.e., GM). Specifically, when the alpha frequency is relatively slow (i.e., longer alpha cycles) to cover both spatially and temporally segregated information segments, temporal grouping between the frames dominates over spatial grouping, resulting in the EM percepts ( , upper panel; ). On the other hand, when the alpha frequency is relatively high (i.e., shorter alpha cycles) to cover only spatially segregated information segments, spatial grouping with the frames dominates, resulting in the GM percepts ( , lower panel; ). The intracranial data further confirmed the above alpha frequency effect in distributed visual areas including both the dorsal and ventral visual stream, such as the primary and secondary visual areas in the lingual gyrus, higher-order areas in the fusiform gyrus, the lateral occipital cortex (LOC), the middle temporal gyrus (MT), and the IPS ( ). The dorsal occipitoparietal areas, such as the inferior IPS, have been associated with perceptual integration of multiple elements and object representations . The ventral visual areas, such as LOC, have been found to be involved in object recognition . Moreover, it has been well documented that the MT area is highly responsive to visual motion and codes highly specialized representations of visual information [ – ], which is putative for generating apparent motion percepts . The present results further suggest that the alpha frequency effect is a ubiquitous property of the visual system, which is involved in representing coherent object motion percepts. It has been revealed that neural oscillations could create temporal windows that favor the communication between neurons . The common alpha frequency effect in distributed visual systems may drive the communication between neuronal groups in these areas to effectively encode and organize the dynamic visual inputs and induce coherent apparent motion percepts. Please note, for both the present EEG and the intracranial results, a very small within-subject variance in the PAF (about 0.1 Hz in the EEG data, , and about 0.2 Hz in the intracranial data, ) was associated with qualitatively different perceptual percepts. Based on our hypothesis, at the within-subject level, the most critical factor that causes different perceptual outcomes is the perceptual inference built through the intrinsic alpha frequency, but not the absolute alpha frequency per se. It has been suggested before that the perceptual inference is very sensitive to subtle changes in the intrinsic brain states . Therefore, in the present study, a slightly lower alpha frequency could be enough to induce a perceptual inference towards the EM percept, while a slightly higher alpha frequency could be enough to induce an inference towards the GM percept. In addition, the small within-subject frequency effect is consistent with previous studies showing small frequency modulations . Technically speaking, this small effect might result from the fact that the alpha frequency data derived from the EEG and intracranial signals reflect the summed activity of both the task-relevant and the task-irrelevant neuronal populations. Therefore, the observed effect could be attenuated by the noises from the task-irrelevant neuronal populations . The generative models of perceptions commonly consider the brain as an unconscious inference machine that uses hidden states to predict observed sensory inputs . Although there have been some detailed theories on the neural basis of the underlying computations of this system , it still lacks direct empirical evidence about how the brain uses its intrinsic states to build up specific prediction signals for perceptual inference. In the present study, even with the sensory inputs (the two frames with a threshold IFI) being kept constant in the bistable condition, the subjective perception varies between the EM and GM percepts on a trial-by-trial base, thus suggesting fluctuations in prior predictions. Please note, the definition of prediction in the present study stands for “internal model’s prediction” under the general perceptual inference framework , which is different from the term of “top-down prediction” manipulated in the field of cognitive neuroscience and psychology . The former one represents the priors in the Bayesian framework and includes any factors that can provide prior information : the perceptual prediction based on intrinsic PAFs in the present study is an example of this type of prediction. On the other hand, the latter term of “top-down prediction” is associated with the top-down control mechanisms in the higher-order brain areas, which is not directly supported by the data in the present study. The present Ternus display puts the brain under the explicit contextual information that there are two possible apparent motion percepts, i.e., GM versus EM. Moreover, the alpha peak frequency, which provides the critical window for perceptual integration, is widely considered as one putative marker of an individual’s intrinsic state . Therefore, perception is able to employ the current intrinsic alpha frequency to build prior probability of predictions about the most possibly perceived apparent motion percepts ( , left panel). The perceptual inference is efficient if it is consistent with the perceptual outcome, i.e., in the high PAF GM trials and the low PAF EM trials; the inference is inefficient if it is inconsistent with the perceptual outcome, i.e., in the high PAF EM trials and the low PAF GM trials ( , right panel). Our results showed that the peak alpha frequency not only predicted the outcome of bistable perceptual grouping (Figs , and ) but also modulated the fidelity of neural representations of the integrated percepts ( ). Compared to the inefficient inference, neural representations of the bistable EM versus GM percepts could be more robustly decoded under the efficient inference ( , ), suggesting that the efficient inference based on intrinsic PAFs enhanced the fidelity of neural representations of the predicted percepts. More interestingly, under the efficient inference, the neural signals evoked by the actual presentation of the second frame could be readily read out from the neural signals even before the presentation of the second frame ( ), suggesting a preactivation of the predicted percepts. These results thus fundamentally advance our mechanistic understanding on how the alpha frequency builds up specific prediction signals for perceptual inference: perceptual predictions on the spatially versus temporally integrated percepts are generated based on variation in the intrinsic PAFs, which induces preactivated neural representations that resemble the neural representations evoked by the actual stimuli. Please note, since time 0 in the decoding analysis was relative to the presentation of frame 2, the significant 0–100 ms poststimulus time window of the decoding analysis ( ) corresponds to about 100 (threshold IFI)–200 ms (threshold IFI + 100 ms) relative to the actual presentation of frame 1. It is thus possible that, about 150–200 ms after the presentation of frame 1, the participants have already generated conscious perceptual experiences of the predicted percepts under the modulation of perceptual prediction . However, since the whole significant poststimulus time window still involves relatively early processing phases around 100–150 ms after the presentation of frame 1, one alternative interpretation is that the present relatively early poststimulus time window might reflect early neural mechanisms such as the iconic memory . Since the explicit EM percepts are observed at the short IFI while the explicit GM percepts are observed at the long IFI (Figs and ), and since the bistable EM percepts involve higher alpha frequency than the bistable GM percepts ( ), it is possible that shorter time frames and accordingly faster temporal processing are involved in the EM percepts, which may demand more efficient communication of information through the brain. It has been correspondingly suggested that attention facilitates fast temporal processing [ – ]. Therefore, one hypothesis is that the bistable EM percepts may require more frontoparietal attentional network involvement than the bistable GM percepts. Alternatively, in contrast to EM, which is more a temporal matching, GM is more a gestalt/global matching of objects, which ignores retinotopic correspondence in favor of object-based grouping [ , , ]. Such high-level, nonretinotopic, gestalt grouping of GM might be expected to require more frontoparietal involvement as opposed to the occipital regions, which might be sufficient for short-lived, retinotopically organized grouping . Our fMRI results provide supporting evidence to the former hypothesis: both increased prestimulus neural activity ( ) and increased prestimulus network dynamics ( ) in the frontoparietal network predicted the subsequent bistable EM (temporal grouping), rather than GM (spatial grouping), percepts. Moreover, the enhanced frontoparietal activations and DMN deactivations during the bistable EM trials ( ) indicated that bistable temporal grouping was more attention-demanding than bistable spatial grouping . Consistent with the fMRI results, the prestimulus alpha power results ( ) also supported this conclusion by showing a lower prestimulus alpha power in the bistable EM than GM trials. Since it has been well documented that alpha power is an effective indicator of the level of attention engaged in a certain cognitive task (the higher the alpha power, the lower the level of attention) [ , – ], the lower prestimulus alpha power during the bistable EM trials indicated higher level of attention. Therefore, the fMRI results, together with the prestimulus alpha power results, suggested that temporal grouping is more attention-demanding than spatial grouping in the Ternus paradigm. To further understand the more general role of alpha frequency in perceptual grouping across both space and time rather than just see the specific effect with a single variant of the Ternus paradigm, future experiments with paradigms examining perceptual grouping at different levels of complexity and with regard to different visual attributes are still needed. In terms of the Ternus paradigm per se, the present study only focused on the temporal window of perceptual integration and the effect of perceptual inference, while there are other possible interpretations of this specific illusion, such as alternations between object versus group processing , between the use of top-down predictions (for example, trial history) , and between the use of different reference frames [ – ]. Moreover, other frequency-band oscillatory activities, such as the theta-band oscillations that have been suggested to be implicated in the perception of apparent motion and temporal integration , might be involved in the present phenomenon as well, but the current research methods may not be sufficient enough to detect these significant theta effects. For example, in the intracranial experiment, since most of the implanted electrodes of the four patients were in the posterior brain regions with few electrodes in the higher-order areas, it remains unknown whether there is a significant theta effect in the higher brain areas, such as the frontal cortex . To summarize, by adopting a Ternus display in which subjective perception fluctuates between temporally versus spatially integrated percepts, we showed that the occipitoparietal alpha frequency defines a temporal window for perceptual integration. Moreover, in the situation of efficient perceptual inference, neural representations of the predicted percepts based on the alpha frequency were preactivated before the actual presentation of the critical stimuli. Therefore, perceptual inference employs PAF-induced predictions to resolve perceptual ambiguity.
Ethics statement All the participants gave their informed consent prior to the experiment in accordance with the Declaration of Helsinki. The fMRI, the EEG, and the patient experiments were all approved by the Ethics Committee of School of Psychology, South China Normal University (06202015_TernusCQ). The placement of the depth electrodes was based solely on the clinical needs for the treatment of the patients and was thus independent of the purpose of the present study. This study did not add any invasive procedure to the intracranial recordings. All the participants were at least 18 years old and gave their written informed consent prior to the experiments. Participants Nineteen adult participants (12 females, mean age of 19.6 years old) took part in the EEG experiment. Another group of 20 adult participants (12 females, mean age of 23.4 years old) took part in the fMRI experiment. Two participants in the EEG experiment were discarded because of excessive eye movement artifacts. One participant in the fMRI experiment was discarded because of low accuracy (less than 70%) in the explicit conditions, and another participant was discarded because of the excessive head movements during the scanning. Therefore, 17 participants in the EEG experiment and 18 participants in the fMRI experiment were included for further analysis. Additionally, four adult patients (two males, mean age of 24 years old) undergoing intracranial recordings with stereotactically implanted multilead electrodes (Guangdong Sanjiu Brain Hospital, China) for epilepsy treatment participated in the present study. Although the anatomical locations of the electrodes were different in each patient, we included the patients whose electrodes were implanted in the occipital and parietal regions. Patients who had destructive lesions such as tumor or encephalomalacia were excluded. All the participants were right-handed, with normal or corrected-to-normal visual acuity. Stimuli Visual stimuli consisted of two consecutively presented frames of stimuli (frame 1 and frame 2), and each frame was presented for 30 ms ( ). There was a blank period between the two frames, i.e., the IFI. The IFI could be either explicitly short at 50 ms or explicitly long at 230 ms or at the transition threshold, which was specific for each subject based on pre-experiment psychophysics. Each frame contained two horizontally arranged black disks (1.6° of visual angle in diameter) on a gray background. The center-to-center spatial distance between the two disks was 3° of visual angle. The two frames shared one common disk location at the center of the display. The location of the lateral disk of the first frame, either on the left or the right side of the shared central disk, was always opposite to the lateral disk of the second frame ( ). Specifically speaking, frame 1 with left and central disks and frame 2 with right and central disks induced rightward apparent motion; frame 1 with right and central disks and frame 2 with left and central disks induced leftward apparent motion. The same set of stimulus parameters was adopted for the fMRI, the EEG, and the patient experiments. Depending on the IFI and participants’ online judgments in the bistable trials, there were four types of experimental trials: 1) the explicit EM trials (“Explicit EM”) with the short IFI of 50 ms; 2) the explicit GM trials (“Explicit GM”), with the long IFI of 230 ms; 3) the bistable trials with the threshold IFI, which were judged by the participants as the EM trials (“Bistable EM”); and 4) the bistable trials with the threshold IFI, which were judged by the participants as the GM trials (“Bistable GM”). Psychophysical procedures To specify the 50% threshold of IFI for the bistable condition for each individual subject, we asked each participant to perform a psychophysical pretest before the main experiment. Prior to the psychophysics test, participants were shown demos of the explicit EM and GM conditions and performed a practice block with only explicit EM and GM trials until the accuracy reached no less than 95%. During the formal psychophysics test, the first frame was presented for 30 ms. After a variable IFI (seven levels: 50, 80, 110, 140, 170, 200, or 230 ms), the second frame was presented for 30 ms as well. Participants were asked to perform a two-alternative forced choice (2AFC) task in which they had to choose between the EM and the GM percept. For each IFI condition, the percentage of GM reports (i.e., “1 –percentage of EM reports”) was collapsed over the leftward and rightward motion directions. The seven data points (one for each IFI) were fitted into a psychometric curve using a logistic function . The transition IFI threshold, i.e., the point at which EM and GM were reported with equal possibility, was calculated by estimating the 50% performance point on the fitted logistic function for each participant . The individual transition threshold derived from the psychophysics test was then used as the IFI in the bistable trials of the subsequent main experiment. Differently from the EEG and fMRI experiment, in the intracranial experiment, an adaptive staircase procedure was adopted to find the individual IFI threshold at which 50% of the stimuli were perceived as GM. Main experiment procedures Participants were instructed to fixate at a central fixation throughout the experiment without moving their eyes. The experimental task was to discriminate the two types of motion by pressing two prespecified buttons on the response pad using the thumb of each hand, respectively. The mapping between the two response buttons and the two types of apparent motion percept was counterbalanced between participants. In each trial, the first frame was presented for 30 ms, and after a variable IFI (50 ms, 230 ms, or the individual IFI threshold), the second frame was presented for another 30 ms. The fMRI experiment consisted of 440 trials in total, including 80 explicit EM trials, 80 explicit GM trials, 160 bistable trials, and 120 null trials. The null trials, in which only the central fixation cross was presented, were used as the implicit baseline. The participants were asked to rest for a short period of time (11 s, i.e., five repetition times [TRs]) after every 6 minutes’ task performance, which made three short periods of rest in total. During the three short rest periods, the scanner kept running, and a visual instruction “rest” was presented on the center of the screen throughout. One TR after the disappearance of the “rest” instruction, the behavioral task resumed. The EEG experiment consisted of four blocks, and each block included 40 explicit EM trials, 40 explicit GM trials, and 80 bistable trials, which were intermixed randomly, resulting in 640 experimental trials in total. A rest break was allowed between blocks. For the fMRI and EEG experiment, each trial was followed by a time interval that was selected randomly among 2,000, 2,250, 2,500, 2,750, and 3,000 ms. In the intracranial experiment, there were four blocks of 80 trials (320 trials in total), 10% of which were explicit EM and GM trials. The intertrial interval varied randomly between 1.5 and 2.5 s. In all the three experiments, the temporal order of all the trials was randomized for each participant individually to avoid potential problems of unbalanced transition probabilities. All participants completed a training section of 5 min before the recording. Recording and preprocessing of the EEG data EEGs were continuously recorded from 64 Ag/AgCl electrodes (10–20 System) with BrainAmp DC amplifiers (low-pass = 100 Hz, high-pass = 0.01 Hz, and sampling frequency = 500 Hz). The vertical electro-oculogram was recorded by one electrode under the participants’ left eyes. All the electrode impedances were kept below 5 kΩ. Signals were referenced online to the unilateral mastoid. Offline processing and analysis were performed using EEGLAB and customized scripts in MATLAB (The MathWorks, Natick, MA, USA). Data were down-sampled to 160 Hz, rereferenced to the average reference, epoched from –800 ms before the first frame to 1,000 ms after the first frame for the subsequent alpha frequency analysis, and re-epoched from –500 ms to 500 ms relative to the presentation of the second frame for the decoding analysis. Trials containing visually identified eye movements or muscle artifacts were excluded manually. Visually identified noisy electrodes were spherically interpolated. Acquisition and preprocessing of the intracranial data Ten to 13 semirigid, multilead electrodes were stereotactically implanted in the four participants, respectively. All the electrodes have a diameter of 0.8 mm and contain 10–16 2-mm–wide and 1.5-mm–apart contacts. The precise anatomical location of each contact was identified by coregistering each participant’s postimplantation CT with the preimplantation 3D T1 image, using rigid affine transformations derived from FSL’s FLIRT algorithm . Intracranial recordings were conducted using commercial video–intracranial monitoring system. The data were bandpass filtered online from 0.1 to 300 Hz and sampled at 1,000 Hz, using a reference contact located in the white matter. For the offline analysis, recording signals were down-sampled to 500 Hz. Contacts in the epileptogenic zones were excluded from further analyses. Each contact was rereferenced with respect to its direct neighbor, i.e., bipolar montage, to achieve high local specificity by removing effects of distant sources that spread equally to adjacent sites through volume conduction. All the data were epoched from –800 to 1,000 ms relative to the presentation of the first frame. Acquisition and preprocessing of the fMRI data A Siemens 3T Trio system with a standard head coil at Beijing MRI Center for Brain Research was utilized to obtain T2*-weighted echo-planar images (EPIs) with blood oxygenation level-dependent contrast. The matrix size was 64 × 64 mm 3 , and the voxel size was 3.4 × 3.4 × 3 mm 3 . Thirty-six transversal slices of 3-mm thickness that covered the whole brain were acquired sequentially with a 0.75-mm gap (TR = 2.2 s, TE = 30 ms, FOV = 220 mm, flip angle = 90°). There was a single run of functional scanning, including 524 EPI volumes. The first five volumes were discarded to allow for T1 equilibration effects. Data were preprocessed with Statistical Parametric Mapping software SPM12 (Wellcome Department of Imaging Neuroscience, London, UK; http://www.fil.ion.ucl.ac.uk ). Images were realigned to the first volume to correct for interscan head movements. The mean EPI of each participant was then computed and spatially normalized to the MNI single-participant template using the “unified segmentation” function in SPM12. This algorithm is built on a probabilistic framework that enables image registration, tissue classification, and bias correction to be combined within the same generative model. The resulting parameters of a discrete cosine transform, which define the deformation field necessary to move individual data into the space of the MNI tissue probability maps, were then combined with the deformation field transforming between the latter and the MNI single participant template. The ensuing deformation was subsequently applied to individual EPI volumes. All images were thus transformed into standard MNI space and resampled to 2 × 2 × 2 mm 3 voxel size. The data were then smoothed with a Gaussian kernel of 8-mm full-width half-maximum to accommodate interparticipant anatomical variability. Data were high-pass filtered at 1/128 Hz and analyzed with a general linear model (GLM) as implemented in SPM12. Temporal autocorrelation was modeled using an AR (1) process. Analysis of the behavioral data For the behavioral data in the EEG, intracranial, and fMRI experiment, omissions and trials with RTs 3 standard deviations (SDs) away from the mean RT in each condition were first excluded from further analysis. For the calculation of accuracy rates in the two explicit conditions, the explicit trials at the short IFI with a judgment of GM and the explicit trials at the long IFI with a judgment of EM were considered as incorrect trials, which were discarded and excluded from further analysis. For both the EEG and fMRI experiment, paired t tests were performed to test the difference in the accuracy rates between the two types of explicit trials, the proportions of EM and GM trials in the bistable condition, and the mean RTs for the two explicit and the two bistable conditions, respectively. Alpha oscillation analysis of the EEG data For all the electrodes in all the participants, a power spectrum (from 5 Hz to 30 Hz) was obtained through a Fast Fourier Transform (FFT) of all the trials (from –800 to 0 ms relative to the presentation of the first frame). An amplitude topographic map of the most prominent frequency band in the power spectrum was obtained. For each participant, the individual peak alpha frequency was determined as the value corresponding to the maximum peak frequency from the 800 ms of data prior to the presentation of the first frame within the 8–13 Hz range for the selected posterior electrodes. The Pearson product–moment correlation between the individual alpha frequency and the individual IFI threshold obtained from the psychophysical procedures was then calculated. Instantaneous PAF and alpha power were analyzed using the methods and code developed by Cohen . We chose only one electrode, which showed the strongest alpha amplitude among all the occipital electrodes in the posterior ROI for each participant, to calculate the PAF and the alpha power . To avoid contaminations by the poststimulus signals, only the prestimulus period (from –800 to 0 ms) of the EEG signals were extracted, and all the poststimulus period signals (starting from 0 ms) were excluded. Furthermore, to avoid edge artifacts at the stimulus onset due to filtering, the prestimulus signals of each bistable trial were copied, flipped from left to right, and appended to the right side of the original data. These epochs were filtered between 8 and 13 Hz with a zero-phase, plateau-shaped bandpass filter with 15% transition zones. Phase angle and amplitude time series were extracted from the filtered data with a Hilbert transform. The alpha power was obtained by calculating the square of the amplitude. For the frequency calculation, the temporal derivative of the phase angle time series describes how phase changes over time and thus corresponds to the instantaneous frequency in Hz (when scaled by the sampling rate and 2 π ). Since noises in the phase angle time series can cause sharp, nonphysiological responses in the derivative, the instantaneous frequency was filtered with a median filter with an order of 10 and a maximum window size of 400 ms: data were median filtered ten times with 10 time windows ranging from 10 to 400 ms prior to averaging across trials. Since this analysis considers changes only in the instantaneous phase of the data, it is mathematically independent from the amplitude of the oscillation, except where amplitude is equal to zero and phase is undefined. Subsequently, the instantaneous PAFs were averaged across bistable EM and GM trials, respectively. Decoding analysis of the EEG data Multivariate decoding techniques were further adopted to investigate how the PAF affects the representation contents of the bistable EM and GM percepts with the progress of time. For each participant, we first calculated the instantaneous alpha frequency for each time point in the prestimulus window (from –800 ms to 0 relative to the onset of frame 2) of each bistable trial, based on the one chosen electrode with the maximal alpha amplitude. Subsequently, statistical tests (paired t test) between the bistable EM and bistable GM conditions were performed at the group level. The significant time points were further selected as the time points of interest, and the PAF for each trial was determined by averaging the instantaneous alpha frequency across these significant time points (–570 to –350 ms relative to the presentation of frame 2; see ). Subsequently, amplitude data epochs of all the bistable trials (–400 to 400 ms relative to the presentation of frame 2), right after the preprocessing steps and without any further processing steps (no spectral analysis applied), were sorted according to the calculated PAF of each trial and half split into the high PAF and the low PAF trial sessions. The bistable GM trials in the high PAF session and the bistable EM trials in the low PAF session were selected as the two types of trials in the efficient inference condition; the bistable EM trials in the high PAF session and the bistable GM trials in the low PAF session were selected as the two types of trials in the inefficient inference condition. Please note, the PAF of each bistable trial was used only as an indicator to categorize the bistable trials into the efficient versus inefficient conditions in a post hoc way but was never used as the actual data fed into the subsequent decoding analysis. To exclude potential confounds caused by different number of trials upon comparing different conditions, we matched the trial count in the above four types of trials by randomly selecting a subsample of trials from the conditions with more trials. We then applied a multivariate linear discriminant analysis to characterize the temporal dynamics that discriminated between the subjectively perceived bistable EM versus GM percepts for the efficient and inefficient inference condition, respectively. Classifications were based on the regularized linear discriminant analysis to identify a projection in the multidimensional EEG data, x , that maximally discriminated between the two representations across all stimulus levels. Each projection is defined by a weight vector, w , which describes a one-dimensional projection y of the EEG data y = ∑ i w i x i + c , with i summing over all channels and c a constant. The regularization parameter was optimized in preliminary tests and kept fixed for all the analyses. The decoding analysis was performed in a time-resolved manner by applying it to each time point sequentially, resulting in an array of classifiers, for example, w (t1), w (t2), w (t3) and so on. To improve the signal-to-noise ratio, the data were first averaged within a time window of 50 ms centered around the time point of interest. This process could introduce some contaminations from the poststimulus signals to the prestimulus signals around the stimuli onset: the signal at time 0 contains the information within –25 to 25 ms. However, the abovementioned contaminations can influence the prestimulus signals about 25 ms at most. Subsequently, the classifier performance was assessed not only at the time point used for training (for example, classifier w (t1) was tested at t1, w (t2) was tested at t2, and so on) but also on data from all the other time points (for example, classifier w (t1) was tested on all the time points t1, t2, t3, and so on). The performance of the classifier was quantified using the receiver operator characteristic (ROC), based on leave-one-out cross-validation within each participant. The above procedure resulted in a (training time) × (decoding time) temporal generalization matrix per condition. Alpha frequency analysis of the intracranial data We first extracted the averaged alpha amplitude during the prestimulus period (–800 to 0 ms relative to the first frame) for each contact in the same manner as for the EEG analysis (using FFT). The first 10 contacts with the highest alpha amplitude (8–13 Hz) were then selected as ROIs for each patient. Subsequently, we adopted similar methods and procedures as in the EEG analysis to calculate the prestimulus instantaneous frequency for the bistable EM and GM trials for each contact, which was subsequently averaged across the 10 contacts. Statistical testing of the neurophysiology data The difference between two conditions was statistically tested using nonparametric cluster-based permutation tests, which were implemented in customized scripts in MATLAB (The MathWorks). Specifically speaking, paired t tests were first calculated between the two conditions, for example, the temporal generalization matrices for the efficient versus inefficient inference conditions. Elements that passed a threshold value corresponding to a p -value of 0.05 were marked, and neighboring marked elements were identified as clusters. Cluster-based correction was applied when multiple time points were tested (Figs , and ): data were first randomly shuffled 1,000 times (500 times in the decoding analysis); for each shuffle, the count of suprathreshold samples within a cluster was used to define the cluster size; and the largest cluster size was entered into a distribution of cluster sizes , which was expected under the null hypothesis. Clusters in the real data were considered as statistically significant only if they exceeded the size of 95th percentile of the null distribution of clusters, at α = 0.05. Statistical analysis of the fMRI data At the individual level, the GLM was used to construct a multiple regression design matrix. The four experimental conditions were modeled as regressors of interest: explicit EM, explicit GM, bistable EM, and bistable GM. The four types of event were time locked to the onset of the first frame in each trial by a canonical synthetic hemodynamic response function and its first-order time derivative with an event duration of 0 s. In addition, all the omission trials and the outlier trials in which RTs were outside of the mean RT ± 3 SD were modeled separately as another regressor. The six head movement parameters derived from the realignment procedure were also included as confounds. Parameter estimates were subsequently calculated for each voxel using weighted least-square analysis to provide maximum likelihood estimators based on the temporal autocorrelation of the data. No global scaling was applied. For each participant, simple main effects for each of the four experimental conditions were computed by applying appropriate “1 0” baseline contrasts, that is, experimental conditions versus implicit baseline (null trials). The four first-level individual contrast images were then fed into a within-participants ANOVA at the second group level employing a random-effects model (flexible factorial design in SPM12 including an additional factor modeling the subject means). In the modeling of variance components, we allowed for violations of sphericity by modeling nonindependence across parameter estimates from the same subject and allowing unequal variances between both conditions and participants using the standard implementation in SPM12. We were particularly interested in the differential neural activity between the two types of bistable trials (bistable EM versus bistable GM). Areas of activation were identified as significant only if they passed a conservative threshold of p < 0.005, family-wise error (FWE) corrected for multiple comparisons at the cluster level, with an underlying voxel level of p < 0.005, uncorrected . Statistical analysis on prestimulus neural activity of the fMRI data To investigate how the prestimulus neural activity predicted the outcome of bistable perceptual grouping, a new GLM model was estimated. Given that the ITI was jittered between 2,000–3,000 ms and one-third of all the trials were null trials, the prestimulus periods of all the experimental trials were long enough and adequately jittered for the present statistical analysis on prestimulus neural activity. In the new GLM model, four types of new events were time locked to the time points after the participants made their responses in the preceding trials (“Trials N-1”) of the four types of experimental trials, i.e., the prestimulus preparation period of the current trial (“Trials N”). All the outliers, errors, and missed trials and trials preceded by outliers and errors were separately modeled as another regressor. In this way, parameter estimates in each of the four newly defined critical neural events indicate the height of prestimulus preparation neural activity prior to the actual presentation of the explicit EM, the explicit GM, the bistable EM, and the bistable GM stimuli. Brain regions of activation were identified as significant only if they passed a conservative threshold of p < 0.005 FWE correction for multiple comparisons at the cluster level, with an underlying voxel level of p < 0.005, uncorrected. PPI analysis on prestimulus neural activity of the fMRI data Since the left IPS exhibited specific selectivity towards the bistable EM percepts during both the pre- ( ) and the poststimulus ( ) period, we used the left IPS as the seed region to perform the PPI analysis, focusing on the prestimulus period. For the PPI analysis, prestimulus neural activity (time locked to the responses in “Trials N-1”) in the left IPS was used as the physiological factor and the contrast of “bistable EM versus bistable GM” as the psychological factor. For each participant, the neural contrast of “bistable EM versus bistable GM” was first calculated in the individual level GLM. Subsequently, each participant’s individual peak voxel in the left IPS was determined as the maximally activated voxel within a sphere of 16-mm radius (i.e., twice smoothing kernel) around the coordinates of the peak voxel from the second-level group analysis ( ). Individual peak voxels from every participant are located in the same anatomical structure (left IPS MNI coordinates: x = –33 ± 6, y = –37 ± 7, z = 42 ± 6). Next, the left IPS time series in every participant were extracted from a sphere of 4-mm radius around the individual peak voxels. The PPI term was created for each participant by multiplying the deconvolved and mean-corrected BOLD signal in the given ROI (i.e., the physiological variable) with the psychological variable of interest (i.e., “bistable EM versus bistable GM”). After convolution with the HRF, mean correction, and orthogonalization, three regressors (the PPI term, the physiological variable, and the psychological variable) were entered into the GLM to reveal areas in which neural activations were predicted by the PPI term, with the physiological and the psychological regressors being treated as confounding variables. The PPI analysis was first carried out for each participant and then entered into a random-effects group analysis. Statistical significance was set to p < 0.005, uncorrected at the voxel level, with the cluster extent exceeding 100 voxels.
All the participants gave their informed consent prior to the experiment in accordance with the Declaration of Helsinki. The fMRI, the EEG, and the patient experiments were all approved by the Ethics Committee of School of Psychology, South China Normal University (06202015_TernusCQ). The placement of the depth electrodes was based solely on the clinical needs for the treatment of the patients and was thus independent of the purpose of the present study. This study did not add any invasive procedure to the intracranial recordings. All the participants were at least 18 years old and gave their written informed consent prior to the experiments.
Nineteen adult participants (12 females, mean age of 19.6 years old) took part in the EEG experiment. Another group of 20 adult participants (12 females, mean age of 23.4 years old) took part in the fMRI experiment. Two participants in the EEG experiment were discarded because of excessive eye movement artifacts. One participant in the fMRI experiment was discarded because of low accuracy (less than 70%) in the explicit conditions, and another participant was discarded because of the excessive head movements during the scanning. Therefore, 17 participants in the EEG experiment and 18 participants in the fMRI experiment were included for further analysis. Additionally, four adult patients (two males, mean age of 24 years old) undergoing intracranial recordings with stereotactically implanted multilead electrodes (Guangdong Sanjiu Brain Hospital, China) for epilepsy treatment participated in the present study. Although the anatomical locations of the electrodes were different in each patient, we included the patients whose electrodes were implanted in the occipital and parietal regions. Patients who had destructive lesions such as tumor or encephalomalacia were excluded. All the participants were right-handed, with normal or corrected-to-normal visual acuity.
Visual stimuli consisted of two consecutively presented frames of stimuli (frame 1 and frame 2), and each frame was presented for 30 ms ( ). There was a blank period between the two frames, i.e., the IFI. The IFI could be either explicitly short at 50 ms or explicitly long at 230 ms or at the transition threshold, which was specific for each subject based on pre-experiment psychophysics. Each frame contained two horizontally arranged black disks (1.6° of visual angle in diameter) on a gray background. The center-to-center spatial distance between the two disks was 3° of visual angle. The two frames shared one common disk location at the center of the display. The location of the lateral disk of the first frame, either on the left or the right side of the shared central disk, was always opposite to the lateral disk of the second frame ( ). Specifically speaking, frame 1 with left and central disks and frame 2 with right and central disks induced rightward apparent motion; frame 1 with right and central disks and frame 2 with left and central disks induced leftward apparent motion. The same set of stimulus parameters was adopted for the fMRI, the EEG, and the patient experiments. Depending on the IFI and participants’ online judgments in the bistable trials, there were four types of experimental trials: 1) the explicit EM trials (“Explicit EM”) with the short IFI of 50 ms; 2) the explicit GM trials (“Explicit GM”), with the long IFI of 230 ms; 3) the bistable trials with the threshold IFI, which were judged by the participants as the EM trials (“Bistable EM”); and 4) the bistable trials with the threshold IFI, which were judged by the participants as the GM trials (“Bistable GM”).
To specify the 50% threshold of IFI for the bistable condition for each individual subject, we asked each participant to perform a psychophysical pretest before the main experiment. Prior to the psychophysics test, participants were shown demos of the explicit EM and GM conditions and performed a practice block with only explicit EM and GM trials until the accuracy reached no less than 95%. During the formal psychophysics test, the first frame was presented for 30 ms. After a variable IFI (seven levels: 50, 80, 110, 140, 170, 200, or 230 ms), the second frame was presented for 30 ms as well. Participants were asked to perform a two-alternative forced choice (2AFC) task in which they had to choose between the EM and the GM percept. For each IFI condition, the percentage of GM reports (i.e., “1 –percentage of EM reports”) was collapsed over the leftward and rightward motion directions. The seven data points (one for each IFI) were fitted into a psychometric curve using a logistic function . The transition IFI threshold, i.e., the point at which EM and GM were reported with equal possibility, was calculated by estimating the 50% performance point on the fitted logistic function for each participant . The individual transition threshold derived from the psychophysics test was then used as the IFI in the bistable trials of the subsequent main experiment. Differently from the EEG and fMRI experiment, in the intracranial experiment, an adaptive staircase procedure was adopted to find the individual IFI threshold at which 50% of the stimuli were perceived as GM.
Participants were instructed to fixate at a central fixation throughout the experiment without moving their eyes. The experimental task was to discriminate the two types of motion by pressing two prespecified buttons on the response pad using the thumb of each hand, respectively. The mapping between the two response buttons and the two types of apparent motion percept was counterbalanced between participants. In each trial, the first frame was presented for 30 ms, and after a variable IFI (50 ms, 230 ms, or the individual IFI threshold), the second frame was presented for another 30 ms. The fMRI experiment consisted of 440 trials in total, including 80 explicit EM trials, 80 explicit GM trials, 160 bistable trials, and 120 null trials. The null trials, in which only the central fixation cross was presented, were used as the implicit baseline. The participants were asked to rest for a short period of time (11 s, i.e., five repetition times [TRs]) after every 6 minutes’ task performance, which made three short periods of rest in total. During the three short rest periods, the scanner kept running, and a visual instruction “rest” was presented on the center of the screen throughout. One TR after the disappearance of the “rest” instruction, the behavioral task resumed. The EEG experiment consisted of four blocks, and each block included 40 explicit EM trials, 40 explicit GM trials, and 80 bistable trials, which were intermixed randomly, resulting in 640 experimental trials in total. A rest break was allowed between blocks. For the fMRI and EEG experiment, each trial was followed by a time interval that was selected randomly among 2,000, 2,250, 2,500, 2,750, and 3,000 ms. In the intracranial experiment, there were four blocks of 80 trials (320 trials in total), 10% of which were explicit EM and GM trials. The intertrial interval varied randomly between 1.5 and 2.5 s. In all the three experiments, the temporal order of all the trials was randomized for each participant individually to avoid potential problems of unbalanced transition probabilities. All participants completed a training section of 5 min before the recording.
EEGs were continuously recorded from 64 Ag/AgCl electrodes (10–20 System) with BrainAmp DC amplifiers (low-pass = 100 Hz, high-pass = 0.01 Hz, and sampling frequency = 500 Hz). The vertical electro-oculogram was recorded by one electrode under the participants’ left eyes. All the electrode impedances were kept below 5 kΩ. Signals were referenced online to the unilateral mastoid. Offline processing and analysis were performed using EEGLAB and customized scripts in MATLAB (The MathWorks, Natick, MA, USA). Data were down-sampled to 160 Hz, rereferenced to the average reference, epoched from –800 ms before the first frame to 1,000 ms after the first frame for the subsequent alpha frequency analysis, and re-epoched from –500 ms to 500 ms relative to the presentation of the second frame for the decoding analysis. Trials containing visually identified eye movements or muscle artifacts were excluded manually. Visually identified noisy electrodes were spherically interpolated.
Ten to 13 semirigid, multilead electrodes were stereotactically implanted in the four participants, respectively. All the electrodes have a diameter of 0.8 mm and contain 10–16 2-mm–wide and 1.5-mm–apart contacts. The precise anatomical location of each contact was identified by coregistering each participant’s postimplantation CT with the preimplantation 3D T1 image, using rigid affine transformations derived from FSL’s FLIRT algorithm . Intracranial recordings were conducted using commercial video–intracranial monitoring system. The data were bandpass filtered online from 0.1 to 300 Hz and sampled at 1,000 Hz, using a reference contact located in the white matter. For the offline analysis, recording signals were down-sampled to 500 Hz. Contacts in the epileptogenic zones were excluded from further analyses. Each contact was rereferenced with respect to its direct neighbor, i.e., bipolar montage, to achieve high local specificity by removing effects of distant sources that spread equally to adjacent sites through volume conduction. All the data were epoched from –800 to 1,000 ms relative to the presentation of the first frame.
A Siemens 3T Trio system with a standard head coil at Beijing MRI Center for Brain Research was utilized to obtain T2*-weighted echo-planar images (EPIs) with blood oxygenation level-dependent contrast. The matrix size was 64 × 64 mm 3 , and the voxel size was 3.4 × 3.4 × 3 mm 3 . Thirty-six transversal slices of 3-mm thickness that covered the whole brain were acquired sequentially with a 0.75-mm gap (TR = 2.2 s, TE = 30 ms, FOV = 220 mm, flip angle = 90°). There was a single run of functional scanning, including 524 EPI volumes. The first five volumes were discarded to allow for T1 equilibration effects. Data were preprocessed with Statistical Parametric Mapping software SPM12 (Wellcome Department of Imaging Neuroscience, London, UK; http://www.fil.ion.ucl.ac.uk ). Images were realigned to the first volume to correct for interscan head movements. The mean EPI of each participant was then computed and spatially normalized to the MNI single-participant template using the “unified segmentation” function in SPM12. This algorithm is built on a probabilistic framework that enables image registration, tissue classification, and bias correction to be combined within the same generative model. The resulting parameters of a discrete cosine transform, which define the deformation field necessary to move individual data into the space of the MNI tissue probability maps, were then combined with the deformation field transforming between the latter and the MNI single participant template. The ensuing deformation was subsequently applied to individual EPI volumes. All images were thus transformed into standard MNI space and resampled to 2 × 2 × 2 mm 3 voxel size. The data were then smoothed with a Gaussian kernel of 8-mm full-width half-maximum to accommodate interparticipant anatomical variability. Data were high-pass filtered at 1/128 Hz and analyzed with a general linear model (GLM) as implemented in SPM12. Temporal autocorrelation was modeled using an AR (1) process.
For the behavioral data in the EEG, intracranial, and fMRI experiment, omissions and trials with RTs 3 standard deviations (SDs) away from the mean RT in each condition were first excluded from further analysis. For the calculation of accuracy rates in the two explicit conditions, the explicit trials at the short IFI with a judgment of GM and the explicit trials at the long IFI with a judgment of EM were considered as incorrect trials, which were discarded and excluded from further analysis. For both the EEG and fMRI experiment, paired t tests were performed to test the difference in the accuracy rates between the two types of explicit trials, the proportions of EM and GM trials in the bistable condition, and the mean RTs for the two explicit and the two bistable conditions, respectively.
For all the electrodes in all the participants, a power spectrum (from 5 Hz to 30 Hz) was obtained through a Fast Fourier Transform (FFT) of all the trials (from –800 to 0 ms relative to the presentation of the first frame). An amplitude topographic map of the most prominent frequency band in the power spectrum was obtained. For each participant, the individual peak alpha frequency was determined as the value corresponding to the maximum peak frequency from the 800 ms of data prior to the presentation of the first frame within the 8–13 Hz range for the selected posterior electrodes. The Pearson product–moment correlation between the individual alpha frequency and the individual IFI threshold obtained from the psychophysical procedures was then calculated. Instantaneous PAF and alpha power were analyzed using the methods and code developed by Cohen . We chose only one electrode, which showed the strongest alpha amplitude among all the occipital electrodes in the posterior ROI for each participant, to calculate the PAF and the alpha power . To avoid contaminations by the poststimulus signals, only the prestimulus period (from –800 to 0 ms) of the EEG signals were extracted, and all the poststimulus period signals (starting from 0 ms) were excluded. Furthermore, to avoid edge artifacts at the stimulus onset due to filtering, the prestimulus signals of each bistable trial were copied, flipped from left to right, and appended to the right side of the original data. These epochs were filtered between 8 and 13 Hz with a zero-phase, plateau-shaped bandpass filter with 15% transition zones. Phase angle and amplitude time series were extracted from the filtered data with a Hilbert transform. The alpha power was obtained by calculating the square of the amplitude. For the frequency calculation, the temporal derivative of the phase angle time series describes how phase changes over time and thus corresponds to the instantaneous frequency in Hz (when scaled by the sampling rate and 2 π ). Since noises in the phase angle time series can cause sharp, nonphysiological responses in the derivative, the instantaneous frequency was filtered with a median filter with an order of 10 and a maximum window size of 400 ms: data were median filtered ten times with 10 time windows ranging from 10 to 400 ms prior to averaging across trials. Since this analysis considers changes only in the instantaneous phase of the data, it is mathematically independent from the amplitude of the oscillation, except where amplitude is equal to zero and phase is undefined. Subsequently, the instantaneous PAFs were averaged across bistable EM and GM trials, respectively.
Multivariate decoding techniques were further adopted to investigate how the PAF affects the representation contents of the bistable EM and GM percepts with the progress of time. For each participant, we first calculated the instantaneous alpha frequency for each time point in the prestimulus window (from –800 ms to 0 relative to the onset of frame 2) of each bistable trial, based on the one chosen electrode with the maximal alpha amplitude. Subsequently, statistical tests (paired t test) between the bistable EM and bistable GM conditions were performed at the group level. The significant time points were further selected as the time points of interest, and the PAF for each trial was determined by averaging the instantaneous alpha frequency across these significant time points (–570 to –350 ms relative to the presentation of frame 2; see ). Subsequently, amplitude data epochs of all the bistable trials (–400 to 400 ms relative to the presentation of frame 2), right after the preprocessing steps and without any further processing steps (no spectral analysis applied), were sorted according to the calculated PAF of each trial and half split into the high PAF and the low PAF trial sessions. The bistable GM trials in the high PAF session and the bistable EM trials in the low PAF session were selected as the two types of trials in the efficient inference condition; the bistable EM trials in the high PAF session and the bistable GM trials in the low PAF session were selected as the two types of trials in the inefficient inference condition. Please note, the PAF of each bistable trial was used only as an indicator to categorize the bistable trials into the efficient versus inefficient conditions in a post hoc way but was never used as the actual data fed into the subsequent decoding analysis. To exclude potential confounds caused by different number of trials upon comparing different conditions, we matched the trial count in the above four types of trials by randomly selecting a subsample of trials from the conditions with more trials. We then applied a multivariate linear discriminant analysis to characterize the temporal dynamics that discriminated between the subjectively perceived bistable EM versus GM percepts for the efficient and inefficient inference condition, respectively. Classifications were based on the regularized linear discriminant analysis to identify a projection in the multidimensional EEG data, x , that maximally discriminated between the two representations across all stimulus levels. Each projection is defined by a weight vector, w , which describes a one-dimensional projection y of the EEG data y = ∑ i w i x i + c , with i summing over all channels and c a constant. The regularization parameter was optimized in preliminary tests and kept fixed for all the analyses. The decoding analysis was performed in a time-resolved manner by applying it to each time point sequentially, resulting in an array of classifiers, for example, w (t1), w (t2), w (t3) and so on. To improve the signal-to-noise ratio, the data were first averaged within a time window of 50 ms centered around the time point of interest. This process could introduce some contaminations from the poststimulus signals to the prestimulus signals around the stimuli onset: the signal at time 0 contains the information within –25 to 25 ms. However, the abovementioned contaminations can influence the prestimulus signals about 25 ms at most. Subsequently, the classifier performance was assessed not only at the time point used for training (for example, classifier w (t1) was tested at t1, w (t2) was tested at t2, and so on) but also on data from all the other time points (for example, classifier w (t1) was tested on all the time points t1, t2, t3, and so on). The performance of the classifier was quantified using the receiver operator characteristic (ROC), based on leave-one-out cross-validation within each participant. The above procedure resulted in a (training time) × (decoding time) temporal generalization matrix per condition.
We first extracted the averaged alpha amplitude during the prestimulus period (–800 to 0 ms relative to the first frame) for each contact in the same manner as for the EEG analysis (using FFT). The first 10 contacts with the highest alpha amplitude (8–13 Hz) were then selected as ROIs for each patient. Subsequently, we adopted similar methods and procedures as in the EEG analysis to calculate the prestimulus instantaneous frequency for the bistable EM and GM trials for each contact, which was subsequently averaged across the 10 contacts.
The difference between two conditions was statistically tested using nonparametric cluster-based permutation tests, which were implemented in customized scripts in MATLAB (The MathWorks). Specifically speaking, paired t tests were first calculated between the two conditions, for example, the temporal generalization matrices for the efficient versus inefficient inference conditions. Elements that passed a threshold value corresponding to a p -value of 0.05 were marked, and neighboring marked elements were identified as clusters. Cluster-based correction was applied when multiple time points were tested (Figs , and ): data were first randomly shuffled 1,000 times (500 times in the decoding analysis); for each shuffle, the count of suprathreshold samples within a cluster was used to define the cluster size; and the largest cluster size was entered into a distribution of cluster sizes , which was expected under the null hypothesis. Clusters in the real data were considered as statistically significant only if they exceeded the size of 95th percentile of the null distribution of clusters, at α = 0.05.
At the individual level, the GLM was used to construct a multiple regression design matrix. The four experimental conditions were modeled as regressors of interest: explicit EM, explicit GM, bistable EM, and bistable GM. The four types of event were time locked to the onset of the first frame in each trial by a canonical synthetic hemodynamic response function and its first-order time derivative with an event duration of 0 s. In addition, all the omission trials and the outlier trials in which RTs were outside of the mean RT ± 3 SD were modeled separately as another regressor. The six head movement parameters derived from the realignment procedure were also included as confounds. Parameter estimates were subsequently calculated for each voxel using weighted least-square analysis to provide maximum likelihood estimators based on the temporal autocorrelation of the data. No global scaling was applied. For each participant, simple main effects for each of the four experimental conditions were computed by applying appropriate “1 0” baseline contrasts, that is, experimental conditions versus implicit baseline (null trials). The four first-level individual contrast images were then fed into a within-participants ANOVA at the second group level employing a random-effects model (flexible factorial design in SPM12 including an additional factor modeling the subject means). In the modeling of variance components, we allowed for violations of sphericity by modeling nonindependence across parameter estimates from the same subject and allowing unequal variances between both conditions and participants using the standard implementation in SPM12. We were particularly interested in the differential neural activity between the two types of bistable trials (bistable EM versus bistable GM). Areas of activation were identified as significant only if they passed a conservative threshold of p < 0.005, family-wise error (FWE) corrected for multiple comparisons at the cluster level, with an underlying voxel level of p < 0.005, uncorrected .
To investigate how the prestimulus neural activity predicted the outcome of bistable perceptual grouping, a new GLM model was estimated. Given that the ITI was jittered between 2,000–3,000 ms and one-third of all the trials were null trials, the prestimulus periods of all the experimental trials were long enough and adequately jittered for the present statistical analysis on prestimulus neural activity. In the new GLM model, four types of new events were time locked to the time points after the participants made their responses in the preceding trials (“Trials N-1”) of the four types of experimental trials, i.e., the prestimulus preparation period of the current trial (“Trials N”). All the outliers, errors, and missed trials and trials preceded by outliers and errors were separately modeled as another regressor. In this way, parameter estimates in each of the four newly defined critical neural events indicate the height of prestimulus preparation neural activity prior to the actual presentation of the explicit EM, the explicit GM, the bistable EM, and the bistable GM stimuli. Brain regions of activation were identified as significant only if they passed a conservative threshold of p < 0.005 FWE correction for multiple comparisons at the cluster level, with an underlying voxel level of p < 0.005, uncorrected.
Since the left IPS exhibited specific selectivity towards the bistable EM percepts during both the pre- ( ) and the poststimulus ( ) period, we used the left IPS as the seed region to perform the PPI analysis, focusing on the prestimulus period. For the PPI analysis, prestimulus neural activity (time locked to the responses in “Trials N-1”) in the left IPS was used as the physiological factor and the contrast of “bistable EM versus bistable GM” as the psychological factor. For each participant, the neural contrast of “bistable EM versus bistable GM” was first calculated in the individual level GLM. Subsequently, each participant’s individual peak voxel in the left IPS was determined as the maximally activated voxel within a sphere of 16-mm radius (i.e., twice smoothing kernel) around the coordinates of the peak voxel from the second-level group analysis ( ). Individual peak voxels from every participant are located in the same anatomical structure (left IPS MNI coordinates: x = –33 ± 6, y = –37 ± 7, z = 42 ± 6). Next, the left IPS time series in every participant were extracted from a sphere of 4-mm radius around the individual peak voxels. The PPI term was created for each participant by multiplying the deconvolved and mean-corrected BOLD signal in the given ROI (i.e., the physiological variable) with the psychological variable of interest (i.e., “bistable EM versus bistable GM”). After convolution with the HRF, mean correction, and orthogonalization, three regressors (the PPI term, the physiological variable, and the psychological variable) were entered into the GLM to reveal areas in which neural activations were predicted by the PPI term, with the physiological and the psychological regressors being treated as confounding variables. The PPI analysis was first carried out for each participant and then entered into a random-effects group analysis. Statistical significance was set to p < 0.005, uncorrected at the voxel level, with the cluster extent exceeding 100 voxels.
S1 Table Patient information and behavioral performance in the intracranial experiment. Following information is reported for each patient: the EZ identified by the clinical investigation; the number of ELs; the total number of CHs. Task performance: ARs of the explicit EM and GM condition; reported rates of GM in the bistable condition. AR, accuracy rate; CH, electrode contact; EL, implanted electrode shaft; EM, element motion; EZ, epileptogenic zone; GM, group motion. (DOCX) Click here for additional data file. S2 Table Brain activations in the main . Brain regions showing significant relative increases of BOLD response associated with the bistable EM and bistable GM trials before and after the presentation of the first frame. BOLD, blood-oxygen–level dependent; EM, element motion; GM, group motion. (DOCX) Click here for additional data file. S3 Table Brain activations in the main . Brain regions that showed higher prestimulus functional connectivity with the left IPS (–32, –38, 38) in the bistable EM than bistable GM trials. EM, element motion; GM, group motion; IPS, intraparietal sulcus. (DOCX) Click here for additional data file. S1 Video Demo of explicit EM. EM, element motion. (MP4) Click here for additional data file. S2 Video Demo of explicit GM. GM, group motion. (MP4) Click here for additional data file. S1 Fig Psychometric fitting results for each participant in fMRI experiment. Underlying data available at https://osf.io/tze94/ . fMRI, functional magnetic resonance imaging. (TIF) Click here for additional data file. S2 Fig Correlation between the occipital alpha cycle in each individual subject and the individual transition IFI threshold. Dashed lines indicate 95% confidence intervals around the linear fit line. Underlying data available at https://osf.io/tze94/ . IFI, interframe interval. (TIF) Click here for additional data file. S3 Fig Instantaneous PAF relative to the representation of the second frame revealed higher alpha frequency preceding the bistable GM trials than the bistable EM trials. Significant time points are indicated by the horizontal black bar (cluster-based correction, p < 0.05). Shaded regions denote ±1 within-subjects SEM. Underlying data available at https://osf.io/tze94/ . EM, element motion; GM, group motion; PAF, prestimulus alpha frequency; SEM, standard error of the mean. (TIF) Click here for additional data file.
|
Quality of life and its associated factors among women diagnosed with pelvic organ prolapse in Gynecology outpatient department Southern Nations, Nationalities, and Peoples region public referral hospitals, Ethiopia | 7e58303e-5066-4f61-ab49-799d2c88589d | 10308721 | Gynaecology[mh] | Pelvic organ prolapse (POP) is the descent of female pelvic organs into or through the vagina, including the bladder, uterus, and rectum . This case results from the defect of the pelvic floor support, caused by many risk factors such as vaginal birth, advancing age, and increasing body mass index; atrophic changes caused by aging or estrogen loss, chronic straining, and abnormalities of connective tissue . The affected women may present with signs and symptoms of urinary or fecal loss or retention, vaginal pressure or heaviness, abdominal, low back, vaginal, or perennial pain or discomfort, a mass sensation, difficulty walking, lifting, sitting, and stress or fear related to anxiety about the problem . It is the leading indication for hysterectomy in postmenopausal women and accounts for 15–18% of procedures in all age groups . In the Southern Nation Nationality and Peoples Region of Ethiopia, there were high burdens of pelvic organ prolapse, and the most common cause was old age, long hours of carrying heavy objects, a high parity, a history of home delivery, a history of chronic constipation, and a history of chronic cough . Pelvic organ prolapse is a significant public health issue that affects the lives of millions of women . On average, POP influenced 19.7% of women in developing countries . Similarly in Ethiopia, it influenced 23.52% of women . It has the potential to severely influence women’s health-related quality of life through restrictions on physical, social, and sexual activities, psychological distress, and increased financial burden related to healthcare . In addition, access to health care to manage these conditions is often limited, and women usually have to live with the consequences for the rest of their lives . The quality of life of women with POP varies from country to country based on economic level, lifestyle, educational level, and culture . However, a study showed that globally, the most common predominant risk factor for worsening QoL in women with POP is symptoms of POP . Overall, pelvic floor disorders harm women’s lives, emotions, and quality of life, and they may be associated with a variety of systemic symptoms such as urinary, bowel, and sexual symptoms, which may significantly confront the quality of life of the women . Pelvic organ prolapse highly affects women’s quality of life in developing countries compared to developed countries . However, in developing countries, the degree and consequences of the burden of the disease due to pelvic floor dysfunction, especially on the quality of life, are more poorly understood . In addition, the QoL of POP on women’s health has not yet been recognized as a public health problem in many developing countries. This is because only a few studies have examined POP about the normal deterioration of general health-related QOL in the general population . In developed countries, the quality of life and its associated factor in women with POP is assessed with P-QoL tool and used as a baseline strategy for the treatment of POP, but in developing countries, especially in Africa, including Ethiopia not applicable, this is due to limited information on the quality of life of pelvic organ prolapse . Despite the quality of life allowing the quantification of morbidity, treatment efficacy and also acting as a measure of how lives are affected, and coping strategies, research has usually concentrated on the prevalence, etiology, diagnosis, and management of pelvic floor dysfunction, with limited work being performed on the effects of chronic conditions or their treatment on QoL . Pelvic organ prolapse is one of the sources of severe morbidity and psychological upheaval for the patient, who is often socially withdrawn and stigmatized and negatively influences the socioeconomic and reproductive activity of affected women . However, in Ethiopia, no reports were showing the quality of life of women with pelvic organ prolapse. Measuring the quality of life of women with POP is an important input for policymakers and program implementers to develop strategies. Study area and period The study was conducted in South Nations, Nationalities, and Peoples Region (SNNPR). The SNNPR is bordered by Kenya to the south, South Sudan to the west, Gambela regional to the northwest, Oromia region to the north and east, South West region to the Southwest, and Sidama region to the east . The region has 3-referral public hospitals(Dilla university referall, Wolaita Sodo university, Wolkite university) 2 specialized and comprehensive hospitals, 3 general hospitals, 44 primary hospitals, 474 health centers, and 2633 health posts . The previous two-month report in the three referral public hospitals showed that there were four hundred seventy POP cases . The data were collected from May 1, 2022-July 4, 2022 in public referral hospitals, SNNPR; Ethiopia 2022. Study population and eligibility criteria All women who were diagnosed with pelvic organ prolapse in the Gynecology outpatient department of SNNPR public referral hospitals during the data collection period. All women diagnosed with pelvic organ prolapse and who had not gotten treatment before in the Gynecology outpatient department in all selected public referral hospitals were included. Sample size and sampling procedure The sample size was calculated using a single population formula ( n = (Z α/2) 2 * P (1-P) ⁄ d 2 ). Based on a study conducted in Uganda, the proportion of poor QoL among POP women was 45.5% . Then, using the following assumptions: 95% of a confidence interval, α = 0.05, and margin of error = 5% the sample size required for this study was calculated. The calculated sample size was 381. By adding 10% of the non-response rate the final sample size was 419. All three public referral hospitals found in SNNPR were included in this study. The previous two months’ average POP report of each hospital was used to proportional allocate the calculated sample size . Finally, study respondents were selected using a consecutive sampling technique. Operational definitions Quality of Life (QoL) WHO defines the quality of Life as an individual’s perception of their position in life in the context of the culture and value systems in which they live and about their goals, expectations, standards, and concern. It is a multi-dimensional concept that includes domains related to physical, mental, emotional, level of independence, and social functioning . Poor quality of life Greater or equal to the median score of the overall (nine) QoL domains . Duration of prolapse: The number of months or years from the time pelvic organ prolapse first occurred until now . Stages of prolapse Based on Baden–Walker Halfway Scoring System: Stage 0: is no prolapse. Stage I: is leading part of the prolapse is more than 1 cm above the hymen. Stage II: is the leading edge less than or equal to 1 cm above or below the hymen; Stage III: is leading edge is more than 1 cm beyond the hymen, but less than or equal to the total vaginal length; Stage IV: is complete eversion . Data collection tool and procedure The data collection tool was adopted from the University of Gondar validated tool on prolapse quality of life questionnaire (P-QoL) . An interviewer-administered validated Amharic version questionnaire was used to collect the data. The questionnaire includes 20 items in nine different domains (General Health Condition and POP on the overall life has one item, Role limitation, Social Limitation, Sleep/ Energy and physical limitation has two items, personal relationship and emotion has three items, intensity or Severity of Pain have four items). The Questions are rated on a four-point Likert scale ranging from one to four, indicating better to worse conditions. This scale was converted into zero to hundred scales. Therefore, the total score is a range of 0–100. Each domain score was obtained by adding the scores of the individual items that comprise the domain. A full-scale score was obtained by the addition of nine-domain items. The lowest total score that can be on the scale is zero, and the highest score is 100. The cut-off point was the median of the total score . Three BSc and MSc midwives were recruited as data collectors and supervisors respectively. The training was given to both data collectors and supervisors by the investigator about the objective of the study, data collection tool, procedure, and how to fill out the questionnaires. All women who were diagnosed with POP in the Gynecology outpatient department of SNNPR public referral hospitals were interviewed after assessing eligibility and obtaining informed written consent and their charts were reviewed to assess the stage of the prolapse. Data quality assurance The training was given to data collectors and supervisors. A data collector was supervised throughout the data collection period. Then, the overall process was coordinated and controlled by the investigator. Investigators, supervisors, and data collectors took a discussion meeting after data collection to ensure completeness. Furthermore, the collected data was Checked and coded and entered into the Epi-data computer program version 3.1 to minimize data entry errors. Data processing and analysis The collected data was entered into the EPI data version 3.1 computer program. Then, it was exported to Statistical Package for social science version 25. Descriptive statistics like frequency and summary statistics were employed to describe the characteristics of the study participants. Multicollinearity of the predictor variable was checked by using the variance inflation factor before binary logistic regression was done, and it was < 2 VIF for all independent variables. All explanatory variables in bivariable logistic regression that fulfill the chi-square static assumption were considered for multivariable logistic regression analysis to control for confounding factors. Adjusted Odds Ratio (AOR) with their corresponding 95% Confidence Intervals (CI) and p -value less than 0.05 was used to declare the association between dependent and independent variables. Model fitness was checked by the Hosmer–Lemeshow test, it was fitted with a p -value of 0.45; finally presented by figure, table, and graph. The study was conducted in South Nations, Nationalities, and Peoples Region (SNNPR). The SNNPR is bordered by Kenya to the south, South Sudan to the west, Gambela regional to the northwest, Oromia region to the north and east, South West region to the Southwest, and Sidama region to the east . The region has 3-referral public hospitals(Dilla university referall, Wolaita Sodo university, Wolkite university) 2 specialized and comprehensive hospitals, 3 general hospitals, 44 primary hospitals, 474 health centers, and 2633 health posts . The previous two-month report in the three referral public hospitals showed that there were four hundred seventy POP cases . The data were collected from May 1, 2022-July 4, 2022 in public referral hospitals, SNNPR; Ethiopia 2022. All women who were diagnosed with pelvic organ prolapse in the Gynecology outpatient department of SNNPR public referral hospitals during the data collection period. All women diagnosed with pelvic organ prolapse and who had not gotten treatment before in the Gynecology outpatient department in all selected public referral hospitals were included. The sample size was calculated using a single population formula ( n = (Z α/2) 2 * P (1-P) ⁄ d 2 ). Based on a study conducted in Uganda, the proportion of poor QoL among POP women was 45.5% . Then, using the following assumptions: 95% of a confidence interval, α = 0.05, and margin of error = 5% the sample size required for this study was calculated. The calculated sample size was 381. By adding 10% of the non-response rate the final sample size was 419. All three public referral hospitals found in SNNPR were included in this study. The previous two months’ average POP report of each hospital was used to proportional allocate the calculated sample size . Finally, study respondents were selected using a consecutive sampling technique. Quality of Life (QoL) WHO defines the quality of Life as an individual’s perception of their position in life in the context of the culture and value systems in which they live and about their goals, expectations, standards, and concern. It is a multi-dimensional concept that includes domains related to physical, mental, emotional, level of independence, and social functioning . Poor quality of life Greater or equal to the median score of the overall (nine) QoL domains . Duration of prolapse: The number of months or years from the time pelvic organ prolapse first occurred until now . Stages of prolapse Based on Baden–Walker Halfway Scoring System: Stage 0: is no prolapse. Stage I: is leading part of the prolapse is more than 1 cm above the hymen. Stage II: is the leading edge less than or equal to 1 cm above or below the hymen; Stage III: is leading edge is more than 1 cm beyond the hymen, but less than or equal to the total vaginal length; Stage IV: is complete eversion . WHO defines the quality of Life as an individual’s perception of their position in life in the context of the culture and value systems in which they live and about their goals, expectations, standards, and concern. It is a multi-dimensional concept that includes domains related to physical, mental, emotional, level of independence, and social functioning . Greater or equal to the median score of the overall (nine) QoL domains . The number of months or years from the time pelvic organ prolapse first occurred until now . Based on Baden–Walker Halfway Scoring System: Stage 0: is no prolapse. Stage I: is leading part of the prolapse is more than 1 cm above the hymen. Stage II: is the leading edge less than or equal to 1 cm above or below the hymen; Stage III: is leading edge is more than 1 cm beyond the hymen, but less than or equal to the total vaginal length; Stage IV: is complete eversion . The data collection tool was adopted from the University of Gondar validated tool on prolapse quality of life questionnaire (P-QoL) . An interviewer-administered validated Amharic version questionnaire was used to collect the data. The questionnaire includes 20 items in nine different domains (General Health Condition and POP on the overall life has one item, Role limitation, Social Limitation, Sleep/ Energy and physical limitation has two items, personal relationship and emotion has three items, intensity or Severity of Pain have four items). The Questions are rated on a four-point Likert scale ranging from one to four, indicating better to worse conditions. This scale was converted into zero to hundred scales. Therefore, the total score is a range of 0–100. Each domain score was obtained by adding the scores of the individual items that comprise the domain. A full-scale score was obtained by the addition of nine-domain items. The lowest total score that can be on the scale is zero, and the highest score is 100. The cut-off point was the median of the total score . Three BSc and MSc midwives were recruited as data collectors and supervisors respectively. The training was given to both data collectors and supervisors by the investigator about the objective of the study, data collection tool, procedure, and how to fill out the questionnaires. All women who were diagnosed with POP in the Gynecology outpatient department of SNNPR public referral hospitals were interviewed after assessing eligibility and obtaining informed written consent and their charts were reviewed to assess the stage of the prolapse. The training was given to data collectors and supervisors. A data collector was supervised throughout the data collection period. Then, the overall process was coordinated and controlled by the investigator. Investigators, supervisors, and data collectors took a discussion meeting after data collection to ensure completeness. Furthermore, the collected data was Checked and coded and entered into the Epi-data computer program version 3.1 to minimize data entry errors. The collected data was entered into the EPI data version 3.1 computer program. Then, it was exported to Statistical Package for social science version 25. Descriptive statistics like frequency and summary statistics were employed to describe the characteristics of the study participants. Multicollinearity of the predictor variable was checked by using the variance inflation factor before binary logistic regression was done, and it was < 2 VIF for all independent variables. All explanatory variables in bivariable logistic regression that fulfill the chi-square static assumption were considered for multivariable logistic regression analysis to control for confounding factors. Adjusted Odds Ratio (AOR) with their corresponding 95% Confidence Intervals (CI) and p -value less than 0.05 was used to declare the association between dependent and independent variables. Model fitness was checked by the Hosmer–Lemeshow test, it was fitted with a p -value of 0.45; finally presented by figure, table, and graph. Socio-demographic characteristics A total of 409 women were enrolled in the study, giving a 97.6% response rate. The age range of the participants is 36 to 72 years with a median of 54 years. Most of the respondents 220(53.8%) were married. Additionally, regarding educational status, 204 (49.87%) were non-educated (Table ). Gynecologic and obstetrics characteristics For most respondents, 251(61.36%) prolapses were greater than two years durations. Regarding stages of prolapse, 153(37.4%) of women had stage III prolapse. Moreover, the parity ranged from 4 to 14 with a mean and standard deviation of (8.23 ± 2.23), and 295(72.1%) of women had ≥ 7 number of childbirth (Table ). The magnitude of quality of life of women with pelvic organ prolapse According to this study, the magnitude of poor quality of life among women with POP was 235(57.5%) with (95% CI: 52.5, 62.3) (Fig. ). Personal relationships 301(73.6%), the impact of pop 248(60.6%), and emotional well-being 239(58.4%) were domains of quality of life that had the highest score, whereas social limitation 142(34.7%) and sleep/energy 99(24.2%) had the lowest score (Table ). Factor associated with QoL of women with pelvic organ prolapse To assess the relationship between various socio-demographic, obstetric, and POP-related factors and quality of life, binary logistic regression was used. The variables that full fill the chi-square static assumption were maternal educational status, residency, marital status, parity, transportation access, menopausal status, duration POP, and stage of POP. Then, using the backward likelihood ratio approach, these variables were subjected to multivariable logistic regression analysis. In the final model, only five variables were present. Model fitness was tested with the Hosmer and Lemeshow Goodness of fit test and fit with a p -value of 0.45. Additionally, all independent variables had no multicollinearity with a variance inflation factor value (VIF) < 2. This study shows that women with stage III/IV prolapse were 2.52 times more likely to have a poor quality of life than those with stage I or II prolapse (AOR = 2.52, 95% CI: 1.34, 4.74). In addition, women in the menopause period were 3.21 times more likely to have a poor quality of life than in the pre-menopause period (AOR = 3.21, 95% CI 1.75, 5.97). Furthermore, the odds of having a poor quality of life were 2.81 times greater for women who were not married (widowed or divorced) than for those who were married (AOR = 2.81, 95% CI:1.48, 5.32). Moreover, women with a longer duration of POP were 5.8 times more likely to have a poor quality of life than their counterparts (AOR = 5.8, 95% CI: 3.13, 10.81) (Table ). A total of 409 women were enrolled in the study, giving a 97.6% response rate. The age range of the participants is 36 to 72 years with a median of 54 years. Most of the respondents 220(53.8%) were married. Additionally, regarding educational status, 204 (49.87%) were non-educated (Table ). For most respondents, 251(61.36%) prolapses were greater than two years durations. Regarding stages of prolapse, 153(37.4%) of women had stage III prolapse. Moreover, the parity ranged from 4 to 14 with a mean and standard deviation of (8.23 ± 2.23), and 295(72.1%) of women had ≥ 7 number of childbirth (Table ). According to this study, the magnitude of poor quality of life among women with POP was 235(57.5%) with (95% CI: 52.5, 62.3) (Fig. ). Personal relationships 301(73.6%), the impact of pop 248(60.6%), and emotional well-being 239(58.4%) were domains of quality of life that had the highest score, whereas social limitation 142(34.7%) and sleep/energy 99(24.2%) had the lowest score (Table ). To assess the relationship between various socio-demographic, obstetric, and POP-related factors and quality of life, binary logistic regression was used. The variables that full fill the chi-square static assumption were maternal educational status, residency, marital status, parity, transportation access, menopausal status, duration POP, and stage of POP. Then, using the backward likelihood ratio approach, these variables were subjected to multivariable logistic regression analysis. In the final model, only five variables were present. Model fitness was tested with the Hosmer and Lemeshow Goodness of fit test and fit with a p -value of 0.45. Additionally, all independent variables had no multicollinearity with a variance inflation factor value (VIF) < 2. This study shows that women with stage III/IV prolapse were 2.52 times more likely to have a poor quality of life than those with stage I or II prolapse (AOR = 2.52, 95% CI: 1.34, 4.74). In addition, women in the menopause period were 3.21 times more likely to have a poor quality of life than in the pre-menopause period (AOR = 3.21, 95% CI 1.75, 5.97). Furthermore, the odds of having a poor quality of life were 2.81 times greater for women who were not married (widowed or divorced) than for those who were married (AOR = 2.81, 95% CI:1.48, 5.32). Moreover, women with a longer duration of POP were 5.8 times more likely to have a poor quality of life than their counterparts (AOR = 5.8, 95% CI: 3.13, 10.81) (Table ). This study showed that the overall poor quality of life among women who had pelvic organ prolapse was 57.5% (95%CI: 52.5, 62.3). This indicated that, although pelvic organ prolapse is a benign case, it highly influences the quality of life of women. This finding is in line with a study conducted in France 54.5% ; Pakistan 60.8% , and Slovakia 52.8% . This is due to the fact that, although the quality of life of women with POP varies from country to country based on economic level, lifestyle, educational level, and culture, the most common predominant risk factors for worsening QoL in women are pelvic organ prolapse symptoms . On the other side, it is higher than a study conducted in Uganda 45.5% , and Ghana 39.4% . This difference is due to QoL domain and item differences and sample size differences. In addition, this study showed that the poor quality of life of women with POP was significantly associated with advanced stages of prolapse. This finding is supported by studies conducted in South Africa , Taiwan , Thailand , Nepal , the USA , Italy , and London . This is due to the fact that as the prolapse advanced, secondary consequences, including decubitus ulcers, bowel symptoms, urinary symptoms, abdominal symptoms, and vaginal symptoms, increased, which worsened the quality of life . Moreover, this study showed that there was a significant association between the poor quality of life of women with POP and longer durations of prolapse. This finding is supported by studies conducted in Pakistan , France , and Bangladesh . This is because, as the duration of the disease increased, there was progressive disability. In addition, as the duration of the prolapse increased, there were advanced stages of prolapse that caused a symptomatic prolapse that worsened women’s health conditions . Furthermore, this study showed that women’s being unmarried was strongly associated with poor quality of life. This finding is in line with studies conducted in Bangladesh . In fact being in a relationship, could be considered a buffer mechanism against psychological illnesses, reducing the likelihood of developing depressive symptoms, and isolation . In addition, lonely women with pelvic organ prolapse had double responsibility to lead their lives; this condition could be worse for their lives. Moreover, this study showed that menopausal women were significantly associated with poor quality of life. This finding is supported by studies conducted in Turkey . This is due to the fact that as the women entered the menopausal stage, there were physiological changes resulting in menopausal symptoms and worsening of the prolapse, which is a double burden that worsens the quality of life . According to this study, more than half of women with pelvic organ prolapse had a poor quality of life. Stage III/IV prolapse, longer duration of prolapse, menopausal period, and unmarried women (widowed, divorced) were significantly associated with poor quality of life. Recommendation To the South Nation Nationalities and Peoples of Ethiopia region health bureau: Offering the possibility of early diagnosis and treatment for women with POP that has been hidden in the community To researcher: Researches need on why women with pelvic organ prolapse are not early attending health care facilities (Time to health-seeking behavior among women with POP and its factors). Further community based study To the South Nation Nationalities and Peoples of Ethiopia region health bureau: Offering the possibility of early diagnosis and treatment for women with POP that has been hidden in the community To researcher: Researches need on why women with pelvic organ prolapse are not early attending health care facilities (Time to health-seeking behavior among women with POP and its factors). Further community based study Additional file 1. |
Unraveling the morphological and molecular profile of Setaria digitata in Aceh cattle | 2a1a2914-b55c-474a-86b2-343015d4a6fb | 11910275 | Cytology[mh] | Aceh cow is one of the most bred livestock types chosen by farmers in Aceh, Indonesia. This cow has diverse body shapes, physical attributes, genetic composition, and good adaptive ability in environments with limitations. Thus, they need to be protected, preserved, and have their advantages improved ( ). Aceh cow has been designated as a cow breed from Indonesia based on the decree of the Ministry of Agriculture of the Republic of Indonesia number 2907/Kpts/OT.140/6/2011 ( ). Aceh cow has also been assigned Indonesia National Standard (Standar Nasional Indonesia–SNI) by National Standard Agency number 7651.3:2013 as a meat-producing and working breed ( ). Aceh cow is a cross of Bos indicus and Bos sondaicus , which have smaller statures compared to other cows. The males commonly have hump and horns. The male and female Aceh cow coat colors that have been assigned in the SNI are brick red and light brown ( ). Aceh cow is also one of the ruminant livestock with high economic value among the farmers ( ). Thus, understanding the rearing condition of Aceh cows and the presence of serious health problems such as worm infestation is crucial. Parasitic disease is a factor that may lower livestock productivity. The presence of worms in livestock is often dismissed by farmers despite having a large influence on livestock health and productivity ( ). Filaria nematodes are parasites in the tissue and tissue chamber of all vertebrates. Among these filaria nematodes, Setaria sp. worm is a parasite residing in the cow peritoneal cavity where it lives, grows, and sexually reproduces. The male Setaria sp. worm has a small and thin body while the female worm has a longer and thicker body than the male. Setaria sp. worm is a type of worm that commonly causes health problems in cows ( ) Setaria sp. worm has a round long form shaped like a soft thread which can be seen with the naked eye when the cow’s abdomen is opened ( ). Setaria sp. worm looks like a milky white thin thread, getting thinner by its posterior with an average length of 62.8 ± 9.89 mm (ranging between 40 and 80 mm), when found in adult form and does not cause some kind of pathogenicity to the cow ( ). The cow is a natural host of Setaria sp. worm. This parasite generally does not cause disease in its natural host. Infective larvae (L3) transmission via mosquitoes as vectors into aberrant hosts such as goats, sheep, or horses and causes serious neuropathological disorders and is often fatal ( ). Mosquitoes are infected by microfilaria upon consuming microfilaria containing blood from a host. These microfilariae grow into infective larvae at the mosquito’s chest muscle within 2 to 3 weeks. Infected mosquitoes transmit these infective larvae to another vulnerable host during their blood feeding ( ). Worm identification was carried out by microscopic observation of the parasite and one of the methods is by observing the worm’s morphology. Worms found in cows usually still cannot be identified macroscopically with the naked eye, so the staining solution should be used to identify the worm with clarity and easier identification. This research used several staining solutions, lugol 2%, lactophenol, and glycerol. According to , these solutions can produce a clear background and clear worm morphology which make the worms transparent. Other than identification with different staining solutions, molecular character analysis is also required to ensure that the species found matches the desired target, which in this case was Setaria digitata worm. Therefore, The objective of this research is to conduct morphological and molecular characterization of S. digitata worms in Aceh cows and to determine a good staining solution for identifying S. digitata worms from cows slaughtered in Banda Aceh and Bireuen slaughterhouse.
The blood and worm sample collection was obtained through direct methods from slaughtered cows and extracted innards. Setaria sp. worms located in the peritoneum and other cow organs were directly taken from the slaughtered cows. Setaria sp., shaped like a milky white thin thread, could be seen by the membrane covering the abdomen in the peritoneum of Setaria sp. infected cows. Worm samples were washed and stored in sample bottles filled with 70% alcohol by using dressing tweezers. Samples were transported and observed in the Parasitology Laboratory of the Veterinary Medicine Faculty, Universitas Syiah Kuala. Setaria sp. microfilaria presence detection in Aceh cow by Knott test One hundred twenty-four Aceh cow blood samples were analyzed for microfilaria by Knott’s test. The whole blood sample was obtained from the jugular vein when the cows were culled. The blood samples were mixed with EDTA for the Knott Test performed in the Parasitology Laboratory. As much as 1 ml of blood was taken and mixed with 9 ml of 2% formalin and then centrifuged at 1,500 rpm. The supernatant was discarded and the pellet was dripped with Methylene blue 10%. One drop of the mixture was smeared on an object glass to observe the presence of microfilaria by a binocular microscope ( ). Identification of adult Setaria sp. in Aceh cow Lugol 2% and lactophenol staining Adult worms found in Aceh cow’s duodenum serous cavity and peritoneum cavity during necropsy were collected and stored in bottles filled with alcohol 70%. The samples were then fixed with Lugol 2% and lactophenol staining and were left alone for around 5–10 minutes. Samples were then put on a petri dish and observed under a microscope. Glycerol staining Adult worms found in the peritoneum cavity of Aceh cow during necropsy were collected and stored in alcohol 70% solution, before being moved into a glass bottle and fixed with glycerol before being observed under a light microscope. Species identification Morphological observation of Setaria sp. worm was performed by binocular light microscope Olympus CX 21 and a capture system according to the method explained by by using HDMI+USB 1080P/2m SIGMA microscope camera. The incidence rate was determined based on the observation by using the following equation: Incidence = The number of infected samples (host) The total number of samples (host) × 100 Molecular identification of Setaria spp. worm DNA extraction The DNA extraction from samples Setaria worm in Aceh cow was conducted using a DNA extraction kit. The procedure started by combining 200 mg of the sample with 320 µl of lysis buffer, 16 µl of RNase A, 16 µl of proteinase K, and 3.2 µl of DTT in a 2 ml tube. The mixture was incubated at 65°C for 1 hour. After incubation, the sample was centrifuged at 16,099 × g for 11 minutes. Following this, 320 µl of binding buffer was added, and the mixture was vortexed and incubated again at 65°C for 10 minutes. The incubated mixture was then rinsed with 320 µl of 100% ethanol, 500 µl of wash buffer, and 500 µl of 75% ethanol. Finally, the DNA was eluted in 100 µl of elution buffer at 65°C. The primer characteristic used in this research is presented in . Phylogenetic analysis Polymerase chain reaction products of cox1 and 12s rDNA were purified using Gel recovery kit (Vivantis, Malaysia) and sequenced on both strands using Big Dye Terminator V.3.1 Cycle Sequencing kit in an ABI 3130 Genetic Analyzer (Applied Biosystems, USA). Sequences were aligned together using ClustalW software ( ) to determine the consensus sequences. Consensus 12S rDNA and cox1 sequences were subjected to BLASTn analysis (http://blast.ncbi.nlm.nih.gov) and compared to all nucleotide sequences of Setaria species available in the current databases. Sequence identities (in %) were calculated by pairwise comparisons. Subsequently, the consensus sequences were aligned with a selected subset of closely related sequences of the genus Setaria . Phylogenetic relationships were inferred based on analyses employing the Neighbor-Joining (NJ) method using MEGA7. The topological stability of the tree was evaluated by 1000 bootstrap replications. Analysis data Data are presented in the form of figure and table, and then data obtained are analyzed descriptively. Ethical approval The experimental protocol was approved by the Animal Care and Use Committee of Faculty Veterinary Medicine, Syiah Kuala University No. 2.KEH.080.07.2023.
One hundred twenty-four Aceh cow blood samples were analyzed for microfilaria by Knott’s test. The whole blood sample was obtained from the jugular vein when the cows were culled. The blood samples were mixed with EDTA for the Knott Test performed in the Parasitology Laboratory. As much as 1 ml of blood was taken and mixed with 9 ml of 2% formalin and then centrifuged at 1,500 rpm. The supernatant was discarded and the pellet was dripped with Methylene blue 10%. One drop of the mixture was smeared on an object glass to observe the presence of microfilaria by a binocular microscope ( ).
Adult worms found in Aceh cow’s duodenum serous cavity and peritoneum cavity during necropsy were collected and stored in bottles filled with alcohol 70%. The samples were then fixed with Lugol 2% and lactophenol staining and were left alone for around 5–10 minutes. Samples were then put on a petri dish and observed under a microscope.
Adult worms found in the peritoneum cavity of Aceh cow during necropsy were collected and stored in alcohol 70% solution, before being moved into a glass bottle and fixed with glycerol before being observed under a light microscope.
Morphological observation of Setaria sp. worm was performed by binocular light microscope Olympus CX 21 and a capture system according to the method explained by by using HDMI+USB 1080P/2m SIGMA microscope camera. The incidence rate was determined based on the observation by using the following equation: Incidence = The number of infected samples (host) The total number of samples (host) × 100
DNA extraction The DNA extraction from samples Setaria worm in Aceh cow was conducted using a DNA extraction kit. The procedure started by combining 200 mg of the sample with 320 µl of lysis buffer, 16 µl of RNase A, 16 µl of proteinase K, and 3.2 µl of DTT in a 2 ml tube. The mixture was incubated at 65°C for 1 hour. After incubation, the sample was centrifuged at 16,099 × g for 11 minutes. Following this, 320 µl of binding buffer was added, and the mixture was vortexed and incubated again at 65°C for 10 minutes. The incubated mixture was then rinsed with 320 µl of 100% ethanol, 500 µl of wash buffer, and 500 µl of 75% ethanol. Finally, the DNA was eluted in 100 µl of elution buffer at 65°C. The primer characteristic used in this research is presented in .
Polymerase chain reaction products of cox1 and 12s rDNA were purified using Gel recovery kit (Vivantis, Malaysia) and sequenced on both strands using Big Dye Terminator V.3.1 Cycle Sequencing kit in an ABI 3130 Genetic Analyzer (Applied Biosystems, USA). Sequences were aligned together using ClustalW software ( ) to determine the consensus sequences. Consensus 12S rDNA and cox1 sequences were subjected to BLASTn analysis (http://blast.ncbi.nlm.nih.gov) and compared to all nucleotide sequences of Setaria species available in the current databases. Sequence identities (in %) were calculated by pairwise comparisons. Subsequently, the consensus sequences were aligned with a selected subset of closely related sequences of the genus Setaria . Phylogenetic relationships were inferred based on analyses employing the Neighbor-Joining (NJ) method using MEGA7. The topological stability of the tree was evaluated by 1000 bootstrap replications. Analysis data Data are presented in the form of figure and table, and then data obtained are analyzed descriptively.
The experimental protocol was approved by the Animal Care and Use Committee of Faculty Veterinary Medicine, Syiah Kuala University No. 2.KEH.080.07.2023.
Setaria sp. microfilaria detection in Aceh cow using Knott Test The blood were taken from 124 Aceh cows slaughtered in Banda Aceh and Bireun Slaughterhouse, Aceh to detect Setaria sp. microfilaria using Microhematocrit Centrifugation Technique (MHCT) ( ). The results can be seen in and . Identification of adult Setaria sp. in Aceh cow The observation of adult Setaria sp. from 124 cows slaughtered in Banda Aceh and Bireuen, Aceh, through direct observation could be seen in and below. Identification result of adult S. digitata worm by using three staining solution to evaluate their morphology can be seen in below. The molecular identification of Setaria spp. worm The DNA amplification result of Setaria spp from Banda Aceh and Bireuen slaughterhouse used the forward and reverse primer for cox1 and the forward and reverse primer for 12S rDNA with master mix PCR Go Tag which specifically attached to the template samples SD1, SD2, SD3, SD4, SD5 and SD6 then produced 450 bp for 12S rDNA and 680 bp for cox1 ( ). The 12S rRNA gene analysis result of Setaria. sp The Blast result of 12S rRNA gene nucleotide from Setaria sp. worm samples from Aceh cows ( ). The genetic distance table The genetic distance of S. digitata found in Aceh cow with those data from GenBank is presented in . Phylogenetic tree analysis based on COX 1 gene The phylogenetic tree analysis result for COX 1 gene of S. digitata seen on . The phylogenetic analysis result 450 bp, with maximum parsimony method in Mega program version 11.0, and Kimura 2 parameter 1,000× bootstraps showed that the Setaria worm samples from this research are still grouped as S. digitata , although it has different clade with other comparison isolates from the genbank sequence data. The Setaria genus in this research formed two groups, which are the S. digitata and Setaria labiatopapilosa group. The S. digitata group is divided into three groups where the S. digitata worm in this research is within their own group along with S. digitata isolate from China, different from other S. digitata and from Dirofilaria and Wucheria worms as the out-group representatives.
The blood were taken from 124 Aceh cows slaughtered in Banda Aceh and Bireun Slaughterhouse, Aceh to detect Setaria sp. microfilaria using Microhematocrit Centrifugation Technique (MHCT) ( ). The results can be seen in and .
The observation of adult Setaria sp. from 124 cows slaughtered in Banda Aceh and Bireuen, Aceh, through direct observation could be seen in and below. Identification result of adult S. digitata worm by using three staining solution to evaluate their morphology can be seen in below.
The DNA amplification result of Setaria spp from Banda Aceh and Bireuen slaughterhouse used the forward and reverse primer for cox1 and the forward and reverse primer for 12S rDNA with master mix PCR Go Tag which specifically attached to the template samples SD1, SD2, SD3, SD4, SD5 and SD6 then produced 450 bp for 12S rDNA and 680 bp for cox1 ( ).
The Blast result of 12S rRNA gene nucleotide from Setaria sp. worm samples from Aceh cows ( ).
The genetic distance of S. digitata found in Aceh cow with those data from GenBank is presented in .
The phylogenetic tree analysis result for COX 1 gene of S. digitata seen on . The phylogenetic analysis result 450 bp, with maximum parsimony method in Mega program version 11.0, and Kimura 2 parameter 1,000× bootstraps showed that the Setaria worm samples from this research are still grouped as S. digitata , although it has different clade with other comparison isolates from the genbank sequence data. The Setaria genus in this research formed two groups, which are the S. digitata and Setaria labiatopapilosa group. The S. digitata group is divided into three groups where the S. digitata worm in this research is within their own group along with S. digitata isolate from China, different from other S. digitata and from Dirofilaria and Wucheria worms as the out-group representatives.
revealed that Setaria sp . incidence rate in Aceh cow was 12.7%. This is lower compared to the result from Sundar and D’Souza (2013) who reported the prevalence of Setaria sp. in cows in India ranged between 77% and 95% with 187 out of 500 livestock samples found positive for Setari a sp. However, This result is lower compared to the incidence rate observed by who reported that 136 nematodes found in cows and water buffaloes in Brazil with the prevalence rates were 24% and 25%, respectively. Based on the results, it indicated that for livestock in the Aceh area the cases are still low, but attention needs to be given to eliminate the Setariosis cases in livestock and to broad the treatment coverage beyond anthelmintic drugs. showed that the incidence rate of Setaria sp. worm in Aceh was 16.9%. The incidence rate in this research is lower compared to the research performed by in Korea where the prevalence rate of Setaria sp. in cows was 70%. Although these worms grow into adulthood in the abdominal cavity and thus mostly are not dangerous for livestock, serious pathogenic results may occur in animals such as goats, sheep, and horses where the Setaria sp. larvae may migrate randomly into the central nervous system. The difference in nematode prevalences in several territories is influenced by several factors: the infectious agent, age, sex, breed, feed, and reading management. The rearing and enclosure management by the farmers and overall farm management greatly influence parasite infection prevalence (Tolistyawati et al ., 2016). Blood is a body liquid that distributes oxygen to all body tissue and sends the nutrition needed by the cells. This allows Setaria sp. worm larvae to grow in their host’s body. Definitive diagnosis in setariosis cases can be done by complete ophthalmic testing, showing worm movement inside the eye, or by Knott test method to detect Setaria microfilariae ( ). Knott test is a technique to detect microfilaria by hemolysis and blood sample concentration to detect Setaria species microfilariae ( ). The morphological observation to identify Setaria sp. worm found in the peritoneum cavity of Aceh Cow from Banda Aceh and Bireuen, Aceh, was performed by glycerol, Lugol 2%, and lactophenol solutions ( ). The morphological observation of the worms with lactophenol staining solution in this research ( ) resulted in clearer and easier-to-identify worm morphology, but the sample did not last as the worm would be damaged. Whereas the use of Lugol 2% ( ) and glycerol ( ) staining solution took a longer time to be clearly seen but lasted longer and the morphology appeared clear. According to Asarina and Haeruni (2019), the three solutions have each’s advantages and disadvantages in identifying worm morphology. Glycerol solution is hygroscopic so it absorbs water molecules. Samples added with glycerol in all concentrations are proven to last longer and not dry. Glycerol is a colorless odorless viscous chemical often used in drug formulation. Glycerol is generally used for preservation media or long-term storage as well as a medium to move microorganisms ( ). Glycerol can be used as a medium because glycerol can protect antimicrobial activity by improving the structural stability of the microbe’s natural protein thus preventing the protein from damage from thermal process and aggregation ( ). Lugol solution is often used in microscopic parasite observation, among them is worm morphology observation. Lugol solution staining gives a clear background with a yellow color on the worm ( ). Lactophenol staining gives a clear view of worm morphology where the worm turns transparent, but this staining may lower the elasticity thus hindering the identification process ( ) or even at times damaging the worm in a moderately long storage span ( ). From , the anterior and posterior parts of S. digitata by several staining solutions could be seen. depicted the anterior of S. digitata , and showed the anterior part stained by Lugol 2%, lactophenol, and glycerol. The anterior part ( ) showed that S. digitata worm has peribuccal crown with central “helmet” at the end of its head and triangle-shaped lateral labia (LL). showed that the tail ends with a smooth knob with oval lateral appendages (ALs). Differences with Setaria sp. worm morphology could be seen at the anterior tip. The anterior tip of S. digitata worm is round with two similar projections. The posterior tip of S. digitata has a smooth or slightly rough knob/dull tip with surface papilae ( ). Setaria sp. morphology observation revealed that S. digitata morphology differs between sexes. The female worm has a peribuccal crown or “helm” in the anterior tips shaped like a triangle with two projecting thorns showing the dorsal projection (DP) and ventral projection (VP) at the peribuccal crown. The posterior of the female S. digitata has a smooth round terminal knob with ALs. Meanwhile, the male S. digitata has a similar anterior to the female worm and a wavy cuticle. The posterior end of the male worm has a thick and sturdy right side with a narrow proximal end while the left spicula has a stem and a rod ( ). Setaria sp. in cows can cause fibrinous peritonitis. This worm can live and reproduce in cows by using the blood circulation system as a source of nutrition for their growth. Moreover, infective larvae transmission through mosquito vectors into hosts such as sheep and horses can cause serious neuropathological problems. Setaria sp. worm can easily infect cows, especially in areas with high mosquito populations ( ). According to , morphological characteristic identification is not enough to detect and differentiate S. digitata with other species like S. equina , S. labiatopapillosa , and S. cervi. This is due to the morphological characteristic similarities between species which makes species identification difficult. Thus, molecular characterization is required to identify S. digitata species and determine the phylogenetic position of S. digitata from Aceh cows in Indonesia within the genus as well as other filaria nematodes genus position based on the analysis of the two genes COX1 and 12S rDNA. Setaria digitata is the most reported in tropical Asian countries, where its vector is generally found ( ; ). South Korea is the only non-tropical country in high latitude where S. digitata has been reported in a horse ( ; ). In this research, S. digitata is identified and confirmed by molecular method in a horse on a farm in North China, where the latitude line is close to South Korea. This incidental case most likely occurred due to the presence of a cow farm in the surrounding environment. Thus, this finding indicates that cows in north China possibly carried S. digitata. In 2017, S. digitata was first identified with molecular method in a water buffalo carcass in South China, but not in living water buffalo or cow ( ). Setariasis is an ophthalmology disorder known in horse, with S. digitata reported to be the most common cause ( ). S. digitata is regarded as a primary health hazard in animals and causes economic losses to the owners. Although traditional identification has been used to differentiate Setaria sp. species, according to Perumal et al . (2016) and this is insufficient because S. digitata worm has similar morphology with S. equina , S. labiatopapillosa and S. cervi . and have conducted research by nucleic acid (DNA) based detection method to ensure the species of filaria isolated from a mare. The result was obtained and confirmed to be S. digitata when compared to other sequences in NCBI GenBank database by BLAST. The genetic sequence extracted from the worm has 99% similarities with S. digitata isolated from cows and water buffaloes in Sri Lanka and from water buffaloes in Greater China. The Blast result of the nucleotide sequences ( ) in this research used the 12S rRNA gene from Setaria sp Aceh isolate and showed 92.78% similarity with S. digitata Nagaland isolate (Accession no. KJ780079.1) and has the most distant similarity with S. digitata isolate from Osaka City University (Accession no. AM779801.1) with 88.36% similarity. When compared with , the sequence analysis of ribosomal DNA 12S, 28S, and ITS-2 showed 98% similarity with S. digitata , while the mitochondrial gene Cox 1 showed 99% similarity with S. digitata Sri Lanka isolate (Acc. No.EF179382), and 87% similar with S. labiatopapillosa . The ITS-2 region showed 98% similarity with Sri Lanka isolate (Accession no. EF196088), 96% similarity with Brugia pahangi (Accession no. EU373652), 95% similarity with B. malayi (EU37361) and 96% similarity with Volvulus onchocerca (AF228572). In the ribosomal DNA sequence analysis of 12S, 28S, and ITS-2 region showed 98% identity with S. digitata , while mitochondrial gene Cox 1 showed 99% similarity with S. digitata Sri Lanka isolate (Acc. No. EF179382,) and 87% similarity with S. labiatopapillosa . The ITS-2 region showed 98% similarity with Sri Lanka isolates (Accession no. EF196088), 96% similarity with Brugia pahangi (Accession no. EU373652), 95% similarity with B. malayi (EU37361), and 96% similarity with Onchocerca volvulus (AF228572). Based on the genetic distance in this research, SD1, SD2, SD3, SD4, SD5, and SD6, samples have 1 to 12 % genetic distance compared to available S. digitata isolate from GenBank (Accession no. GU138699.1), while SD3 samples have 26 genetic distance value compared to S. digitata isolate from the GenBank (Accession no. GU138699.1), and has 18 to 21 % genetic distance when compared to the other research samples. This analysis result also showed the genetic distance value of Setaria sp sample Aceh isolate (SD1-SD6) with other parasitic nematodes out of the Setaria group with 37% against Wuchereria bancrofti (Accession no. JF775522.1) and 39% against Dirofilaria sp (Accession no. KX2650933.1). The phylogenetic tree rooted at further mind point confirmed that the S. digitata isolated in this research is part of the same large clade as those isolated from Japan, India, and Sri Lanka. S. equina and S. tundra are formed on different clusters. S. digitata and S. labiatopapillosa appeared to be sister species with a bootstrap value of 67% ( ). The 12S rDNA sequence is a useful marker for phylogenetic analysis because it has slower evolution compared with other protein-coding mitochondrial genes (Peng et al ., 2017). The phylogenetic analysis of S. digitata showed that the isolate sequence from different host species produces closely related clades with high bootstrap support within the Setaria genus. Nucleotide sequence variation observation is few. This finding indicates that the S. digitata isolates detected worldwide have similar molecular characteristics (Maharana et al. , 2010). The phylogenetic tree analysis result of 12S rDNA gene is shown in . The phylogenetic analysis result 680 bp, using the maximum parsimony method in Mega Program version 11.0, with Kimura 2 parameter model with 1,000× bootstrap, showed that the Setaria worm sample of this research is still grouped S. digitata species, despite having different similarity level with other isolates taken from genebank as a comparison. The Setaria genus in this research formed two groups, which are the S. digitata and S. labiatopapilosa group. The S. digitata group is divided into two big groups, the S. digitata and Setaria cervi group. The S. digitata worm in this research is in its own group with S. digitata isolates from China and Thailand, different from other S. digitata which are isolates from Sri Lanka and India as well as with Wucheria worm as an out-group representative. The phylogenetic analysis of S. digitata based on COX 1 gene and 12 S rDNA gene has showed the isolate and sequence from different host species has produced separate clade with high bootstrap support in Setaria genus and the nucleotide sequence variation is observed. This showed that the S. digitata isolates have molecular characteristic similarities worldwide. The trend of wide molecular phylogenetics which appears in this analysis (based on 12S rDNA and COX I gene sequence) was similar to previous findings (Casiraghi et al ., 2004; Yatawara et al ., 2007; ; ). Moreover, the phylogeny appears to be partly consistent with the classification based on morphological and biological characteristics of filaria nematodes ( ). Analysis of Setaria sp from Aceh cattle using the COX 1 gene showed the presence of S. digitata and S. labiatopapilosa . These results are expected to contribute to the additional information regarding the Setaria species that infect cattle, and can also be applied in terms of prevention and control of setariosis in livestock. In addition, considering that humans can act as accidental hosts for S. digitata and S. labiatopapillosa parasites, it is also necessary to anticipate whether this parasite is zoonotic or not. Considering that microfilariae and adult S. digitata worms have been found in Aceh cattle, further studies need to be carried out both on other ruminant livestock and mosquitoes as vectors that carry setariaosis.
The identification of S. digitata through the cox1 and 12S rDNA genes produced 450 bp and 680 bp fragments, respectively. To observe these worms effectively, three staining agents lactophenol, 2% lugol, and glycerol were tested. While each staining method has specific benefits, glycerol proved to be the most advantageous, offering the clearest visualization and enhancing the ease of detailed examination. This makes glycerol a highly suitable choice for reliable identification, balancing both practicality and clarity in studying the morphology of S. digitata .
|
Gene metabolite relationships revealed metabolic adaptations of rice salt tolerance | 56fdfc9e-2734-4796-9f92-2de41e2e5455 | 11742878 | Biochemistry[mh] | Abiotic stress describes the negative impact of non-living environmental factors on plants, leading to a variety of responses that can alter biological processes like gene expression and cellular metabolism, as well as affect growth and development. This type of stress encompasses issues such as extreme temperatures, drought, flooding, salinity, metal toxicity, and nutrient deficiencies, each prompting distinct reactions. Key environmental challenges, particularly extreme temperatures, drought, and saline soils, significantly restrict plant survival and their distribution in natural ecosystems , . Soil salinization adversely affects both the yield and quality of crops. Salt stress poses a significant threat to plant growth, leading to decreased leaf expansion, stoma closure, a reduced photosynthetic rate, and a loss of biomass , . Salinization currently affects more than 800 million hectares of land on Earth, and projections suggested that by 2050, approximately half of cultivated land could be impacted by salinity , . Plants have developed adaptive mechanisms to counter salt stress by making changes at the morphological, physiological, biochemical, and molecular levels. They also adjust metabolite and gene expression to combat stress and minimize damage , . Rice ( Oryza sativa L.), a glycophyte, is highly sensitive to salt stress . The extent of salt tolerance varies among genotypes and stages of development. While rice is particularly vulnerable to salt stress during the seedling stage, it shows moderate tolerance during the tillering stage , . Certain genotypes have been recognized for their salt tolerance . In various studies, the genotype IR28 has been utilized as salt-sensitive rice for molecular investigations of salinity tolerance , . Additionally, there are reports indicating the genetic analysis and high salinity tolerance of the genotype CSR28 – . Reactive oxygen species (ROS) are a group of highly reactive molecules that contain oxygen, such as superoxide, hydrogen peroxide, and hydroxyl radicals. They are produced as natural byproducts of cellular metabolism and play important roles in cellular signaling and defense against pathogens. However, under severe abiotic stress conditions, excess of ROS is produced, causing damage to various cellular components, such as DNA, proteins, carbohydrates, lipids, and enzymes, ultimately triggering programmed cell death , . To prevent injuries, plants regulate ROS production effectively by employing a range of enzymatic and nonenzymatic antioxidants. Enzymatic antioxidants belonging to the plant defense system include peroxidase (POD), superoxide dismutase (SOD), glutathione reductase (GR), catalase (CAT), dehydroascorbate reductase (DHAR), ascorbate peroxidase (APX), and monodehydroascorbate reductase (MDHAR), while nonenzymatic antioxidants include ascorbate (AsA), flavonoids, carotenoids, stilbenes, tocopherols, and various vitamins , . Omics technologies, such as metabolomics, enable a system-wide analysis of metabolic processes, for example in response to salinity stress . Metabolite profiling is conducted by instruments such as gas chromatography–mass spectrometry (GC‒MS) and permits the study of plant responses to environmental stresses at the molecular level. The comprehensive quantitative and qualitative measurements of the cellular metabolites acquired from stress-treated tissues provide a broad view of plant physiological and molecular reactions to stresses. Furthermore, metabolites are considered the final product of gene expression and are closely related to phenotype, which doubles the value of their study – . Recent reports have indicated the role of metabolites such as amino acids, sugar alcohols, and organic acids in osmotic adjustment as osmolytes, ionic homeostasis, photosynthesis and leaf senescence in salt-treated rice – . Analyzing the metabolome and transcriptome together can provide precise insights into how genes and metabolites interact, allowing for a systematic exploration of metabolic pathway synthesis and regulation. This approach helps overcome the limitations of individual omics studies, providing a more detailed explanation of the expression patterns and involvement of key genes in metabolic adaptations . Wang et al. reported several genes in OsDRAP1 -mediated salt tolerance in rice by Pearson correlation analysis of transcript and metabolite levels. These genes were involved in key biosynthetic pathways of amino acids (proline, valine), organic acids (glyceric acid, phosphoenolpyruvic acid and ascorbic acid) and carbohydrate metabolism. In the present study, we used an association approach of metabolomics and gene expression data to elucidate metabolic adaptations of rice salt tolerance using various genotypes/organs/timepoints. Phenotypic evaluation confirmed the contrasting salinity tolerances of IR28 and CSR28 Phenotypic evaluation of IR28 and CSR28 rice seedlings 1 week after exposure to high salinity stress confirmed differences in their salinity tolerance (Fig. ). The differences between the genotypes were more prominent in the shoots than in the roots. The difference in shoot length among the genotypes increased from 3.7% under control conditions to 47.1% under salinity stress. Furthermore, the difference in shoot dry weight increased from 1% in the control treatment to 57.9% in the salinity treatment. Compared with CSR28, IR28 exhibited greater reductions in both shoot length and dry weight under salinity stress. The leaf RWC of salt-stressed sensitive IR28 plants decreased significantly (23.4%) compared to that of the salt-tolerant CSR28 plants, while no significant difference was detected between the genotypes under control conditions. Brown and tubular leaves appeared in most IR28 seedlings after 1 week of salt stress, while CSR28 seedlings displayed more green leaves. The CSR28 genotype had a significantly lower mean salinity score than the IR28 genotype, which indicated that CSR28 was more salinity tolerant. Effects of salt stress on H 2 O 2 and MDA contents and antioxidant enzyme activity Changes in the levels of H 2 O 2 and MDA provide insights into the capacity to combat ROS and lipid peroxidation under stress. Both H 2 O 2 (Fig. a) and MDA (Fig. b) levels increased in response to salinity stress in the organs of both genotypes. However, the increases were notably more pronounced under long-term stress in the sensitive genotype. The levels of H 2 O 2 and MDA in the roots of IR28 at 54-h timepoint increased compared to CSR28 by 206.6% and 164.7%, respectively, while the increases in the shoots were 216.6% and 166.6%, respectively. The results of the antioxidant enzyme activity revealed that the levels of both CAT and SOD increased in all conditions in response to salinity. Although no significant difference between the two genotypes in both organs at the 6-h timepoint, however CAT and SOD enzyme levels in the roots of CSR28 at 54-h timepoint, were elevated compared to IR28 by 226.6% and 162.1%, respectively, while those increases in the shoots were 214.6% and 180.1%, respectively (Fig. c,d). The metabolic responses of salt-stressed rice seedlings revealed the specific functions of metabolites in salinity tolerance GC‒MS analysis revealed 37 primary metabolites, including 18 amino acids (AAs), 5 sugars and sugar alcohols and 14 organic acids (OAs), in the roots and shoots of the CSR28 and IR28 genotypes at the 6-h and 54-h timepoints (the mean and standard variation values of relative metabolite levels are shown in Supplementary Table ). ANOVA revealed significant differences between 35 metabolites in roots and shoots, with an average phenotypic variation of 35.6% based on the salinity/control ratio (Supplementary Table ). Of the 37 identified metabolites, 26 presented a significant difference between two timepoints of 6 h and 54 h, explaining 8.2% of the phenotypic variation. The two genotypes displayed significant differences in 28 metabolites, which explained 9.7% of the phenotypic variation. Furthermore, the interaction of genotype × timepoint × organ was significant for 25 metabolites, which explained 4% of the phenotypic variation. The relative changes of the metabolites are shown as ratio of salinity/control in all conditions (Table ). In general, 89.3% of the metabolite changes were significant in response to salinity, of which 56.5% and 32.8% represented increased and decreased accumulation, respectively. Lactate had the smallest response to salinity under the different conditions. Amino acids (AAs) Among the 18 identified AAs, 94.4% exhibited significant changes in response to salinity stress, of which 83.3% and 11.1% exhibited increased and decreased accumulation, respectively. The greatest increases were observed for the metabolites in the salt-stressed shoots of CSR28 at the 54-h timepoint. These metabolites included isoleucine (42.8-fold), leucine (31.06-fold) and proline (36.05-fold). Compared with those in the roots of the genotypes at the 6-h timepoint, only three AAs (α-alanine, GABA and methionine) were increased in CSR28 compared to those in IR28, while 14 AAs were remarkably increased in CSR28 compared to those in IR28 at the 54-h timepoint. Furthermore, the accumulation of six and 12 AAs was greater in the shoots of CSR28 than in those of IR28 at the 6-h and 54-h timepoints, respectively (Table ). Sugars and sugar alcohols Out of the 5 sugars and sugar alcohols identified through GC‒MS analysis, 90% exhibited significant changes in response to salinity, with 55% and 35% increased and decreased accumulation, respectively. In the roots of CSR28, raffinose (45.2-fold) and fructose (-11.1-fold) had the greatest and greatest changes, respectively, at the 54-h timepoint. After 6 h of salinity treatment, the glucose and raffinose contents in the roots of IR28 were greater, and the fructose content was lower than those in the roots of CSR28. However, at the 54-h timepoint, the raffinose and myoinositol contents in CSR28 were significantly greater, and the glucose and glycerol contents were lower than those in IR28. In the shoots, the values in IR28 were greater than those in CSR28 at both timepoints (Table ). Organic acids (OAs) Among the 14 OAs identified via metabolite profiling, 84.2% of the changes were significant in response to salinity under all conditions, including 22.4% and 61.8% increased and decreased accumulation, respectively. Furthermore, the maximum (8.4-fold) and minimum (-4.9-fold) changes were due to citrate in the roots of IR28 and quinate in the shoots of CSR28 at the 54-h timepoint, respectively. After 6 h of exposure to salinity, the concentrations of six OAs in CSR28 roots were greater than those in IR28 roots, while only the hydroxyglutarate concentration in CSR28 roots was greater than that in IR28 roots. Between CSR28 and IR28, nine OAs were differentially accumulated after 54 h of salinity treatment. After 6 h of salt exposure, the contents of six OAs in the shoots of IR28 were greater than those in the shoots of CSR28, while only the fumarate content in the shoots of CSR28 was greater than that in the shoots of IR28. Under long-term stress, a greater reduction in OAs was observed in CSR28 than in IR28 (Table ). Aspartate among AAs, myo-inositol among sugars and sugar alcohols and citrate, glycerate, isocitrate and shikimate among OAs showed organ-specific accumulation and increased only in roots in response to salinity stress. Among the OAs, only α-ketoglutarate and pyruvate were specifically accumulated between the genotypes in the salt-stressed shoots, so both decreased and increased in CSR28 and IR28, respectively. Hierarchical cluster analysis (HCA) grouped the metabolites and samples Heatmap was conducted to obtain an overview of metabolite profiling under different conditions. HCA grouped the metabolite data into two major clusters, roots and shoot, and each cluster into two distinct control and salinity stress subclusters (Fig. ). Furthermore, each subcluster was classified with respect to the timepoints of 6 h and 54 h, and each subcluster included two tolerant and sensitive genotypes. Maximum similarity of the timepoints of 6 h and 54 h was observed under the control condition in both organs, and this similarity was greater in the shoots than in the roots, while the difference between the metabolites at both timepoints increased significantly under salinity stress. The genotypes in both organs exhibited a maximum correlation under the control condition, and the difference in their metabolome increased under salinity stress. In the roots, the difference between the two genotypes at 54 h was greater than that at 6 h, while the difference in the shoots was greater at 6 h than at 54 h. In general, the correlations between the samples were as follows: genotype > timepoint > treatment > organ. Correlations between metabolites To further explain the relationships between metabolite contents in response to salinity stress, the correlations between amino acids, and between organic acids and carbohydrates were analyzed (Fig. ). The results indicated that there was a significant positive correlation between the content of most amino acids, except for the correlations of glycine with putrescine ( r = − 0.91, P value = 0.001), aspartate ( r = − 0.85, P value = 0.02), asparagine ( r = − 0.82, P value = 0.02), and β-alanine ( r = − 0.81, P value = 0.04), and the correlations of putrescine with proline ( r = − 0.85, P value = 0.03) and threonine ( r = − 0.81, P value = 0.001), which were negatively correlated (Fig. a). In the correlation analysis between organic acids and carbohydrates, diverse patterns of both positive and negative correlations were observed. For example, except for glycerol, the other carbohydrates were positively correlated with each other. Furthermore, glycerol followed a pattern similar to that of organic acids such as lactate and pyruvate (Fig. b). Expression of genes involved in the metabolism of metabolites and antioxidant enzyme activity under salinity stress Analysis of genes related to the accumulation of metabolites and antioxidant enzymes is highly important for understanding the synthesis of these compounds in response to salinity stress. Therefore, we focused on the key genes associated with the metabolites and antioxidant enzymes identified in this research (Fig. ). The expression of key genes involved in proline biosynthesis demonstrated that salinity stress led to the up-regulation of the genes OsP5CS2 , OsP5CR , and OsP5CS1 in most of the experimental samples. OsP5CS2 showed a significant increase in expression under all conditions except at the 6-h timepoint in the roots. The expression of this gene down-regulated in IR28, but did not change in CSR28. OsP5CR and OsP5CS1 showed elevated expression at both timepoints in the roots of CSR28, while a notable increase in the expression of OsP5CS1 occurred in all conditions in the shoots. The results of the expression of three genes involved in raffinose biosynthesis showed that OsRS2 had a significant increase in expression in response to salinity under all conditions, while OsNIN7 and OsEno5 were up-regulated in response to salinity in the roots, especially at the 54-h timepoint. Our findings also indicated an increase in the expression of the OsIMP-2 and OsMIOX genes involved in myoinositol biosynthesis in the roots. Among the four genes involved in glycolate metabolism, the expression of the OsGLO1 , OsGLO6 and OsPLGG1 genes significantly increased in response to salinity in the roots of CSR28 at the 54-h timepoint. Finally, key genes involved in the synthesis of antioxidant enzymes were studied, and the results showed that except of OsCatB which encodes CAT and is specifically expressed in the shoots, other genes were up-regulated in the roots. Remarkably, OsSOD-Fe and OsNCA1a exhibited a significant increase in their expression in response to salinity only in the roots of the tolerant genotype CSR28 at the 54-h timepoint. Linear regression analysis reveals the relationships between metabolites and antioxidant enzymes with their relevant genes in response to salinity stress Linear regression analysis was used to identify significant relationships between the contents of metabolites and antioxidant enzymes and their encoding genes. The results indicated that the proline and myoinositol contents were positively correlated with the expression of OsP5CS2 (R 2 = 0.81, P value = 0.03) and OsIMP (R 2 = 0.82, P value = 0.02), respectively. Among the three genes related to the CAT synthesis, only OsNCA1a (R 2 = 0.84, P value = 0.01) was significantly correlated with the enzyme content, while OsSOD-Fe (R 2 = 0.88, P value = 0.001) was positively related to the SOD content (Fig. ). Phenotypic evaluation of IR28 and CSR28 rice seedlings 1 week after exposure to high salinity stress confirmed differences in their salinity tolerance (Fig. ). The differences between the genotypes were more prominent in the shoots than in the roots. The difference in shoot length among the genotypes increased from 3.7% under control conditions to 47.1% under salinity stress. Furthermore, the difference in shoot dry weight increased from 1% in the control treatment to 57.9% in the salinity treatment. Compared with CSR28, IR28 exhibited greater reductions in both shoot length and dry weight under salinity stress. The leaf RWC of salt-stressed sensitive IR28 plants decreased significantly (23.4%) compared to that of the salt-tolerant CSR28 plants, while no significant difference was detected between the genotypes under control conditions. Brown and tubular leaves appeared in most IR28 seedlings after 1 week of salt stress, while CSR28 seedlings displayed more green leaves. The CSR28 genotype had a significantly lower mean salinity score than the IR28 genotype, which indicated that CSR28 was more salinity tolerant. 2 O 2 and MDA contents and antioxidant enzyme activity Changes in the levels of H 2 O 2 and MDA provide insights into the capacity to combat ROS and lipid peroxidation under stress. Both H 2 O 2 (Fig. a) and MDA (Fig. b) levels increased in response to salinity stress in the organs of both genotypes. However, the increases were notably more pronounced under long-term stress in the sensitive genotype. The levels of H 2 O 2 and MDA in the roots of IR28 at 54-h timepoint increased compared to CSR28 by 206.6% and 164.7%, respectively, while the increases in the shoots were 216.6% and 166.6%, respectively. The results of the antioxidant enzyme activity revealed that the levels of both CAT and SOD increased in all conditions in response to salinity. Although no significant difference between the two genotypes in both organs at the 6-h timepoint, however CAT and SOD enzyme levels in the roots of CSR28 at 54-h timepoint, were elevated compared to IR28 by 226.6% and 162.1%, respectively, while those increases in the shoots were 214.6% and 180.1%, respectively (Fig. c,d). GC‒MS analysis revealed 37 primary metabolites, including 18 amino acids (AAs), 5 sugars and sugar alcohols and 14 organic acids (OAs), in the roots and shoots of the CSR28 and IR28 genotypes at the 6-h and 54-h timepoints (the mean and standard variation values of relative metabolite levels are shown in Supplementary Table ). ANOVA revealed significant differences between 35 metabolites in roots and shoots, with an average phenotypic variation of 35.6% based on the salinity/control ratio (Supplementary Table ). Of the 37 identified metabolites, 26 presented a significant difference between two timepoints of 6 h and 54 h, explaining 8.2% of the phenotypic variation. The two genotypes displayed significant differences in 28 metabolites, which explained 9.7% of the phenotypic variation. Furthermore, the interaction of genotype × timepoint × organ was significant for 25 metabolites, which explained 4% of the phenotypic variation. The relative changes of the metabolites are shown as ratio of salinity/control in all conditions (Table ). In general, 89.3% of the metabolite changes were significant in response to salinity, of which 56.5% and 32.8% represented increased and decreased accumulation, respectively. Lactate had the smallest response to salinity under the different conditions. Amino acids (AAs) Among the 18 identified AAs, 94.4% exhibited significant changes in response to salinity stress, of which 83.3% and 11.1% exhibited increased and decreased accumulation, respectively. The greatest increases were observed for the metabolites in the salt-stressed shoots of CSR28 at the 54-h timepoint. These metabolites included isoleucine (42.8-fold), leucine (31.06-fold) and proline (36.05-fold). Compared with those in the roots of the genotypes at the 6-h timepoint, only three AAs (α-alanine, GABA and methionine) were increased in CSR28 compared to those in IR28, while 14 AAs were remarkably increased in CSR28 compared to those in IR28 at the 54-h timepoint. Furthermore, the accumulation of six and 12 AAs was greater in the shoots of CSR28 than in those of IR28 at the 6-h and 54-h timepoints, respectively (Table ). Sugars and sugar alcohols Out of the 5 sugars and sugar alcohols identified through GC‒MS analysis, 90% exhibited significant changes in response to salinity, with 55% and 35% increased and decreased accumulation, respectively. In the roots of CSR28, raffinose (45.2-fold) and fructose (-11.1-fold) had the greatest and greatest changes, respectively, at the 54-h timepoint. After 6 h of salinity treatment, the glucose and raffinose contents in the roots of IR28 were greater, and the fructose content was lower than those in the roots of CSR28. However, at the 54-h timepoint, the raffinose and myoinositol contents in CSR28 were significantly greater, and the glucose and glycerol contents were lower than those in IR28. In the shoots, the values in IR28 were greater than those in CSR28 at both timepoints (Table ). Organic acids (OAs) Among the 14 OAs identified via metabolite profiling, 84.2% of the changes were significant in response to salinity under all conditions, including 22.4% and 61.8% increased and decreased accumulation, respectively. Furthermore, the maximum (8.4-fold) and minimum (-4.9-fold) changes were due to citrate in the roots of IR28 and quinate in the shoots of CSR28 at the 54-h timepoint, respectively. After 6 h of exposure to salinity, the concentrations of six OAs in CSR28 roots were greater than those in IR28 roots, while only the hydroxyglutarate concentration in CSR28 roots was greater than that in IR28 roots. Between CSR28 and IR28, nine OAs were differentially accumulated after 54 h of salinity treatment. After 6 h of salt exposure, the contents of six OAs in the shoots of IR28 were greater than those in the shoots of CSR28, while only the fumarate content in the shoots of CSR28 was greater than that in the shoots of IR28. Under long-term stress, a greater reduction in OAs was observed in CSR28 than in IR28 (Table ). Aspartate among AAs, myo-inositol among sugars and sugar alcohols and citrate, glycerate, isocitrate and shikimate among OAs showed organ-specific accumulation and increased only in roots in response to salinity stress. Among the OAs, only α-ketoglutarate and pyruvate were specifically accumulated between the genotypes in the salt-stressed shoots, so both decreased and increased in CSR28 and IR28, respectively. Among the 18 identified AAs, 94.4% exhibited significant changes in response to salinity stress, of which 83.3% and 11.1% exhibited increased and decreased accumulation, respectively. The greatest increases were observed for the metabolites in the salt-stressed shoots of CSR28 at the 54-h timepoint. These metabolites included isoleucine (42.8-fold), leucine (31.06-fold) and proline (36.05-fold). Compared with those in the roots of the genotypes at the 6-h timepoint, only three AAs (α-alanine, GABA and methionine) were increased in CSR28 compared to those in IR28, while 14 AAs were remarkably increased in CSR28 compared to those in IR28 at the 54-h timepoint. Furthermore, the accumulation of six and 12 AAs was greater in the shoots of CSR28 than in those of IR28 at the 6-h and 54-h timepoints, respectively (Table ). Out of the 5 sugars and sugar alcohols identified through GC‒MS analysis, 90% exhibited significant changes in response to salinity, with 55% and 35% increased and decreased accumulation, respectively. In the roots of CSR28, raffinose (45.2-fold) and fructose (-11.1-fold) had the greatest and greatest changes, respectively, at the 54-h timepoint. After 6 h of salinity treatment, the glucose and raffinose contents in the roots of IR28 were greater, and the fructose content was lower than those in the roots of CSR28. However, at the 54-h timepoint, the raffinose and myoinositol contents in CSR28 were significantly greater, and the glucose and glycerol contents were lower than those in IR28. In the shoots, the values in IR28 were greater than those in CSR28 at both timepoints (Table ). Among the 14 OAs identified via metabolite profiling, 84.2% of the changes were significant in response to salinity under all conditions, including 22.4% and 61.8% increased and decreased accumulation, respectively. Furthermore, the maximum (8.4-fold) and minimum (-4.9-fold) changes were due to citrate in the roots of IR28 and quinate in the shoots of CSR28 at the 54-h timepoint, respectively. After 6 h of exposure to salinity, the concentrations of six OAs in CSR28 roots were greater than those in IR28 roots, while only the hydroxyglutarate concentration in CSR28 roots was greater than that in IR28 roots. Between CSR28 and IR28, nine OAs were differentially accumulated after 54 h of salinity treatment. After 6 h of salt exposure, the contents of six OAs in the shoots of IR28 were greater than those in the shoots of CSR28, while only the fumarate content in the shoots of CSR28 was greater than that in the shoots of IR28. Under long-term stress, a greater reduction in OAs was observed in CSR28 than in IR28 (Table ). Aspartate among AAs, myo-inositol among sugars and sugar alcohols and citrate, glycerate, isocitrate and shikimate among OAs showed organ-specific accumulation and increased only in roots in response to salinity stress. Among the OAs, only α-ketoglutarate and pyruvate were specifically accumulated between the genotypes in the salt-stressed shoots, so both decreased and increased in CSR28 and IR28, respectively. Heatmap was conducted to obtain an overview of metabolite profiling under different conditions. HCA grouped the metabolite data into two major clusters, roots and shoot, and each cluster into two distinct control and salinity stress subclusters (Fig. ). Furthermore, each subcluster was classified with respect to the timepoints of 6 h and 54 h, and each subcluster included two tolerant and sensitive genotypes. Maximum similarity of the timepoints of 6 h and 54 h was observed under the control condition in both organs, and this similarity was greater in the shoots than in the roots, while the difference between the metabolites at both timepoints increased significantly under salinity stress. The genotypes in both organs exhibited a maximum correlation under the control condition, and the difference in their metabolome increased under salinity stress. In the roots, the difference between the two genotypes at 54 h was greater than that at 6 h, while the difference in the shoots was greater at 6 h than at 54 h. In general, the correlations between the samples were as follows: genotype > timepoint > treatment > organ. To further explain the relationships between metabolite contents in response to salinity stress, the correlations between amino acids, and between organic acids and carbohydrates were analyzed (Fig. ). The results indicated that there was a significant positive correlation between the content of most amino acids, except for the correlations of glycine with putrescine ( r = − 0.91, P value = 0.001), aspartate ( r = − 0.85, P value = 0.02), asparagine ( r = − 0.82, P value = 0.02), and β-alanine ( r = − 0.81, P value = 0.04), and the correlations of putrescine with proline ( r = − 0.85, P value = 0.03) and threonine ( r = − 0.81, P value = 0.001), which were negatively correlated (Fig. a). In the correlation analysis between organic acids and carbohydrates, diverse patterns of both positive and negative correlations were observed. For example, except for glycerol, the other carbohydrates were positively correlated with each other. Furthermore, glycerol followed a pattern similar to that of organic acids such as lactate and pyruvate (Fig. b). Analysis of genes related to the accumulation of metabolites and antioxidant enzymes is highly important for understanding the synthesis of these compounds in response to salinity stress. Therefore, we focused on the key genes associated with the metabolites and antioxidant enzymes identified in this research (Fig. ). The expression of key genes involved in proline biosynthesis demonstrated that salinity stress led to the up-regulation of the genes OsP5CS2 , OsP5CR , and OsP5CS1 in most of the experimental samples. OsP5CS2 showed a significant increase in expression under all conditions except at the 6-h timepoint in the roots. The expression of this gene down-regulated in IR28, but did not change in CSR28. OsP5CR and OsP5CS1 showed elevated expression at both timepoints in the roots of CSR28, while a notable increase in the expression of OsP5CS1 occurred in all conditions in the shoots. The results of the expression of three genes involved in raffinose biosynthesis showed that OsRS2 had a significant increase in expression in response to salinity under all conditions, while OsNIN7 and OsEno5 were up-regulated in response to salinity in the roots, especially at the 54-h timepoint. Our findings also indicated an increase in the expression of the OsIMP-2 and OsMIOX genes involved in myoinositol biosynthesis in the roots. Among the four genes involved in glycolate metabolism, the expression of the OsGLO1 , OsGLO6 and OsPLGG1 genes significantly increased in response to salinity in the roots of CSR28 at the 54-h timepoint. Finally, key genes involved in the synthesis of antioxidant enzymes were studied, and the results showed that except of OsCatB which encodes CAT and is specifically expressed in the shoots, other genes were up-regulated in the roots. Remarkably, OsSOD-Fe and OsNCA1a exhibited a significant increase in their expression in response to salinity only in the roots of the tolerant genotype CSR28 at the 54-h timepoint. Linear regression analysis was used to identify significant relationships between the contents of metabolites and antioxidant enzymes and their encoding genes. The results indicated that the proline and myoinositol contents were positively correlated with the expression of OsP5CS2 (R 2 = 0.81, P value = 0.03) and OsIMP (R 2 = 0.82, P value = 0.02), respectively. Among the three genes related to the CAT synthesis, only OsNCA1a (R 2 = 0.84, P value = 0.01) was significantly correlated with the enzyme content, while OsSOD-Fe (R 2 = 0.88, P value = 0.001) was positively related to the SOD content (Fig. ). The present study assessed the responses of the roots and shoots of rice seedlings of two contrasting genotypes to high salinity. After 1 week of high salinity treatment, the length, biomass and dry weight of the IR28 shoots were lower than those of the CSR28 shoots (Fig. ). This is explained by the osmotic phase of salinity stress and consequently ionic toxicity, which accelerates the aging of older leaves and their necrosis due to salt accumulation . It seems that IR28 experienced both osmotic and ionic toxicity phases earlier and more severely. The CSR28 genotype exhibited greater growth vigor than that of the IR28 genotype under salinity stress, which indicated greater salinity tolerance. A faster growth can transfer Na + ions to shoots more slowly , . Furthermore, the rapid growth and development of cells prevent the accumulation of high salt concentrations , . The RWC, which is used to describe the water status of plant cells was significantly greater in salt-stressed CSR28 than in IR28. Numerous studies have reported that the RWC of tolerant genotypes is greater than that of sensitive ones , . An increased ability of plants to maintain water potential allows them to sustain photosynthetic activity, increase water use efficiency (WUE), and enhance their osmotic adjustment ability , . Plants exposed to salt stress undergo diverse physiological alterations . ROS such as H 2 O 2 and O 2 − are extremely reactive molecules that can accumulate at elevated levels during environmental stresses such as salt, drought, and cold, causing oxidative damage to plant cells . MDA is produced through lipid peroxidation and serves as a marker for oxidative damage in plant cell membranes induced by stress . The H 2 O 2 and MDA contents were greater in the roots and shoots of the sensitive genotype than in those of the tolerant plants in response to long-term salinity stress (Fig. a,b), which is an indication of greater oxidative stress damage in IR28. ROS-scavenging enzymes and antioxidants such as CAT and SOD play important roles in reducing oxidative stress , . In the present study, the results revealed that the tolerant genotype had greater CAT and SOD contents than that of the sensitive genotype in response to salinity stress, particularly under long-term exposure (Fig. c and d), suggesting that these enzymes play vital roles in ROS scavenging and alleviating stress. Furthermore, our findings revealed the expression of the key encoding genes of the antioxidant enzymes (Fig. ). Remarkably, linear regression analysis revealed that OsNCA1a and OsSOD-Fe had significant positive relationships with the contents of CAT and SOD enzymes, respectively (Fig. ). The GC-MS analysis revealed increased accumulation of AAs in both salt-stressed organs of the two genotypes (83.3%) in response to salinity. AAs act as osmolytes that maintain cellular turgor and protect molecules against damage caused by oxidative stresses through osmotic adjustment . In the present study, the accumulation of AAs increased in both organs after 54 h of salinity treatment, indicating that long-term salinity stress results in increased Na + accumulation and doubling of the role of the osmotic protection of AAs. The difference in AA accumulation between the two genotypes increased in both organs under long-term salinity stress. More AAs were detected in the roots of CSR28 than in those of IR28 at the 54-h timepoint (Table ), suggesting the specific role of the metabolic pathways of roots in promoting salinity tolerance. Proline, as one of the key primary metabolites, possesses antioxidant activity and protects macromolecules against ROS, along with playing the role of osmolyte and osmotic adjustment , . Proline accumulation is directly related to abiotic stress tolerance . Here, proline levels increased in response to salinity in both organs, genotypes and timepoints (Table ). The tolerant genotype CSR28 possessed greater potential for coping with osmotic challenges via proline accumulation in the shoots of CSR28 than in those of IR28. On the other hand, a significant increase in GABA was observed in response to salinity stress in the roots of the tolerant genotype CSR28 at the 6-h timepoint. GABA, a non-protein amino acid, quickly builds up in plants under stress conditions , helping to alleviate plant stress by regulating osmotic balance . In general, the results of the present study were in agreement with previous findings on the role of AAs in inducing the salinity tolerance of rice at the seedling stage , . Numerous studies have shown that genes related to proline biosynthesis are up-regulated under salt stress , . This study showed that the genes OsP5CS2 , OsP5CR , and OsP5CS1 were up-regulated in response to salinity stress under most of the experimental conditions (Fig. ). However, a significant gene-metabolite relationship was observed between the expression of the OsP5CS2 gene and the content of proline (Fig. ); therefore, this gene is considered to play a key role in increasing proline under salinity stress. The overexpression of P5CS (pyrroline-5-carboxylate synthetase 5) could increase the proline content in potato and rice and enhanced salt tolerance of plants. Furthermore, p5cs1-4 mutants exhibited strongly impaired proline accumulation in response to NaCl, suggesting that P5CS1 contributes greater to stress-induced proline accumulation . Sugars and sugar alcohols act as osmolytes and antioxidants, in addition to being resources for metabolism and structural support , . Raffinose increased in response to salinity under all conditions, especially in roots, where its maximum accumulation was observed at the 54-h timepoint in CSR28 roots (Table ). Nishizawa et al. reported that galactinol and raffinose protect plant cells against oxidative stress by scavenging hydroxyl radicals. Myo-inositol accumulated more in the roots of CSR28 under long-term salinity than in those of IR28. Using external myo-inositol in Malus hupehensis Rehd under salinity stress prevented the damage caused by salt accumulation through the support of the plant antioxidant defense system, Na + and K + ion homeostasis and osmotic balance . IMP (L-myo-inositol monophosphatase) is a key enzyme in the last process of myoinositol biosynthesis. The present study revealed a significant correlation between the myoinositol content and the gene expression of OsIMP in response to salinity stress (Fig. ). It has been reported that the overexpression of OsIMP in transgenic tobacco led to elevated inositol levels and improved cold tolerance by regulating antioxidant enzymes . Based on the assessment of primary metabolite data, although most OAs (61.8%) decreased in both salt-stressed organs, the value and pattern of their accumulation differed among organs, genotypes and timepoints. The lower reduction in OAs in the roots of both genotypes under long-term salinity stress could be due to the compensation of ionic imbalance . Increasing the amount of citrate and isocitrate anions affects the maintenance of the ionic balance caused by the excessive entrance of the toxic cations of Na + . In addition, the accumulation of OAs in roots can play a role in osmotic adjustment. The results of the present study were consistent with those of Zhao et al. . Our study showed that three genes involved in glycolate metabolism were up-regulated in the roots of CSR28 in response to salinity stress (Fig. ). Glycolate oxidase (GLO) is a key enzyme for photorespiratory metabolism in plants. The overexpression of four GLO-encoding genes has been shown in rice transgenic lines to enhance photosynthesis under conditions of high light and high temperature. Furthermore, H 2 O 2 , which can serve as a signaling molecule, was induced upon GLO overexpression . Since, H 2 O 2 and GLO were induced in the present study, we hypothesized that stress defense responses were triggered by the signaling function of H 2 O 2 cooperated with GLO gene expression. In this study, the impact of high salinity on rice genotypes was investigated at the seedling stage. The tolerant genotype (CSR28) exhibited better salt tolerance than the sensitive genotype (IR28). The osmoprotectants such as AAs and sugars increased, while OAs decreased in response to salinity stress. Strong correlations were observed between key genes and important compounds such as proline, myoinositol, CAT, and SOD under salt stress. This study highlighted the importance of gene expression and metabolomics data for understanding salt tolerance mechanisms and identified potential biomarkers for developing new salt-tolerant rice varieties. Plant materials and growth conditions Seeds of two rice ( Oryza sativa L. ssp. Indica ) genotypes with varying salt tolerances were procured from the International Rice Research Institute (IRRI) in the Philippines. The sensitive genotype IR28 was developed at the IRRI, while the tolerant genotype CSR28 (IR51485-AC6534-4) was developed at the Central Soil Salinity Research Institute (CSSRI) in Karnal, India. The plants were cultivated hydroponically in the greenhouse at Heinrich-Heine-University (HHU) in Düsseldorf, Germany. Initially, the seeds were treated with 2.5% sodium hypochlorite for sterilization and then germinated at 28 °C in the absence of light. Subsequently, the seedlings were transplanted into 4-liter pots containing Yoshida culture medium and were grown under a light regime of 14 h light and 10 h dark at a temperature of 28 ± 2 °C. The culture medium at a pH of 5.5 was replaced every 3 days. After 2 weeks, the seedlings were subjected to 150 mM (15 dS/m) NaCl. The roots and shoots of both the untreated and salt-treated plants were collected at 6 h, 54 h, and 1 week after salt treatment. Phenotypic evaluations of salinity tolerance To evaluate the salinity tolerance of the IR28 and CSR28 genotypes, the length and the fresh and dry weights of roots and shoots, the leaf relative water content (RWC) and the salinity tolerance scores were assessed (in three replications of five seedlings each) 1 week after 150 mM salt treatment. Root and shoot dry weight Dry weight was determined after placing the samples in a 72 °C oven for 48 h. Leaf RWC Leaf RWC was calculated for the youngest fully developed leaves with the following equation. [12pt]{minimal} $${}\; = 100$$ where FW, DW and TW represent the fresh, dry, and turgid weights, respectively. Salt score 20 seedlings subjected to salinity treatment for 1 week were used to score the salinity tolerance of genotypes based on the method of Gregoria et al. in which 1, 3, 5, 7 and 9 refer to very tolerant (normal growth), tolerant (relatively normal growth), relatively tolerant (delayed growth), sensitive (completely stopped growth) and very sensitive (death of all plants), respectively. Determination of H 2 O 2 and MDA contents and antioxidant enzyme activity The H 2 O 2 and malondialdehyde (MDA) contents act as ROS and are indicators of stressful environments. The H 2 O 2 content in the root and shoot samples was measured using the method described by Ghiazdowska et al. . The method of Heath and Pacher was used to measure MDA as measure of lipid peroxidation. Catalase (CAT) and superoxide dismutase (SOD) are essential antioxidant enzymes that are required for ROS scavenging when plants experience salt stress. The CAT and SOD contents were measured according to the methods described by Scebba et al. and Giannopolitis , respectively. Metabolite profiling The topmost parts of the plants were harvested (from five replications of 10 plants each) after 6 h and 54 h of salinity treatment, shock-frozen in liquid nitrogen and stored at 80 °C until further processing. The samples were ground in a mortar and freeze-dried. 10 mg of lyophilized material were extracted with 1.5 ml of a water, methanol and chloroform (1:2.5:1, v/v) mixture including 5 µM ribitol as an internal standard and stored at − 20 °C. GC–MS analysis was conducted using protocols adapted from Lisec et al. and Gu et al. , as described previously by Shim et al. . For relative quantification, the peak areas of all metabolites were normalized against sample weight and the peak area of the internal standard ribitol, which was added prior to extraction buffer. Quantitative real-time PCR (qRT-PCR) analysis qRT-PCR analysis was used to evaluate the expression of key genes encoding metabolites and antioxidant enzymes. Total RNA from the control and stressed samples was extracted with a P-Biozol kit (manufactured by Bio Flux-Bioer, Tokyo, Japan). Spectrophotometry and agarose gel electrophoresis were used to determine the quantity and quality of the extracted RNA after DNaseI treatment. cDNA was synthesized from 1 µg of total RNA by a cDNA reverse transcription kit (Applied Biosystems, California, USA), according to the manufacturer’s protocol. The primers (Table ) were designed with Primer Express v3.0 software (Applied Biosystems, Foster City, CA). qRT-PCR analysis of three biological and two technical replicates was performed with an iCycler iQ5 thermocycler (Bio Rad Company) and SYBR Green I (SBP, Iran). All reactions were performed with the default parameters. The expression level of each gene was normalized to that of the internal control gene, elongation factor 1 alpha ( OseEF-1a ). The method of 2 −ΔΔCT and log 2 fold change (FC) were used to calculate the relative expression as a salinity/control ratio. The statistical significance of the ratios was considered to be │Log2 FC│≥ 1 and P value ≤ 0.05 (as calculated by Student’s t -test). Data analysis The salinity tolerance of the two genotypes was assessed using phenotypic and physiological data through Student’s t-test ( P value ≤ 0.05). The ROS and antioxidant contents and metabolite profiles were analyzed as factorial in a completely randomized design (CRD), and the significance level was tested using ANOVA in SAS v9.2 software. Relative metabolite abundances in roots and shoots of the two genotypes were compared at two timepoints based on the ratio of salinity/control, and their significance was determined using Student’s t-test. Furthermore, the means were compared through Duncan’s Multiple Range test ( P value ≤ 0.05). MeV v4.9.0 (Multiple Experiment Viewer) software was used for heatmap and cluster analysis. We also used the “cor” function in R to calculate the Pearson correlation coefficient of metabolites with a threshold greater than 0.80 and a P value < 0.05. A linear regression analysis through R was performed to determine whether there was a significant relationship ( P value ≤ 0.05) between metabolites and antioxidant enzymes with their corresponding coding genes. Seeds of two rice ( Oryza sativa L. ssp. Indica ) genotypes with varying salt tolerances were procured from the International Rice Research Institute (IRRI) in the Philippines. The sensitive genotype IR28 was developed at the IRRI, while the tolerant genotype CSR28 (IR51485-AC6534-4) was developed at the Central Soil Salinity Research Institute (CSSRI) in Karnal, India. The plants were cultivated hydroponically in the greenhouse at Heinrich-Heine-University (HHU) in Düsseldorf, Germany. Initially, the seeds were treated with 2.5% sodium hypochlorite for sterilization and then germinated at 28 °C in the absence of light. Subsequently, the seedlings were transplanted into 4-liter pots containing Yoshida culture medium and were grown under a light regime of 14 h light and 10 h dark at a temperature of 28 ± 2 °C. The culture medium at a pH of 5.5 was replaced every 3 days. After 2 weeks, the seedlings were subjected to 150 mM (15 dS/m) NaCl. The roots and shoots of both the untreated and salt-treated plants were collected at 6 h, 54 h, and 1 week after salt treatment. To evaluate the salinity tolerance of the IR28 and CSR28 genotypes, the length and the fresh and dry weights of roots and shoots, the leaf relative water content (RWC) and the salinity tolerance scores were assessed (in three replications of five seedlings each) 1 week after 150 mM salt treatment. Root and shoot dry weight Dry weight was determined after placing the samples in a 72 °C oven for 48 h. Leaf RWC Leaf RWC was calculated for the youngest fully developed leaves with the following equation. [12pt]{minimal} $${}\; = 100$$ where FW, DW and TW represent the fresh, dry, and turgid weights, respectively. Salt score 20 seedlings subjected to salinity treatment for 1 week were used to score the salinity tolerance of genotypes based on the method of Gregoria et al. in which 1, 3, 5, 7 and 9 refer to very tolerant (normal growth), tolerant (relatively normal growth), relatively tolerant (delayed growth), sensitive (completely stopped growth) and very sensitive (death of all plants), respectively. 2 O 2 and MDA contents and antioxidant enzyme activity The H 2 O 2 and malondialdehyde (MDA) contents act as ROS and are indicators of stressful environments. The H 2 O 2 content in the root and shoot samples was measured using the method described by Ghiazdowska et al. . The method of Heath and Pacher was used to measure MDA as measure of lipid peroxidation. Catalase (CAT) and superoxide dismutase (SOD) are essential antioxidant enzymes that are required for ROS scavenging when plants experience salt stress. The CAT and SOD contents were measured according to the methods described by Scebba et al. and Giannopolitis , respectively. The topmost parts of the plants were harvested (from five replications of 10 plants each) after 6 h and 54 h of salinity treatment, shock-frozen in liquid nitrogen and stored at 80 °C until further processing. The samples were ground in a mortar and freeze-dried. 10 mg of lyophilized material were extracted with 1.5 ml of a water, methanol and chloroform (1:2.5:1, v/v) mixture including 5 µM ribitol as an internal standard and stored at − 20 °C. GC–MS analysis was conducted using protocols adapted from Lisec et al. and Gu et al. , as described previously by Shim et al. . For relative quantification, the peak areas of all metabolites were normalized against sample weight and the peak area of the internal standard ribitol, which was added prior to extraction buffer. qRT-PCR analysis was used to evaluate the expression of key genes encoding metabolites and antioxidant enzymes. Total RNA from the control and stressed samples was extracted with a P-Biozol kit (manufactured by Bio Flux-Bioer, Tokyo, Japan). Spectrophotometry and agarose gel electrophoresis were used to determine the quantity and quality of the extracted RNA after DNaseI treatment. cDNA was synthesized from 1 µg of total RNA by a cDNA reverse transcription kit (Applied Biosystems, California, USA), according to the manufacturer’s protocol. The primers (Table ) were designed with Primer Express v3.0 software (Applied Biosystems, Foster City, CA). qRT-PCR analysis of three biological and two technical replicates was performed with an iCycler iQ5 thermocycler (Bio Rad Company) and SYBR Green I (SBP, Iran). All reactions were performed with the default parameters. The expression level of each gene was normalized to that of the internal control gene, elongation factor 1 alpha ( OseEF-1a ). The method of 2 −ΔΔCT and log 2 fold change (FC) were used to calculate the relative expression as a salinity/control ratio. The statistical significance of the ratios was considered to be │Log2 FC│≥ 1 and P value ≤ 0.05 (as calculated by Student’s t -test). The salinity tolerance of the two genotypes was assessed using phenotypic and physiological data through Student’s t-test ( P value ≤ 0.05). The ROS and antioxidant contents and metabolite profiles were analyzed as factorial in a completely randomized design (CRD), and the significance level was tested using ANOVA in SAS v9.2 software. Relative metabolite abundances in roots and shoots of the two genotypes were compared at two timepoints based on the ratio of salinity/control, and their significance was determined using Student’s t-test. Furthermore, the means were compared through Duncan’s Multiple Range test ( P value ≤ 0.05). MeV v4.9.0 (Multiple Experiment Viewer) software was used for heatmap and cluster analysis. We also used the “cor” function in R to calculate the Pearson correlation coefficient of metabolites with a threshold greater than 0.80 and a P value < 0.05. A linear regression analysis through R was performed to determine whether there was a significant relationship ( P value ≤ 0.05) between metabolites and antioxidant enzymes with their corresponding coding genes. Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 |
Use of Feeding Tubes Among Hospitalized Older Adults With Dementia | 2bdf087a-6de7-471e-b2a8-13a6639f42ef | 11843365 | Surgical Procedures, Operative[mh] | Dementia refers to a group of disorders related to cognitive decline and neurodegeneration, ultimately affecting everyday functioning. , Older individuals with more advanced stages of dementia often develop issues with chewing and swallowing. , , A previous study reported 86% of nursing home residents experienced eating problems, while a scoping review reported prevalence of dysphagia as high as 93%. Consequently, family caregivers and health care professionals may need to decide whether to use a percutaneous endoscopic gastrostomy (PEG) feeding tube, which is inserted into the stomach through an incision in the abdomen. However, use of PEG tubes poses risks of developing pressure ulcers and aspiration pneumonia , , and has been considered of low value among individuals with dementia, with previous work showing these interventions were not associated with improved quality of life or lengthened survival time. , , , , , , , , , , Previous studies primarily focused on nursing home residents in the US with advanced dementia and reported that feeding tube use was associated with no improvements in health, with worse survival, and with greater use of health care resources. , , , , , Notably, current guidelines from the Canadian Geriatrics Society, American Geriatrics Society, European Society for Clinical Nutrition and Metabolism, and the Canadian Choosing Wisely Campaign clearly recommend that PEG tubes should not be offered to individuals with advanced dementia. However, many previous studies were limited to individuals with advanced dementia, and no previous population-based study, to our knowledge, has examined feeding tube use and outcomes among community-dwelling individuals with dementia. Anecdotally, feeding tube placement still occurs for individuals with dementia, despite existing recommendations, yet gaps exist in the population-level incidence of feeding tube use and factors associated with its use. Given these knowledge gaps, we sought to describe the incidence of feeding tube placement among hospitalized older adults (aged ≥65 years) with dementia in Ontario, Canada, regardless of dementia severity or place of residence. We also aimed to verify whether individuals with dementia who received a feeding tube (henceforth referred to as feeding tube recipients ) had better or worse in-hospital and postdischarge outcomes than those who did not (henceforth referred to as feeding tube nonrecipients ). Finally, we sought to identify factors associated with receipt of feeding tubes among the subgroups of hospitalized older adults with dementia who were receiving home care or admitted to long-term care (LTC) to identify reasons for feeding tube placement despite lack of demonstrated benefit. Study Design, Setting, and Data Sources We conducted a population-based retrospective cohort study in Ontario using linked databases held at ICES, an independent nonprofit research institute whose legal status under section 45 of Ontario’s Personal Health Information Protection Act allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement. Given this, this study did not require further ethics approval. Our study population included all adults with dementia 65 years or older at the time of admission to an acute care hospital and discharged between April 1, 2014, and March 31, 2018. Individuals’ first hospitalization within the study period served as their index hospitalization, with date of admission as the index date. Individuals were excluded if they were ineligible for the Ontario Health Insurance Plan (OHIP) at the index date or had missing or invalid birth date, sex, or death date. Detailed information about each dataset is provided in eTable 1 in . This study followed the Strengthening the Reporting of Observational Studies in Epidemiology ( STROBE ) reporting guideline. Physician-diagnosed dementia was identified using a previously validated algorithm and data on physician claims from the OHIP; prescription claims from the Ontario Drug Benefit; hospitalizations from the Canadian Institute of Health Information’s Discharge and Abstract database; same-day surgical procedures from the Canadian Institute of Health Same Day Surgery database; and age from the Registered Persons Database. eMethods in presents detailed definitions. We identified 2 subcohorts who were receiving publicly funded home care or LTC based on receipt of a Resident Assessment Instrument in Home Care (RAI-HC) or Resident Assessment Instrument in Long-Term Care (RAI-MDS) within the 6 months prior to the index hospitalization. In Ontario, the RAI-HC is a standardized routine assessment provided to home-care recipients every 6 months, and the RAI-MDS is a standardized routine assessment provided to LTC residents every 3 months. They assess individuals’ functional status, health profiles, service use, advance directives, and caregiver information (eTable 2 in ). The assessment closest to the admission date of their index hospitalization was selected. The interrater reliabilities of items on both assessments are high, and the disease diagnoses items have been validated. , , Receipt of Feeding Tubes Receipt of a feeding tube was defined using a combination of Canadian Classification of Health Interventions codes and OHIP fee codes billed by physicians indicating insertion of a gastrostomy, gastrostomy-jejunostomy, or jejunostomy tube during index hospitalization. Further details are given in eTable 3 in . Cohort Characteristics We described the following characteristics of all individuals in this cohort: age at admission, sex, neighborhood income quintile (based on postal codes linked to the 2016 census data), and rurality (defined by Statistics Canada as living in a rural or small community with a population size <10 000 persons outside the commuting zone of larger urban centers). Among those who received an RAI-HC or RAI-MDS assessment, we also reported the following: marital status, presence of chewing or swallowing problems, relationship with their primary caregiver (only RAI-HC), and do-not-resuscitate (DNR) and do-not-hospitalize (DNH) orders (only RAI-MDS). We also reported their health profiles and functional and cognitive status as measured by the Activities of Daily Living Self-Performance Hierarchy Scale (ADL-H; scores range from 0-6), the Cognitive Performance Scale (scores range from 0-6), and the Changes in Health, End-Stage Disease and Signs and Symptoms Scale (CHESS; scores range from 0-5), which is a measure of health instabilities and an indicator of being at the end of life. Details of these scales have been previously reported and found to be valid , , , and reliable. , , Higher scores on these measures indicate higher levels of impairment or instability. Characteristics of the Hospitalization and Outcomes Post Hospitalization We examined the following characteristics captured during individuals’ index hospitalization: (1) the proportion admitted to the intensive care unit (ICU) during hospitalization and mean length of stay (in days) in the ICU; (2) among those who received a feeding tube, mean time from admission to receipt of feeding tube and hospital length of stay (in days); and (3) discharge disposition, specifically whether individuals died in hospital or were discharged home, to the ICU, or to inpatient care. We reported 14-, 30-, 90-, 180-, and 360-day mortality rates from admission date of index hospitalization. Among those who survived until discharge from index hospitalization, we examined rates of mortality and rehospitalization or emergency department visits within 1 year post discharge. Statistical Analysis Data were analyzed between October 2021 and November 2024. We used descriptive statistics (ie, mean [SD] and median [IQR] for continuous variables and proportions for categorical variables) to describe the cohort characteristics (ie, individuals’ sociodemographic characteristics, health, and functioning prior to hospitalization) and the main outcomes of our descriptive analyses (hospitalization characteristics and outcomes post hospitalization). We performed sensitivity analyses stratifying these descriptive analyses by home care vs LTC subcohorts and by CHESS scores of 3 or greater (indicative of high health instability) vs less than 3. We performed logistic regression with receipt of feeding tubes as the outcome, separately for subcohorts of older adults who were receiving home care (RAI-HC) or admitted to LTC (RAI-MDS), to examine the associations between patient and care-related characteristics and their likelihood of receiving a feeding tube. Factors included in the regressions were prespecified, that is, chosen if there was evidence of associations with the receipt of feeding tubes in the literature or identified as potentially relevant by clinical experts. For both the home care and LTC subcohorts, we included individuals’ sociodemographic characteristics (eg, sex, age) and their health and functioning (eg, ADL-H, Cognitive Performance Scale) in the modeling. As the RAI-HC and RAI-MDS differed in assessment items, the following factors were included in only one of the models: marital status (only included from the RAI-HC due to the large amount of nonrandom missingness in the RAI-MDS) and DNR and DNH advance directives (only RAI-MDS). Age (in 5-year groups) was entered as a continuous variable since a plot of the restricted cubic spline of age against the logit of the outcome showed a linear association. All factors were kept in the model since no collinearity issues were detected using the SAS VARCLUS procedure. Statistical tests were 2 tailed with P < .05 indicating statistical significance. We also performed sensitivity analyses that added the comorbidities of stroke, cancer, chronic obstructive pulmonary disorder, chronic coronary syndrome, and acute myocardial infarction into both regressions. All analyses were performed using SAS Enterprise, version 9.4 (SAS Institute Inc). We conducted a population-based retrospective cohort study in Ontario using linked databases held at ICES, an independent nonprofit research institute whose legal status under section 45 of Ontario’s Personal Health Information Protection Act allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement. Given this, this study did not require further ethics approval. Our study population included all adults with dementia 65 years or older at the time of admission to an acute care hospital and discharged between April 1, 2014, and March 31, 2018. Individuals’ first hospitalization within the study period served as their index hospitalization, with date of admission as the index date. Individuals were excluded if they were ineligible for the Ontario Health Insurance Plan (OHIP) at the index date or had missing or invalid birth date, sex, or death date. Detailed information about each dataset is provided in eTable 1 in . This study followed the Strengthening the Reporting of Observational Studies in Epidemiology ( STROBE ) reporting guideline. Physician-diagnosed dementia was identified using a previously validated algorithm and data on physician claims from the OHIP; prescription claims from the Ontario Drug Benefit; hospitalizations from the Canadian Institute of Health Information’s Discharge and Abstract database; same-day surgical procedures from the Canadian Institute of Health Same Day Surgery database; and age from the Registered Persons Database. eMethods in presents detailed definitions. We identified 2 subcohorts who were receiving publicly funded home care or LTC based on receipt of a Resident Assessment Instrument in Home Care (RAI-HC) or Resident Assessment Instrument in Long-Term Care (RAI-MDS) within the 6 months prior to the index hospitalization. In Ontario, the RAI-HC is a standardized routine assessment provided to home-care recipients every 6 months, and the RAI-MDS is a standardized routine assessment provided to LTC residents every 3 months. They assess individuals’ functional status, health profiles, service use, advance directives, and caregiver information (eTable 2 in ). The assessment closest to the admission date of their index hospitalization was selected. The interrater reliabilities of items on both assessments are high, and the disease diagnoses items have been validated. , , Receipt of a feeding tube was defined using a combination of Canadian Classification of Health Interventions codes and OHIP fee codes billed by physicians indicating insertion of a gastrostomy, gastrostomy-jejunostomy, or jejunostomy tube during index hospitalization. Further details are given in eTable 3 in . We described the following characteristics of all individuals in this cohort: age at admission, sex, neighborhood income quintile (based on postal codes linked to the 2016 census data), and rurality (defined by Statistics Canada as living in a rural or small community with a population size <10 000 persons outside the commuting zone of larger urban centers). Among those who received an RAI-HC or RAI-MDS assessment, we also reported the following: marital status, presence of chewing or swallowing problems, relationship with their primary caregiver (only RAI-HC), and do-not-resuscitate (DNR) and do-not-hospitalize (DNH) orders (only RAI-MDS). We also reported their health profiles and functional and cognitive status as measured by the Activities of Daily Living Self-Performance Hierarchy Scale (ADL-H; scores range from 0-6), the Cognitive Performance Scale (scores range from 0-6), and the Changes in Health, End-Stage Disease and Signs and Symptoms Scale (CHESS; scores range from 0-5), which is a measure of health instabilities and an indicator of being at the end of life. Details of these scales have been previously reported and found to be valid , , , and reliable. , , Higher scores on these measures indicate higher levels of impairment or instability. We examined the following characteristics captured during individuals’ index hospitalization: (1) the proportion admitted to the intensive care unit (ICU) during hospitalization and mean length of stay (in days) in the ICU; (2) among those who received a feeding tube, mean time from admission to receipt of feeding tube and hospital length of stay (in days); and (3) discharge disposition, specifically whether individuals died in hospital or were discharged home, to the ICU, or to inpatient care. We reported 14-, 30-, 90-, 180-, and 360-day mortality rates from admission date of index hospitalization. Among those who survived until discharge from index hospitalization, we examined rates of mortality and rehospitalization or emergency department visits within 1 year post discharge. Data were analyzed between October 2021 and November 2024. We used descriptive statistics (ie, mean [SD] and median [IQR] for continuous variables and proportions for categorical variables) to describe the cohort characteristics (ie, individuals’ sociodemographic characteristics, health, and functioning prior to hospitalization) and the main outcomes of our descriptive analyses (hospitalization characteristics and outcomes post hospitalization). We performed sensitivity analyses stratifying these descriptive analyses by home care vs LTC subcohorts and by CHESS scores of 3 or greater (indicative of high health instability) vs less than 3. We performed logistic regression with receipt of feeding tubes as the outcome, separately for subcohorts of older adults who were receiving home care (RAI-HC) or admitted to LTC (RAI-MDS), to examine the associations between patient and care-related characteristics and their likelihood of receiving a feeding tube. Factors included in the regressions were prespecified, that is, chosen if there was evidence of associations with the receipt of feeding tubes in the literature or identified as potentially relevant by clinical experts. For both the home care and LTC subcohorts, we included individuals’ sociodemographic characteristics (eg, sex, age) and their health and functioning (eg, ADL-H, Cognitive Performance Scale) in the modeling. As the RAI-HC and RAI-MDS differed in assessment items, the following factors were included in only one of the models: marital status (only included from the RAI-HC due to the large amount of nonrandom missingness in the RAI-MDS) and DNR and DNH advance directives (only RAI-MDS). Age (in 5-year groups) was entered as a continuous variable since a plot of the restricted cubic spline of age against the logit of the outcome showed a linear association. All factors were kept in the model since no collinearity issues were detected using the SAS VARCLUS procedure. Statistical tests were 2 tailed with P < .05 indicating statistical significance. We also performed sensitivity analyses that added the comorbidities of stroke, cancer, chronic obstructive pulmonary disorder, chronic coronary syndrome, and acute myocardial infarction into both regressions. All analyses were performed using SAS Enterprise, version 9.4 (SAS Institute Inc). Cohort Characteristics During the study period, 143 331 individuals with dementia met the inclusion criteria (83 536 [58.3%] female and 59 795 [41.7%] male; mean [SD] age, 83.8 [7.5] years). describes the characteristics of our cohort, among whom 1312 (0.9%) received a PEG feeding tube insertion during their hospitalization. Feeding tube recipients were younger than nonrecipients (mean [SD] age, 80.8 [7.6] vs 83.8 [7.5] years). Male individuals were more likely than female individuals to receive a feeding tube (722 [1.2%] vs 590 [0.7%]). Residents in urban settings were also more likely to receive a feeding tube than those in rural settings (1238 [1.0%] vs 67 [0.4%]). The proportions of individuals with various comorbidities based on receipt of a feeding tube are provided in eTable 4 in . The main reasons for hospitalization among feeding tube recipients are provided in eTable 5 in . Characteristics of Feeding Tube Recipients in Home Care or in LTC shows the sociodemographic, health, and care-related characteristics of individuals residing in the community and receiving home care or LTC at the time of hospitalization who received a feeding tube. Among 42 441 home care recipients (ie, assessed with the RAI-HC), 383 (0.9%) received a feeding tube, of whom 101 (26.4%) had a moderate level of ADL impairment and 109 (28.5%) had a maximal level (indicating moderate to high levels of functional dependency), and 217 (56.7%) had moderate to total cognitive impairment. Nearly one-quarter of these individuals had moderate to high health instability (90 [23.5%]) as indicated by a CHESS score of 3 or greater; about one-third were widowed, separated, or divorced (140 [36.6%]); and most relied on a spouse (178 [46.5%]) or a child (168 [43.9%]) as their primary caregiver and did not have chewing (335 [87.5%]) or swallowing problems (218 [56.9%]) prior to hospitalization. Among the 38 028 LTC residents (ie, assessed with the RAI-MDS), 452 (1.2%) received a feeding tube, of whom 158 (35.0%) had moderate levels of ADL impairment and 129 (28.5%) had maximal levels, and 342 (75.7%) had moderate to total cognitive impairment. Only 30 (6.6%) had moderate to high health instability (CHESS score of ≥3). Prior to hospitalization, about one-third of these individuals did not have chewing problems (158 [35.0%]), and half did not have swallowing problems (223 [49.3%]). A total of 229 (50.7%) had in place DNR advanced directives and 76 (16.8%) had in place DNH advanced directives. Inpatient Hospitalization Characteristics shows the characteristics of the hospitalization among all individuals in our study, by receipt of feeding tube. Feeding tube recipients (compared with nonrecipients) were 4 times more often admitted into the ICU (557 [42.5%] vs 14 423 [10.2%]), stayed 6 times longer in the ICU (mean [SD], 26.8 [48.0] vs 4.3 [8.1] days), and stayed more than 4 times longer in hospital (mean [SD], 65.6 [120.8] vs 14.8 [35.2] days). Deaths in hospital occurred in 294 feeding tube recipients (22.4%), compared with 14 698 nonrecipients (10.3%). Within 1 year of hospital admission, 743 feeding tube recipients (56.6%) died, compared with 49 987 nonrecipients (35.2%). Mortality and Rehospitalization Patterns After Discharge and the eFigure in show the mortality and rehospitalization patterns of survivors of the hospitalization. Mortality and rehospitalization rates were consistently higher among feeding tube recipients at all time points within 1 year of discharge from hospital. At 360 days post discharge, 509 of 1018 feeding tube recipients (50.0%) had died, compared with 36 162 of 127 321 nonrecipients (28.4%), while 512 of 1018 feeding tube recipients (50.3%) were rehospitalized, compared with 54 575 of 127 321 nonrecipients (43.0%). Sensitivity analyses on mortality and rehospitalization for the home care and LTC subcohorts as well as by CHESS scores showed similar patterns (eTables 6-9 in ). Multivariable Analysis presents results of 2 logistic regression models (model 1 among home care recipients; model 2 among LTC residents) examining factors associated with the likelihood of receiving a feeding tube in hospital. Among home care recipients, being female (odds ratio [OR], 0.66; 95% CI, 0.52-0.84), older (OR for every 5-year increase in age, 0.75; 95% CI, 0.70-0.81), widowed (OR, 0.66; 95% CI, 0.47-0.94), and living in rural settings (OR, 0.38; 95% CI, 0.22-0.66) were associated with reduced odds of receiving a feeding tube. In contrast, having an ADL-H score of 5 or 6 (OR, 2.75; 95% CI, 1.80-4.20) or swallowing problems (OR, 2.22; 95% CI, 1.99-2.49) was associated with higher odds of receiving a feeding tube. Similar patterns were observed for LTC residents. Being female (OR, 0.77; 95% CI, 0.63-0.94), older (OR for every 5-year increase in age, 0.82; 95% CI, 0.77-0.87), and living in rural settings (OR, 0.51; 95% CI, 0.31-0.83) also were associated with reduced odds of receiving a feeding tube, while having an ADL-H score of 5 or 6 (OR, 3.52; 95% CI, 1.10-11.27) or swallowing problems (OR, 2.29; 95% CI, 1.81-2.89) were associated with increased odds. Additionally, having a DNR order was associated with reduced odds of receiving a feeding tube (OR, 0.38; 95% CI, 0.31-0.47), while having chewing problems was associated with increased odds (OR, 1.88; 95% CI, 1.47-2.40). Other factors were not associated with and did not increase the likelihood of receiving a feeding tube. Our sensitivity analyses (eTable 10 in ) showed that none of the comorbidities was statistically significant, nor did the comorbidities impact the associations with other factors. During the study period, 143 331 individuals with dementia met the inclusion criteria (83 536 [58.3%] female and 59 795 [41.7%] male; mean [SD] age, 83.8 [7.5] years). describes the characteristics of our cohort, among whom 1312 (0.9%) received a PEG feeding tube insertion during their hospitalization. Feeding tube recipients were younger than nonrecipients (mean [SD] age, 80.8 [7.6] vs 83.8 [7.5] years). Male individuals were more likely than female individuals to receive a feeding tube (722 [1.2%] vs 590 [0.7%]). Residents in urban settings were also more likely to receive a feeding tube than those in rural settings (1238 [1.0%] vs 67 [0.4%]). The proportions of individuals with various comorbidities based on receipt of a feeding tube are provided in eTable 4 in . The main reasons for hospitalization among feeding tube recipients are provided in eTable 5 in . shows the sociodemographic, health, and care-related characteristics of individuals residing in the community and receiving home care or LTC at the time of hospitalization who received a feeding tube. Among 42 441 home care recipients (ie, assessed with the RAI-HC), 383 (0.9%) received a feeding tube, of whom 101 (26.4%) had a moderate level of ADL impairment and 109 (28.5%) had a maximal level (indicating moderate to high levels of functional dependency), and 217 (56.7%) had moderate to total cognitive impairment. Nearly one-quarter of these individuals had moderate to high health instability (90 [23.5%]) as indicated by a CHESS score of 3 or greater; about one-third were widowed, separated, or divorced (140 [36.6%]); and most relied on a spouse (178 [46.5%]) or a child (168 [43.9%]) as their primary caregiver and did not have chewing (335 [87.5%]) or swallowing problems (218 [56.9%]) prior to hospitalization. Among the 38 028 LTC residents (ie, assessed with the RAI-MDS), 452 (1.2%) received a feeding tube, of whom 158 (35.0%) had moderate levels of ADL impairment and 129 (28.5%) had maximal levels, and 342 (75.7%) had moderate to total cognitive impairment. Only 30 (6.6%) had moderate to high health instability (CHESS score of ≥3). Prior to hospitalization, about one-third of these individuals did not have chewing problems (158 [35.0%]), and half did not have swallowing problems (223 [49.3%]). A total of 229 (50.7%) had in place DNR advanced directives and 76 (16.8%) had in place DNH advanced directives. shows the characteristics of the hospitalization among all individuals in our study, by receipt of feeding tube. Feeding tube recipients (compared with nonrecipients) were 4 times more often admitted into the ICU (557 [42.5%] vs 14 423 [10.2%]), stayed 6 times longer in the ICU (mean [SD], 26.8 [48.0] vs 4.3 [8.1] days), and stayed more than 4 times longer in hospital (mean [SD], 65.6 [120.8] vs 14.8 [35.2] days). Deaths in hospital occurred in 294 feeding tube recipients (22.4%), compared with 14 698 nonrecipients (10.3%). Within 1 year of hospital admission, 743 feeding tube recipients (56.6%) died, compared with 49 987 nonrecipients (35.2%). and the eFigure in show the mortality and rehospitalization patterns of survivors of the hospitalization. Mortality and rehospitalization rates were consistently higher among feeding tube recipients at all time points within 1 year of discharge from hospital. At 360 days post discharge, 509 of 1018 feeding tube recipients (50.0%) had died, compared with 36 162 of 127 321 nonrecipients (28.4%), while 512 of 1018 feeding tube recipients (50.3%) were rehospitalized, compared with 54 575 of 127 321 nonrecipients (43.0%). Sensitivity analyses on mortality and rehospitalization for the home care and LTC subcohorts as well as by CHESS scores showed similar patterns (eTables 6-9 in ). presents results of 2 logistic regression models (model 1 among home care recipients; model 2 among LTC residents) examining factors associated with the likelihood of receiving a feeding tube in hospital. Among home care recipients, being female (odds ratio [OR], 0.66; 95% CI, 0.52-0.84), older (OR for every 5-year increase in age, 0.75; 95% CI, 0.70-0.81), widowed (OR, 0.66; 95% CI, 0.47-0.94), and living in rural settings (OR, 0.38; 95% CI, 0.22-0.66) were associated with reduced odds of receiving a feeding tube. In contrast, having an ADL-H score of 5 or 6 (OR, 2.75; 95% CI, 1.80-4.20) or swallowing problems (OR, 2.22; 95% CI, 1.99-2.49) was associated with higher odds of receiving a feeding tube. Similar patterns were observed for LTC residents. Being female (OR, 0.77; 95% CI, 0.63-0.94), older (OR for every 5-year increase in age, 0.82; 95% CI, 0.77-0.87), and living in rural settings (OR, 0.51; 95% CI, 0.31-0.83) also were associated with reduced odds of receiving a feeding tube, while having an ADL-H score of 5 or 6 (OR, 3.52; 95% CI, 1.10-11.27) or swallowing problems (OR, 2.29; 95% CI, 1.81-2.89) were associated with increased odds. Additionally, having a DNR order was associated with reduced odds of receiving a feeding tube (OR, 0.38; 95% CI, 0.31-0.47), while having chewing problems was associated with increased odds (OR, 1.88; 95% CI, 1.47-2.40). Other factors were not associated with and did not increase the likelihood of receiving a feeding tube. Our sensitivity analyses (eTable 10 in ) showed that none of the comorbidities was statistically significant, nor did the comorbidities impact the associations with other factors. Our population-based retrospective cohort study examined characteristics associated with receipt of a feeding tube among hospitalized older adults with dementia in Ontario. Overall, 0.9% of hospitalized individuals with dementia received a feeding tube, which is similar to previous findings on the incidence of feeding tube use among individuals with advanced dementia. Compared with a study on older adults at a geriatric medical center in the US that found the incidence of enteral feeding use to be 1.35%, the incidence in our study of individuals with dementia was lower, possibly in (at least partial) compliance with guidelines. Feeding tube recipients were 4 times more likely to have been admitted to the ICU, had a mean length of stay of longer than 2 months (4 times longer than nonrecipients), and were more than twice as likely to die in hospital than nonrecipients, suggesting that feeding tube recipients likely had greater frailty than nonrecipients. We found that large proportions of feeding tube recipients had severe cognitive and functional impairments, and more than half of those in LTC also had a DNH order prior to admission. These findings show that a substantial portion of feeding tube recipients were likely incapable of making decisions, leaving decision making to their family caregivers and physicians. Past studies have shown that, despite guidelines and clinical judgments against feeding tube insertion, substitute decision-makers and even physicians continue to hold misperception of the efficacy of tube feeding in improving patient outcomes. , , , This is exacerbated by the lack of education and conversations on feeding tube use or other options for patients and their families , ; a study on family caregivers of individuals who died with dementia showed that more than half of the caregivers reported no or less than 15 minutes of discussion regarding feeding tube insertion. Physicians also often experience a lack of control in this decision. Most physicians surveyed in a study by Gieniusz et al stated that families often request feeding tube insertion even when physicians counsel against it, and many physicians have concerns about potential litigation. Advanced care planning and education on possible adverse results and alternatives for feeding are particularly important and recommended for these individuals and their families. , Consistent with most previous studies, , , , , , , , , , most of which focused on patients with advanced dementia, our study found no survival benefit of feeding tube insertion in hospitalized patients with dementia, even without restricting to an advanced dementia stage. We found that older adults with dementia who received a feeding tube had higher rates of mortality, rehospitalization, and admission to the emergency department post discharge, in both the short and long term, which concur with past findings. , , , , , , Given high and increased rates of disability and mortality, questions of appropriateness—including consideration of goal-concordant care—should be explored prior to feeding tube insertion in hospitalized patients with dementia. While we acknowledge that a feeding tube may be the appropriate intervention in some cases, the consistently demonstrated lack of benefits to survival and health outcomes for feeding tube recipients suggests that they may benefit more through a holistic palliative care approach rather than this invasive intervention. , Appropriate less invasive alternatives do exist, such as assisted feeding interventions (eg, hand feeding, postural changes to head and/or body position, dietary modification including specialized dysphagia diet or high-calorie supplement), which constitute a part of palliative care. Focused education for the clinical care teams, as well as individualized and patient-centered discussions, could help individuals and their families make informed decisions about appropriate care options. There may also be opportunities for protocol changes in clinical settings. For example, referrals for feeding tube insertion for individuals with dementia could automatically trigger a palliative care consultation to initiate the aforementioned discussions and establish goals of care. We also found that being male, having high functional impairment, being younger, and having swallowing problems were associated with increased odds of receiving a feeding tube, while having a DNR order and living in rural areas were associated with reduced odds. Some findings are consistent with previous studies; for example, it has previously been reported that prehospitalization dysphagia increased the odds of feeding tube placement. Given that some variables (eg, sex, urbanicity) should not be associated with medical indications for feeding tube insertion, further consideration should be given to the potential impacts of sociocultural considerations and system-level barriers (eg, constraints on access to necessary resources and medical care in rural settings) on clinical decision-making processes. Furthermore, we found no association for some factors that could be expected to impact feeding tube insertion (eg, level of cognitive impairment, DNH order). This points to a potential need for the use of decision aids to enable informed decision making by both clinicians and patients and/or their caregivers. There is clear evidence that decision aids can help make information explicit, provide information about options and their associated benefits and harms, and identify congruence between decisions and personal values. Limitations To our knowledge, this study is the first to use comprehensive population-based health care data to examine factors and outcomes associated with receiving a feeding tube among all hospitalized patients with dementia in Ontario. However, our study has several limitations. First, our regression models, which required data from RAI assessments, were only performed on data from individuals in home care or LTC. While this may have limited the generalizability of our findings to the broader population with dementia, this allowed us to examine many factors associated with individuals’ health, functioning, and care needs that were previously not addressed. We also acknowledge there may be residual confounding in our modeling; for example, we did not have data on individuals’ race or ethnicity, which was previously identified to be associated with receipt of a feeding tube. A recent study also demonstrated that racial and cultural differences exist in the perception of feeding tube use, including misconceptions about their effectiveness and the clinical course of dementia. Finally, there may be misclassifications of individuals’ functional, health, and care-related status, since the assessments were conducted within 6 months of hospitalization and these statuses could have changed in the time between assessment and hospitalization. We note this limitation since a sudden change in these statuses, such as developing chewing and swallowing problems or a rapid deterioration in health, could have provoked the hospitalization. Such misclassification would likely bias the effects of these factors toward the null, which could result in factors such as higher CHESS scores to seemingly have no effect on receipt of a feeding tube. To our knowledge, this study is the first to use comprehensive population-based health care data to examine factors and outcomes associated with receiving a feeding tube among all hospitalized patients with dementia in Ontario. However, our study has several limitations. First, our regression models, which required data from RAI assessments, were only performed on data from individuals in home care or LTC. While this may have limited the generalizability of our findings to the broader population with dementia, this allowed us to examine many factors associated with individuals’ health, functioning, and care needs that were previously not addressed. We also acknowledge there may be residual confounding in our modeling; for example, we did not have data on individuals’ race or ethnicity, which was previously identified to be associated with receipt of a feeding tube. A recent study also demonstrated that racial and cultural differences exist in the perception of feeding tube use, including misconceptions about their effectiveness and the clinical course of dementia. Finally, there may be misclassifications of individuals’ functional, health, and care-related status, since the assessments were conducted within 6 months of hospitalization and these statuses could have changed in the time between assessment and hospitalization. We note this limitation since a sudden change in these statuses, such as developing chewing and swallowing problems or a rapid deterioration in health, could have provoked the hospitalization. Such misclassification would likely bias the effects of these factors toward the null, which could result in factors such as higher CHESS scores to seemingly have no effect on receipt of a feeding tube. In this cohort study of hospitalized individuals with dementia in Ontario, only 0.9% received a feeding tube; however, this still represented more than 1000 insertions. Many feeding tube recipients showed high levels of cognitive and functional impairments and had higher mortality and rehospitalization rates than those who did not receive a feeding tube. We found no associations for certain factors that could be expected to have impacted feeding tube insertion, while factors such as sex and urbanicity were associated. These findings showed a clear need for effective and timely conversations around goals of care and alternate intervention options, improved policy and clinical protocols, and potential use of decision aids to help the care team as well as the patients and their families make informed decisions. |
Tumor Staging of Endocervical Adenocarcinoma: Recommendations From the International Society of Gynecological Pathologists | d9a43b40-368d-4c8d-9d94-78a08fabf550 | 7969160 | Gynaecology[mh] | In 2018, the FIGO revised the staging system for carcinoma of the uterine cervix, for the first time allowing incorporation of imaging and/or pathologic findings in the assessment of tumor size and disease extent . There were a number of reasons for this according to FIGO, including the technological advances in imaging modalities, and the fact that the pathological findings were sometimes not concordant with clinical staging, leading to an increase in para-aortic lymph node sampling to determine the need for extended field radiation. The major changes to the revised FIGO 2018 staging include (1) the removal of the 7 mm horizontal extent as a parameter for stage IA carcinoma; (2) dividing stage IB into 3 subgroups (IB1, IB2, IB3); (3) including lymph node involvement in stage assignment (stage IIIC1 with only pelvic lymph node involvement, stage IIIC2 with para-aortic nodal involvement); (4) allowing the use of imaging to assess lymph nodes and (5) denoting “p” for pathologic or “r” for radiologic to indicate the method used to derive the stage. The intent was to allow the staging system to be applicable to all resource levels, including low- and middle-income countries, acknowledging the concern that access to imaging modalities or surgico-pathologic documentation of disease extent may not always be feasible. Unfortunately, there were errors in the original publication of the new FIGO staging in the International Journal of Gynecology and Obstetrics . The “=” in measurement cut-offs was erroneously placed in 12 instances in stages I and II (Table ), and there were also several statements that were confusing or vague. These were pointed out by users and in particular by members of the ICCR data set authoring committee on cervical cancer reporting. As a result, a corrigendum was subsequently published with corrections , . The key amendments to the staging are delineated in Table , and the FIGO response to most of the queries raised by ICCR are listed in Table . A recent manuscript by Salvo et al. highlights various flaws in the revised FIGO 2018 staging and particularly points to the gaps in pathologic evaluation of cervical cancers. In the cervix, as in other gynecologic organs, both FIGO and TNM [Union for International Cancer Control (UICC) or American Joint Committee on Cancer (AJCC)] staging systems exist. With regard to updating of staging systems, there is collaboration between FIGO and those agencies responsible for TNM with an agreement to adopt FIGO staging but there is no coordination of timing of revisions and generally following the introduction of a new FIGO staging system, this is incorporated into TNM (both UICC and AJCC versions) at a later date. Apart from minor discrepancies in terminology, the UICC and AJCC systems are broadly concurrent. Tumor size is an important parameter in tumor staging which dictates treatment and patient prognosis. There are several difficult issues regarding tumor size measurements that pathologists often face. With a grossly visible tumor, which measurements should be reported and used for staging purposes? Macroscopic, microscopic or a combination? Opening a surgical specimen of the cervix longitudinally may demonstrate a maximum dimension that is not evident clinically or radiologically, especially in tumors which circumferentially involve the cervix. Should this measurement supersede the maximum tumor diameter if this is the greater of the 2 measurements? In cases with multiple specimens containing tumor (loop(s), cone(s), trachelectomy, hysterectomy), how does one incorporate the measurements from the various specimens to give the most accurate size? How does one measure “depth of invasion” in purely or mostly exophytic tumors with no or limited stromal infiltration? How should multifocal carcinomas be measured and reported? How many dimensions of the tumor should be reported? What is meant by the term microinvasive carcinoma? In lesions composed of an admixture of adenocarcinoma in situ (AIS)/cervical glandular intraepithelial neoplasia (CGIN, terminology used in United Kingdom and some other jurisdictions) and adenocarcinoma, it can be difficult to delineate invasive from noninvasive tumor. In these cases, should the whole lesion be measured, or an attempt made to separate only the invasive component? When tumor is present at the margin of a loop or cone excision, how should the tumor be staged? Measuring Grossly Visible Tumors The purpose of providing tumor measurements is ultimately to accurately stage for treatment planning and prognostication. Therefore, whatever measurements are provided should give the treating physician enough information without adding extraneous information. In the 2009 FIGO staging, grossly visible tumors were automatically allotted to stage IB, regardless of tumor size or depth of invasion. This led to neoplasms being upstaged inappropriately in some small tumors and those with only superficial invasion. The revised 2018 FIGO system now states that although a tumor is clinically visible, final stage should be based on all the information available at the time, which includes pathology and radiology, with pathologic findings being the ultimate arbiter of stage. In those situations where pathology and radiology are not available clinical examination should be used to stage tumors. Therefore, we recommend using all available information to best determine true tumor size, which pathologically may require a combination of gross and microscopic measurements. Separate pathologic gross and microscopic measurements should not be provided but a single set of measurements based on a combination of gross and microscopic examination; in some cases, gross examination may be more important (for example in larger neoplasms), while in others microscopic examination is more important and many smaller tumors can only be measured microscopically since they are not grossly visible. Multiple Specimens It is often the case that the patient undergoes multiple procedures—loop excision(s), cone(s), trachelectomy, and hysterectomy with tumor in more than 1 specimen. In these situations, the important question arises of how best to measure tumor size and depending on the method used this may result in a different stage being assigned with important management implications. Some (eg, the ICCR) advocate adding all maximum horizontal dimensions from each specimen, while others recommend using the largest horizontal measurement in any one specimen. Each situation poses its own problems. If adding all horizontal dimensions across multiple specimens, it is not possible to accurately align where 1 edge of the tumor from 1 specimen lines up with the correct edge on a different specimen. This will almost certainly result in an overestimation of tumor size. Given that the new FIGO staging no longer uses horizontal extent as a criterion for staging microscopic tumors, this is perhaps less of an issue. If only the largest horizontal dimension in any 1 specimen is used, this may underestimate the maximum horizontal dimension. Horizontal extent can be reported as additive across multiple specimens, recognizing that this likely results in overestimating tumor size, or only the current size can be reported with a comment on the size in prior specimens. This also applies to multifocal tumors across multiple specimens, as can be seen in adenocarcinoma (see multifocal section below). In such cases, discussion at the tumor board/multidisciplinary team meeting should determine the final tumor stage. It is recommended that depth of invasion should be reported as measured on each specimen but for final staging purposes, the deepest invasion in any single specimen should be used. Exophytic Tumors In purely or predominantly exophytic adenocarcinomas, the tumor can be quite large yet have minimal cervical wall infiltration (Fig. ). This results in 2 separate dimensions: (1) tumor thickness as measured from the top-most part of the tumor to the bottom-most tip of the infiltrating front or base of the tumor if there is no stromal infiltration; this may constitute the maximum tumor dimension but does NOT equate to invasive depth, and (2) depth of invasion measured from the normal epithelial-stromal junction to the base of the infiltrating front if stromal infiltration is present. In the survey conducted by ISGyP, there was marked variability with more pathologists reporting the actual depth of invasion (only the invasion in the cervical wall) than the tumor thickness with or without the depth of invasion. As recommended by FIGO, both maximum tumor dimension and depth of stromal infiltration should be provided in the pathology report, with a comment detailing how each measurement was derived. For exophytic tumors with cervical wall infiltration of ≤5 mm (stage IA), but tumor thickness >5 mm, we recommend basing the final FIGO stage on the maximum tumor thickness, that is stage IB. Recording both measurements will facilitate the accrual of information regarding the clinical importance of tumor thickness and depth of invasion in these uncommon neoplasms. The thickness of the cervical wall in the area of deepest invasion should also be provided. Although we recommend staging such tumors based on the thickness and equating it with depth of invasion for staging purposes, the clinical significance should be discussed at the tumor board/multidisciplinary team meeting; although there is limited evidence thus far, tumors with minimal stromal invasion may not have the same metastatic potential as tumors of similar thickness invading the stroma . Multifocal Tumors Multifocal carcinomas are invasive tumor foci that occur discontinuously, separated by uninvolved stroma. The distance between invasive foci that constitutes multifocality is not clearly defined or agreed upon, but some have suggested any foci >2 mm apart (the blocks should be leveled to ensure that the separate foci do not join) should be considered multifocal; these criteria have been endorsed by the ICCR , , . The ISGyP survey results show how varied pathologists are in determining multifocality. Some use 2 mm separation, others 5 mm, while still others require the involvement of different quadrants or cervical lips. Anecdotally, multifocality is less common in cervical adenocarcinomas as compared with squamous cell carcinomas and with the new FIGO system eliminating the horizontal extent as part of the staging, it may have different clinical implications. Without specific studies in adenocarcinoma, one must extrapolate from the squamous cell carcinoma literature. In two small, retrospective studies evaluating outcomes in multifocal squamous cell carcinomas defined as being at least 2 mm apart, Day et al. and McIlwaine et al. found that there were no recurrences or metastases in a total of 47 cases with median follow-up periods of 7 yr and 45.5 mo, respectively, after excisions with clear margins. McIlwaine and colleagues pointed out that half of their cases would have been upstaged to IB1 had the tumors been measured contiguously rather than as separate foci. This provides some evidence for eliminating horizontal extent for staging purposes in microscopic disease. Based on these limited findings and the revised FIGO staging, the recommendation is in line with that of the ICCR which is to measure each invasive focus individually if they are (1) located in different blocks separated by intervening uninvolved blocks; (2) located on separate cervical lips with discontinuous tumor (not involving the curvature of the canal); (3) situated apart from each other in the same section using 2 mm distance between the invasive foci. Only the depth of invasion should be used for staging purposes with the deepest invasion reported and the tumor staged accordingly. An important point is leveling blocks to ensure that the multiple foci do not “join up” to form a contiguous mass. It should also be noted than the 2 mm designation is completely arbitrary and this is an area which requires more study. Multiple specimens with multifocal disease should be treated similarly. How Many Tumor Dimensions Should Be Reported? Regarding the number of tumor dimensions to be reported, as per FIGO recommendations, most pathologists report at least 2 dimensions—the deepest invasion and the largest tumor dimension. This provides adequate information for tumor staging which best informs treatment planning. The third dimension can still be reported if local protocols mandate this (and is currently recommended by the ICCR) but this is unnecessary since tumor volume is not routinely taken into account in patient management. Circumferential tumors that result in a “barrel” shaped cervix can be difficult to measure since they are sometimes not grossly visible yet involve all quadrants. Measuring each section with tumor in a linear manner would grossly overestimate tumor size; therefore, it is recommended that such tumors be measured grossly if possible and if not, the diameter of the cervix be used as the closest approximation of tumor size in these situations. This only applies to tumors that invade the full thickness of the cervical wall and involve all quadrants. In tumors that do not invade the full thickness of the cervix and/or do not involve all quadrants, only the depth of cervical wall infiltration and the largest tumor dimension as measured histologically should be provided, with a comment on the extent of tumor (eg, how many quadrants are involved) and the absence of a grossly visible tumor to measure. Occasionally, radiology may provide the most accurate tumor size and should be incorporated into final staging when appropriate. In addition to reporting the absolute depth of invasion, the total thickness of the cervical wall in the area of deepest invasion should be reported in conjunction such that the percentage of stromal infiltration can be assessed and the presence of tumor within the inner, middle or outer third of the cervical stroma determined. This is to help guide the use of Sedlis criteria for adjuvant external pelvic radiation therapy following radical hysterectomy based on certain high-risk features (lymphovascular invasion, depth of stromal invasion by thirds, tumor size) . Microinvasive Carcinoma The term “microinvasive carcinoma” does not appear in the FIGO staging system for cervical cancer. Furthermore, use of the term “microinvasive carcinoma” has different connotations in different geographical areas. For example, in the United Kingdom, microinvasive carcinoma was considered to be synonymous with FIGO stage IA1 and IA2 disease in most, but not all, institutions (some used the term “microinvasive carcinoma” to denote only FIGO stage IA1 tumors). Thus, in order to avoid confusion, it is recommended to avoid using the term “microinvasive carcinoma” but to accurately measure the tumor and use the specific FIGO or TNM stage. Measurement of Tumors that Are a Combination of AIS and Adenocarcinoma The assessment of tumors that are a combination of AIS and adenocarcinoma where it is difficult to delineate the 2 components should follow the guidance in another review in the series (Alvarado-Cabrero). Measurement of these lesions is somewhat analogous to that of exophytic tumors and should incorporate maximum horizontal tumor dimension and an assessment of invasive depth as accurately as possible, with these measurements forming the basis of staging. Where necessary this may be accompanied by a comment detailing the diagnostic problems; the distinction of in situ from invasive disease in such cases can be extremely subjective and problematic and this is an area where referral for a specialist opinion may be useful. Tumor at Margins of Excision Specimen When tumor is present at the margins of a loop or cone excision, the true size of the tumor cannot be assessed accurately. In the new FIGO system, there is a statement that tumors that reach the margin should be staged as IB based on the positive margin. This was queried by ICCR and FIGO conceded that in these cases a repeat excision would be required to accurately assign a stage to the cancer. Therefore, it is recommended tumors should not be staged IB based only on positive margins. In most such cases, there will be a subsequent excision specimen and the stage should be determined based on the findings in all specimens. As an alternative, one can give a provisional stage, for example “at least stage x based on the measurements of the incomplete excision.” The purpose of providing tumor measurements is ultimately to accurately stage for treatment planning and prognostication. Therefore, whatever measurements are provided should give the treating physician enough information without adding extraneous information. In the 2009 FIGO staging, grossly visible tumors were automatically allotted to stage IB, regardless of tumor size or depth of invasion. This led to neoplasms being upstaged inappropriately in some small tumors and those with only superficial invasion. The revised 2018 FIGO system now states that although a tumor is clinically visible, final stage should be based on all the information available at the time, which includes pathology and radiology, with pathologic findings being the ultimate arbiter of stage. In those situations where pathology and radiology are not available clinical examination should be used to stage tumors. Therefore, we recommend using all available information to best determine true tumor size, which pathologically may require a combination of gross and microscopic measurements. Separate pathologic gross and microscopic measurements should not be provided but a single set of measurements based on a combination of gross and microscopic examination; in some cases, gross examination may be more important (for example in larger neoplasms), while in others microscopic examination is more important and many smaller tumors can only be measured microscopically since they are not grossly visible. It is often the case that the patient undergoes multiple procedures—loop excision(s), cone(s), trachelectomy, and hysterectomy with tumor in more than 1 specimen. In these situations, the important question arises of how best to measure tumor size and depending on the method used this may result in a different stage being assigned with important management implications. Some (eg, the ICCR) advocate adding all maximum horizontal dimensions from each specimen, while others recommend using the largest horizontal measurement in any one specimen. Each situation poses its own problems. If adding all horizontal dimensions across multiple specimens, it is not possible to accurately align where 1 edge of the tumor from 1 specimen lines up with the correct edge on a different specimen. This will almost certainly result in an overestimation of tumor size. Given that the new FIGO staging no longer uses horizontal extent as a criterion for staging microscopic tumors, this is perhaps less of an issue. If only the largest horizontal dimension in any 1 specimen is used, this may underestimate the maximum horizontal dimension. Horizontal extent can be reported as additive across multiple specimens, recognizing that this likely results in overestimating tumor size, or only the current size can be reported with a comment on the size in prior specimens. This also applies to multifocal tumors across multiple specimens, as can be seen in adenocarcinoma (see multifocal section below). In such cases, discussion at the tumor board/multidisciplinary team meeting should determine the final tumor stage. It is recommended that depth of invasion should be reported as measured on each specimen but for final staging purposes, the deepest invasion in any single specimen should be used. In purely or predominantly exophytic adenocarcinomas, the tumor can be quite large yet have minimal cervical wall infiltration (Fig. ). This results in 2 separate dimensions: (1) tumor thickness as measured from the top-most part of the tumor to the bottom-most tip of the infiltrating front or base of the tumor if there is no stromal infiltration; this may constitute the maximum tumor dimension but does NOT equate to invasive depth, and (2) depth of invasion measured from the normal epithelial-stromal junction to the base of the infiltrating front if stromal infiltration is present. In the survey conducted by ISGyP, there was marked variability with more pathologists reporting the actual depth of invasion (only the invasion in the cervical wall) than the tumor thickness with or without the depth of invasion. As recommended by FIGO, both maximum tumor dimension and depth of stromal infiltration should be provided in the pathology report, with a comment detailing how each measurement was derived. For exophytic tumors with cervical wall infiltration of ≤5 mm (stage IA), but tumor thickness >5 mm, we recommend basing the final FIGO stage on the maximum tumor thickness, that is stage IB. Recording both measurements will facilitate the accrual of information regarding the clinical importance of tumor thickness and depth of invasion in these uncommon neoplasms. The thickness of the cervical wall in the area of deepest invasion should also be provided. Although we recommend staging such tumors based on the thickness and equating it with depth of invasion for staging purposes, the clinical significance should be discussed at the tumor board/multidisciplinary team meeting; although there is limited evidence thus far, tumors with minimal stromal invasion may not have the same metastatic potential as tumors of similar thickness invading the stroma . Multifocal carcinomas are invasive tumor foci that occur discontinuously, separated by uninvolved stroma. The distance between invasive foci that constitutes multifocality is not clearly defined or agreed upon, but some have suggested any foci >2 mm apart (the blocks should be leveled to ensure that the separate foci do not join) should be considered multifocal; these criteria have been endorsed by the ICCR , , . The ISGyP survey results show how varied pathologists are in determining multifocality. Some use 2 mm separation, others 5 mm, while still others require the involvement of different quadrants or cervical lips. Anecdotally, multifocality is less common in cervical adenocarcinomas as compared with squamous cell carcinomas and with the new FIGO system eliminating the horizontal extent as part of the staging, it may have different clinical implications. Without specific studies in adenocarcinoma, one must extrapolate from the squamous cell carcinoma literature. In two small, retrospective studies evaluating outcomes in multifocal squamous cell carcinomas defined as being at least 2 mm apart, Day et al. and McIlwaine et al. found that there were no recurrences or metastases in a total of 47 cases with median follow-up periods of 7 yr and 45.5 mo, respectively, after excisions with clear margins. McIlwaine and colleagues pointed out that half of their cases would have been upstaged to IB1 had the tumors been measured contiguously rather than as separate foci. This provides some evidence for eliminating horizontal extent for staging purposes in microscopic disease. Based on these limited findings and the revised FIGO staging, the recommendation is in line with that of the ICCR which is to measure each invasive focus individually if they are (1) located in different blocks separated by intervening uninvolved blocks; (2) located on separate cervical lips with discontinuous tumor (not involving the curvature of the canal); (3) situated apart from each other in the same section using 2 mm distance between the invasive foci. Only the depth of invasion should be used for staging purposes with the deepest invasion reported and the tumor staged accordingly. An important point is leveling blocks to ensure that the multiple foci do not “join up” to form a contiguous mass. It should also be noted than the 2 mm designation is completely arbitrary and this is an area which requires more study. Multiple specimens with multifocal disease should be treated similarly. Regarding the number of tumor dimensions to be reported, as per FIGO recommendations, most pathologists report at least 2 dimensions—the deepest invasion and the largest tumor dimension. This provides adequate information for tumor staging which best informs treatment planning. The third dimension can still be reported if local protocols mandate this (and is currently recommended by the ICCR) but this is unnecessary since tumor volume is not routinely taken into account in patient management. Circumferential tumors that result in a “barrel” shaped cervix can be difficult to measure since they are sometimes not grossly visible yet involve all quadrants. Measuring each section with tumor in a linear manner would grossly overestimate tumor size; therefore, it is recommended that such tumors be measured grossly if possible and if not, the diameter of the cervix be used as the closest approximation of tumor size in these situations. This only applies to tumors that invade the full thickness of the cervical wall and involve all quadrants. In tumors that do not invade the full thickness of the cervix and/or do not involve all quadrants, only the depth of cervical wall infiltration and the largest tumor dimension as measured histologically should be provided, with a comment on the extent of tumor (eg, how many quadrants are involved) and the absence of a grossly visible tumor to measure. Occasionally, radiology may provide the most accurate tumor size and should be incorporated into final staging when appropriate. In addition to reporting the absolute depth of invasion, the total thickness of the cervical wall in the area of deepest invasion should be reported in conjunction such that the percentage of stromal infiltration can be assessed and the presence of tumor within the inner, middle or outer third of the cervical stroma determined. This is to help guide the use of Sedlis criteria for adjuvant external pelvic radiation therapy following radical hysterectomy based on certain high-risk features (lymphovascular invasion, depth of stromal invasion by thirds, tumor size) . The term “microinvasive carcinoma” does not appear in the FIGO staging system for cervical cancer. Furthermore, use of the term “microinvasive carcinoma” has different connotations in different geographical areas. For example, in the United Kingdom, microinvasive carcinoma was considered to be synonymous with FIGO stage IA1 and IA2 disease in most, but not all, institutions (some used the term “microinvasive carcinoma” to denote only FIGO stage IA1 tumors). Thus, in order to avoid confusion, it is recommended to avoid using the term “microinvasive carcinoma” but to accurately measure the tumor and use the specific FIGO or TNM stage. The assessment of tumors that are a combination of AIS and adenocarcinoma where it is difficult to delineate the 2 components should follow the guidance in another review in the series (Alvarado-Cabrero). Measurement of these lesions is somewhat analogous to that of exophytic tumors and should incorporate maximum horizontal tumor dimension and an assessment of invasive depth as accurately as possible, with these measurements forming the basis of staging. Where necessary this may be accompanied by a comment detailing the diagnostic problems; the distinction of in situ from invasive disease in such cases can be extremely subjective and problematic and this is an area where referral for a specialist opinion may be useful. When tumor is present at the margins of a loop or cone excision, the true size of the tumor cannot be assessed accurately. In the new FIGO system, there is a statement that tumors that reach the margin should be staged as IB based on the positive margin. This was queried by ICCR and FIGO conceded that in these cases a repeat excision would be required to accurately assign a stage to the cancer. Therefore, it is recommended tumors should not be staged IB based only on positive margins. In most such cases, there will be a subsequent excision specimen and the stage should be determined based on the findings in all specimens. As an alternative, one can give a provisional stage, for example “at least stage x based on the measurements of the incomplete excision.” Parametrial/Paracervical Soft Tissue Involvement Extrauterine involvement by cervical cancer is usually stage II or higher (see the next section on Adnexal involvement) and parametrial involvement is stage IIB. As formally defined, the parametrium consists of the connective tissue surrounding branches of the hypogastric vessels during their course toward the uterus and vagina and are located lateral to the uterus. There are occasions where the tumor extends beyond the cervix in the anterior or posterior plane, which technically is not encompassed in the anatomic delineation of parametria. Since any tumor that invades beyond the uterus without involving the lower third of the vagina or extending to the pelvic side wall is considered stage II, any anterior or posterior extension of tumor beyond the cervix should be staged as IIB since it is analogous to parametrial involvement for management purposes. It should be noted that parametrial lymph node involvement should be designated as stage IIIC1 in the revised FIGO system. Adnexal Involvement FIGO 2018 explicitly states that ovarian involvement by cervical carcinoma does not change the stage. There are limited data on the prognostic implications of tubo-ovarian involvement, and it is a rare event but more common in adenocarcinoma than squamous cell carcinoma. In early stage cervical cancer, the incidence of metastases to the ovaries is <1% for squamous cell carcinoma and <5% for adenocarcinoma – . In addition, since it is often associated with other high-risk factors, there are limited data on its impact on survival as an independent risk factor. One study by Shimada et al. showed that patients with ovarian metastases had poor outcomes unrelated to FIGO stage and the presence of ovarian metastases did not correlate with lymph node involvement or parametrial invasion, suggesting it may be an independent prognostic factor. It is likely that the clinical and prognostic implications of adnexal involvement differ between HPV-associated and HPV-independent cervical adenocarcinomas, and also depending on the route of spread and the pattern/extent of involvement – . There still are not enough robust data and in line with the FIGO directive, it is recommended to report ovarian and tubal involvement but not to alter stage based solely on their involvement. It is also the recommendation to specify the pattern of involvement in the fallopian tubes (mucosal epithelial, mucosal stromal, mural, serosal) as this will allow for prospective data collection for future studies to assess their impact on outcomes. Extrauterine involvement by cervical cancer is usually stage II or higher (see the next section on Adnexal involvement) and parametrial involvement is stage IIB. As formally defined, the parametrium consists of the connective tissue surrounding branches of the hypogastric vessels during their course toward the uterus and vagina and are located lateral to the uterus. There are occasions where the tumor extends beyond the cervix in the anterior or posterior plane, which technically is not encompassed in the anatomic delineation of parametria. Since any tumor that invades beyond the uterus without involving the lower third of the vagina or extending to the pelvic side wall is considered stage II, any anterior or posterior extension of tumor beyond the cervix should be staged as IIB since it is analogous to parametrial involvement for management purposes. It should be noted that parametrial lymph node involvement should be designated as stage IIIC1 in the revised FIGO system. FIGO 2018 explicitly states that ovarian involvement by cervical carcinoma does not change the stage. There are limited data on the prognostic implications of tubo-ovarian involvement, and it is a rare event but more common in adenocarcinoma than squamous cell carcinoma. In early stage cervical cancer, the incidence of metastases to the ovaries is <1% for squamous cell carcinoma and <5% for adenocarcinoma – . In addition, since it is often associated with other high-risk factors, there are limited data on its impact on survival as an independent risk factor. One study by Shimada et al. showed that patients with ovarian metastases had poor outcomes unrelated to FIGO stage and the presence of ovarian metastases did not correlate with lymph node involvement or parametrial invasion, suggesting it may be an independent prognostic factor. It is likely that the clinical and prognostic implications of adnexal involvement differ between HPV-associated and HPV-independent cervical adenocarcinomas, and also depending on the route of spread and the pattern/extent of involvement – . There still are not enough robust data and in line with the FIGO directive, it is recommended to report ovarian and tubal involvement but not to alter stage based solely on their involvement. It is also the recommendation to specify the pattern of involvement in the fallopian tubes (mucosal epithelial, mucosal stromal, mural, serosal) as this will allow for prospective data collection for future studies to assess their impact on outcomes. FIGO 2018 addresses the issue of LVSI in parametrial tissue and states that tumors should not be upstaged based only on parametrial vascular invasion. In general, when tumor is present only in vessels outside the cervix (eg, parametrial, adnexal, etc.), this should not be considered involvement of that tissue and should not alter the stage, although it should be mentioned in the pathology report (Fig. ). The issue of quantifying the number of involved vessels has been a recent topic of interest in gynecologic pathology, particularly in endometrial carcinomas where extent of LVSI has been found to be a strong independent prognostic factor for recurrence and overall survival – . Comparable, but less numerous, studies in cervical cancer have shown the importance of LVSI in outcome, although quantification of LVSI was not generally studied – . One study by Alvarado-Cabrero et al. evaluating micropapillary architecture in cervical adenocarcinomas (a marker of aggressive behavior) quantified LVSI as negative, low (1–4 spaces involved), moderate (5–19 spaces involved) and extensive (≥20 spaces involved) and found significant association with overall survival (100%, 51.9%, and 0% for low, moderate, and extensive LVSI, respectively). However, micropapillary type endocervical adenocarcinomas are aggressive neoplasms which characteristically exhibit extensive LVSI and it remains to be proven in future studies if quantifying LVSI is clinically significant in cervical carcinoma, including adenocarcinoma, and what values, if any, should be used as cut-offs. It is recommended at this time that LVSI be reported as present or absent, but that quantification is not necessary, although this can be done locally which may facilitate accrual of information for future studies. One of the major changes to the revised FIGO 2018 system is the incorporation of lymph node involvement in assigning stage. It allows for assessment of retroperitoneal lymph nodes by imaging and/or pathologic findings and if deemed metastatic, is designated stage IIIC1 (pelvic/parametrial lymph node involvement) or IIIC2 (para-aortic lymph node involvement). Sentinel lymph nodes are now commonly removed in cervical carcinoma and have been incorporated into the NCCN guidelines – . Tumor involvement of lymph nodes, in line with the recommendations in TNM8, is reported as: negative, isolated tumor cells (ITCs, ≤0.2 mm), micrometastases (>0.2 and ≤2.0 mm), and macrometastases (>2 mm). Studies have shown that the presence of micrometastases in patients with early stage cervical cancer is significantly associated with reduced overall survival, equivalent to macrometastases, while no prognostic significance has generally been found for ITCs , . A statement in the original publication of the revised FIGO staging stated that the presence of micrometastases or ITCs may be recorded but “their presence does not change the stage.” The ICCR group queried FIGO regarding the designation of micrometastases in the same category as ITCs and FIGO did respond that this was an error and made the corrections in the subsequent corrigendum. In line with other organ systems, it is recommended that the presence of ITCs should be recorded but this does not count as nodal involvement, and should not result in tumor upstaging, designated pN0(i+) in TNM. The issue of how to measure low-volume metastases in lymph nodes can be problematic. There are occasions where single cells and small clusters (≤0.2 mm) are present scattered throughout the entire lymph node (and therefore spans >0.2 mm in aggregate) yet the tumor foci are not contiguous suggesting they are not the same volumetrically as solid metastatic foci. The recommendation is to measure only contiguous tumor cells to designate ITCs and not aggregate all scattered clusters into a single measurement; however, if multiple collections of ITCs are present, this should be documented in the pathology report. The number of lymph nodes harboring ITCs/micro/macrometastases should also be recorded. When there is nodal involvement, the presence or absence of extracapsular/extranodal spread should be documented . The revised FIGO 2018 staging system includes major changes which better incorporate clinical, pathologic, and radiologic data, as well as including lymph node involvement in assigning stage. The horizontal measurement in microscopic disease is no longer a factor in assigning stage, which may have alleviated some of the problematic issues in tumor measurements. There are still controversial issues in pathologic reporting of cervical cancer, including adenocarcinoma, and we have provided recommendations for best practice based on the available evidence and sometimes on expert opinion. It is hoped that the ISGyP international data collection initiative will provide provisional answers to some of the remaining questions and form the basis for future prospective validation. It is further hoped that in future revisions of the FIGO staging system, international pathology organizations, such as ISGyP, will be more actively involved. |
Comparative physiological and proteomic analysis indicates lower shock response to drought stress conditions in a self-pollinating perennial ryegrass | a8ffb2fc-7981-436e-97f9-b35cccbbd604 | 7302502 | Physiology[mh] | Cool-season grasses, emerging from and adapted to cool climatic conditions are economically and ecologically among the most important species due to their critical role in CO 2 assimilation as well as forage production for livestock farming utilization . Perennial ryegrass ( Lolium perenne L.) is the most frequently cultivated species considered properly for farming as it combines high growth rates and supplies the nutritive value of fodder. However, Norris and Turgeon mentioned that perennial ryegrass is a drought-susceptible grass species, although there seems to be a noticeable genotypic variation for drought stress responses. Drought stress is the main abiotic element limiting plant growth and crop output throughout the world , which is roughly limiting plant production in 25% of the world's land. Due to its geographical location, Iran has an arid (65%) to semi-arid (25%) climate condition and is generally considered a dry country. Therefore, drought stress is the most critical and common environmental stress that limits agricultural production in the country and reduces the efficiency of using semi-arid and rain-fed areas. In cool-season grass species, drought stress adversely affects leaf water potential, photosynthetic and carbohydrate accumulation, respiration, production of reactive oxygen species, lipid peroxidation, and denaturation of proteins (reviewed by Loka et al. ). However, many cool-season grasses can resist moderate drought and maintain their aerial growth to avoid and tolerate leaf dehydration. Also, they may be equipped with the strategies facilitating their survival of severe drought, mainly associated with dehydration avoidance and tolerance, which primarily occurs in meristematic tissues . In some species and genotypes of grasses, summer dormancy is also another strategy that confers efficient survival of meristematic tissues through dehydration avoidance and tolerance . The grass family (Poaceae) exhibits different reproductive strategies among more than 10,000 species, some promoting self-pollination, and others cross-pollination, due to mechanisms such as self-incompatibility and being dioecious . Lolium perenne is commonly considered as an out-crossing species . This breeding system maintains a high genetic variation within the progenies and keeps the plasticity of the variety against environmental instabilities . However, this will also prohibit the repeatability of results in scientific studies since keeping the background of genetic material is impossible. Recently, Torkian et al. reported that some of the Iranian perennial ryegrasses have appropriate reproductive characters of seed production as well as turf characteristics, which are also self-pollinating. Therefore, they may be considered as a new source of germplasm for future turfgrass breeding programs. Before this, it needs to confirm the potential of this newly emerged material, especially their performance in drought stress environments. Under drought stress conditions, numerous alterations are induced in plants' physiological and molecular characteristics. Physiological characters of plants, particularly those related to plant water status, have a great deal of importance in the growth and development of plants during water-stress conditions. In the present study, relative water content (RWC) has been used as an example of the characters which determine drought-stress tolerance of Lolium genotypes. Biomass is a crucial parameter that influences plant growth; therefore, the accumulation and distribution of dry mass are significant considerations when investigating the effect of drought stress on plant growth. The role of glutathione in the antioxidative defense system also provides a rationale for its use as a stress marker. In general, an initial stress response is related to changes in the glutathione redox state. By this, acclimation is marked by increased glutathione concentrations, increase in related enzyme activities, and a more reduced redox state of glutathione. Among phytohormones, abscisic acid is the most important one that confers abiotic stress tolerance in crop plants . In drought conditions, ABA content in plants increases considerably, inspiring stress-tolerance effects that help plants adapt and survive under these stressful situations . The 'omics' approaches are often applied to reveal molecular changes due to drought stress and to investigate samples collected from different populations or upon different physiological states, aiming to find molecules that would differentiate between these classes of samples. These molecular studies in L . perenne for understanding some underlying molecular mechanisms in response to drought stress have been somewhat limited. In other species, recently proteome analyses have opened the way to the identification of proteins involved in drought stress tolerance in rice, Oryza sativa , barley, Hordeum vulgare , Stipa purpurea , Poa pratensis and wheat, Triticum aestivum . The effort described here related to the comparative physiological and proteomic analysis was applied to investigate the molecular events–due to drought stress tolerance- in three perennial ryegrass genotypes including S10 as a newly introduced breeding line from Iran with self-pollinating reproductive habit with superior seed and turf characteristics compared to Speedy (Speedy Green) and Vigor as cross-pollinating commercial genotypes.
2.1 Plant material and experimental design Lolium perenne plant material of this experiment consisted of one well-performed genotype of S10, selected from the previous experiment and two commercial populations of Vigor and Speedy Green or in short Speedy from Barenbrug (Barenbrug Co., the Netherland). Thus, three genotypes in this research were examined in a pot culture in a greenhouse according to a completely randomized design with three replications. Greenhouse conditions consisted of 16h/8h of light/dark photoperiod at a temperature range from 18.3 to 25.5°C. Ten tillers (with approximately equal sizes) of each genotype previously grown in a field were selected and transplanted into every 60 pots (size: 20*15 cm) filled with a 1:3 (v/v) mixture of sand and soil. The soil was sandy loam in texture with pH = 7.78 and Electrical Conductivity of 1.34 Ds.m -1 . All three genotypes were then propagated for two months before being used in the experiment. Then, plants were regularly irrigated and grown in controlled greenhouse conditions for two weeks before starting the drought stress treatment. For this treatment, pots were divided into two groups: well-watered plants were irrigated every two days (control), and water-stressed plants. Drought stress was induced on plants by limiting irrigation to ensure water depletion to 20% of the field capacity of the soil using theta probe (AT Delta-T Devices SM300, Cambridge, England) before re-irrigation for 30 days. After seven days of the first period of drought stress treatment, leaves were cut off and stored at -70°C for later protein extraction. Biochemical measurements for the three genotypes of Lolium under drought stress were carried out on 1 g of fresh leaves. Antioxidant enzymes activities, ABA hormone and glutathione contents were measured one week after treatment; and proline, total N and amino acid contents, and shoot dry weight were calculated three weeks after the initial drought stress. At the end of the experiment, the harvested plants in pots were washed with distilled water, followed by surface drying with filter paper. For each plant, the leaves and roots were separated. Then leaves and roots were dried for 15 min at 105°C and then at 65°C to a constant weight in an oven. The dry weights were then determined. 2.2 Relative water content (RWC) Leaf relative water contents (RWC) were calculated with the method of Barrs and Weatherley . The relative leaf water content (RWC) was calculated as 100*(FW-DW)/(TW-DW), where FW = fresh weight, DW = dry weight, and TW = turgid weight. TW was measured after saturating the water content of leaf discs for 24 hours. DW was measured after desiccating leaf discs at 60°C to air-dry state. 2.3 Proline content A 0.1 g sample of the fresh leaf was ground in liquid nitrogen from all the Lolium genotypes which were subjected to drought treatment. Leaf proline content was analyzed at the 3rd week post-stress treatment. The content of free proline in leaves was determined as described by Troll and Lindsley . 2.4 Estimation of total nitrogen For estimation of total N, 1 g of fine-ground leaf dry samples was digested with sulfuric acid, and assays were carried out according to the Kjeldahl method . 2.5 Extraction and measurement of ABA ABA was extracted according to the procedure developed by Kelen et al. with some modifications. Leaf samples (1g) were used for extraction. Leaf tissue was ground with a mortar and a pestle in liquid nitrogen and homogenized in 10 cm 3 extraction solution containing butylated hydroxy-toluene (0.25 g) and ascorbic acid (0.24 g) dissolved in 90% methanol and stirred overnight at 4°C. The extract was filtered through a Whatman filter, and methanol was evaporated under vacuum. Then, pH of the aqueous phase was adjusted to 8.5 with 0.1 M phosphate buffer and then partitioned with ethyl acetate three times. After removal of the ethyl acetate phase, pH of the aqueous phase was adjusted to 2.5 with 0.2 M HCl. The solution was partitioned with ethyl acetate three times and then passed through anhydrous sodium sulfate. After, the ethyl acetate phase was evaporated under vacuum, and the dry residue containing hormones was dissolved in 0.5 cm 3 of HPLC grade methanol and stored in vials at 4°C. An aliquot (20 mm 3 ) of the filtered samples was injected into a Waters Symmetry-C18 chromatographic column (250 × 4.6 mm) (USA) with isocratic elution at a flow rate of 0.7 cm 3 min -1 at 25°C using a mobile phase containing acetic acid solution (0.2%) and methanol (100%) (50:50 v/v). Detection of ABA was carried out at 265 nm by a UV detector (Waters 2487) with known concentrations of (±)-abscisic acid (Sigma A1049). 2.6 Antioxidant enzymes and glutathione assays Activities of catalase (CAT), superoxide dismutase (SOD), ascorbate peroxidase (APX), glutathione reductase (GR) and glutathione S-transferase (GST) enzymes were detected in leaf tissue extracts spectrophotometrically as described in our previous work . Glutathione (GSH) content of the leaves was measured according to May and Leaver . 2.7 Amino acids assay Amino acids were extracted from one gram of dry leaves according to the procedure described by Lisiewska et al. . The chromatographic separation of 13 amino acids was achieved by the ion exchange chromatography in an amino acid analyzer (ARACUS, Germany). The identification of amino acids was carried out based on the retention time at the chromatographic separation. All determinations were carried out in two replications for each sample. 2.8 Proteomic analysis Approximately 100 mg of plant material (fresh weight) was ground in liquid nitrogen using a pre-cooled mortar and pestle. The dry powder was transferred to 1.5 cm 3 of precipitation solution (10% TCA w/v, 0.07% DTT w/v), according to Yang et al. . The extract was left to precipitate at -20°C overnight. Precipitated proteins were then pelleted at 16,000 ×g for 15 min at 4°C. The pellet was washed at least three times with acetone containing 0.07% (w/v) DTT at -20°C for 30 min, then, the mixture was vortexed and centrifuged at 16,000 ×g for 20 min, at 4°C. The pellet was recovered and air-dried at room temperature and dissolved in 1 cm 3 of 1x SDS sample buffer . The solutions were vigorously vortex-mixed and kept at room temperature for one h and centrifuged at 14,000 ×g for 10 min at 4°C. The supernatant was transformed into a new tube and stored at 4°C overnight and run on a 5% stacking gel (in an SDS PAGE system with an equal volume of stacking (5%) and resolving (10%) gel) according to Raorane et al. . The single bands formed in stacking gel, containing condensed and purified total soluble proteins, were stained with coomassie brilliant blue (G-250) as described by Neuhoff et al. and used in the following steps. 2.8.1 Mass spectrometry and data analysis After gel staining using coomassie, the gel was cut in 10 slices. Next, gel digestion was performed using trypsin, and peptides were extracted and desalted. The protocol reported by Rappsilber et al. was used for peptide extraction . Then, liquid chromatography-mass spectrometry (LC/MS) analysis was performed on an LTQ Orbitrap XL (Thermo Fisher Scientific, Bremen, Germany) coupled to an Eksigent 2D nano flow-HPLC equipped with in-house packed C18 columns of approximately 20 cm length (Reprosil-Pur 120 C-18-AQ, 3 μm, Ammerbuch, Germany) without a pre-column. Because of an unavailable database for the proteins of L . perenne and limited protein database of Lolium genus, Brachypodium distachyon protein database was selected for the identification of proteins in this study. The analysis of MS raw data was performed using MaxQuant (v. 1.5.3.30 ) against the UniProt Brachypodium distachyon protein sequence database (released June 20, 2018; 44,786 entries in total). A false discovery rate of 1% was applied for both protein lists and peptide-spectrum matches (on modified peptides, separately). Analysis of protein groups was performed with the Perseus software . Normalized ratios were log 2 -transformed, and a mean log 2 was calculated across all three replicates. Finally, proteins with a p-value below 0.05 and fold change higher or lower than +1.5 and -0.5, respectively, were considered. The differentially abundant proteins were used for COG (Cluster of Orthologous Groups of proteins) category and annotation. 2.8.2 Bioinformatics analysis The proteomic data visualization in schematic metabolism pathways, as well as functional classification of proteins, was performed by using the MapMan (Version 3.6.0RC1, downloaded from http://mapman.gabipd.org/ Web Site) software program. 2.9 Statistical analyses Statistical analyses were carried out with three biological and two technical replicates for proteomic analyses. The results of protein abundance were statistically evaluated by student's t-test by using the Multi Experiment Viewer (MEW) software considering P <0.05 as the critical significance level. Relative water content, shoot dry weight, total nitrogen, proline and glutathione contents, antioxidant enzymes activities, and ABA content were analyzed using ANOVA by the SAS software . Significant differences between means were defined by LSD post-hoc test at P <0.05. PCA (Principal Component Analysis) was conducted utilizing STATGRAPHICS Centurion ver. 17 software (Statpoint Technologies, Inc., Warrenton, Virginia, USA).
Lolium perenne plant material of this experiment consisted of one well-performed genotype of S10, selected from the previous experiment and two commercial populations of Vigor and Speedy Green or in short Speedy from Barenbrug (Barenbrug Co., the Netherland). Thus, three genotypes in this research were examined in a pot culture in a greenhouse according to a completely randomized design with three replications. Greenhouse conditions consisted of 16h/8h of light/dark photoperiod at a temperature range from 18.3 to 25.5°C. Ten tillers (with approximately equal sizes) of each genotype previously grown in a field were selected and transplanted into every 60 pots (size: 20*15 cm) filled with a 1:3 (v/v) mixture of sand and soil. The soil was sandy loam in texture with pH = 7.78 and Electrical Conductivity of 1.34 Ds.m -1 . All three genotypes were then propagated for two months before being used in the experiment. Then, plants were regularly irrigated and grown in controlled greenhouse conditions for two weeks before starting the drought stress treatment. For this treatment, pots were divided into two groups: well-watered plants were irrigated every two days (control), and water-stressed plants. Drought stress was induced on plants by limiting irrigation to ensure water depletion to 20% of the field capacity of the soil using theta probe (AT Delta-T Devices SM300, Cambridge, England) before re-irrigation for 30 days. After seven days of the first period of drought stress treatment, leaves were cut off and stored at -70°C for later protein extraction. Biochemical measurements for the three genotypes of Lolium under drought stress were carried out on 1 g of fresh leaves. Antioxidant enzymes activities, ABA hormone and glutathione contents were measured one week after treatment; and proline, total N and amino acid contents, and shoot dry weight were calculated three weeks after the initial drought stress. At the end of the experiment, the harvested plants in pots were washed with distilled water, followed by surface drying with filter paper. For each plant, the leaves and roots were separated. Then leaves and roots were dried for 15 min at 105°C and then at 65°C to a constant weight in an oven. The dry weights were then determined.
Leaf relative water contents (RWC) were calculated with the method of Barrs and Weatherley . The relative leaf water content (RWC) was calculated as 100*(FW-DW)/(TW-DW), where FW = fresh weight, DW = dry weight, and TW = turgid weight. TW was measured after saturating the water content of leaf discs for 24 hours. DW was measured after desiccating leaf discs at 60°C to air-dry state.
A 0.1 g sample of the fresh leaf was ground in liquid nitrogen from all the Lolium genotypes which were subjected to drought treatment. Leaf proline content was analyzed at the 3rd week post-stress treatment. The content of free proline in leaves was determined as described by Troll and Lindsley .
For estimation of total N, 1 g of fine-ground leaf dry samples was digested with sulfuric acid, and assays were carried out according to the Kjeldahl method .
ABA was extracted according to the procedure developed by Kelen et al. with some modifications. Leaf samples (1g) were used for extraction. Leaf tissue was ground with a mortar and a pestle in liquid nitrogen and homogenized in 10 cm 3 extraction solution containing butylated hydroxy-toluene (0.25 g) and ascorbic acid (0.24 g) dissolved in 90% methanol and stirred overnight at 4°C. The extract was filtered through a Whatman filter, and methanol was evaporated under vacuum. Then, pH of the aqueous phase was adjusted to 8.5 with 0.1 M phosphate buffer and then partitioned with ethyl acetate three times. After removal of the ethyl acetate phase, pH of the aqueous phase was adjusted to 2.5 with 0.2 M HCl. The solution was partitioned with ethyl acetate three times and then passed through anhydrous sodium sulfate. After, the ethyl acetate phase was evaporated under vacuum, and the dry residue containing hormones was dissolved in 0.5 cm 3 of HPLC grade methanol and stored in vials at 4°C. An aliquot (20 mm 3 ) of the filtered samples was injected into a Waters Symmetry-C18 chromatographic column (250 × 4.6 mm) (USA) with isocratic elution at a flow rate of 0.7 cm 3 min -1 at 25°C using a mobile phase containing acetic acid solution (0.2%) and methanol (100%) (50:50 v/v). Detection of ABA was carried out at 265 nm by a UV detector (Waters 2487) with known concentrations of (±)-abscisic acid (Sigma A1049).
Activities of catalase (CAT), superoxide dismutase (SOD), ascorbate peroxidase (APX), glutathione reductase (GR) and glutathione S-transferase (GST) enzymes were detected in leaf tissue extracts spectrophotometrically as described in our previous work . Glutathione (GSH) content of the leaves was measured according to May and Leaver .
Amino acids were extracted from one gram of dry leaves according to the procedure described by Lisiewska et al. . The chromatographic separation of 13 amino acids was achieved by the ion exchange chromatography in an amino acid analyzer (ARACUS, Germany). The identification of amino acids was carried out based on the retention time at the chromatographic separation. All determinations were carried out in two replications for each sample.
Approximately 100 mg of plant material (fresh weight) was ground in liquid nitrogen using a pre-cooled mortar and pestle. The dry powder was transferred to 1.5 cm 3 of precipitation solution (10% TCA w/v, 0.07% DTT w/v), according to Yang et al. . The extract was left to precipitate at -20°C overnight. Precipitated proteins were then pelleted at 16,000 ×g for 15 min at 4°C. The pellet was washed at least three times with acetone containing 0.07% (w/v) DTT at -20°C for 30 min, then, the mixture was vortexed and centrifuged at 16,000 ×g for 20 min, at 4°C. The pellet was recovered and air-dried at room temperature and dissolved in 1 cm 3 of 1x SDS sample buffer . The solutions were vigorously vortex-mixed and kept at room temperature for one h and centrifuged at 14,000 ×g for 10 min at 4°C. The supernatant was transformed into a new tube and stored at 4°C overnight and run on a 5% stacking gel (in an SDS PAGE system with an equal volume of stacking (5%) and resolving (10%) gel) according to Raorane et al. . The single bands formed in stacking gel, containing condensed and purified total soluble proteins, were stained with coomassie brilliant blue (G-250) as described by Neuhoff et al. and used in the following steps. 2.8.1 Mass spectrometry and data analysis After gel staining using coomassie, the gel was cut in 10 slices. Next, gel digestion was performed using trypsin, and peptides were extracted and desalted. The protocol reported by Rappsilber et al. was used for peptide extraction . Then, liquid chromatography-mass spectrometry (LC/MS) analysis was performed on an LTQ Orbitrap XL (Thermo Fisher Scientific, Bremen, Germany) coupled to an Eksigent 2D nano flow-HPLC equipped with in-house packed C18 columns of approximately 20 cm length (Reprosil-Pur 120 C-18-AQ, 3 μm, Ammerbuch, Germany) without a pre-column. Because of an unavailable database for the proteins of L . perenne and limited protein database of Lolium genus, Brachypodium distachyon protein database was selected for the identification of proteins in this study. The analysis of MS raw data was performed using MaxQuant (v. 1.5.3.30 ) against the UniProt Brachypodium distachyon protein sequence database (released June 20, 2018; 44,786 entries in total). A false discovery rate of 1% was applied for both protein lists and peptide-spectrum matches (on modified peptides, separately). Analysis of protein groups was performed with the Perseus software . Normalized ratios were log 2 -transformed, and a mean log 2 was calculated across all three replicates. Finally, proteins with a p-value below 0.05 and fold change higher or lower than +1.5 and -0.5, respectively, were considered. The differentially abundant proteins were used for COG (Cluster of Orthologous Groups of proteins) category and annotation. 2.8.2 Bioinformatics analysis The proteomic data visualization in schematic metabolism pathways, as well as functional classification of proteins, was performed by using the MapMan (Version 3.6.0RC1, downloaded from http://mapman.gabipd.org/ Web Site) software program.
After gel staining using coomassie, the gel was cut in 10 slices. Next, gel digestion was performed using trypsin, and peptides were extracted and desalted. The protocol reported by Rappsilber et al. was used for peptide extraction . Then, liquid chromatography-mass spectrometry (LC/MS) analysis was performed on an LTQ Orbitrap XL (Thermo Fisher Scientific, Bremen, Germany) coupled to an Eksigent 2D nano flow-HPLC equipped with in-house packed C18 columns of approximately 20 cm length (Reprosil-Pur 120 C-18-AQ, 3 μm, Ammerbuch, Germany) without a pre-column. Because of an unavailable database for the proteins of L . perenne and limited protein database of Lolium genus, Brachypodium distachyon protein database was selected for the identification of proteins in this study. The analysis of MS raw data was performed using MaxQuant (v. 1.5.3.30 ) against the UniProt Brachypodium distachyon protein sequence database (released June 20, 2018; 44,786 entries in total). A false discovery rate of 1% was applied for both protein lists and peptide-spectrum matches (on modified peptides, separately). Analysis of protein groups was performed with the Perseus software . Normalized ratios were log 2 -transformed, and a mean log 2 was calculated across all three replicates. Finally, proteins with a p-value below 0.05 and fold change higher or lower than +1.5 and -0.5, respectively, were considered. The differentially abundant proteins were used for COG (Cluster of Orthologous Groups of proteins) category and annotation.
The proteomic data visualization in schematic metabolism pathways, as well as functional classification of proteins, was performed by using the MapMan (Version 3.6.0RC1, downloaded from http://mapman.gabipd.org/ Web Site) software program.
Statistical analyses were carried out with three biological and two technical replicates for proteomic analyses. The results of protein abundance were statistically evaluated by student's t-test by using the Multi Experiment Viewer (MEW) software considering P <0.05 as the critical significance level. Relative water content, shoot dry weight, total nitrogen, proline and glutathione contents, antioxidant enzymes activities, and ABA content were analyzed using ANOVA by the SAS software . Significant differences between means were defined by LSD post-hoc test at P <0.05. PCA (Principal Component Analysis) was conducted utilizing STATGRAPHICS Centurion ver. 17 software (Statpoint Technologies, Inc., Warrenton, Virginia, USA).
3.1 Physiological characteristics of ryegrass genotypes under drought stress The results of statistical analyses for RWC, shoot dry weight (SDW), proline, ABA, total N, and glutathione contents of the three genotypes of Lolium perenne are presented in . RWC content of the three genotypes decreased significantly under drought stress (P<0.05). However, the RWC content of the S10 genotype was much higher compared to the other genotypes under both stress and control conditions. Also, the SDW of the genotypes decreased significantly under drought stress (P<0.05). However, like RWC, SDW was significantly higher in S10 than that of the other two genotypes under both stress and control conditions. At stress conditions, dry weights of Vigor, Speedy, and S10 genotypes decreased by 9.7%, 12.5%, and 5.5%, respectively, compared to the control condition. Proline content of the three genotypes increased under drought stress (P<0.05). However, proline accumulation in S10 genotype was much higher than Vigor and Speedy genotypes under both drought and control conditions. The ABA concentration in the genotypes also increased under drought stress (P<0.05) and, like proline, the concentration of this hormone in S10 was significantly higher than that in the two commercial genotypes under both drought and control conditions. In all genotypes, the content of leaf total N decreased under drought stress (P<0.05). In contrast, drought stress significantly increased glutathione content in Vigor and S10 genotypes. All of the tested antioxidant enzymes, including APX, SOD, CAT, GR, and GST, showed similar activity changes under drought stress in the three genotypes. However, the induction of APX activity in Vigor and Speedy was much higher than that of the S10 genotype under drought stress. In contrast, CAT and GR activities were higher in S10 genotype under drought stress conditions . The content of amino acids, including aspartic acid, threonine, serine, glutamic acid, valine, methionine, isoleucine, leucine, tyrosine, arginine, phenylalanine, histidine and lysine of the three genotypes was measured three weeks after drought stress . The results showed that the majority of the amino acid content of all genotypes increased under drought conditions. However, the total amino acid content of S10 was much higher than that of Vigor and Speedy under both control and drought stress conditions. Under stress conditions, the total amino acid content of S10 was 262 (mg/100 g DW) as against 219 and 217 (mg/100 g DW) in Vigor and Speedy genotypes, respectively. 3.2 Proteomic characteristics of ryegrass genotypes under drought stress The small grass species, Brachypodium distachyon , is potentially an ideal model plant system for grass research. Therefore, the Brachypodium protein database was used for protein identification to dissect tolerance to drought stress in Lolium perenne genotypes. Based on the Brachypodium protein database, a total of 915 proteins were identified in each genotype. The number of proteins whose relative abundance level was significantly altered under drought stress conditions was 467, 456, and 97 in Vigor, Speedy, and S10 genotypes, respectively. Both “up-regulated” and “on (New)” proteins are considered together as “differentially increased proteins." Similarly, the term “differentially decreased proteins” included both “down-regulated” and “off (disappeared)” proteins. Among the proteins, 230, 282 and 40 differentially increased proteins (fold change >1.5, P < 0.05) and 237, 174 and 57 differentially decreased proteins (fold change <0.5, P < 0.05) were found under drought stress compared to control condition in Vigor, Speedy, and S10 genotypes, respectively, . 3.3 Functional annotation of identified and differentially abundant proteins To simplify the analysis of protein expression, differentially abundant proteins were classified into 12 COG categories using Mapman ontology and the Brachypodium mapping file . Among 230 differentially increased proteins in the Vigor genotype, the most are involved in carbon and energy metabolism, protein, redox, and transport categories. Similarly, in Speedy genotype, among 282 differentially increased proteins, the most are involved in carbon and energy metabolism, protein, redox, stress, and transport categories. In contrast, in the S10 genotype, most of the proteins are involved in carbon and energy metabolism, and protein and stress categories decreased under drought stress . Based on the COG annotation, drought stress induced changes in the abundance of proteins which are constituting the photosystem I (PSI) and PSII reaction centers and PSI and PSII polypeptide subunits. We found that proteins constituting reaction centers of the two photosystems (no. 6, 9, 17, 31, 54) increased in both Vigor and Speedy genotypes. In contrast, some proteins of reaction centers decreased in Vigor (no. 455) and S10 (no. 54) genotypes. Several polypeptide subunits of photosystems highly decreased in the three genotypes of Lolium under drought stress. We observed a drought-induced increase in the levels of small and large subunits of RuBisCO in Vigor and Speedy genotypes, whereas, in S10, such a change was not detected. In the Vigor genotype, RuBisCO activase exhibited drought-induced accumulation. The Expression level of the protein sedoheptulose-bisphosphatase, which participates in the Calvin cycle, was up-regulated under drought in Speedy genotype . The chloroplastic isoform of fructose-1,6-bisphosphate aldolase (FBA) was down-regulated in Vigor and Speedy genotypes; however, no changes were detected in S10 genotype. In contrast, the cytoplasmic isoform of FBA was up-regulated in Vigor and Speedy genotypes, whereas it was down-regulated in the S10 genotype. Alpha and beta subunits of pyruvate dehydrogenase complex, malate dehydrogenase, malic enzyme, citrate synthase, succinyl-CoA ligase, succinate dehydrogenase, isocitrate dehydrogenase, and aconitate hydratase, which are involved in the glycolysis pathway and Krebs cycle, increased in both Vigor and Speedy genotypes but did not change in the S10 genotype. However, the NADP dependent malic enzyme (NADP-ME) increased in all three genotypes. Based on the COG annotation, the differentially abundant proteins involved in starch and sucrose metabolisms were up- or down-regulated in the three genotypes under drought stress. Regarding sucrose synthase-related proteins, we found that in S10 genotype, sucrose synthase (SS) increased, while fructose-1,6-bisphosphatase showed reduced levels . We also found the up-regulation of starch-degradation enzymes, starch-related R1 protein, and starch phosphorylase in the three genotypes of Lolium in drought stress conditions . On the contrary, the isoforms of ADP-glucose pyrophosphorylase involved in the pathway of starch synthesis were up-regulated only in Vigor and Speedy genotypes. In the S10 genotype, no changes were detected in some of the isoforms, and other isoforms showed down-regulated expression levels . Based on the COG annotation, the differentially abundant proteins in the cluster of cell walls, including cellulose, pectinase, β-xylosidase and β-1, 4-glucanase enzymes were down-regulated in Vigor and Speedy genotypes . A significant increase in the protein abundance of arabinogalactan-rich proteins (AGPs) has also been found in the S10 genotype. Our proteomic analysis revealed that drought-induced changes in concentration of proteins involved in lipid metabolism and two isoforms of the phospholipase enzyme increased in both commercial genotypes. Interestingly, opposite to these genotypes, proteins involved in lipid metabolism did not change or even were reduced in the S10 genotype under drought stress conditions . In the functional group of nitrogen and amino acid metabolism, some of the differentially abundant proteins increased in Vigor and Speedy genotypes and the proteins involved in the metabolism of methionine, cysteine, aspartate, glycine, and proline were up-regulated in these two commercial genotypes. Δ-1-pyrroline-5-carboxylate synthase (P5CS) enzyme, which catalyzes the first two steps in proline biosynthesis in plants, did not exhibit significant changes in the three genotypes of Lolium . However, drought-induced increase in the level of pyrroline-5-carboxylate reductase (P5CR), which is a part of the L-proline biosynthesis pathway, was significantly higher in Vigor and Speedy genotypes. Also, the abundance of 1-pyrroline-5-carboxylate dehydrogenase enzyme, which is a part of the pathway of L-proline degradation into L-glutamate, significantly increased in these two genotypes. We observed drought-induced reduction in the levels of nitrite reductase 1, glutamine synthetase, and glutamate dehydrogenase (involved in nitrogen flow) in Vigor and Speedy genotypes . Concerning the ROS-generation proteins, we found two differentially increased proteins, glycolate oxidase, and oxalate oxidase, in the two commercial genotypes. However, they decreased or did no change in drought-stressed S10 genotype. Under drought stress, plants respond to ROS-over abundance in cells by producing enzymatic and nonenzymatic ROS scavenging antioxidants. The increased level of dehydroascorbate reductase (DHAR), monodehydroascorbate reductase (MDHAR), catalase, ascorbate peroxidase (APX) and SOD was observed in leaves of both Vigor and Speedy genotypes. However, GR and glutathione peroxidase was up-regulated only in the Vigor genotype. Under drought stress, the up-regulation of glutathione S-transferase was found in the three genotypes of Lolium . The other hydrogen peroxide-decomposing enzymes, glutathione peroxidase, and peroxiredoxin and thioredoxin reductase increased only in Vigor and Speedy genotypes in response to drought stress . Given that ABA metabolism is a crucial aspect of the plant response to drought stress, proteins involved in ABA metabolism were searched in the regulated proteins. We found that two isoforms of 9-cis epoxy carotenoid dioxygenase (NCED1 and NCED4) increased during drought stress in Vigor and Speedy genotypes; however, the S10 genotype presented no alteration in this respect . Based on the COG annotation, the levels of four heat shock proteins (HSPs) (HSP60, chaperonin-60ALPHA, chaperonin 20 and DNAJ) increased under drought stress only in Vigor and Speedy genotypes; and no protein abundance changes of HSP-like proteins were detected in S10 genotype . Also, based on the COG annotation, some proteins of the transport system were up-regulated only in the two commercial genotypes. The PIP1;4 and PIP2B aquaporin proteins (AQPs) were up-regulated in Speedy and Vigor genotypes; however, the up-regulation level of these proteins was higher in Speedy genotype. Furthermore, the PIP2A protein and two isoforms of PIP were also up-regulated in the Speedy genotype . 3.4 Multivariate data analysis Principal Component Analysis (PCA) was conducted to determine the correlation between the three genotypes of ryegrass, drought stress, and all the examined characteristics. Noteworthy, the PCA analysis displayed a high correlation between biomass, total N, and RWC content ; however, no positive correlation between biomass and total N with amino acids total content and antioxidant enzymes was found. The separation of S10 genotype from the two commercial genotypes (Vigor and Speedy) can be easily deduced from the data provided. Also, we found that the S10 genotype in drought conditions had a combination of higher dry weight (second component) and antioxidant enzymes (first component) than those in the other two genotypes. Therefore, under drought, better biomass and antioxidant potential along with higher drought tolerance are undoubtedly ascribed to this genotype in comparison to the two commercial genotypes. The PCA results also revealed the separation of control and drought conditions for all tested genotypes and indicated that drought stress was particularly distinctive by induction in the production of antioxidant enzymes. This PCA analysis may suggest that the defense strategy utilized by the two commercial genotypes to assist plants against drought is different from the mechanism used by the well-performed S10 genotype.
The results of statistical analyses for RWC, shoot dry weight (SDW), proline, ABA, total N, and glutathione contents of the three genotypes of Lolium perenne are presented in . RWC content of the three genotypes decreased significantly under drought stress (P<0.05). However, the RWC content of the S10 genotype was much higher compared to the other genotypes under both stress and control conditions. Also, the SDW of the genotypes decreased significantly under drought stress (P<0.05). However, like RWC, SDW was significantly higher in S10 than that of the other two genotypes under both stress and control conditions. At stress conditions, dry weights of Vigor, Speedy, and S10 genotypes decreased by 9.7%, 12.5%, and 5.5%, respectively, compared to the control condition. Proline content of the three genotypes increased under drought stress (P<0.05). However, proline accumulation in S10 genotype was much higher than Vigor and Speedy genotypes under both drought and control conditions. The ABA concentration in the genotypes also increased under drought stress (P<0.05) and, like proline, the concentration of this hormone in S10 was significantly higher than that in the two commercial genotypes under both drought and control conditions. In all genotypes, the content of leaf total N decreased under drought stress (P<0.05). In contrast, drought stress significantly increased glutathione content in Vigor and S10 genotypes. All of the tested antioxidant enzymes, including APX, SOD, CAT, GR, and GST, showed similar activity changes under drought stress in the three genotypes. However, the induction of APX activity in Vigor and Speedy was much higher than that of the S10 genotype under drought stress. In contrast, CAT and GR activities were higher in S10 genotype under drought stress conditions . The content of amino acids, including aspartic acid, threonine, serine, glutamic acid, valine, methionine, isoleucine, leucine, tyrosine, arginine, phenylalanine, histidine and lysine of the three genotypes was measured three weeks after drought stress . The results showed that the majority of the amino acid content of all genotypes increased under drought conditions. However, the total amino acid content of S10 was much higher than that of Vigor and Speedy under both control and drought stress conditions. Under stress conditions, the total amino acid content of S10 was 262 (mg/100 g DW) as against 219 and 217 (mg/100 g DW) in Vigor and Speedy genotypes, respectively.
The small grass species, Brachypodium distachyon , is potentially an ideal model plant system for grass research. Therefore, the Brachypodium protein database was used for protein identification to dissect tolerance to drought stress in Lolium perenne genotypes. Based on the Brachypodium protein database, a total of 915 proteins were identified in each genotype. The number of proteins whose relative abundance level was significantly altered under drought stress conditions was 467, 456, and 97 in Vigor, Speedy, and S10 genotypes, respectively. Both “up-regulated” and “on (New)” proteins are considered together as “differentially increased proteins." Similarly, the term “differentially decreased proteins” included both “down-regulated” and “off (disappeared)” proteins. Among the proteins, 230, 282 and 40 differentially increased proteins (fold change >1.5, P < 0.05) and 237, 174 and 57 differentially decreased proteins (fold change <0.5, P < 0.05) were found under drought stress compared to control condition in Vigor, Speedy, and S10 genotypes, respectively, .
To simplify the analysis of protein expression, differentially abundant proteins were classified into 12 COG categories using Mapman ontology and the Brachypodium mapping file . Among 230 differentially increased proteins in the Vigor genotype, the most are involved in carbon and energy metabolism, protein, redox, and transport categories. Similarly, in Speedy genotype, among 282 differentially increased proteins, the most are involved in carbon and energy metabolism, protein, redox, stress, and transport categories. In contrast, in the S10 genotype, most of the proteins are involved in carbon and energy metabolism, and protein and stress categories decreased under drought stress . Based on the COG annotation, drought stress induced changes in the abundance of proteins which are constituting the photosystem I (PSI) and PSII reaction centers and PSI and PSII polypeptide subunits. We found that proteins constituting reaction centers of the two photosystems (no. 6, 9, 17, 31, 54) increased in both Vigor and Speedy genotypes. In contrast, some proteins of reaction centers decreased in Vigor (no. 455) and S10 (no. 54) genotypes. Several polypeptide subunits of photosystems highly decreased in the three genotypes of Lolium under drought stress. We observed a drought-induced increase in the levels of small and large subunits of RuBisCO in Vigor and Speedy genotypes, whereas, in S10, such a change was not detected. In the Vigor genotype, RuBisCO activase exhibited drought-induced accumulation. The Expression level of the protein sedoheptulose-bisphosphatase, which participates in the Calvin cycle, was up-regulated under drought in Speedy genotype . The chloroplastic isoform of fructose-1,6-bisphosphate aldolase (FBA) was down-regulated in Vigor and Speedy genotypes; however, no changes were detected in S10 genotype. In contrast, the cytoplasmic isoform of FBA was up-regulated in Vigor and Speedy genotypes, whereas it was down-regulated in the S10 genotype. Alpha and beta subunits of pyruvate dehydrogenase complex, malate dehydrogenase, malic enzyme, citrate synthase, succinyl-CoA ligase, succinate dehydrogenase, isocitrate dehydrogenase, and aconitate hydratase, which are involved in the glycolysis pathway and Krebs cycle, increased in both Vigor and Speedy genotypes but did not change in the S10 genotype. However, the NADP dependent malic enzyme (NADP-ME) increased in all three genotypes. Based on the COG annotation, the differentially abundant proteins involved in starch and sucrose metabolisms were up- or down-regulated in the three genotypes under drought stress. Regarding sucrose synthase-related proteins, we found that in S10 genotype, sucrose synthase (SS) increased, while fructose-1,6-bisphosphatase showed reduced levels . We also found the up-regulation of starch-degradation enzymes, starch-related R1 protein, and starch phosphorylase in the three genotypes of Lolium in drought stress conditions . On the contrary, the isoforms of ADP-glucose pyrophosphorylase involved in the pathway of starch synthesis were up-regulated only in Vigor and Speedy genotypes. In the S10 genotype, no changes were detected in some of the isoforms, and other isoforms showed down-regulated expression levels . Based on the COG annotation, the differentially abundant proteins in the cluster of cell walls, including cellulose, pectinase, β-xylosidase and β-1, 4-glucanase enzymes were down-regulated in Vigor and Speedy genotypes . A significant increase in the protein abundance of arabinogalactan-rich proteins (AGPs) has also been found in the S10 genotype. Our proteomic analysis revealed that drought-induced changes in concentration of proteins involved in lipid metabolism and two isoforms of the phospholipase enzyme increased in both commercial genotypes. Interestingly, opposite to these genotypes, proteins involved in lipid metabolism did not change or even were reduced in the S10 genotype under drought stress conditions . In the functional group of nitrogen and amino acid metabolism, some of the differentially abundant proteins increased in Vigor and Speedy genotypes and the proteins involved in the metabolism of methionine, cysteine, aspartate, glycine, and proline were up-regulated in these two commercial genotypes. Δ-1-pyrroline-5-carboxylate synthase (P5CS) enzyme, which catalyzes the first two steps in proline biosynthesis in plants, did not exhibit significant changes in the three genotypes of Lolium . However, drought-induced increase in the level of pyrroline-5-carboxylate reductase (P5CR), which is a part of the L-proline biosynthesis pathway, was significantly higher in Vigor and Speedy genotypes. Also, the abundance of 1-pyrroline-5-carboxylate dehydrogenase enzyme, which is a part of the pathway of L-proline degradation into L-glutamate, significantly increased in these two genotypes. We observed drought-induced reduction in the levels of nitrite reductase 1, glutamine synthetase, and glutamate dehydrogenase (involved in nitrogen flow) in Vigor and Speedy genotypes . Concerning the ROS-generation proteins, we found two differentially increased proteins, glycolate oxidase, and oxalate oxidase, in the two commercial genotypes. However, they decreased or did no change in drought-stressed S10 genotype. Under drought stress, plants respond to ROS-over abundance in cells by producing enzymatic and nonenzymatic ROS scavenging antioxidants. The increased level of dehydroascorbate reductase (DHAR), monodehydroascorbate reductase (MDHAR), catalase, ascorbate peroxidase (APX) and SOD was observed in leaves of both Vigor and Speedy genotypes. However, GR and glutathione peroxidase was up-regulated only in the Vigor genotype. Under drought stress, the up-regulation of glutathione S-transferase was found in the three genotypes of Lolium . The other hydrogen peroxide-decomposing enzymes, glutathione peroxidase, and peroxiredoxin and thioredoxin reductase increased only in Vigor and Speedy genotypes in response to drought stress . Given that ABA metabolism is a crucial aspect of the plant response to drought stress, proteins involved in ABA metabolism were searched in the regulated proteins. We found that two isoforms of 9-cis epoxy carotenoid dioxygenase (NCED1 and NCED4) increased during drought stress in Vigor and Speedy genotypes; however, the S10 genotype presented no alteration in this respect . Based on the COG annotation, the levels of four heat shock proteins (HSPs) (HSP60, chaperonin-60ALPHA, chaperonin 20 and DNAJ) increased under drought stress only in Vigor and Speedy genotypes; and no protein abundance changes of HSP-like proteins were detected in S10 genotype . Also, based on the COG annotation, some proteins of the transport system were up-regulated only in the two commercial genotypes. The PIP1;4 and PIP2B aquaporin proteins (AQPs) were up-regulated in Speedy and Vigor genotypes; however, the up-regulation level of these proteins was higher in Speedy genotype. Furthermore, the PIP2A protein and two isoforms of PIP were also up-regulated in the Speedy genotype .
Principal Component Analysis (PCA) was conducted to determine the correlation between the three genotypes of ryegrass, drought stress, and all the examined characteristics. Noteworthy, the PCA analysis displayed a high correlation between biomass, total N, and RWC content ; however, no positive correlation between biomass and total N with amino acids total content and antioxidant enzymes was found. The separation of S10 genotype from the two commercial genotypes (Vigor and Speedy) can be easily deduced from the data provided. Also, we found that the S10 genotype in drought conditions had a combination of higher dry weight (second component) and antioxidant enzymes (first component) than those in the other two genotypes. Therefore, under drought, better biomass and antioxidant potential along with higher drought tolerance are undoubtedly ascribed to this genotype in comparison to the two commercial genotypes. The PCA results also revealed the separation of control and drought conditions for all tested genotypes and indicated that drought stress was particularly distinctive by induction in the production of antioxidant enzymes. This PCA analysis may suggest that the defense strategy utilized by the two commercial genotypes to assist plants against drought is different from the mechanism used by the well-performed S10 genotype.
The decisive role of plant genotype for improving drought tolerance in Lolium perenne was reflected in higher RWC, dry weight and N content in S10 genotype under drought stress than those in Vigor and Speedy genotypes, indicating increased drought tolerance in this genotype of Lolium . In this study, the accumulation of proline and some other amino acids increased in the three genotypes of Lolium under drought stress. However, the content of proline, aspartate, glutamate, phenylalanine, and tyrosine was higher in S10 genotypes than that in the other two genotypes . These results were consistent with the previous studies on tall fescue that showed up-regulation of proline and some amino acids under drought stress . Therefore, the results of this study indicate an increase in physiological response capacity of S10 genotype to drought stress, which may enable the plant to have excellent stomatal control and accordingly reduce water utilization. Based on the Brachypodium protein database, we identified 467, 456, and 97 differentially abundant proteins in Vigor, Speedy, and S10 genotypes, respectively, under drought stress . These results demonstrate that drought stress coincides with changes in proteins. We also found that the number of differentially decreased proteins (57) was higher than that of differentially increased proteins (40) in the S10 genotype, showing that protein breakdown is the primary function during drought stress in this genotype. Some previous studies also showed that drought resistance was involved in an alteration in the gene expression responsive to a decrease in the transcript abundance . In the present study, we found that the increased proteins in the two commercial genotypes were mostly involved in carbon and energy metabolism, photosynthesis, TCA cycle, redox, and transport categories. However, in the S10 genotype, they were in N and amino acid metabolism category . Also, in Vigor and Speedy genotypes, the decreased proteins were related to cell wall, N, and amino acid metabolism category and redox. However, in the S10 genotype, they were in carbon and energy metabolisms. Preventing photosynthesis due to the closure of stomata is one of the destructive effects of drought stress which in turn results in a total decrease in the expression level of the proteins involved in photosynthesis . In contrast, in our study, the majority of photosynthesis-related proteins increased as an influence of drought stress in the two commercial genotypes. This increase in abundance mainly involved carbon fixation enzymes, including RuBisCO, RuBisCO activase, FBA, glyceraldehyde-3-phosphate dehydrogenase, sedoheptulose-1,7-bisphosphatase, and phosphoribulokinase. While light reaction-involved proteins were down-regulated under drought in all three genotypes of Lolium ; therefore, the increase in carbon fixation enzymes may remain useless. Damage to photosystems usually occurs when plants are subjected to drought stress . This statement was confirmed in our study by the identification of reduction in light-harvesting complexes in Vigor and Speedy genotypes, and somehow in S10 genotype. The decrease in abundance of these polypeptides has been reported before in several species under drought stress . All of the enzymes involved in the Krebs cycle were enhanced under drought stress in the two commercial genotypes. The up-regulation of the proteins involved in the Krebs cycle under drought stress is in accordance with the results of previous researches . Enhancing TCA cycle activity may ensure energy for various processes towards tolerance to drought stress. The TCA pathway is one of the main pathways of energy production, and the up-regulation of the enzymes involved in this pathway can be due to the cell's demand for energy. Sucrose synthase (SS), fructose-1,6-bisphosphatase, and ADP-glucose pyrophosphorylase are the key enzymes involved in the regulation of sucrose synthesis in the cytoplasm and starch synthesis in the chloroplast. Our data showed that under drought stress, the abundance of SS in the S10 genotype increased to 3-fold compared to the control condition. However, the abundance of fructose-1,6-bisphosphatase was reduced in this genotype. SS belongs to a large family of glycosyltransferase enzymes in plants that catalyzes the synthesis of sucrose from fructose and UDP-glucose, providing carbon for respiration and the biosynthesis of starch and cellulose . During drought stress, biosynthesis of starch is generally emptied in several plant species . The ADP-glucose pyrophosphorylase enzyme plays a crucial role in the biosynthesis regulation of starch from glucose-1-phosphate. In the genotype S10, we found a reduction in the abundance of the ADP-glucose pyrophosphorylase. In contrast, the abundance of this protein and starch synthase increased in the two commercial genotypes of Lolium . There is evidence that at the beginning of drought stress, a short-term increase in starch synthesis may happen . Then, the up-regulated expression levels of starch-degradation enzymes could lead to a reduced level of starch in the plants. On the one hand, the plant requires energy to cope with the stress; on the other hand, the rate of photosynthesis in the plant decreases under stress conditions. Therefore, to provide the required energy for various activated pathways against stress, the plant starts metabolizing starch. Also, cell wall invertase, which is involved in sucrose degradation, decreased under drought stress in the two commercial genotypes of Lolium . Invertase catalyzes the hydrolysis of sucrose into glucose and fructose, plays a crucial role in primary metabolism and plant development. Down-regulation of cell wall invertase and sucrose synthase was reported from Piriformospora indica -inoculated barley exposed to drought stress compared to non-inoculated plants . Our results imply a crucial role of SS and invertase in Lolium genotypes in the regulation of sugar biosynthesis when exposed to drought stress. Since sucrose is involved in the stability of proteins as a compatible osmolyte, thus in stressed plants, enhancement of sucrose may play an essential regulatory effect in the alleviation of damage . The amounts of most of the proteins in the two commercial genotypes were down-regulated at the cell wall level in order to cope with drought stress. Plants may down-regulate some enzymes, such as cell wall hydrolysates, in order to reduce the hydrolysis of cell wall polysaccharides, which may be an approach to save their energy and preserve their carbohydrate reserves, and thus to survive under stress. An increase in cell wall synthesis, under drought stress conditions, is probably due to the increased mechanical strength as a strategy to reduce dehydration . Arabinogalactan proteins (AGPs) belonging to the hydroxyproline-rich cell wall glycoprotein superfamily, are implicated in various aspects of plant growth and development, including cell differentiation, cell expansion and abiotic stress response through modulating cell wall expansion and possible interaction of the plant with microbial endophytes . We identified FLA11 as a fasciclin-like arabinogalactan protein that was up-regulated in the S10 genotype under drought stress. Similarly, the up-regulation of five FLA genes in Populus trichocarpa was reported under salt stress . Accumulation of AGPs in the S10 genotype could be stimulating for extreme amounts of cell walls expansion under drought stress. The enlargement of cell wall, in turn, can cause a considerably higher water absorption expanding the cell volume of plants. In plants, phospholipases hydrolyze membrane phospholipids to phosphatidic acid (a membranous second messenger molecule); the target of phosphatidic acid is ABI1 that is implicated in ABA signaling in stomatal responses . The latest studies demonstrate that phospholipase plays an essential role in plant drought stress tolerance . In this study, we found that phospholipases were considerably accumulated in the two commercial genotypes of Lolium under drought stress. The activity and transcript levels of PLD (phospholipase D) in cowpea and peanut increased under drought stress in cultivars that were drought susceptible compared to drought-tolerant cultivars . Hence, we propose that phospholipases may be implicated in the membrane rearrangement and first happenings in ABA signal transduction in stomatal movement in Vigor and Speedy genotypes under drought stress. Various studies have reported that drought stress alters the volume of free amino acids in plants' cells . These alterations can be due to increased protein degradation under stress . The participation of metabolic pathways of amino acids in growth and regulation of tolerance and adaption to stresses has been shown in plants. In this study, drought stress enhanced synthesis and degradative pathways of some amino acids, especially methionine, cysteine, aspartate, glycine, and proline. It strongly enhanced the synthesis of proline, as biochemical data also confirm that proline content was up-regulated in drought-stressed plants in all three genotypes . Drought stress can induce proteolysis of protein and thereby leads to an accumulation of amino acids . In our study, drought led to a rise in amino acid content in all three genotypes. The difference in amino acid contents resulted from drought stress is more likely to reflect the turnover of amino acids. The accumulation of P5CR, which participates in proline biosynthesis, was significantly higher under drought stress in the two commercial genotypes. Notwithstanding with no overall change at the protein level of P5CS in the examined genotypes, our biochemical analysis showed enhancement of proline level in leaves of all three genotypes. In the S10 genotype, we found a more significant increase than 2.1-fold in the accumulation of this amino acid, but only 1.2-fold alteration was recorded for Vigor and Speedy genotypes . Drought stress decreased photosynthesis, coupled with the reduction in nitrogen metabolism . Nitrogen assimilation occurs in plants via two key enzymes, glutamine synthetase (GS) and glutamate synthase (GOGAT). Singh et al. showed that soil drying decreased the activities of these enzymes together with a decrease in N absorption and content of nitrogenous compounds in leaves of sunflower seedlings. Our results show that the abundance of GS and GOGAT decreased in the two commercial genotypes under stress, but no change was observed in the abundance of these two enzymes in the S10 genotype. Similarly, Ghaffari et al. reported that drought stress declined NR, GS, and GOGAT enzymes in non-inoculated barley, but this was not the case for inoculated plants with Piriformospora indica . It was demonstrated in numerous studies that leaf nitrogen reduces progressively as the drought proceeds, and this can be correlated to photosynthetic machinery damage . Our biochemical data show that total nitrogen content decreased in Vigor and Speedy genotypes under drought stress, which confirms the lower activity of GS and GOGAT enzymes in this condition. Meanwhile, total nitrogen content did not change in the S10 genotype during drought stress, which is in accordance with the proteomic data. Numerous studies have reported crop productivity reduction due to drought stress as one of the most significant challenges . One of the reasons for the decreased growth rate of the plants exposed to drought stress can be due to the prevention of nitrogen fixation by drought stress. In this study, a down-regulation in the expression level of nitrite reductase (NR) protein was also detected in the two genotypes of Speedy and Vigor. The relative reduction of shoot dry weight under drought stress was similar for these two genotypes (about 11%), probably showing a lower NR activity in both genotypes. However, this reduction for the S10 genotype under drought was about 5.4 times smaller than the reduction in the two commercial genotypes. Abscisic acid (ABA) is a phytohormone that plays a crucial role in plant tolerance to drought stress. ABA regulates many aspects of plant development under abiotic stresses, permitting the plant to resist drought conditions. Drought-induced increase in ABA content usually results in leaf stomata closure, stimulation of stress-related proteins, and provides different metabolites for the response to stress . An increased ABA level has been formerly informed in several species under drought stress . We found an approximately 6-fold increase of ABA concentration in the two genotypes of Speedy and Vigor two days after drought stress. The ABA content is controlled by the relative values of biosynthesis, catabolism, conjugation, and its distribution through the plant. In our results, both the content of endogenous ABA and the expression of rate-limiting ABA biosynthetic protein (NCED) increased in the two commercial genotypes under drought stress. Similar proteomic analysis findings have been documented for several plant species, for instance, in Phaseolus , Arabidopsis , and Stipa showing the up-regulation of NCED genes under drought stress. However, our findings showed that although no protein abundance changes of ABA biosynthesis-related enzymes were identified in the S10 genotype, the endogenous ABA increased in the leaves of this genotype, suggesting that the increase could be possibly due to the decreased catabolism of this hormone. ROSs are generated as a routine by-product of aerobic metabolism in plant cells. Besides, drought stress is recognized to accelerate the accumulation of ROS in plant cells, lastly resulting in cell damage. Therefore, ROS levels need to be actively controlled for the protection of plants from drought-induced oxidative stress . The redox signaling also regulates a large number of transcripts that are necessary for stress acclimatization. Ascorbic acid (AsA), which is a constituent of the AsA-glutathione (GSH) cycle , maintains redox homeostasis in the plant cell. In order to maintain the antioxidative potential of AsA, rapid restitution of AsA is arranged by DHAR and MDHAR . Activated enhancement of DHAR and MDHAR in these stress responses has been found in various plants. In our study, up-regulation of DHAR and MDHAR in the two commercial genotypes of Lolium under stress suggests that th ese genotypes may increase AsA content by increasing DHAR and MDHAR enzymes. Also, glutathione (GSH; γ-glutamyl-cysteinyl-glycine) plays an essential function in sequestration and transport of reduced sulfur participating directly or indirectly in the detoxification of ROS . Regarding glutathione-related proteins, we identified that glutathione reductase and glutathione peroxidase were up-regulated only in commercial genotypes; but not in the S10 genotype. However, the up-regulation of glutathione S-transferase was observed in all three genotypes. We observed that reduced glutathione content increased in the three genotypes of Lolium , probably due to the action of glutathione-related proteins. Thioredoxins (Trxs) are well-conserved disulfide reductases that regulate the redox state of target proteins. In plants, Trxs are proved to be essential for plant tolerance to oxidative stresses . In Vigor and Speedy genotypes, the up-regulation of thioredoxin M-type was detected under drought stress. Peroxiredoxins (Prx) are a family of thiol dependent peroxidases found in plant cells, which use thioredoxin as their electron donor for the catalysis of H 2 O 2 . They can act as redox sensors, signal transducers, and molecular chaperones through the alterations in their oligomeric structures . Also, we found up-regulation of the two isoforms of peroxiredoxin type 2 in Vigor and Speedy genotypes under drought stress. Generally, high numbers of detected leaf proteins related to ROS scavengings such as APX, SOD, GR and CAT, glutathione peroxidase, and peroxiredoxin and thioredoxin reductase were up-regulated. Up-regulation of most antioxidant enzymes has been formerly documented in different species under drought stress . However, in the S10 genotype under drought stress, the antioxidant system did not respond as vigorously as expected, except for SOD and GST enzymes, which were enhanced to play defensive roles against ROS. Ghaffari et al. reported that four isoforms of GST and three isoforms of ascorbate peroxidase were up-regulated only in the non-inoculated barley; but not in the inoculated ones with mycorrhiza. Glutathione S-transferase was also up-regulated in all three genotypes of Lolium . Possibly because of the up-regulation of ROS-generating proteins in commercial genotypes, it seems that the amount of produced ROS in these genotypes was higher than that in the S10 genotype. Thus they increased the number of activated enzymes in order to maintain their homeostasis. The up-regulated proteins included heat shock proteins (HSPs) and aquaporins (AQP), which are known to be critical for stress acclimatization and water transport . Heat shock proteins are implicated in plant acclimatization to different stresses . Members of various heat shock protein families play substantial roles in trafficking and folding of proteins and assembling of macromolecular in the cell . Our results showed induction of heat shock protein 60, and increased abundance was found for the chaperonin-60ALPHA, chaperonin 20, and DNAJ, which is named HSP40, under drought treatment only in the two commercial genotypes. The expression of HSP has been indicated to increase in many plant species under drought stress . Conversely, Ghaffari et al. have reported that under drought stress, the expression of HSPs decreased in the non-mycorrhizal plants of barley. Our findings highlight that the HSPs may play an essential role in maintaining protein stability in some genotypes under drought stress. Strikingly, in the two genotypes of Vigor and Speedy under drought stress, five plasma membrane intrinsic proteins were found, and all these exhibited increased levels. However, these proteins in the S10 genotype remained unchanged under stress. Several investigators have formerly found the up-regulation of aquaporins in response to water stress in different plant species . Also, the transcript level of aquaporin was enhanced by Trichoderma harzianum colonization in rice . In our study, the evaluation of RWC content during drought stress indicated significantly higher values of RWC in the S10 genotype in comparison to the two commercial genotypes . The better water status of the S10 genotype ensured that the expression of aquaporins gene is unnecessary in this genotype and permitted to maintain higher leaf net photosynthesis and water use efficiency than the two commercial genotypes when exposed to drought stress. Comparably, low genetic gains characterize current population-based breeding methods for essential traits such as biomass yield in many plants. However, grass species have adopted a range of various breeding systems, some promoting self-pollination, some cross-pollination, and some asexual propagation. Our results demonstrated that the self-pollinating genotype of Lolium perenne (S10) might provide a better buffer in response to drought stress. Therefore, self-pollinating genotypes of the plant might be advantageous over cross-pollinating ones due to having more gain for desired traits, including drought stress resistance after fixation in genetic loci and selection for the well-performed characteristics . In our study, it seems that the S10 genotype is a potential new genetic resource with a self-pollinating reproductive system that could be introduced to different genetic backgrounds for better resistance to biotic and abiotic stresses.
This study provides perceptions into the effects of drought stress on the abundance of proteins in the leaves of three genotypes of Lolium perenne . The results of this study indeed showed that RWC and SDW, proline, ABA, N, and amino acid contents, and antioxidant enzyme activities were significantly higher in the drought-tolerant genotype of S10 under drought stress in comparison with the two commercial genotypes. Results of the proteomics analysis revealed significant differences between the examined genotypes of Lolium under drought stress conditions , and this result seems to be related to the higher ability of the S10 genotype in response to drought stress. Intriguingly, we found only slight alterations in the protein profiles of S10 genotypes under drought stress, proposing that protein abundance was less influenced, possibly due to the tolerance in this genotype. That consequence may be related to the self-compatibility mechanism that can improve grass characteristics and alleviate some adverse effects of drought stress. Also, the results showed that acclimatization to drought stress in the two commercial genotypes involved specific responses that resembled between Vigor and Speedy. In this study, some underlying mechanisms were identified through which the plant responses may be improved to drought. These mechanisms included (1) significant increases in the abundance of antioxidant enzymes to scavenge ROS; and (2) changes in the expression of various proteins, including those involved in TCA cycle, amino acid and ABA metabolisms, aquaporins, and HSPs. These changes increased stress tolerance through ABA-mediated regulation of stomatal control to diminish water loss and assist in decreasing the adverse effects of drought by inhibiting damage to fundamental cell building blocks and improving osmoregulation. Additional analyses may be also required to clarify the root proteome that seems to participate in this response.
S1 Table Proteins showing significant changes in responses to drought stress in Vigor, Speedy, and S10 genotypes. B.C. is the abbreviation of the Bin Code (i.e., major functional categories). No.: identification number. ID: Δ: fold changes in drought-stressed plants (D) with respect to the control ones (C) (up: D/C, down:—C/D). new: not present in C; D.: disappeared, not present in drought stress. (DOCX) Click here for additional data file. S1 Fig One-dimensional SDS-Polyacrylamide Gel Electrophoresis (1D SDS-PAGE) of total protein of three genotypes of Lolium perenne . Left to right: Protein marker (lane 1), Control sample of Vigor genotype (lane 2,3), Drought sample of Vigor genotype (lane 4,5), Drought sample of Speedy genotype (lane 6), Control sample of Speedy genotype (lane 7), Control sample of S10 genotype (lane 8,9), Drought sample of S10 genotype (lane 10, 11). (TIF) Click here for additional data file. S1 Data (XLSX) Click here for additional data file.
|
How Are We Preparing Australian and Aotearoa New Zealand Teachers to Be Health Promotors? Examining Physical Activity, Sleep and Sun Safety in Initial Teacher Education | 9d326265-2cfd-40eb-9ac7-5fe82847ed0c | 11880413 | Health Promotion[mh] | Introduction Physical activity (PA), sleep and sun safety have all been identified as modifiable lifestyle behaviours in young people to improve health outcomes and reduce the risk of chronic disease, including cancer in later life [ , , ]. Physical inactivity is now the fourth leading risk factor associated with mortality globally . In Australia, less than one in 10 (8.9%) young people meet the national PA guidelines with COVID‐19 exacerbating the downward trend [ , , ]. In Aotearoa New Zealand (NZ), the 2022 PA report card data indicates that less than half (47%) of adolescents (aged 15–18) and only three in five (62%) children (aged 5–14 years) meet recommended PA guidelines . The link between PA, optimal sleep and improvements in health and well‐being is now well established . Poor sleep is also increasingly recognised as a significant risk factor for preventable morbidity and mortality, and is prevalent in young people in Australia with 27% of 12–13 years. olds and 52% of 16–17 years. olds not meeting sleep guidelines on school nights . In NZ, only 7 in 10 adults (69.1%) and almost 4 in 5 (78.1%) children are reported to be meeting the recommended amount of sleep, with disproportionately lower rates reported in disadvantaged groups . While the promotion of these behaviours remains vitally important for the maintenance of young people's health, PA often occurs outdoors, so the management of ultraviolet radiation exposure and associated health impacts is an important consideration when promoting this health behaviour. Despite decades of education and the fact that skin cancer is noted as Australasia's most preventable cancer , Australia and NZ have the highest rates of skin cancer in the world . There is consensus that sun exposure in the early years of life significantly increases the likelihood of developing skin cancer during adulthood, and for this reason, young people remain a particularly important group for primary prevention . It is estimated that young people spend one‐third of their time at school, making them an ideal setting for promoting health and engendering preventative health behaviours from an early age . Developing physical and health literacy skills in schooling years has been identified as a global health promotion strategy that could significantly reduce the risk of morbidity for the future population, with the World Health Organization (WHO) and United Nations Educational, Scientific and Cultural Organization (UNESCO) launching a global initiative to make all schools ‘health‐promoting schools’ . Health‐promoting schools create conditions that are conducive to health at all levels of the school through the implementation of policies, practices and targeted health curriculum, environment modification and connection with supports and local health networks . Central to this are teachers who remain highly influential role models in young people to develop healthy routines and preventative behaviours . However, research indicates that there are significant barriers for teachers in schools when attempting to increase young people's PA levels and reduce ultraviolet radiation exposure and improve sleep . Supplementing existing initial teacher education programs to include targeted health promotion content to upskill preservice teachers before they enter schools has been reported to be a potentially feasible and sustainable strategy to improve the provision of knowledge and practices for preventative health behaviours in Australian and NZ schools [ , , ]. In Victoria, researchers have trialled the ‘Transform Ed!’ program which is embedded into the initial teacher education degree and was designed to equip primary preservice teachers with the skills to be able to provide more meaningful PA experiences and outcomes. After the program, preservice teachers reported they were more willing, confident and competent to implement PA pedagogical strategies and perceived less barriers . Other research with preservice teachers ( n = 275) at one Australian University indicated significant gaps in ultraviolet radiation awareness and perceived knowledge, skills and confidence to teach sun safety . After attending a brief university‐based intervention, preservice teachers ( n = 161) indicated they felt the intervention increased their awareness of the dangers of overexposure to ultraviolet radiation, with many feeling more knowledgeable, skilled and confident to teach sun safety . Preservice teachers supported a need for more consistent sun protection messaging across Australian schools with greater emphasis on ultraviolet radiation education and tailored curriculum implementation, rather than compliance management. The intervention was well‐received by preservice teachers and there was unanimous agreement among participants that all initial teacher education programs across Australia should include this targeted health promotion material so that all graduate teachers are properly equipped to protect young people from overexposure to ultraviolet radiation . While research supports the value of sleep education for schoolteachers and young people , there is limited research evaluating the value of sleep education in initial teacher education for preservice teachers warranting further research. While PA, sleep and sun safety have been identified as important modifiable health behaviours to prevent the onset of chronic disease and schools have been identified as an important setting for health promotion and primary prevention; little is known about how teacher education programs across Australia and NZ are preparing future teachers to deliver physical activity, sleep and sun safety (PASS) education in early childhood and school settings. Hence, this study investigated teacher educators' perceptions of existing early childhood, primary and secondary teacher education programs in Australia and NZ to obtain insights into how their programs are preparing future teachers to teach PASS education.
Methodology 2.1 Aims This study aimed to: Examine the estimated number of courses and amount of time being spent on PASS within existing initial teacher education programs in Australia and NZ. Summarise Australian and NZ teacher educators' perceptions of graduates' knowledge, confidence and competence in delivering PASS content. Explore differences according to country (Australia/NZ), teaching specialisation (Health and Physical Education/other), and discipline (early childhood/primary/secondary) in the number of courses and time spent teaching PASS content, as well as teacher educators' perceptions of graduates' knowledge, confident and competence. 2.2 Study Design and Participants This study used an electronic cross‐sectional survey administered Nov 2023‐June 2024. Recruitment invitations were circulated via various social media platforms, relevant professional associations and emailed to leads of initial teacher education programs in Australia ( n = 48) and NZ ( n = 25). From here forth, entire initial teacher education programs or degrees will be referred to as ‘program’ and courses or units within the initial teacher education programs will be referred to as ‘course’. Invitations contained an online link and QR code to the project description and consent information page. Participants could also opt‐in to go in a prize draw to win a smartwatch. A total of ( n = 187) consented to being involved in the study, however, ( n = 6) were excluded from the study as they confirmed they were not employed at an institution that offered an initial teacher education program in Australia or NZ. A further ( n = 83) were excluded due to incomplete data where they did not answer a minimum of five survey items. The final sample of ( n = 98) was included in the analysis. 2.3 Survey Participants completed the anonymous electronic Qualtrics survey that was designed for the purposes of this study. The survey contained 30 questions (seven × demographic questions, 23 × attitudinal questions related to teacher education). To ensure that participants could remain anonymous, they were not required to provide the name of their institution, only their state/country, role, experience and teaching specialisation. For questions where participants had to estimate numerical values for time spent teaching content or the number of courses included in programs, a response slider was provided. For attitudinal questions, participants were provided with statements and a 7‐point scale (ranging from strongly disagree/disagree/don't know/agree/strongly agree category) as has been suggested as the most useful for accurately collecting attitudinal data . 2.4 Analysis Survey data was downloaded from Qualtrics and imported into IBM SPSS software 29 (SPSS Inc. Chicago, IL) for quantitative analysis. Shapiro–Wilk tests were used for normality testing. Descriptive and frequency analysis was used to examine features of the data. For analysis of the teacher educators' attitudinal response data, variables were collapsed into three categories (disagree/don't know/agree). To examine differences between respondents from health and physical education backgrounds and those from other teaching areas, the teaching specialisation variable was collapsed into two groups (health and physical education/other). For non‐parametric ordinal data, independent‐samples Kruskal‐Wallis H tests with pairwise comparisons ( α > 0.05) were used to explore group differences in the data based on country, state, teaching specialisation and discipline. Friedman ANOVA with Wilcoxon post hoc tests ( α > 0.05) were used to explore differences in teacher educators' perceptions of PASS content and delivery in the initial teacher education programs.
Aims This study aimed to: Examine the estimated number of courses and amount of time being spent on PASS within existing initial teacher education programs in Australia and NZ. Summarise Australian and NZ teacher educators' perceptions of graduates' knowledge, confidence and competence in delivering PASS content. Explore differences according to country (Australia/NZ), teaching specialisation (Health and Physical Education/other), and discipline (early childhood/primary/secondary) in the number of courses and time spent teaching PASS content, as well as teacher educators' perceptions of graduates' knowledge, confident and competence.
Study Design and Participants This study used an electronic cross‐sectional survey administered Nov 2023‐June 2024. Recruitment invitations were circulated via various social media platforms, relevant professional associations and emailed to leads of initial teacher education programs in Australia ( n = 48) and NZ ( n = 25). From here forth, entire initial teacher education programs or degrees will be referred to as ‘program’ and courses or units within the initial teacher education programs will be referred to as ‘course’. Invitations contained an online link and QR code to the project description and consent information page. Participants could also opt‐in to go in a prize draw to win a smartwatch. A total of ( n = 187) consented to being involved in the study, however, ( n = 6) were excluded from the study as they confirmed they were not employed at an institution that offered an initial teacher education program in Australia or NZ. A further ( n = 83) were excluded due to incomplete data where they did not answer a minimum of five survey items. The final sample of ( n = 98) was included in the analysis.
Survey Participants completed the anonymous electronic Qualtrics survey that was designed for the purposes of this study. The survey contained 30 questions (seven × demographic questions, 23 × attitudinal questions related to teacher education). To ensure that participants could remain anonymous, they were not required to provide the name of their institution, only their state/country, role, experience and teaching specialisation. For questions where participants had to estimate numerical values for time spent teaching content or the number of courses included in programs, a response slider was provided. For attitudinal questions, participants were provided with statements and a 7‐point scale (ranging from strongly disagree/disagree/don't know/agree/strongly agree category) as has been suggested as the most useful for accurately collecting attitudinal data .
Analysis Survey data was downloaded from Qualtrics and imported into IBM SPSS software 29 (SPSS Inc. Chicago, IL) for quantitative analysis. Shapiro–Wilk tests were used for normality testing. Descriptive and frequency analysis was used to examine features of the data. For analysis of the teacher educators' attitudinal response data, variables were collapsed into three categories (disagree/don't know/agree). To examine differences between respondents from health and physical education backgrounds and those from other teaching areas, the teaching specialisation variable was collapsed into two groups (health and physical education/other). For non‐parametric ordinal data, independent‐samples Kruskal‐Wallis H tests with pairwise comparisons ( α > 0.05) were used to explore group differences in the data based on country, state, teaching specialisation and discipline. Friedman ANOVA with Wilcoxon post hoc tests ( α > 0.05) were used to explore differences in teacher educators' perceptions of PASS content and delivery in the initial teacher education programs.
Results A final sample of ( n = 98) teacher educators were included in the data analysis; 82% of were employed at Australian institutions with the remaining participants from NZ, which is representative of each country's populations (26 million and 5 million, respectively) . Role in the institutions included: teacher/tutor (20%), course coordinator (46%), program coordinator (13%), discipline lead (7%) and Head/Dean of School (10%). Their years of teaching experience varied from 5 to 15 years and their specialisation was diverse and included: English, Maths, Health and Physical Education, Science, Humanities and Social Sciences, the Arts, Technologies and Languages among others (Table ). 3.1 Number of Courses and Time Spent on PASS Content in Initial Teacher Education Programs Considering entire initial teacher education programs at institutions, Friedman ANOVA with post hoc Wilcoxon signed rank tests indicated significant differences between PA, sleep and sun safety in terms of the number of courses which include PASS content, time spent in hours teaching PASS content, and time spent teaching preservice teachers how to plan and teach PASS content ( p < 0.001) (Table ). The number of courses within the entire initial teacher education programs that focus on PA (mean = 4.96, SD = 4.57), was significantly higher than for sun safety education (mean = 2.61, SD = 4.07) and sleep education (mean = 1.09, SD = 3.10). While only 2% of participants reported that programs at their institution include no courses which contain PA education, 20% of participants reported no courses included sleep education and 11% reported none that included sun safety education. Time in hours spent on PA education (mean = 26.09, SD = 25.69) across entire programs far outweighed time spent on sun safety education (mean = 8.48, SD = 18.77) and sleep education (mean = 5.79, SD = 15.21). When estimating the amount of time spent in hours educating preservice teachers on how to plan and effectively deliver education, PA again was significantly higher (mean = 27.09, SD = 28.02) than sun safety (mean = 5.95, SD = 14.29) and sleep (mean = 4.18, SD = 9.91). A total of 4% of participants reported their programs spent no time on educating teachers how to effectively plan and deliver lessons that focus on PA, whereas over 21% of participants reported that their entire programs spent no time at all teaching preservice teachers how to plan and effectively delivery sun safety and sleep content. 3.2 Graduates' Knowledge, Confidence and Competence to Teach PASS Content While most teacher educators believed that their graduates received sufficient training on PA education (65.5%) and the associated health benefits (78.6%), the burden associated with inactivity (72.4%), and understanding of current guidelines (58%); these numbers were far lower for sun safety and sleep ( p < 0.001) (Table ). In addition, > 40% of participants reported they did not know if their graduates received sufficient sun safety or sleep training, nor if they understood the associated benefits (35% and 37%, respectively), risks (34% and 34%, respectively) and guidelines (38% and 48%, respectively). Most participants also indicated that they believed their graduates were confident (71.4%) and competent (70.4%) to plan and teach lessons that focused on PA education. However, this was significantly different to their perceptions of graduates' confidence and competence to teach lessons focused on sun safety ( p < 0.001) and sleep ( p < 0.001) content. In terms of sun safety, only 41.8% felt their graduates were confident and 45.9% felt they were competent. When asked about sleep, most of the sample (> 70%) reported they did not agree or did not know if their graduates were confident to plan and teach lessons that focus on sleep education. 3.3 Group Differences Based on Country, Specialisation and Discipline 3.3.1 Country There were no significant differences between Australia and NZ other than the number of courses offered in the program that focused on sleep ( H = 4.47; p = 0.035); and how many hours were spent on PA content ( H = 3.87; p = 0.049), with Australia having higher means in both than NZ. There was no difference between states within Australia in terms of time spent on PASS content or perceptions of their graduates' knowledge, confidence or competence ( p > 0.05). 3.3.2 Teaching Specialisation Teacher educators with health and physical education specialisation had more favourable perspectives when compared to those from other learning areas about their own graduates' knowledge of PA benefits ( p = 0.002), risks ( p < 0.001) and guidelines ( p = 0.008) and ability to plan and teach lessons focusing on this content ( p = 0.008). However, this was not the case for sun safety ( p > 0.05) or sleep ( p > 0.05) and there were no other differences in perspectives based on time spent on PASS content or graduates' knowledge, confidence and competence to teach sun safety or sleep content ( p > 0.05). 3.3.3 Discipline There was no overall difference in the number of courses offered that included PASS content between early childhood, primary and secondary disciplines for PA ( p > 0.05) or sun safety ( p > 0.05). However, early childhood teacher educators reported higher number of courses that included sleep content ( p = 0.02) and higher number of hours allocated to both sleep ( p < 0.01) and sun safety ( p < 0.01) education. There was no difference across disciplines in terms of graduates' knowledge of PA benefits, risks and guidelines ( p > 0.05). However, teacher educators from the early childhood discipline perceived their graduates to have higher knowledge of sun safety and sleep benefits, risks and guidelines than both primary ( p < 0.01) and secondary ( p = 0.02) programs.
Number of Courses and Time Spent on PASS Content in Initial Teacher Education Programs Considering entire initial teacher education programs at institutions, Friedman ANOVA with post hoc Wilcoxon signed rank tests indicated significant differences between PA, sleep and sun safety in terms of the number of courses which include PASS content, time spent in hours teaching PASS content, and time spent teaching preservice teachers how to plan and teach PASS content ( p < 0.001) (Table ). The number of courses within the entire initial teacher education programs that focus on PA (mean = 4.96, SD = 4.57), was significantly higher than for sun safety education (mean = 2.61, SD = 4.07) and sleep education (mean = 1.09, SD = 3.10). While only 2% of participants reported that programs at their institution include no courses which contain PA education, 20% of participants reported no courses included sleep education and 11% reported none that included sun safety education. Time in hours spent on PA education (mean = 26.09, SD = 25.69) across entire programs far outweighed time spent on sun safety education (mean = 8.48, SD = 18.77) and sleep education (mean = 5.79, SD = 15.21). When estimating the amount of time spent in hours educating preservice teachers on how to plan and effectively deliver education, PA again was significantly higher (mean = 27.09, SD = 28.02) than sun safety (mean = 5.95, SD = 14.29) and sleep (mean = 4.18, SD = 9.91). A total of 4% of participants reported their programs spent no time on educating teachers how to effectively plan and deliver lessons that focus on PA, whereas over 21% of participants reported that their entire programs spent no time at all teaching preservice teachers how to plan and effectively delivery sun safety and sleep content.
Graduates' Knowledge, Confidence and Competence to Teach PASS Content While most teacher educators believed that their graduates received sufficient training on PA education (65.5%) and the associated health benefits (78.6%), the burden associated with inactivity (72.4%), and understanding of current guidelines (58%); these numbers were far lower for sun safety and sleep ( p < 0.001) (Table ). In addition, > 40% of participants reported they did not know if their graduates received sufficient sun safety or sleep training, nor if they understood the associated benefits (35% and 37%, respectively), risks (34% and 34%, respectively) and guidelines (38% and 48%, respectively). Most participants also indicated that they believed their graduates were confident (71.4%) and competent (70.4%) to plan and teach lessons that focused on PA education. However, this was significantly different to their perceptions of graduates' confidence and competence to teach lessons focused on sun safety ( p < 0.001) and sleep ( p < 0.001) content. In terms of sun safety, only 41.8% felt their graduates were confident and 45.9% felt they were competent. When asked about sleep, most of the sample (> 70%) reported they did not agree or did not know if their graduates were confident to plan and teach lessons that focus on sleep education.
Group Differences Based on Country, Specialisation and Discipline 3.3.1 Country There were no significant differences between Australia and NZ other than the number of courses offered in the program that focused on sleep ( H = 4.47; p = 0.035); and how many hours were spent on PA content ( H = 3.87; p = 0.049), with Australia having higher means in both than NZ. There was no difference between states within Australia in terms of time spent on PASS content or perceptions of their graduates' knowledge, confidence or competence ( p > 0.05). 3.3.2 Teaching Specialisation Teacher educators with health and physical education specialisation had more favourable perspectives when compared to those from other learning areas about their own graduates' knowledge of PA benefits ( p = 0.002), risks ( p < 0.001) and guidelines ( p = 0.008) and ability to plan and teach lessons focusing on this content ( p = 0.008). However, this was not the case for sun safety ( p > 0.05) or sleep ( p > 0.05) and there were no other differences in perspectives based on time spent on PASS content or graduates' knowledge, confidence and competence to teach sun safety or sleep content ( p > 0.05). 3.3.3 Discipline There was no overall difference in the number of courses offered that included PASS content between early childhood, primary and secondary disciplines for PA ( p > 0.05) or sun safety ( p > 0.05). However, early childhood teacher educators reported higher number of courses that included sleep content ( p = 0.02) and higher number of hours allocated to both sleep ( p < 0.01) and sun safety ( p < 0.01) education. There was no difference across disciplines in terms of graduates' knowledge of PA benefits, risks and guidelines ( p > 0.05). However, teacher educators from the early childhood discipline perceived their graduates to have higher knowledge of sun safety and sleep benefits, risks and guidelines than both primary ( p < 0.01) and secondary ( p = 0.02) programs.
Country There were no significant differences between Australia and NZ other than the number of courses offered in the program that focused on sleep ( H = 4.47; p = 0.035); and how many hours were spent on PA content ( H = 3.87; p = 0.049), with Australia having higher means in both than NZ. There was no difference between states within Australia in terms of time spent on PASS content or perceptions of their graduates' knowledge, confidence or competence ( p > 0.05).
Teaching Specialisation Teacher educators with health and physical education specialisation had more favourable perspectives when compared to those from other learning areas about their own graduates' knowledge of PA benefits ( p = 0.002), risks ( p < 0.001) and guidelines ( p = 0.008) and ability to plan and teach lessons focusing on this content ( p = 0.008). However, this was not the case for sun safety ( p > 0.05) or sleep ( p > 0.05) and there were no other differences in perspectives based on time spent on PASS content or graduates' knowledge, confidence and competence to teach sun safety or sleep content ( p > 0.05).
Discipline There was no overall difference in the number of courses offered that included PASS content between early childhood, primary and secondary disciplines for PA ( p > 0.05) or sun safety ( p > 0.05). However, early childhood teacher educators reported higher number of courses that included sleep content ( p = 0.02) and higher number of hours allocated to both sleep ( p < 0.01) and sun safety ( p < 0.01) education. There was no difference across disciplines in terms of graduates' knowledge of PA benefits, risks and guidelines ( p > 0.05). However, teacher educators from the early childhood discipline perceived their graduates to have higher knowledge of sun safety and sleep benefits, risks and guidelines than both primary ( p < 0.01) and secondary ( p = 0.02) programs.
Discussion To the authors’ knowledge, this study was the first study to investigate initial teacher educators' perceptions of their existing programs in relation to the amount of content and delivery of PASS education in Australia and NZ. Findings indicate there were minimal differences between Australian and NZ programs, other than the number of courses that focus on sleep and how many hours were spent on PA content. There was also no difference between states within Australia in terms of time spent on PASS content or perceptions of their graduates' knowledge, confidence or competence. However, there was significant variance in what is being offered in terms of the amount of content, curriculum focus and perceived teacher preparedness to teach PASS‐related content. Based on the findings, most initial teacher education programs in Australia and NZ spent very little (< 3 h to no time) in the entire initial teacher education program for each of the PASS areas. With between 10%–32% of participants reporting that they do not feel that their graduates are confident and competent to teach lessons that focus on PASS content, it is indicated that the current amount of content within existing programs in Australia and NZ is insufficient. In both Australia and NZ, PA had the highest coverage in programs, followed by sun safety and sleep. Teacher educators also perceived their graduates had the least knowledge of benefits, risks and guidelines and perceived confidence and competence to plan and deliver sleep‐related content. This may be due to the fact that the promotion of PA and sun safety and related policies has been a focus in schools for the last couple of decades , whereas sleep is relatively new on the public health agenda with Australia only including sleep in the children and young people's 24‐h movement guidelines in 2018 . These findings could speak to the need for tertiary institutions to examine and modify their existing initial teacher education programs to include more comprehensive sleep education content. Of concern, a large proportion of the sample (ranging from 15%–36% depending on the question) reported they “did not know” how many courses or hours were allocated to PASS education in their program. However, as a large proportion of the sample (approximately 65%) were not program coordinators, discipline leads or heads of school and based on their role in the institution may only have knowledge of the courses they teach and not deep knowledge of the whole program. Regardless, these findings could highlight a potential need for initial teacher education providers to cross reference all course content via regular review to ensure that all teachers and educators who are teaching these programs are aware of what is covered in the broader program and identify existing gaps. Examination of group differences between perceptions of health and physical education teacher educators and those in other specialisations revealed no significant difference in their estimates of the amount of PA content in their programs. This was an unexpected finding as those with a health and physical education specialisation had more favourable perceptions of their preservice teachers' preparedness to plan and deliver PA‐related lessons and their knowledge of PA benefits, risks and guidelines. However, there were no differences in perspectives on sun safety or sleep when educators were grouped by specialisation. Our findings also indicated no overall differences between early childhood, primary and secondary programs in the number of courses that include PA or sun safety education, but there was for sleep with early childhood and primary having higher means than secondary. In addition, early childhood educators reported a higher number of allocated hours for both sleep and sun safety and were also more favourable in their reporting about their graduates' knowledge of the risks, guidelines and benefits of these health behaviours. In Australia, this could be a result of the fact that early childhood settings are regulated by national law in which the Australian Children's Education & Care Quality Authority (ACECQA) sets a national quality standard for children's care with particular standards for health, well‐being and safety . Accreditation requirements for early childhood initial teacher education programs may reflect these requirements. While primary and secondary have similar accreditation requirements and professional teaching standards set by the Australian Institute for Teaching and School Leadership (AITSL) , there is potentially less prescriptive guidance in terms of some of these health behaviours leading to institutions potentially placing larger focuses on other content more closely aligned to the wording in the teaching standards. For national curriculum used by Australian and NZ school teachers set by the Australian Curriculum, Assessment and Reporting Authority (ACARA) and the Ministry of Education (MoE) in NZ , PA and personal safety (including sun) are most explicitly mentioned in the health and physical education learning area. In Australia, PA is a requirement at every year level; however, sun safety is a ‘suggested topic’ and not a requirement at any year level. Cancer Council Australia and Cancer Society New Zealand's SunSmart Schools and Early Childhood Program require registered members to include sun protection in the curriculum for all year levels . However, not all schools or early childhood services are registered members and sun education and protection strategies remain inconsistent across schools with school communities requiring additional support and engagement to effectively implement programs . Sleep as a topic is not explicitly mentioned by ACARA or MoE in either health or physical education curriculum . This limits guidance for schools for when and where sleep education should (or could), be taught in a school program and potentially could lead to sleep education being omitted from school curriculum programs. Health and physical education as a curriculum learning area has historically had the acquisition of movement skills at its core , which is likely the reason for schools prioritising PA education over health behaviours as it is more prescriptive. As the ITE program curriculum is modified based on sociocultural changes in schools and national curriculum, this may be a reason for the prioritisation of PA over other health behaviours in ITE programs in Australia and NZ. With contemporary advancements in technology, media and their relationship with health behaviours and risks, all school curriculum learning areas have gone through significant reform and now include education focusing on young peoples' personal and social capability and their relationships with the outside world . We therefore argue that knowledge of current guidelines, risks and benefits related to health behaviours is particularly important for all preservice teachers regardless of where they study, the program, discipline or specialisation. However, findings of this study indicate that different institutions across Australia are offering different amounts of PASS content within initial teacher education programs and also how much time is spent teaching preservice teachers to plan and teach PASS‐related content. This gap may lead to graduate teachers entering schools with little understanding of some, or all of health behaviours, associated risks and broader health implications for young people which is of significant concern, as teachers have been shown to be influential health promotors in both primary and secondary schools . Our study provides interesting insights into existing initial teacher educator programs in Australia and NZ. However, there are some limitations that should be noted. First, as the overarching aim of the study was to examine PASS education in existing initial teacher education programs, rather than compare institutional offerings, participant institution data was not collected. While it allowed anonymity, it is a noted limitation of this study as it did not allow direct comparison of what individual institutions are offering within their initial teacher education program. Furthermore, it did not allow us to determine if multiple staff from one institution have completed the survey, leading to potential overlap of data. Second, the study included a relatively small sample of teacher educators ( n = 98), which may limit the generalisability of findings. Third, due to the limited scope of this small study, we were only able to focus on three behaviours which are known to reduce risk of developing cancer or chronic disease in later life [ , , ]. However, it is important to note that there are other modifiable health behaviours that reduce cancer and chronic disease risk including nutrition, smoking, alcohol and other substance use and mental health are also important and warrant further research.
Conclusion This novel study was the first to investigate initial teacher educators' perceptions of their existing early childhood, primary and secondary initial teacher education programs in relation to the amount of content and delivery of PASS education in Australia and NZ. The findings revealed there was significant variance in what is being offered in terms of the amount of content, curriculum focus and perceived teacher preparedness to teach PASS‐related content. Consistently, time in hours spent on PA education across the entire initial teacher education programs far outweighed time spent on sun safety education and sleep education. Many teacher educators revealed their programs had little or no sleep or sun safety content and did not agree their graduates were confident to plan and deliver sun safety and sleep content. As all teachers need to have good knowledge of lifestyle‐related health behaviours to support young people's health effectively and efficiently, initial teacher education providers are encouraged to review their existing programs to explore how they can best prepare their graduate teachers to be health promotors including resourcing, planning, reviewing and modifying existing programs. Supplementing existing initial teacher education programs to include greater PASS education to upskill preservice teachers before they enter schools may provide a feasible and sustainable health promotion and cancer prevention strategy in Australian and NZ schools, however further research is needed to further examine institutional differences and the most efficient and effective ways to integrate lifestyle behaviour‐related content into existing initial teacher education programs. There is a need to examine, refine and modify public health content in Australian and NZ teacher education programs to assist with delivering consistent public health messages in schools via teachers. Further research investigating barriers and facilitators for modifying initial teacher education programs is needed.
All authors are responsible for reported research and have participated in the concept and design and/or data collection and/or analysis and interpretation of data, drafting or revising, and have approved this manuscript as submitted.
The study was approved by the relevant university Human Research Ethics Committees in Australia and New Zealand (Approval number: A231855).
The authors declare no conflicts of interest.
|
The predictive value of changes in | e3c48b51-e9ca-4a09-ae00-8813905ec398 | 11867325 | Cardiovascular System[mh] | Cancer and heart diseases are the primary causes of mortality, and chemotherapy has substantially enhanced the survival rates of cancer patients by reducing deaths and improving overall prognosis . Nevertheless, the medications utilized in cancer treatment can have adverse effects on patient health, particularly cardiac toxicity, which stands as one of the most detrimental complications of cancer therapy. Hence, close monitoring of cardiac toxicity during treatment is recommended. Cancer therapy-related cardiovascular toxicity (CTR-CVT) encompasses a spectrum of conditions such as cancer therapy-related cardiac dysfunction (CTRCD), coronary artery disease, valvular heart disease, arrhythmias, hypertension, thrombosis and thromboembolic diseases, peripheral artery disease, bleeding complications, pulmonary hypertension, and pericardial diseases . CTRCD captures the broad spectrum of possible presentations, such as cardiac injury, cardiomyopathy, and heart failure . CTRCD commonly affects 9.3–43.8% of patients undergoing anthracycline-based chemotherapy . Anthracycline-based chemotherapies (AC) are prevalent anti-tumor medications and pivotal components of various chemotherapy protocols, rendering them extensively employed. AC-induced cardiac and endothelial dysfunction exhibit dose-dependent impacts, leading to diverse forms of cardiovascular impairment . Anthracycline drugs induce cardiotoxicity by instigating the binding of reactive oxygen species and topoisomerase IIβ, lipid peroxidation, inflammatory responses, and mitochondrial impairment. Consequently, this prompts apoptosis, necrosis, and interstitial fibrosis of myocardial cells, elevating the risk of coronary endothelial dysfunction, left ventricular dysfunction, and heart failure. These ramifications notably heighten patient mortality rates and impact prognosis . The occurrence of CTR-CVT, particularly CTRCD, hinges on factors such as the cumulative dosage of anthracycline drugs, cardiovascular risk elements, and the duration of follow-up. However, findings from the SUSPOUR study unveiled a negligible alteration in left ventricular ejection fraction (LVEF) by ‒0.03% ± 7.9% over three years, with 85% of patients not meeting CTRCD criteria . Although data concerning early-onset CTRCD during anthracycline-based drug therapy remains scarce, current CTRCD incidence rates are relatively low. Alongside LVEF monitoring, assessing left ventricular global longitudinal strain (LV-GLS) via echocardiography has emerged as a fundamental imaging modality pre-, during-, and post-chemotherapy. Throughout cumulative anthracycline-based drug exposure, LV-GLS deteriorates earlier than LVEF, and a relative LV-GLS reduction exceeding 15% in asymptomatic patients serves as a criterion for asymptomatic CTRCD . However, in numerous investigations, this threshold is gauged by contrasting post-chemotherapy LV-GLS values with baseline strain assessments over an extended period, thus presenting limitations in early CTR-CVT prediction during the anthracycline-based drug cycle . Subcellular scrutiny of myocardial irregularities proves indispensable for the prompt and sensitive identification of cardiac dysfunction induced by anthracycline-based drugs. To monitor chemotherapy-induced cardiac toxicity, a range of methods are available to assess cardiovascular function, including echocardiography, electrocardiography (ECG), biomarkers, cardiac CT, and cardiac MRI . Nuclear cardiac imaging techniques have demonstrated significant value in tumor diagnosis and treatment evaluation. Specifically, positron emission tomography with 2-deoxy-2-[fluorine-18]fluoro-D-glucose integrated with computed tomography ( 18 F-FDG PET/CT) is widely utilized for baseline and follow-up assessments in cancer patients, notably those with lymphoma. The uptake and distribution of 18 F-FDG in tissues are influenced by glucose levels, fasting duration, and medications. Recent research has indicated that myocardial 18 F-FDG uptake is not solely reliant on glucose consumption. The retention of the tracer is influenced by the activity of hexose-6-phosphate dehydrogenase (H6PD) in the endoplasmic reticulum, which processes various hexoses, including FDG, thereby initiating the pentose phosphate pathway to maintain NADPH levels in response to oxidative stress induced by chemotherapy . Consequently, 18 F-FDG PET/CT can serve as an early screening tool for cardiac toxicity in lymphoma patients. Dourado et al. analyzed seventy lymphoma patients who underwent 18 F-FDG PET/CT examinations and assessed three myocardial uptake parameters—maximum standardized uptake value (SUVmax) of the left ventricle, heart-to-blood pool ratio, and heart-to-liver ratio—at baseline, mid-term, and post-treatment. Their findings revealed a significant increase in myocardial 18 F-FDG uptake during and after chemotherapy in lymphoma patients, highlighting the sensitivity and reliability of 18 F-FDG PET/CT as an imaging modality for detecting early metabolic changes indicative of cardiac toxicity . Presently, research on assessing cardiac toxicity related to tumor radio-chemotherapy utilizing 18 F-FDG PET/CT predominantly focuses on analyzing disparities in myocardial FDG uptake metabolic parameters. However, there exists limited investigation into the precise location, pattern, and significance of myocardial FDG uptake. Thus, this study endeavors to elucidate the supplementary utility of 18 F-FDG PET/CT in monitoring cardiac toxicity during tumor treatment by scrutinizing alterations in the location, pattern, and metabolic parameters of cardiac uptake before and after chemotherapy.
Study subjects A total of 366 patients diagnosed with lymphoma and treated with a 6-cycle anthracycline-based chemotherapy regimen at the Second Affiliated Hospital of Dalian Medical University between 1 July 2017 and 31 December 2022 were enrolled in this study. Each patient underwent baseline and post-6-cycle chemotherapy 18 F-FDG PET/CT scans, and complete case data were available. Data were accessed for research purposes on 30 June 2023. Inclusion criteria were as follows: (1) Pathologically confirmed lymphoma patients undergoing an anthracycline-based chemotherapy regimen, with baseline and post-6-cycle chemotherapy 18 F-FDG PET/CT imaging; (2) Comprehensive evaluation of cardiac medical history, including echocardiography, ECG, cardiac biomarkers (creatine kinase, troponin), and brain natriuretic peptide (BNP). Exclusion criteria included: (1) Poor quality of PET/CT images; (2) Incomplete collection of patient case data; (3) History of previous tumors and receipt of radiation or chemotherapy; (4) Poorly controlled severe diabetes; (5) Severe liver or kidney dysfunction; (6) Patients with cardiac lesions (e.g., tumors, granulomatous diseases, etc.). This study had approval from the Ethics Committee of the Second Hospital of Dalian Medical University. All enrolled patients provided written informed consent for their participation in the study. The study was conducted according to the Declaration of Helsinki. Additionally, access to information that could potentially identify individual participants post data collection was secured. Collection of clinical data Reviewing outpatient and inpatient medical records, as well as PET/CT examination records, to gather patients’ general clinical data, including age, gender, hypertension, diabetes, cardiac disease history, history of radiation and/or chemotherapy, purpose of PET/CT imaging, laboratory parameters, including White Blood Cell (WBC), Erythrocyte Sedimentation Rate (ESR), Lactate Dehydrogenase (LDH), Albumin (ALB), β2-microglobulin (β2-MG), Total Cholesterol (TC), Triglyceride (TG), High-density lipoprotein cholesterol (HDL-C), Low density lipoprotein cholesterol (LDL-C), calculating Neutrophil/lymphocyte ratio (NLR), final diagnosis outcomes, chemotherapy regimens, etc. Additionally, collect cardiac-related imaging data, such as ECG, echocardiography, coronary artery CT imaging, and coronary angiography results. 18 F-FDG PET/CT examination We employed the Philips Ingenuity TF PET/CT scanner for the assessments. The 18 F-FDG was produced and synthesized using the Sumitomo HM-10 cyclotron accelerator and the chemical synthesis module from PET CO., LTD. (Beijing), ensuring a radiochemical purity exceeding 95%. Patients refrained from eating for at least 12 hours before the procedure. Following the administration of 18 F-FDG at a dosage of 3.7–5.55 MBq/kg, patients rested in a dimly lit room for 60 minutes before undergoing PET/CT scans post-bladder voiding. The scan ranged from the skull base to the foot. Initially, CT scans were performed with parameters set at a voltage of 120 kV, current of 90 mA, rotation speed of 0.75s/rotation, and a matrix of 512 × 512. Subsequently, PET imaging followed with conditions set at a matrix of 144 × 144 and 1-minute acquisition for each bed position, totaling 8–10 bed positions. After attenuation correction and OSEM reconstruction, PET images were co-registered with CT images on the image processing workstation. Myocardial glucose uptake analysis According to the guidelines of the American Society of Nuclear Cardiology (ASNC) and based on previous research, two experienced nuclear medicine doctors reviewed 18 F-FDG PET/CT images of myocardial glucose uptake retrospectively [ – ]. They classified myocardial glucose uptake patterns into four types based on visual analysis: (1) No uptake, where overall uptake in the left ventricular myocardium was the same as or lower than uptake in the cardiac blood pool; (2) Diffuse uptake, where the left ventricular myocardium showed overall uptake with a fairly even distribution, without focal or significantly higher uptake; (3) Focal uptake, where certain areas of the left ventricular myocardium had higher uptake, while the rest had uptake equal to or lower than the blood pool; (4) Focal uptake on a diffuse uptake background, where certain segments of the myocardium showed higher uptake against a diffuse background of myocardial uptake. Myocardial glucose uptake could be categorized based on its location as left ventricular myocardial uptake, right ventricular myocardial uptake, or atrial myocardial uptake. Based on the patterns, locations, and characteristics of myocardial glucose uptake, it could be classified as either normal or abnormal cardiac uptake. Abnormal cardiac uptake was defined as: (1) Focal or focal uptake on a diffuse uptake in the left ventricle, excluding basal circumferential or semicircular uptake, focal uptake in the papillary muscle, and uniform uptake in the left ventricular lateral wall; (2) Right ventricular uptake higher than left ventricular uptake; (3) Atrial uptake higher than blood pool uptake when there was no or low uptake in the left ventricle; (4) Excluding normal uptake in specific locations, such as lipomatous hypertrophy of the interatrial septum (LHIS) or crista terminalis of the right atrium. Based on comparison of PET/CT images pre- and post-chemotherapy, changes in cardiac uptake patterns were classified as normal-normal, normal-abnormal, abnormal-normal, or abnormal-abnormal patterns. The normal-normal pattern constituted the normal change group, while the normal-abnormal, abnormal-normal, and abnormal-abnormal patterns comprised the abnormal change group. Quantitative analysis of cardiac and epicardial adipose tissue uptake On the fused PET/CT images, delineation of the epicardial adipose tissue (EAT) region of interest was performed using CT values ranging from ‒190 to ‒45 Hounsfield units (HU) between myocardium and pericardium . We selected the region adjacent to the origin of right coronary artery for measuring the activity of EAT. This site was less affected by ventricular FDG activity . The maximum adipose tissue ROI was outlined on three consecutive cross sections, and the SUVmax of ROI at each layer was recorded respectively, and the maximum value in the three layers was recorded. The MedEx software, provided by Beijing MedEx Technology Co., Ltd., was utilized to measure the SUVmax and the average standardized uptake value (SUVmean) in both the left ventricle and epicardial adipose tissue. Clinical follow-up From the initiation of tumor chemotherapy until January 2024, clinical follow-up was conducted to document CTR-CVT. CTR-CVT was defined as encompassing CTRCD, coronary artery disease, valvular heart disease, arrhythmias, hypertension, thrombosis and thromboembolic diseases, peripheral artery disease, bleeding complications, pulmonary hypertension, and pericardial diseases . Statistical analysis We conducted statistical analyses using SPSS Statistics 26.0. The normality of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed data are presented as mean ± standard deviation (X ± S), while non-normally distributed data are expressed as median (P25, P75). Categorical variables are depicted as frequencies and percentages (%). To compare differences between two independent samples, we utilized the t-test or Mann-Whitney U test, while the chi-square test was employed to compare rates among these samples. Spearman or Pearson correlation analysis was performed to examine the correlation between two independent samples. Multiple logistic regression analysis was utilized to establish a model for variable selection and intergroup predictive factors. The significance threshold was set at P < 0.05.
A total of 366 patients diagnosed with lymphoma and treated with a 6-cycle anthracycline-based chemotherapy regimen at the Second Affiliated Hospital of Dalian Medical University between 1 July 2017 and 31 December 2022 were enrolled in this study. Each patient underwent baseline and post-6-cycle chemotherapy 18 F-FDG PET/CT scans, and complete case data were available. Data were accessed for research purposes on 30 June 2023. Inclusion criteria were as follows: (1) Pathologically confirmed lymphoma patients undergoing an anthracycline-based chemotherapy regimen, with baseline and post-6-cycle chemotherapy 18 F-FDG PET/CT imaging; (2) Comprehensive evaluation of cardiac medical history, including echocardiography, ECG, cardiac biomarkers (creatine kinase, troponin), and brain natriuretic peptide (BNP). Exclusion criteria included: (1) Poor quality of PET/CT images; (2) Incomplete collection of patient case data; (3) History of previous tumors and receipt of radiation or chemotherapy; (4) Poorly controlled severe diabetes; (5) Severe liver or kidney dysfunction; (6) Patients with cardiac lesions (e.g., tumors, granulomatous diseases, etc.). This study had approval from the Ethics Committee of the Second Hospital of Dalian Medical University. All enrolled patients provided written informed consent for their participation in the study. The study was conducted according to the Declaration of Helsinki. Additionally, access to information that could potentially identify individual participants post data collection was secured.
Reviewing outpatient and inpatient medical records, as well as PET/CT examination records, to gather patients’ general clinical data, including age, gender, hypertension, diabetes, cardiac disease history, history of radiation and/or chemotherapy, purpose of PET/CT imaging, laboratory parameters, including White Blood Cell (WBC), Erythrocyte Sedimentation Rate (ESR), Lactate Dehydrogenase (LDH), Albumin (ALB), β2-microglobulin (β2-MG), Total Cholesterol (TC), Triglyceride (TG), High-density lipoprotein cholesterol (HDL-C), Low density lipoprotein cholesterol (LDL-C), calculating Neutrophil/lymphocyte ratio (NLR), final diagnosis outcomes, chemotherapy regimens, etc. Additionally, collect cardiac-related imaging data, such as ECG, echocardiography, coronary artery CT imaging, and coronary angiography results.
F-FDG PET/CT examination We employed the Philips Ingenuity TF PET/CT scanner for the assessments. The 18 F-FDG was produced and synthesized using the Sumitomo HM-10 cyclotron accelerator and the chemical synthesis module from PET CO., LTD. (Beijing), ensuring a radiochemical purity exceeding 95%. Patients refrained from eating for at least 12 hours before the procedure. Following the administration of 18 F-FDG at a dosage of 3.7–5.55 MBq/kg, patients rested in a dimly lit room for 60 minutes before undergoing PET/CT scans post-bladder voiding. The scan ranged from the skull base to the foot. Initially, CT scans were performed with parameters set at a voltage of 120 kV, current of 90 mA, rotation speed of 0.75s/rotation, and a matrix of 512 × 512. Subsequently, PET imaging followed with conditions set at a matrix of 144 × 144 and 1-minute acquisition for each bed position, totaling 8–10 bed positions. After attenuation correction and OSEM reconstruction, PET images were co-registered with CT images on the image processing workstation.
According to the guidelines of the American Society of Nuclear Cardiology (ASNC) and based on previous research, two experienced nuclear medicine doctors reviewed 18 F-FDG PET/CT images of myocardial glucose uptake retrospectively [ – ]. They classified myocardial glucose uptake patterns into four types based on visual analysis: (1) No uptake, where overall uptake in the left ventricular myocardium was the same as or lower than uptake in the cardiac blood pool; (2) Diffuse uptake, where the left ventricular myocardium showed overall uptake with a fairly even distribution, without focal or significantly higher uptake; (3) Focal uptake, where certain areas of the left ventricular myocardium had higher uptake, while the rest had uptake equal to or lower than the blood pool; (4) Focal uptake on a diffuse uptake background, where certain segments of the myocardium showed higher uptake against a diffuse background of myocardial uptake. Myocardial glucose uptake could be categorized based on its location as left ventricular myocardial uptake, right ventricular myocardial uptake, or atrial myocardial uptake. Based on the patterns, locations, and characteristics of myocardial glucose uptake, it could be classified as either normal or abnormal cardiac uptake. Abnormal cardiac uptake was defined as: (1) Focal or focal uptake on a diffuse uptake in the left ventricle, excluding basal circumferential or semicircular uptake, focal uptake in the papillary muscle, and uniform uptake in the left ventricular lateral wall; (2) Right ventricular uptake higher than left ventricular uptake; (3) Atrial uptake higher than blood pool uptake when there was no or low uptake in the left ventricle; (4) Excluding normal uptake in specific locations, such as lipomatous hypertrophy of the interatrial septum (LHIS) or crista terminalis of the right atrium. Based on comparison of PET/CT images pre- and post-chemotherapy, changes in cardiac uptake patterns were classified as normal-normal, normal-abnormal, abnormal-normal, or abnormal-abnormal patterns. The normal-normal pattern constituted the normal change group, while the normal-abnormal, abnormal-normal, and abnormal-abnormal patterns comprised the abnormal change group.
On the fused PET/CT images, delineation of the epicardial adipose tissue (EAT) region of interest was performed using CT values ranging from ‒190 to ‒45 Hounsfield units (HU) between myocardium and pericardium . We selected the region adjacent to the origin of right coronary artery for measuring the activity of EAT. This site was less affected by ventricular FDG activity . The maximum adipose tissue ROI was outlined on three consecutive cross sections, and the SUVmax of ROI at each layer was recorded respectively, and the maximum value in the three layers was recorded. The MedEx software, provided by Beijing MedEx Technology Co., Ltd., was utilized to measure the SUVmax and the average standardized uptake value (SUVmean) in both the left ventricle and epicardial adipose tissue.
From the initiation of tumor chemotherapy until January 2024, clinical follow-up was conducted to document CTR-CVT. CTR-CVT was defined as encompassing CTRCD, coronary artery disease, valvular heart disease, arrhythmias, hypertension, thrombosis and thromboembolic diseases, peripheral artery disease, bleeding complications, pulmonary hypertension, and pericardial diseases .
We conducted statistical analyses using SPSS Statistics 26.0. The normality of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed data are presented as mean ± standard deviation (X ± S), while non-normally distributed data are expressed as median (P25, P75). Categorical variables are depicted as frequencies and percentages (%). To compare differences between two independent samples, we utilized the t-test or Mann-Whitney U test, while the chi-square test was employed to compare rates among these samples. Spearman or Pearson correlation analysis was performed to examine the correlation between two independent samples. Multiple logistic regression analysis was utilized to establish a model for variable selection and intergroup predictive factors. The significance threshold was set at P < 0.05.
Clinical data of lymphoma patients This study included a total of 366 patients who met the specified inclusion and exclusion criteria. Among them, there were 185 male patients and 181 female patients. Hodgkin lymphoma was observed in 36 cases, whereas non-Hodgkin lymphoma was identified in 330 cases. Please refer to for more details. Comparison of clinical and cardiac ultrasound parameters before and after treatment in lymphoma patients Lymphoma patients underwent 6 cycles of chemotherapy, and the clinical and cardiac ultrasound parameters were compared before and after treatment. It was observed that after treatment, the levels of WBC, NLR, ESR, and β2-MG decreased significantly, with statistically significant differences (P < 0.05). After treatment, ALB, TC, TG, HDL-C, LDL-C, LAD, and LVDD significantly increased compared to before treatment, with statistically significant differences ( P < 0.05). See . Comparison of cardiac metabolic parameters before and after treatment in lymphoma patients using 18 F-FDG PET/CT Lymphoma patients underwent 6 cycles of chemotherapy, and the cardiac metabolic parameters using 18 F-FDG PET/CT were compared before and after treatment. It was observed that after treatment, both the left ventricular SUVmax and SUVmean significantly increased compared to before treatment, with statistically significant differences (P < 0.05). However, there was no significant difference in EAT SUVmax and EAT SUVmean before and after treatment (P > 0.05). Please refer to for details. Comparison of cardiac uptake in lymphoma patients before and after treatment using 18 F-FDG PET/CT Lymphoma patients underwent 6 cycles of chemotherapy, and the cardiac uptake patterns using 18 F-FDG PET/CT were compared before and after treatment. It was observed that after treatment, there was a significant decrease in the number of patients with no uptake in the left ventricle and a significant increase in the number of patients with diffuse uptake, with statistically significant differences (P < 0.05). Additionally, the number of abnormal cardiac uptakes significantly increased after treatment compared to before treatment, with statistically significant differences (P < 0.05). However, there was no significant change in the proportion of abnormal cardiac uptake sites before and after treatment (P > 0.05). Please refer to for details. Changes in cardiac uptake patterns before and after treatment in lymphoma The alterations in cardiac uptake patterns before and after treatment in lymphoma were assessed using 18 F-FDG PET/CT. The variations in uptake patterns in the left ventricle, right ventricle, and atrium are presented in . illustrates an instance of a lymphoma patient displaying abnormal uptake in the left ventricle (highlighted by the red arrow) after completing 6 cycles of chemotherapy. Male, 67 years old, diagnosed with diffuse large B-cell lymphoma, treated with R-CHOP for 6 cycles. – (MIP image, axial PET, CT and PET/CT fusion image): No abnormal uptake in the heart before lymphoma treatment (red arrow). – (MIP image, axial PET, CT and PET/CT fusion image): Focal abnormal uptake in the left ventricular apex after lymphoma treatment (red arrow). Analysis of different cardiac uptake pattern changes after treatment using 18 F-FDG PET/CT in lymphoma patients Following lymphoma treatment, patients were categorized into two groups based on changes in cardiac uptake patterns using 18 F-FDG PET/CT. The normal change group comprised 330 patients, while the abnormal change group consisted of 36 patients. Cardiac ultrasound results were then compared between these groups. The analysis revealed that the abnormal change group exhibited a higher left atrial diameter and lower left ventricular ejection fraction compared to the normal change group, with statistically significant differences (P < 0.05), as demonstrated in . The comparison between the normal change group and abnormal change group indicated that the SUVmax of the epicardial adipose tissue was notably higher in the abnormal change group compared to the normal change group, with statistically significant differences (P < 0.05), as outlined in . Analysis of different CTR-CVT groups after lymphoma treatment After lymphoma treatment, patients were categorized based on the presence or absence of CTR-CVT. The CTR-CVT group comprised 41 patients, while the non-CTR-CVT group consisted of 325 patients. Subsequently, cardiac ultrasound results were compared between these groups. The analysis revealed that the CTR-CVT group displayed higher left atrial diameter and left ventricular end-diastolic diameter compared to the non-CTR-CVT group. Additionally, the left ventricular ejection fraction was lower in the CTR-CVT group compared to the non-CTR-CVT group, with statistically significant differences (P < 0.05), as depicted in . illustrates an example of a newly developed atrial fibrillation patient after lymphoma treatment, showcasing changes in cardiac uptake patterns before and after treatment. Male, 51 years old, diagnosed with diffuse large B-cell lymphoma, treated with R-CHOP for 6 cycles. – (MIP image, axial PET, CT and PET/CT fusion image): No abnormal uptake in the atrium before lymphoma treatment (red arrow). – (MIP image, axial PET, CT and PET/CT fusion image): Abnormal uptake in the left atrium higher than the blood pool uptake after lymphoma treatment (red arrow). Following the categorization of patients based on the presence or absence of CTR-CVT, it was discovered that the proportion of abnormal uptake pattern changes was notably higher in the CTR-CVT group compared to the non-CTR-CVT group. Furthermore, the SUVmax and SUVmean of the epicardial adipose tissue were significantly elevated in the CTR-CVT group compared to the non-CTR-CVT group, with statistically significant differences (P < 0.05), as outlined in . Logistic regression analysis of risk factors for CTR-CVT after lymphoma treatment Logistic regression analysis was performed to identify the risk factors for CTR-CVT after lymphoma treatment. The results revealed that left ventricular diameter [OR = 1.177 (95% CI: 1.038, 1.335)], left ventricular ejection fraction [OR = 0.537 (95% CI: 0.382, 0.756)], epicardial adipose tissue (EAT) SUVmax [OR = 2.668 (95% CI: 0.842, 8.457)], and abnormal uptake pattern changes [OR = 3.564 (95% CI: 1.156, 10.994)] were identified as risk factors for CTR-CVT in lymphoma patients after treatment. Refer to for further details.
This study included a total of 366 patients who met the specified inclusion and exclusion criteria. Among them, there were 185 male patients and 181 female patients. Hodgkin lymphoma was observed in 36 cases, whereas non-Hodgkin lymphoma was identified in 330 cases. Please refer to for more details.
Lymphoma patients underwent 6 cycles of chemotherapy, and the clinical and cardiac ultrasound parameters were compared before and after treatment. It was observed that after treatment, the levels of WBC, NLR, ESR, and β2-MG decreased significantly, with statistically significant differences (P < 0.05). After treatment, ALB, TC, TG, HDL-C, LDL-C, LAD, and LVDD significantly increased compared to before treatment, with statistically significant differences ( P < 0.05). See .
18 F-FDG PET/CT Lymphoma patients underwent 6 cycles of chemotherapy, and the cardiac metabolic parameters using 18 F-FDG PET/CT were compared before and after treatment. It was observed that after treatment, both the left ventricular SUVmax and SUVmean significantly increased compared to before treatment, with statistically significant differences (P < 0.05). However, there was no significant difference in EAT SUVmax and EAT SUVmean before and after treatment (P > 0.05). Please refer to for details.
18 F-FDG PET/CT Lymphoma patients underwent 6 cycles of chemotherapy, and the cardiac uptake patterns using 18 F-FDG PET/CT were compared before and after treatment. It was observed that after treatment, there was a significant decrease in the number of patients with no uptake in the left ventricle and a significant increase in the number of patients with diffuse uptake, with statistically significant differences (P < 0.05). Additionally, the number of abnormal cardiac uptakes significantly increased after treatment compared to before treatment, with statistically significant differences (P < 0.05). However, there was no significant change in the proportion of abnormal cardiac uptake sites before and after treatment (P > 0.05). Please refer to for details.
The alterations in cardiac uptake patterns before and after treatment in lymphoma were assessed using 18 F-FDG PET/CT. The variations in uptake patterns in the left ventricle, right ventricle, and atrium are presented in . illustrates an instance of a lymphoma patient displaying abnormal uptake in the left ventricle (highlighted by the red arrow) after completing 6 cycles of chemotherapy. Male, 67 years old, diagnosed with diffuse large B-cell lymphoma, treated with R-CHOP for 6 cycles. – (MIP image, axial PET, CT and PET/CT fusion image): No abnormal uptake in the heart before lymphoma treatment (red arrow). – (MIP image, axial PET, CT and PET/CT fusion image): Focal abnormal uptake in the left ventricular apex after lymphoma treatment (red arrow).
18 F-FDG PET/CT in lymphoma patients Following lymphoma treatment, patients were categorized into two groups based on changes in cardiac uptake patterns using 18 F-FDG PET/CT. The normal change group comprised 330 patients, while the abnormal change group consisted of 36 patients. Cardiac ultrasound results were then compared between these groups. The analysis revealed that the abnormal change group exhibited a higher left atrial diameter and lower left ventricular ejection fraction compared to the normal change group, with statistically significant differences (P < 0.05), as demonstrated in . The comparison between the normal change group and abnormal change group indicated that the SUVmax of the epicardial adipose tissue was notably higher in the abnormal change group compared to the normal change group, with statistically significant differences (P < 0.05), as outlined in .
After lymphoma treatment, patients were categorized based on the presence or absence of CTR-CVT. The CTR-CVT group comprised 41 patients, while the non-CTR-CVT group consisted of 325 patients. Subsequently, cardiac ultrasound results were compared between these groups. The analysis revealed that the CTR-CVT group displayed higher left atrial diameter and left ventricular end-diastolic diameter compared to the non-CTR-CVT group. Additionally, the left ventricular ejection fraction was lower in the CTR-CVT group compared to the non-CTR-CVT group, with statistically significant differences (P < 0.05), as depicted in . illustrates an example of a newly developed atrial fibrillation patient after lymphoma treatment, showcasing changes in cardiac uptake patterns before and after treatment. Male, 51 years old, diagnosed with diffuse large B-cell lymphoma, treated with R-CHOP for 6 cycles. – (MIP image, axial PET, CT and PET/CT fusion image): No abnormal uptake in the atrium before lymphoma treatment (red arrow). – (MIP image, axial PET, CT and PET/CT fusion image): Abnormal uptake in the left atrium higher than the blood pool uptake after lymphoma treatment (red arrow). Following the categorization of patients based on the presence or absence of CTR-CVT, it was discovered that the proportion of abnormal uptake pattern changes was notably higher in the CTR-CVT group compared to the non-CTR-CVT group. Furthermore, the SUVmax and SUVmean of the epicardial adipose tissue were significantly elevated in the CTR-CVT group compared to the non-CTR-CVT group, with statistically significant differences (P < 0.05), as outlined in .
Logistic regression analysis was performed to identify the risk factors for CTR-CVT after lymphoma treatment. The results revealed that left ventricular diameter [OR = 1.177 (95% CI: 1.038, 1.335)], left ventricular ejection fraction [OR = 0.537 (95% CI: 0.382, 0.756)], epicardial adipose tissue (EAT) SUVmax [OR = 2.668 (95% CI: 0.842, 8.457)], and abnormal uptake pattern changes [OR = 3.564 (95% CI: 1.156, 10.994)] were identified as risk factors for CTR-CVT in lymphoma patients after treatment. Refer to for further details.
As cancer survival rates continue to improve, the cardiovascular side effects of cancer treatments significantly impact patient prognosis. Previous studies have highlighted a clear correlation between factors such as pre-existing cardiovascular diseases, cardiovascular risk factors, genetic predisposition, treatment regimens, age, and the risk of cardiovascular complications post-cancer treatment . The spectrum of CTR-CVT varies widely, ranging from asymptomatic reversible changes to life-threatening complications, encompassing heart failure, acute coronary syndrome, arrhythmias, valvular heart disease, pericardial disease, myocarditis, and thromboembolic events . Consequently, there’s a pressing need for in-depth exploration of the mechanisms underlying cancer and cancer therapy-induced cardiovascular diseases, establishment of appropriate diagnostic protocols, early identification of CTR-CVT, and implementation of effective treatment and prevention strategies. These endeavors constitute the primary focus of current research in onco-cardiology and serve as the cornerstone for interdisciplinary discussions among cardiology experts and oncology treatment teams. Cancer patients, particularly those at high risk for cardiovascular complications, necessitate standardized monitoring of CTR-CVT throughout the treatment journey to promptly identify associated side effects. This entails conducting baseline risk assessments prior to treatment initiation, monitoring for acute complications during treatment, and ensuring long-term follow-up for late-stage cardiovascular effects . Pre-treatment assessment involves gathering medical history, conducting physical examinations, performing ECGs, assessing cardiac biomarkers, including N-terminal pro-brain natriuretic peptide (NT-proBNP), and conducting echocardiography. Moreover, comprehensive documentation of the cancer treatment plan, encompassing conventional chemotherapy, targeted therapy, immunotherapy, or radiation therapy, is imperative. During treatment, cardiac oncology monitoring is tailored according to the patient’s cardiac risk profile, with echocardiography serving as the preferred imaging modality for tracking cardiac damage in cancer patients. In cases where cardiac toxicity is suspected, additional diagnostic modalities such as cardiac magnetic resonance imaging or cardiac catheterization may be warranted. Particularly, patients at elevated risk for late-stage complications from radiation therapy may undergo coronary angiography to evaluate coronary artery disease . Nuclear imaging techniques can unveil metabolic alterations at the molecular level and play a pivotal role in detecting myocardial cell toxicity before irreversible damage occurs. 18 F-FDG PET/CT can identify abnormal myocardial cell survival and assess myocardial inflammatory reactions induced by cancer treatment, though often necessitating suppression of myocardial glucose metabolism . Anthracycline drugs can disrupt normal mitochondrial oxidative metabolism, thereby causing aberrant energy metabolism in myocardial cells. Anthracyclines’ cardiotoxicity involves an accelerated generation of reactive oxygen species. This oxidative damage has been found to accelerate the expression of hexose-6P-dehydrogenase (H6PD), that channels glucose-6-phosphate (G6P) through the pentose phosphate pathway (PPP) confined within the endoplasmic/sarcoplasmic reticulum (SR) . The direct correlation between cardiac FDG uptake and oxidative stress indexes supports the potential role of FDG-PET as an early biomarker of Doxorubicin oxidative damage . Consequently, 18 F-FDG PET/CT is well-suited for detecting anthracycline-induced myocardial cell damage . From an economic and practical standpoint, only a small subset of cancer patients undergoes dedicated 18 F-FDG cardiac metabolic or inflammation imaging. However, can oncologic 18 F-FDG imaging serve as a means to evaluate the toxic effects of tumor drugs on the cardiovascular system while concurrently assessing tumor diagnosis and efficacy? Dourado et al. conducted a study comparing the cardiac uptake SUV values of 70 lymphoma patients before and after treatment using 18 F-FDG PET/CT imaging and observed a significant increase in cardiac uptake of 18 F-FDG post-treatment . Sarocchi et al. found DOX-containing chemotherapy causes an increase in cardiac 18 FDG uptake, which is associated with a decline in LVEF . Similarly, our research demonstrated a notable increase in left ventricular 18 F-FDG uptake in lymphoma patients treated with anthracycline-based chemotherapy compared to pre-treatment levels. In our study, abnormal cardiac uptake was classified based on location, revealing changes in uptake in the left ventricle, right ventricle, and atrium. Kim et al. investigated 121 breast cancer patients and compared right ventricular wall 18 F-FDG uptake before and after treatment using 18 F-FDG PET/CT. They discovered that breast cancer patients receiving anthracycline drugs or trastuzumab exhibited increased uptake in the right ventricular wall, which correlated with cardiac toxicity . Hence, abnormal uptake and alterations in SUV values in different cardiac regions may reflect the cardiotoxic effects of tumor drugs to some extent. Additionally, Vinogradskiy et al. explored the relationship between radiation dose to the heart and overall survival in lung cancer patients undergoing chemotherapy. They found that changes in SUV in the heart post-treatment were significant factors for dose response and predictive of overall survival. This suggests that alterations in cardiac metabolism may serve as early predictive indicators of clinical outcomes . In our study, we observed a significant decrease in the proportion of patients exhibiting no uptake pattern in the left ventricle after treatment, while those with a diffuse uptake pattern significantly increased. This suggests that post-chemotherapy, there’s a heightened myocardial uptake of 18 F-FDG among patients, albeit possibly representing physiological uptake in the case of diffuse uptake. Subsequently, we noted a notable increase in the number of patients displaying abnormal uptake patterns post-treatment, indicating potential abnormal myocardial 18 F-FDG metabolism during the treatment process. These uptake patterns and changes in SUV may indeed mirror the cardiotoxic effects of tumor drugs. Based on the alterations in patterns before and after treatment, we segregated patients into the normal change group and the abnormal change group. Remarkably, we observed that patients in the abnormal change group exhibited an augmented left atrial diameter and reduced left ventricular ejection fraction post-treatment compared to the normal change group, suggesting potential impairment of cardiac function in the abnormal change group. Furthermore, in the abnormal change group, the EAT SUVmax surpassed that of the normal change group, indicating shifts in the uptake of 18 F-FDG in the epicardial adipose tissue. Epicardial adipose tissue has been implicated in the release of inflammatory factors, fostering atrial remodeling. Activation of inflammatory cells triggers the release of numerous cytokines, upregulates glucose transporters, and enhances the uptake of 18 F-FDG by inflammatory cells . These findings may signify an association with damage and cardiac inflammatory activity. Based on the occurrence of CTR-CVT during the treatment process, our findings revealed that patients in the CTR-CVT group exhibited larger left atrial diameter and left ventricular end-systolic diameter compared to those in the non-CTR-CVT group, along with a lower left ventricular ejection fraction. Additionally, the EAT SUVmax and EAT SUVmean post-treatment were higher in the CTR-CVT group than in the non-CTR-CVT group, indicating that the cardiotoxicity of tumor drugs had repercussions on the left atrium, left ventricle, and ejection fraction. Through regression analysis, we identified several risk factors for CTR-CVT, including left ventricular end-systolic diameter, ejection fraction, EAT SUVmax, and changes in uptake patterns post-treatment. Although left ventricular end-systolic diameter and ejection fraction hold predictive potential for future cardiac toxicity, their changes during treatment may be subtle and fall within the normal range, making them easily overlooked. However, during the follow-up of tumor lesions using 18 F-FDG PET/CT, alterations in cardiac uptake patterns and epicardial adipose tissue uptake values, particularly noteworthy differences observed in visual analysis of cardiac uptake patterns, may signify the likelihood of future cardiovascular toxicity. This highlights the additional value of predicting cardiovascular toxicity in oncologic imaging. This study is retrospective in nature, and the fasting protocol for PET/CT scans is tailored towards oncology, differing from the fasting protocol used for evaluating cardiac inflammation imaging. Although a prolonged fasting period (>12 hours) is mandated for tumor imaging patients at our center, it may not entirely suppress the physiological uptake of myocardium for 18 F-FDG PET/CT. Consequently, further validation of lesions exhibiting changes in cardiac uptake patterns is warranted through prospective experiments or animal studies to elucidate pathological or histological characteristics. However, notwithstanding these limitations, the current research findings instill optimism.
In our study, we suggest that during the follow-up process of oncologic 18 F-FDG PET/CT, early prediction of future CTR-CVT may be feasible by observing alterations in cardiac uptake patterns and metabolic parameters of the heart and epicardial adipose tissue. This is a new aid for early identification and intervention of cardiac-related complications in oncology patients to improve the quality of survival.
|
QUALITY OF LIFE OF OSTOMATES – A QUALITATIVE STUDY | d204b219-cb77-4192-b154-2d34351be5eb | 11827400 | Patient Education as Topic[mh] | The prevalence of people undergoing ostomy is 0.12% worldwide ( ). According to recently published data, the estimated number is about 750,000 people in the United States living with some form of stoma, whereas in Europe their number is 700,000 ( ). In the Republic of Croatia, there are approximately 7,500 people living with a stoma (0.2%) ( ). According to the Croatian National Cancer Registry, which was established in 1959 with the aim to collect, manage and analyze cancer incidence data, colorectal cancer was the third most frequent cancer by incidence in men (15%) and second in women (13%) in 2019 ( ), and represents a huge problem within the Croatian health care system ( ). Worldwide, colorectal cancer was estimated to be the third most common cancer and second leading cause of cancer death in 2018 ( ). Ostomy is usually performed during the treatment of inflammatory bowel diseases, traumas, acute diverticulitis, but most commonly in colorectal cancer ( ). Although the national screening for colorectal cancer exists in the Republic of Croatia since 2007, the response rate is still low, approximately 20% ( , ). The life with stoma has a lot of challenges which impact everyday life. By the World Health Organization definition, quality of life (QoL) is “an individual’s perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns” ( ). According to the aforementioned definition, QoL represents the multidimensional concept which is related to physical and mental health of an individual, independence, social interactions, personal beliefs and relation with the environment. Results of a systematic review by Vonk-Klaassen et al ., which included studies that used validated instruments to measure QoL in stoma patients, have shown that living with stoma negatively affects QoL ( ). People with stoma are more prone to develop anxiety and depression due to body image appearance and physical functioning ( ). Interdisciplinary care is needed to improve overall health and QoL, with nurses playing a central role in the successful management of health in this population ( , ). Besides nursing professionals that help the patients overcome mental, physical, social, and emotional hurdles, community support is also an important factor. In Croatia, such support is provided by the Croatian Association of Disabled Societies (ILCO) ( ). The Association is organized in 15 branches throughout Croatia and its mission is to support the patients with information regarding social and health care rights, as well as to provide psychological support before and after ostomy. In addition, ILCO is actively involved in raising awareness and removing the stigma associated with having a stoma. The aim of this study was to qualitatively analyze QoL in stoma patients and to gain an insight into patient experiences in different QoL areas, as well as to determine patient satisfaction with health education provided by hospital nurses.
Participants This study involved 15 participants aged between 40 and 86 years. Of these 15 participants, 6 were women. Inclusion criteria were age 18+ and presence of stoma of more than one year. Exclusion criteria were terminal illness, patients with cognitive impairment, and patients who did not speak the Croatian language. All participants signed an informed consent form before participating in the study. This study was approved by the Ethics Committee of the Faculty of Health Studies, University of Rijeka. Methods Participants were purposively selected using the chain sampling method. The study was conducted using a semistructured interview. All interviews were conducted by telephone due to the COVID-19 pandemic. Interviews were recorded, transcribed, and analyzed by three independent researchers, after which a consensus was reached ( ). Thematic analysis was conducted ( , ).
This study involved 15 participants aged between 40 and 86 years. Of these 15 participants, 6 were women. Inclusion criteria were age 18+ and presence of stoma of more than one year. Exclusion criteria were terminal illness, patients with cognitive impairment, and patients who did not speak the Croatian language. All participants signed an informed consent form before participating in the study. This study was approved by the Ethics Committee of the Faculty of Health Studies, University of Rijeka.
Participants were purposively selected using the chain sampling method. The study was conducted using a semistructured interview. All interviews were conducted by telephone due to the COVID-19 pandemic. Interviews were recorded, transcribed, and analyzed by three independent researchers, after which a consensus was reached ( ). Thematic analysis was conducted ( , ).
Sociodemographic data The sample consisted of 6 women and 9 men who participated in the study. Their age ranged from 40 to 86 years. The mean years of life with stoma was 7.7 (range 1-27) years. Thirteen participants were married, while one was divorced, and one widowed. With the exception of one participant who worked part-time, all others were on sick leave (n=4) or received disability pension (n=10). Satisfaction with health education on stoma hygiene at hospital and inclusion of family members in education Most participants were very satisfied with health education on stoma hygiene provided by the hospital nurse professionals, while some felt that the education could be improved, and one participant was unable to attend the education session at the hospital due to a serious health condition. Most participants involved family members in the education, some chose not to involve family members, and one did not have the opportunity to involve family members. In addition to the education in the hospital, most participants had positive experiences with community nurses. Quotes: “I don`t know, when you come back home and you have an incident like diarrhea, you don’t know how to behave in such specific occasion”. “There are ways to improve education. Only one time is not enough”. Knowledge about health and social care rights All participants were aware of health-related services and accessories needed for routine stoma care. The lack of knowledge and rights were related to social care services. The results showed that participants who were members of ILCO had better insight into health and social care rights. On the other hand, those who were not members of ILCO were less familiar with social care rights. Quotes: “There is no place for patients where you can ask questions about rights and other relevant documentation. I found out everything from other patients. None of the information was available at the hospital”. Inclusion of informal caregivers in stoma hygiene education in the hospital setting Most participants included family members in health education on stoma hygiene. Some of the informal caregivers were overwhelmed with stoma care despite their best will. Those who did not involve informal caregivers said that it was their own problem. Quotes: “I was alone. They also insisted, but I wanted to be alone”. “The wife was there, but only the second or third time when the nurse came”. Adaptation and acceptance of life with surgical stoma For those who had already suffered from severe health problems before, adaptation and acceptance were faster because the stoma was a relief for them. On average, five to six months were sufficient to accept life with the stoma. The most frequently mentioned motives for acceptance were family, personal strengths, religion, and healthcare providers. From the participants’ perspective, challenges in adjustment were digestive problems, fears related to stoma bag leakage in public, body self-perception, and significant psychological challenges such as depression, anxiety and fears in the first 6 months to a year. Quotes: “I cut down trees and prepare the wood for winter. I do everything. So, I don’t mind stoma at all”. “The hardest part was accepting that I am not the same old me anymore”. “... I was really sad. It takes some time to plan the trip and other things. The biggest fear was that stoma bag would wear off. That’s a horrible feeling”. “The worst thing is that you cannot go anywhere because you have to empty the stoma bag”. Physical functioning Participants are able to perform most daily activities, but with limitations such as bending and lifting weights. The most commonly mentioned challenges related to participants’ physical functioning were digestive problems and handling stoma bags during their stay outside the household. Fourteen participants did not participate in any type of exercise program. Clothing adjustments were required for normal physical functioning. Quotes: “When you have diarrhea, there are a lot of problems”. “I cannot do all the activities. If I want to mop the floor, I cannot. The shoes have to be without laces”. “If I just had regular bowel movements, I would not feel the stoma. So, I have to plan my days according to the bowel movement schedule”. Social inclusion, family life and travel Twelve participants reported having a fulfilling social life, with the rest limiting themselves to family as their social environment. Some of the participants experienced negative reactions from the acquaintances such as pity and indignation, but most of them had positive experiences and reactions to their condition. Half of them had discomfort related to the emission of gases in social environment (noise and smell). Fourteen participants stated that stoma did not affect their family life so far, but they experienced some challenges at the beginning. Seven participants stated that when they travelled, they limited themselves to their own vehicle, which they found to be the most suitable means of transport. Most of the participants were affected by COVID-19 pandemic so they missed social life. Quotes: “It affects family life a bit when grandchildren see the stoma bag. I have to be careful, so they don’t see it”. “People can think that I am disabled, but if I think I might be dead, this is a better solution”. “It becomes uncomfortable when uncontrolled emission of gasses happens, people don’t have to listen to that”. “...sometimes it affects family life, if I turn onto my left side in my sleep, and the base plate eases up and leaks, I have to change all bedsheets”. Nutrition adaptation All participants stated that they ate the same food as before the ostomy surgery or with minor adjustments due to gas emission. Quotes: “I eat normal, and have normal body weight”. “I start to eat normally relatively soon after the operation”. “Everything as before the stoma, light food”. Economic impact due to stoma Eight participants stated that they did not have additional costs due to stoma hygiene and supplies. The others had additional costs mainly for anti-odor sprays, nutritional supplements, hygiene supplies (specialized cleaning products), and medications. Quote: “At least HRK 250 up to HRK 400 per month. I am afraid of odors”.
The sample consisted of 6 women and 9 men who participated in the study. Their age ranged from 40 to 86 years. The mean years of life with stoma was 7.7 (range 1-27) years. Thirteen participants were married, while one was divorced, and one widowed. With the exception of one participant who worked part-time, all others were on sick leave (n=4) or received disability pension (n=10).
Most participants were very satisfied with health education on stoma hygiene provided by the hospital nurse professionals, while some felt that the education could be improved, and one participant was unable to attend the education session at the hospital due to a serious health condition. Most participants involved family members in the education, some chose not to involve family members, and one did not have the opportunity to involve family members. In addition to the education in the hospital, most participants had positive experiences with community nurses. Quotes: “I don`t know, when you come back home and you have an incident like diarrhea, you don’t know how to behave in such specific occasion”. “There are ways to improve education. Only one time is not enough”.
All participants were aware of health-related services and accessories needed for routine stoma care. The lack of knowledge and rights were related to social care services. The results showed that participants who were members of ILCO had better insight into health and social care rights. On the other hand, those who were not members of ILCO were less familiar with social care rights. Quotes: “There is no place for patients where you can ask questions about rights and other relevant documentation. I found out everything from other patients. None of the information was available at the hospital”.
Most participants included family members in health education on stoma hygiene. Some of the informal caregivers were overwhelmed with stoma care despite their best will. Those who did not involve informal caregivers said that it was their own problem. Quotes: “I was alone. They also insisted, but I wanted to be alone”. “The wife was there, but only the second or third time when the nurse came”.
For those who had already suffered from severe health problems before, adaptation and acceptance were faster because the stoma was a relief for them. On average, five to six months were sufficient to accept life with the stoma. The most frequently mentioned motives for acceptance were family, personal strengths, religion, and healthcare providers. From the participants’ perspective, challenges in adjustment were digestive problems, fears related to stoma bag leakage in public, body self-perception, and significant psychological challenges such as depression, anxiety and fears in the first 6 months to a year. Quotes: “I cut down trees and prepare the wood for winter. I do everything. So, I don’t mind stoma at all”. “The hardest part was accepting that I am not the same old me anymore”. “... I was really sad. It takes some time to plan the trip and other things. The biggest fear was that stoma bag would wear off. That’s a horrible feeling”. “The worst thing is that you cannot go anywhere because you have to empty the stoma bag”.
Participants are able to perform most daily activities, but with limitations such as bending and lifting weights. The most commonly mentioned challenges related to participants’ physical functioning were digestive problems and handling stoma bags during their stay outside the household. Fourteen participants did not participate in any type of exercise program. Clothing adjustments were required for normal physical functioning. Quotes: “When you have diarrhea, there are a lot of problems”. “I cannot do all the activities. If I want to mop the floor, I cannot. The shoes have to be without laces”. “If I just had regular bowel movements, I would not feel the stoma. So, I have to plan my days according to the bowel movement schedule”.
Twelve participants reported having a fulfilling social life, with the rest limiting themselves to family as their social environment. Some of the participants experienced negative reactions from the acquaintances such as pity and indignation, but most of them had positive experiences and reactions to their condition. Half of them had discomfort related to the emission of gases in social environment (noise and smell). Fourteen participants stated that stoma did not affect their family life so far, but they experienced some challenges at the beginning. Seven participants stated that when they travelled, they limited themselves to their own vehicle, which they found to be the most suitable means of transport. Most of the participants were affected by COVID-19 pandemic so they missed social life. Quotes: “It affects family life a bit when grandchildren see the stoma bag. I have to be careful, so they don’t see it”. “People can think that I am disabled, but if I think I might be dead, this is a better solution”. “It becomes uncomfortable when uncontrolled emission of gasses happens, people don’t have to listen to that”. “...sometimes it affects family life, if I turn onto my left side in my sleep, and the base plate eases up and leaks, I have to change all bedsheets”.
All participants stated that they ate the same food as before the ostomy surgery or with minor adjustments due to gas emission. Quotes: “I eat normal, and have normal body weight”. “I start to eat normally relatively soon after the operation”. “Everything as before the stoma, light food”.
Eight participants stated that they did not have additional costs due to stoma hygiene and supplies. The others had additional costs mainly for anti-odor sprays, nutritional supplements, hygiene supplies (specialized cleaning products), and medications. Quote: “At least HRK 250 up to HRK 400 per month. I am afraid of odors”.
The aim of this study was to obtain information from participants about health education after ostomy in the hospital setting and perceptions of QoL domains. Satisfaction with health education on stoma management provided in the hospital was found to be very satisfactory. The importance of lifelong learning for nurses in stoma care is necessary as they are a link between the patient and the multidisciplinary team ( , ). Enterostomal therapy is a specialized education for nurses caring for people with stoma, wound, or continence needs ( ). The first enterostomal therapists in the Republic of Croatia were educated in 2017 through a lifelong learning program according to the guidelines of the World Council of Enterostomal Therapists ( ). The results of this study showed that patients had confidence in the hospital nurse professionals that provided health education. Those who were familiar with their health care rights were more satisfied with health care services received inside and outside the hospital by community nurses. Most participants involved informal caregivers (family members) in the process of health education, which was found to be important for various aspects of social and health systems ( , ). In addition to reducing costs in the health and social system, family caregivers play an important role in supporting patients to meet their needs during the adjustment and acceptance process ( ). One of the most important findings in this study was that participants who were not members of ILCO were not sufficiently informed about their social care rights. ILCO provides support to stoma patients in meeting their specific needs ( ). The results of this research show that hospital and out-of-hospital nurse professionals should provide education and resources not only in the health care system but also in social care system ( - ). High levels of self-efficacy can improve the patient’s QoL, adjustment to illness, mental health, and health literacy of the patient ( ). Studies show that women, unlike men, have greater challenges in accepting life with stoma ( ). Acceptance of living with stoma leads to a higher QoL ( ). The results of this study show that patients need about five to seven months to overcome the physical, mental, and psychological barriers. Other studies have also confirmed that the barriers in adapting to stoma are due to the psychological effects (odor, noise, leakage, body image) and physical functioning which affect social activities ( , , - ). According to Verweij et al ., the adaptation is more complex and difficult in younger patients than in older ones, although older patients have more difficulties with physical functioning. The explanation for this finding may lie in differences in coping and body image between these two groups ( ). The results of the study by Anaraki et al . showed that about 70% of stoma patients suffered from depression during the process of adaptation ( ). Supporting the needs of stoma patients is an ongoing process and should therefore be ensured even after discharge from the hospital, as another study has shown ( ). Social activities of the participants in this study were fulfilled but limited to the company of family and friends who were familiar with their condition (gas discharge, noises). Other studies report on reduction in social activities among participants ( , , ). Half of the participants in this study reported difficulty with transport and travel due to fecal excretions, noise, and exhaust fumes, and therefore only travelled by their own car. Other studies did not find significant results in this domain of QoL ( ). Fear of leakage, gas emission and noises were perceived as the biggest obstacles to social activities and travel, but still none of the participants adjusted their diet. Digestive problems were common among most participants. These problems lead to limitations in social activities (due to fear of fecal leakage, noises, and gases) and also in traveling. Nutrition is essential not only during recovery but also for normal functioning at long term, as it prevents electrolyte imbalance and malnutrition ( , ). While most patient information suggests return to a normal diet, the literature and this study show that ostomates should adapt their diet to their individual needs ( ). This finding indicates that health education and literacy in ostomates should be improved not only in the hospital but also in the out-of-hospital setting, with an additional focus on individual nutritional needs to improve QoL. Living with a stoma did not incur additional costs for half of the participants, others faced significant costs as their income came mainly from disability pension or sick leave. LeBlanc et al . also pointed at the financial burden that was impacted by stoma ( ). The savings in the health care system, but also in ostomates can be achieved through continuous education of patients by nurse professionals in the hospital, in the home environment, and through the ILCO ( ).
This qualitative research provided an insight to the QoL of stoma patients and determined patient satisfaction with health education provided by hospital nurses. Although the study included a small number of respondents, this kind of QoL research is not that often in Croatia and provides an overview of patient experiences and contributes to patient trust in the health care system. The results of this study can be used as a guide for further education of nurses and also for further research on the patient QoL.
|
Integrating preventive dental care into general Paediatric practice for Indigenous communities: paediatric residents’ perceptions | 0ec1193a-9d2e-459d-b970-0f171a7c6649 | 6352953 | Pediatrics[mh] | Dental decay remains the most common chronic childhood disease. It is five times more common than asthma, four times more common than early childhood obesity and 20 times more common than diabetes . For children aged 2 to 5 years, 70% of caries are found in 8% of the population . Compared to the general Canadian population, Indigenous 1 Used interchangeably with Aboriginal Peoples. Indigenous Peoples include First Nations, Inuit and Métis Peoples. children are reported to have poorer oral health and a higher frequency of dental pain . Research from across Canada has found that children from First Nations communities suffer needlessly from poor oral health, often as a result of barriers and challenges that exist in the delivery of efficient and effective dental caries prevention programmes [ – ]. The prevalence of tooth decay among Indigenous children is three to five times higher than the national average in Canada . About 86% of pre-schoolers and 90% of school children suffer from dental decay . Indigenous populations are less likely to access preventive dental health care services because of more limited access to health information resources and lack of availability of dental care providers in remote areas . As a result, on-reserve indigenous children are three times more likely to experience dental decay than their off-reserve counterparts . The burden of poor oral health and its associated costs are considerable and may compromise overall well-being and quality of life . Due to challenges associated with securing paediatric patient cooperation and the severity of dental disease, most children receive comprehensive oral rehabilitation under general anaesthesia. In fact, dental surgery constitutes 31% of all day surgeries for children aged 1 to 5 years, making it the leading cause of day surgery for children in many Canadian hospitals at an annual cost of $22 million each year . Even more troubling are the many indigenous children who receive repeated dental surgeries because of relapse (8.6 times higher than the general population) . In addition, it has been shown that 99% of general dental practitioners provide preventive dental health care to children 5 years of age or older, but only 9% see children 1 year of age or younger . Due to the shortage of and difficulties associated with recruiting dentists and barriers to accessing dental care in indigenous communities, pre-school children are less likely to visit a dentist as opposed to visiting primary care providers such as paediatricians . Therefore, this study aimed to explore paediatric residents’ perceptions of the feasibility of incorporating preventive dental care into a general paediatric outreach clinic for a First Nations community.
Study design A qualitative approach of inquiry was used in this study. Qualitative data was collected through focus groups using a semi-structured interview guide. The study protocol was approved by the University of Alberta Research Ethics Board (Pro00058627). Setting As part of their general paediatric residency programme at the University of Alberta, paediatric residents provide medical services in a general paediatric outreach clinic that serves children of four First Nations communities. The paediatric clinic runs once a week. It is staffed with alternating two primary paediatricians and rotating paediatric residents. The average appointment times are between 15 to 30 min. The total population of the four communities is approximately 15,000 and half are less than 18 years of age. The division of Pediatric Dentistry and General Pediatrics in the Faculty of Medicine and Dentistry at the University of Alberta have partnered with the Maskwacis Health Centre to integrate dental screenings and fluoride varnish applications into their existing paediatric outreach clinic. The director of the clinic who is a community member facilitated the integration and the conduction of this study. Sampling A total of 34 paediatric residents were invited to participate in the present study. Recruitment was initiated through the general paediatric chief resident. Invitations were sent by email with a link to select the time slot suitable for them to attend. All residents who accepted the invitation to participate in the study were included. Informed consent was collected from all participants. No incentive was offered to participants. Data collection Questions asked during the focus groups revolved around paediatric resident perceptions regarding the feasibility of incorporating preventive dental care into the general paediatric outreach clinic. Focus groups were conducted by a single interviewer (MA) and attended by those who conducted the analysis (MG and ME). Focus groups were approximately 40 to 60 min in length and were held in quiet rooms at the Edmonton Clinic Health Academy. The discussions held throughout the duration of the focus groups were recorded and transcribed verbatim. Researchers avoided any misleading comments and refrained from distorting responses presented by participants. A semi-structured interview guide ( ) was used to reduce bias introduced by the researchers. Depending on the participants’ response to questions, prompts were utilised whenever deemed suitable. Participants were de-identified and each received an identification number. Data management and analysis A basic interpretive inductive approach was used for data analysis. Two team members independently coded the transcripts. Each code was assigned textual quotes and given a specific definition. Discussions were then held between all researchers until a consensus was reached for the initial codes and their definitions. Codes were then grouped into themes and categories and discussed in detail among all researchers. Reflections about personal expectations and biases were discussed during regular meetings and debriefing sessions through all phases of the study.
A qualitative approach of inquiry was used in this study. Qualitative data was collected through focus groups using a semi-structured interview guide. The study protocol was approved by the University of Alberta Research Ethics Board (Pro00058627).
As part of their general paediatric residency programme at the University of Alberta, paediatric residents provide medical services in a general paediatric outreach clinic that serves children of four First Nations communities. The paediatric clinic runs once a week. It is staffed with alternating two primary paediatricians and rotating paediatric residents. The average appointment times are between 15 to 30 min. The total population of the four communities is approximately 15,000 and half are less than 18 years of age. The division of Pediatric Dentistry and General Pediatrics in the Faculty of Medicine and Dentistry at the University of Alberta have partnered with the Maskwacis Health Centre to integrate dental screenings and fluoride varnish applications into their existing paediatric outreach clinic. The director of the clinic who is a community member facilitated the integration and the conduction of this study.
A total of 34 paediatric residents were invited to participate in the present study. Recruitment was initiated through the general paediatric chief resident. Invitations were sent by email with a link to select the time slot suitable for them to attend. All residents who accepted the invitation to participate in the study were included. Informed consent was collected from all participants. No incentive was offered to participants.
Questions asked during the focus groups revolved around paediatric resident perceptions regarding the feasibility of incorporating preventive dental care into the general paediatric outreach clinic. Focus groups were conducted by a single interviewer (MA) and attended by those who conducted the analysis (MG and ME). Focus groups were approximately 40 to 60 min in length and were held in quiet rooms at the Edmonton Clinic Health Academy. The discussions held throughout the duration of the focus groups were recorded and transcribed verbatim. Researchers avoided any misleading comments and refrained from distorting responses presented by participants. A semi-structured interview guide ( ) was used to reduce bias introduced by the researchers. Depending on the participants’ response to questions, prompts were utilised whenever deemed suitable. Participants were de-identified and each received an identification number.
A basic interpretive inductive approach was used for data analysis. Two team members independently coded the transcripts. Each code was assigned textual quotes and given a specific definition. Discussions were then held between all researchers until a consensus was reached for the initial codes and their definitions. Codes were then grouped into themes and categories and discussed in detail among all researchers. Reflections about personal expectations and biases were discussed during regular meetings and debriefing sessions through all phases of the study.
Participants Four focus groups were conducted with ten residents and two attending paediatricians. Five residents were in year 1, three in year 2 and two in year 3 and 4. All residents were female 20–30 years old. All residents attended medical schools in Canada. Six residents attended a Paediatric Oral Health and Fluoride Varnish workshop delivered by the University of Alberta dental school specifically designed for paediatric residents. All residents had worked at least one full day in the First Nations outreach paediatric clinic. Three major themes emerged from the data: advantages of integration, barriers to integration and strategies for integration. Advantages of integration Two major categories related to the advantages of integration were identified: comprehensive care (i.e. concurrent prevention and management of multiple health needs) and service delivery (i.e. provision of needed care). Residents highlighted the importance of learning about preventive dental care and delivering preventive dental care to their patients. They believed that integration of preventive dental care into paediatric practice would optimise patient health and contribute to well-rounded care. Also, it would provide caregivers with a sense of relief that their child would have good oral health in the future: ‘[It would be a] well-rounded health care and that you’re doing both in one shot for them so you’re optimizing their visit…. I think they are excited that maybe you’re going to get on top of it early for their child and maybe they will have better outcomes’. [FG1/P4]. In addition, participants mentioned that integration of preventive dental care would help identify at-risk children: ‘…we can refer the ones that actually need [dental care] rather than saying every patient should see a dentist … If you can get 5 people to go see a dentist in the first year instead of referring them when they are three and you see holes, you could actually improve some of the patient population from an oral health standpoint’ [FG4/P2]. Another resident stated: When I put fluoride on the teeth that’s a good time to have a discussion because you’re focusing on the teeth’. [FG3/P3]. Participants also highlighted that integration of preventive dental care might help reach vulnerable populations and deliver needed services to them: We have this point of contact; all preschool kids get sick and they will eventually come to see a doctor/paediatrician, so we have that opportunity to put fluoride on’. [FG3/P1]. Participants noted that preventive dental care would be convenient for caregivers who would not have to make another appointment to see a dentist. They believed that the convenience of seeing one health care professional would facilitate service delivery and would be appreciated by caregivers who may have limited time or access to transportation. Barriers to integration Three categories of barriers emerged from the data: patient and caregiver-related barriers, resident-related barriers and setting-related barriers. For patient and caregiver-related barriers, four subcategories were identified: knowledge, attitude, behaviours and socioeconomic factors. Patient and caregiver-related barriers Residents identified caregivers’ lack of knowledge and misconceptions as an obstacle to the successful delivery of dental care to their young children. They believed that caregivers do not always recognise the value of having their child see a dentist. They were also concerned about caregiver’s uptake when multiple competing issues are discussed during one visit: ‘You’re sitting in the room with the mom …. you’re talking about hearing, early literacy, healthy food…I wonder how much the family actually takes away’ [FG3/P1]. Furthermore, residents suspected that by delivering preventive dental care services, caregivers might overlook the services as the only dental care their child needs. Participants also noted that some counselling recommendations like changes in diet or oral health habits may be beyond the caregivers’ control: ‘It may not be feasible for them like if it’s the issue of having [clean] running water… talking about brushing teeth may be out of the realm of possibilities for them unfortunately’ [FG2/P2]. They were concerned about the family’s receptiveness and the child’s resistance to the dental examination and application of fluoride varnish: ‘the [child] was upset and wandering all over the place… crying which was not good…’ [FG4/P1]. Paediatric residents mentioned socioeconomic factors that may affect the success of integrating preventive dental care including affordability and accessibility of healthy food and clean water: ‘… healthy food is expensive … in a place like …, [there] is no food security and so there isn’t a lot of healthy food available to the family’ [FG3/P1]. Other identified socioeconomic factors were lack of transportation, family size and affordability of oral hygiene supplies: ‘They might not be able to buy toothpaste or toothbrushes because a lot of the people are on social assistance and don’t have the money’. [FG1/P1] Resident-related barriers Three subcategories of resident-related barriers were identified: resident’s receptiveness, lack of knowledge and skills and scope of practice. Residents were generally very receptive to incorporating oral hygiene and diet counselling as well as fluoride varnish application into their daily paediatric care. However, they felt conducting a dental examination was beyond their ability. They believed that their training on oral health was insufficient and their current scope of practice does not involve a dental examination or oral health counselling: ‘The only screening we do is asking how many times do you brush your teeth, have you seen a dentist and those type of things’ [FG1/P1]. They stressed that doing preventive dental care during the regular visit is not always achievable because they need to prioritise the visit for many children: ‘… depends on the appointment. If they are there because they are sick and have pneumonia like I usually wouldn’t be spending extra time to counsel them’ [FG2/P2]. In addition, some residents believed that dental care is not part of their scope of practice and should be delivered by dentists: ‘You certainly think like oh its oral health we will leave it to the people who are more experienced’ [FG2/P3]. Setting-related barriers Four subcategories of setting-related barriers were identified: patient load, workforce, records and policy. Residents mentioned that numerous tasks such as seeing multiple patients and their families, completing numerous forms and charting could be too overwhelming. Adding additional services to their practice may disrupt the overall flow of the clinic. They commented that support staff could make a significant impact on the success of integrating preventive dental care. For instance, in the outreach clinic, support staff may not be readily available, and even if they are, they may not have the necessary skills. Medical record keeping was another setting-related barrier. Residents highlighted a lack of clear medical records on whether the child had received fluoride varnish or preventive counselling in another setting, for example, in a public health clinic or dental office, and caregivers often could not recall this information. Residents identified potential policy-related issues that might affect the successful integration of preventive dental care. Lack of a defined referral system between physicians and dentists was one of these issues. In addition, there is no fee schedule or billing codes for paediatricians who provide preventive dental care and fluoride applications, a service which prolongs the medical appointment and may have a financial impact on the service provider. ‘…then you can’t bill for it and it takes up visiting time for something else that you could be seeing a kid for’ [FG4/P2]. Strategies for integration Four categories related to strategies for integrating preventive dental care emerged: training and practice, patient education, support and policy. The training and practice category was composed of three subcategories: curriculum, simplicity and practice. Residents recognised that they play an important role in providing preventive dental care to an at-risk population. Residents were willing to learn; however, they identified the need for more training. Participants recommended that a formal education module be developed on site at the outreach clinic so that residents could learn more effectively and have the opportunity to then practice the skills they learned. In addition, the module and added services should be simple enough to incorporate into the regular practice and should not take more than a few minutes. They also highlighted that preventive dental care training early in their career would mean they would be more likely to carry on doing it in the future: ‘I think that in my exposure to general paediatrics over the next couple of years, it starts becoming something habitually; I will be more likely to integrate. It’s just hard for me to imagine right now what it would look like’ [FG1/P3]. Patient education strategies included five subcategories: encouraging dental visits, discussing caries as infectious diseases, graphic presentation of caries, specific targeted messages and ‘shock value’ education. Participants suggested advocating for an early dental visit so that children can see a dentist earlier rather than conducting the dental assessment themselves. They also believed that discussing caries as an infectious disease can help patients and caregivers understand the serious nature of the condition. They emphasised that the right vocabulary could make a significant impact. In addition, they felt that having these targeted messages through a paediatrician as opposed to other healthcare providers would have a greater impact: ‘having a discussion about dental caries as an infectious disease was quite powerful for parents’ [FG4/P1]. Participants also suggested that a graphical presentation of caries can help patients and caregivers view caries from a different perspective. Delivering messages that are specific and brief were believed to have the highest impact by reducing the amount of information that caregivers need to absorb. Some participants believed that delivering messages forcefully may be the best way to get the point across to some patients: ‘It’s almost a bit of a shock value… I hate to say that… but it’s true you remember what’s uncomfortable’ [FG4/P1]. Support strategies comprised three subcategories: record system, supplies and buy-in. Because paediatric residents felt that it was hard to keep track of which patients had received fluoride, a record system is needed. The system could be as simple as a coloured sticker or if available an electronic medical record. To track patients accurately one resident mentioned how: ‘If you had a sticker and you put it in the chart like a bright pink sticker and the next time if it’s the same chart and it’s been done properly you should be able to flip back and be like “oh pink sticker no varnish or oh no pink sticker I can varnish’ [FG3/P2]. Another suggested support strategy was the need for free dental care supplies including toothbrushes, paste and fluoride. In addition, they added the need for paediatrician buy-in, which could be facilitated by creating a billing code for dental care prevention and fluoride varnish application. Policy-related strategies comprised three subcategories: policy changes, advocating for universal dental coverage and community-based approaches. Participants noted that there is a need for a Canadian Paediatric Society position statement that addresses and supports the integration of oral health into paediatric care. They also suggested advocating for universal dental care coverage so that children and youth can access dental care free of charge. Another suggestion was to use community-based approaches that promote oral health at community events, immunisation appointments and within schools.
Four focus groups were conducted with ten residents and two attending paediatricians. Five residents were in year 1, three in year 2 and two in year 3 and 4. All residents were female 20–30 years old. All residents attended medical schools in Canada. Six residents attended a Paediatric Oral Health and Fluoride Varnish workshop delivered by the University of Alberta dental school specifically designed for paediatric residents. All residents had worked at least one full day in the First Nations outreach paediatric clinic. Three major themes emerged from the data: advantages of integration, barriers to integration and strategies for integration.
Two major categories related to the advantages of integration were identified: comprehensive care (i.e. concurrent prevention and management of multiple health needs) and service delivery (i.e. provision of needed care). Residents highlighted the importance of learning about preventive dental care and delivering preventive dental care to their patients. They believed that integration of preventive dental care into paediatric practice would optimise patient health and contribute to well-rounded care. Also, it would provide caregivers with a sense of relief that their child would have good oral health in the future: ‘[It would be a] well-rounded health care and that you’re doing both in one shot for them so you’re optimizing their visit…. I think they are excited that maybe you’re going to get on top of it early for their child and maybe they will have better outcomes’. [FG1/P4]. In addition, participants mentioned that integration of preventive dental care would help identify at-risk children: ‘…we can refer the ones that actually need [dental care] rather than saying every patient should see a dentist … If you can get 5 people to go see a dentist in the first year instead of referring them when they are three and you see holes, you could actually improve some of the patient population from an oral health standpoint’ [FG4/P2]. Another resident stated: When I put fluoride on the teeth that’s a good time to have a discussion because you’re focusing on the teeth’. [FG3/P3]. Participants also highlighted that integration of preventive dental care might help reach vulnerable populations and deliver needed services to them: We have this point of contact; all preschool kids get sick and they will eventually come to see a doctor/paediatrician, so we have that opportunity to put fluoride on’. [FG3/P1]. Participants noted that preventive dental care would be convenient for caregivers who would not have to make another appointment to see a dentist. They believed that the convenience of seeing one health care professional would facilitate service delivery and would be appreciated by caregivers who may have limited time or access to transportation.
Three categories of barriers emerged from the data: patient and caregiver-related barriers, resident-related barriers and setting-related barriers. For patient and caregiver-related barriers, four subcategories were identified: knowledge, attitude, behaviours and socioeconomic factors. Patient and caregiver-related barriers Residents identified caregivers’ lack of knowledge and misconceptions as an obstacle to the successful delivery of dental care to their young children. They believed that caregivers do not always recognise the value of having their child see a dentist. They were also concerned about caregiver’s uptake when multiple competing issues are discussed during one visit: ‘You’re sitting in the room with the mom …. you’re talking about hearing, early literacy, healthy food…I wonder how much the family actually takes away’ [FG3/P1]. Furthermore, residents suspected that by delivering preventive dental care services, caregivers might overlook the services as the only dental care their child needs. Participants also noted that some counselling recommendations like changes in diet or oral health habits may be beyond the caregivers’ control: ‘It may not be feasible for them like if it’s the issue of having [clean] running water… talking about brushing teeth may be out of the realm of possibilities for them unfortunately’ [FG2/P2]. They were concerned about the family’s receptiveness and the child’s resistance to the dental examination and application of fluoride varnish: ‘the [child] was upset and wandering all over the place… crying which was not good…’ [FG4/P1]. Paediatric residents mentioned socioeconomic factors that may affect the success of integrating preventive dental care including affordability and accessibility of healthy food and clean water: ‘… healthy food is expensive … in a place like …, [there] is no food security and so there isn’t a lot of healthy food available to the family’ [FG3/P1]. Other identified socioeconomic factors were lack of transportation, family size and affordability of oral hygiene supplies: ‘They might not be able to buy toothpaste or toothbrushes because a lot of the people are on social assistance and don’t have the money’. [FG1/P1] Resident-related barriers Three subcategories of resident-related barriers were identified: resident’s receptiveness, lack of knowledge and skills and scope of practice. Residents were generally very receptive to incorporating oral hygiene and diet counselling as well as fluoride varnish application into their daily paediatric care. However, they felt conducting a dental examination was beyond their ability. They believed that their training on oral health was insufficient and their current scope of practice does not involve a dental examination or oral health counselling: ‘The only screening we do is asking how many times do you brush your teeth, have you seen a dentist and those type of things’ [FG1/P1]. They stressed that doing preventive dental care during the regular visit is not always achievable because they need to prioritise the visit for many children: ‘… depends on the appointment. If they are there because they are sick and have pneumonia like I usually wouldn’t be spending extra time to counsel them’ [FG2/P2]. In addition, some residents believed that dental care is not part of their scope of practice and should be delivered by dentists: ‘You certainly think like oh its oral health we will leave it to the people who are more experienced’ [FG2/P3]. Setting-related barriers Four subcategories of setting-related barriers were identified: patient load, workforce, records and policy. Residents mentioned that numerous tasks such as seeing multiple patients and their families, completing numerous forms and charting could be too overwhelming. Adding additional services to their practice may disrupt the overall flow of the clinic. They commented that support staff could make a significant impact on the success of integrating preventive dental care. For instance, in the outreach clinic, support staff may not be readily available, and even if they are, they may not have the necessary skills. Medical record keeping was another setting-related barrier. Residents highlighted a lack of clear medical records on whether the child had received fluoride varnish or preventive counselling in another setting, for example, in a public health clinic or dental office, and caregivers often could not recall this information. Residents identified potential policy-related issues that might affect the successful integration of preventive dental care. Lack of a defined referral system between physicians and dentists was one of these issues. In addition, there is no fee schedule or billing codes for paediatricians who provide preventive dental care and fluoride applications, a service which prolongs the medical appointment and may have a financial impact on the service provider. ‘…then you can’t bill for it and it takes up visiting time for something else that you could be seeing a kid for’ [FG4/P2].
Residents identified caregivers’ lack of knowledge and misconceptions as an obstacle to the successful delivery of dental care to their young children. They believed that caregivers do not always recognise the value of having their child see a dentist. They were also concerned about caregiver’s uptake when multiple competing issues are discussed during one visit: ‘You’re sitting in the room with the mom …. you’re talking about hearing, early literacy, healthy food…I wonder how much the family actually takes away’ [FG3/P1]. Furthermore, residents suspected that by delivering preventive dental care services, caregivers might overlook the services as the only dental care their child needs. Participants also noted that some counselling recommendations like changes in diet or oral health habits may be beyond the caregivers’ control: ‘It may not be feasible for them like if it’s the issue of having [clean] running water… talking about brushing teeth may be out of the realm of possibilities for them unfortunately’ [FG2/P2]. They were concerned about the family’s receptiveness and the child’s resistance to the dental examination and application of fluoride varnish: ‘the [child] was upset and wandering all over the place… crying which was not good…’ [FG4/P1]. Paediatric residents mentioned socioeconomic factors that may affect the success of integrating preventive dental care including affordability and accessibility of healthy food and clean water: ‘… healthy food is expensive … in a place like …, [there] is no food security and so there isn’t a lot of healthy food available to the family’ [FG3/P1]. Other identified socioeconomic factors were lack of transportation, family size and affordability of oral hygiene supplies: ‘They might not be able to buy toothpaste or toothbrushes because a lot of the people are on social assistance and don’t have the money’. [FG1/P1]
Three subcategories of resident-related barriers were identified: resident’s receptiveness, lack of knowledge and skills and scope of practice. Residents were generally very receptive to incorporating oral hygiene and diet counselling as well as fluoride varnish application into their daily paediatric care. However, they felt conducting a dental examination was beyond their ability. They believed that their training on oral health was insufficient and their current scope of practice does not involve a dental examination or oral health counselling: ‘The only screening we do is asking how many times do you brush your teeth, have you seen a dentist and those type of things’ [FG1/P1]. They stressed that doing preventive dental care during the regular visit is not always achievable because they need to prioritise the visit for many children: ‘… depends on the appointment. If they are there because they are sick and have pneumonia like I usually wouldn’t be spending extra time to counsel them’ [FG2/P2]. In addition, some residents believed that dental care is not part of their scope of practice and should be delivered by dentists: ‘You certainly think like oh its oral health we will leave it to the people who are more experienced’ [FG2/P3].
Four subcategories of setting-related barriers were identified: patient load, workforce, records and policy. Residents mentioned that numerous tasks such as seeing multiple patients and their families, completing numerous forms and charting could be too overwhelming. Adding additional services to their practice may disrupt the overall flow of the clinic. They commented that support staff could make a significant impact on the success of integrating preventive dental care. For instance, in the outreach clinic, support staff may not be readily available, and even if they are, they may not have the necessary skills. Medical record keeping was another setting-related barrier. Residents highlighted a lack of clear medical records on whether the child had received fluoride varnish or preventive counselling in another setting, for example, in a public health clinic or dental office, and caregivers often could not recall this information. Residents identified potential policy-related issues that might affect the successful integration of preventive dental care. Lack of a defined referral system between physicians and dentists was one of these issues. In addition, there is no fee schedule or billing codes for paediatricians who provide preventive dental care and fluoride applications, a service which prolongs the medical appointment and may have a financial impact on the service provider. ‘…then you can’t bill for it and it takes up visiting time for something else that you could be seeing a kid for’ [FG4/P2].
Four categories related to strategies for integrating preventive dental care emerged: training and practice, patient education, support and policy. The training and practice category was composed of three subcategories: curriculum, simplicity and practice. Residents recognised that they play an important role in providing preventive dental care to an at-risk population. Residents were willing to learn; however, they identified the need for more training. Participants recommended that a formal education module be developed on site at the outreach clinic so that residents could learn more effectively and have the opportunity to then practice the skills they learned. In addition, the module and added services should be simple enough to incorporate into the regular practice and should not take more than a few minutes. They also highlighted that preventive dental care training early in their career would mean they would be more likely to carry on doing it in the future: ‘I think that in my exposure to general paediatrics over the next couple of years, it starts becoming something habitually; I will be more likely to integrate. It’s just hard for me to imagine right now what it would look like’ [FG1/P3]. Patient education strategies included five subcategories: encouraging dental visits, discussing caries as infectious diseases, graphic presentation of caries, specific targeted messages and ‘shock value’ education. Participants suggested advocating for an early dental visit so that children can see a dentist earlier rather than conducting the dental assessment themselves. They also believed that discussing caries as an infectious disease can help patients and caregivers understand the serious nature of the condition. They emphasised that the right vocabulary could make a significant impact. In addition, they felt that having these targeted messages through a paediatrician as opposed to other healthcare providers would have a greater impact: ‘having a discussion about dental caries as an infectious disease was quite powerful for parents’ [FG4/P1]. Participants also suggested that a graphical presentation of caries can help patients and caregivers view caries from a different perspective. Delivering messages that are specific and brief were believed to have the highest impact by reducing the amount of information that caregivers need to absorb. Some participants believed that delivering messages forcefully may be the best way to get the point across to some patients: ‘It’s almost a bit of a shock value… I hate to say that… but it’s true you remember what’s uncomfortable’ [FG4/P1]. Support strategies comprised three subcategories: record system, supplies and buy-in. Because paediatric residents felt that it was hard to keep track of which patients had received fluoride, a record system is needed. The system could be as simple as a coloured sticker or if available an electronic medical record. To track patients accurately one resident mentioned how: ‘If you had a sticker and you put it in the chart like a bright pink sticker and the next time if it’s the same chart and it’s been done properly you should be able to flip back and be like “oh pink sticker no varnish or oh no pink sticker I can varnish’ [FG3/P2]. Another suggested support strategy was the need for free dental care supplies including toothbrushes, paste and fluoride. In addition, they added the need for paediatrician buy-in, which could be facilitated by creating a billing code for dental care prevention and fluoride varnish application. Policy-related strategies comprised three subcategories: policy changes, advocating for universal dental coverage and community-based approaches. Participants noted that there is a need for a Canadian Paediatric Society position statement that addresses and supports the integration of oral health into paediatric care. They also suggested advocating for universal dental care coverage so that children and youth can access dental care free of charge. Another suggestion was to use community-based approaches that promote oral health at community events, immunisation appointments and within schools.
This qualitative inquiry explored paediatric resident perceptions about the integration of preventive dental care into a paediatric outreach clinic serving First Nations children. Paediatric residents recognised the importance of their role in improving the oral health of indigenous children and acknowledged that integrated preventive oral care would reach at-risk children and provide more comprehensive care. These findings were similar to a US national survey which showed that the majority of paediatricians frequently observed dental caries, and acknowledged that they had an important role in identifying dental problems and counselling families on the prevention of caries . They were also interested in being more involved in managing children with oral health problems, in particular for underserved children . Similarly, three-quarters of primary care providers in another study reported that they frequently identified children with signs of early decay and referred them to a dentist . Although medical providers have shown great interest in contributing to improving the oral health of underserved populations, their limited knowledge and skills due to lack of training in oral health during their medical education have been reported as a common barrier to integrating oral health into paediatric care . This barrier, also mentioned by our participants, could be overcome by adequate training and practice . Lack of knowledge and difficulty applying fluoride varnish were both considered to be significant barriers. Close et al. suggested that difficulties related to fluoride varnish application should be anticipated, and multiple possible solutions should be part of the training. In addition, the use of interactive workshops and small group trainings to overcome these barriers were suggested in addition to in-office training on actual patients in the same setting where future applications are anticipated . Participants in the present study had similar insights on how training should be delivered. They suggested developing an on-site education module specific to the outreach clinic so that they could learn the required skills efficiently and get enough practice to feel comfortable doing it regularly. Although incorporating oral health care into primary care has been advocated by the Canadian Paediatric Society and the American Academy of Paediatrics, prioritising patient’s needs and scope of practice as medical care providers were barriers identified in the present as well as previous studies . These barriers affect ‘buy-in’ by health care providers and staff. In a survey involving 76 primary care practices, 14.8% of providers highlighted that their infant and toddler patients had so many problems other than tooth decay . In addition, nursing staff believed that oral health should not be part of the medical office role and should be done by dentists . A qualitative inquiry involving primary care providers showed that the main reason behind this attitude is the perception that adding oral health counselling and fluoride varnish into daily practice is time-consuming, and is considered unpractical because of an increase in workload with no financial incentive . However, this attitude was overcome in 40% of providers after they adopted primary dental services into their regular care . A suggested approach was to use an incentive system where physicians perform oral screening, and an assigned personnel applies fluoride varnish and receives a monetary incentive for every application . Our participants similarly believed that attaching a billing code to fluoride varnish application would improve the buy-in among health care providers. Participants in the present study expressed concerns about a proper referral system, which was also reported as a reason for not performing dental screening by medical providers in a previous study . Although this barrier was overcome by 40% of the providers , it may be more complicated for on-reserve children who may not have access to a paediatrician or dentist within a reasonable distance. The difficulty of finding a dentist for children either younger than 2 years old, with significant developmental disabilities, or with no insurance was another barrier reported by paediatric primary care providers . It was suggested that the training of primary care providers in oral health should also include detailed referral resources for both providers and patients . Our participants’ suggestion on the referral system was the development of an integral universal health care system with dental care included. This was a general suggestion for most centres and not specific to this outreach clinic, as the outreach clinic has a dental office within the same medical centre. Accurate and complete records of patients who previously received fluoride varnish at the clinic and at other sites was also a barrier. This barrier was a result of inconsistent record keeping in the on-reserve clinic, lack of sharing of information from other sites including immunisation clinics and schools and lack of caregiver recollection on whether their child had received a fluoride treatment previously. Similar concerns were highlighted by primary care providers in another qualitative study . In this study, a successful alternative was to have front office staff determine eligibility and flag the chart for those who qualified. Primary care providers then provided oral health counselling and applied fluoride varnish. In addition, office staff documented the fluoride varnish application in the patient’s record of preventive care . Although such approach seems easy to implement in primary care offices, it may be challenging if resources are limited and support staff are not available. Multiple approaches suggested by our caregiver participants ranged from simple colour-coded stickers to an electronic medical record system. Our study limitations included small focus groups of three to four participants due to limited resident availability to attend a focus group. Having a larger number of participants would have created a richer dialogue. However, we believe that the residents who did attend were motivated and fully engaged in the focus group discussions. A limitation of the study is not seeking views of other health care staff on their feedback on barrier and facilitators to implementation. Another limitation is that this study did not include community focus groups of caregivers to assess local perceptions and feedback on the experience of oral health care provision during paediatrician appointments. In conclusion, paediatric residents and attending paediatricians were interested in integrating preventive dental care into their paediatric care for on-reserve children. However, several barriers may impede the success of this integration. Multiple strategies need to be considered and implemented to facilitate the integration. Exploring other stakeholders’ perceived barriers and proposed improvement strategies using a qualitative approach is an essential next step in creating a unified action plan for improved oral health care for indigenous children.
|
Female Urinary Incontinence Evidence-Based Treatment Pathway: An Infographic for Shared Decision-Making | 7244d249-b265-4df8-b036-f31811efdd4d | 8972010 | Gynaecology[mh] | Urinary incontinence (UI) in women is a highly prevalent health condition. Approximately 50% of women age >50 years experience UI, and >20 million women in the United States have bothersome UI. , These estimates are projected to rise, due in large part to the size of the aging demographic and the national obesity epidemic, both of which are associated with increased risk of UI. The scale of the problem, as well as the associated economic, psychosocial, and physical burdens of UI, establish this clearly as a public health issue necessitating a population-focused perspective. As few as 25% of women with UI in the United States seek care for their condition. Inversely, most women with UI in the United States do not discuss their symptoms with a health care provider (HCP), and thus do not initiate care. At an individual level, barriers to care-seeking include embarrassment, lack of knowledge of treatment options, feeling that the symptoms are not bad enough, and unwillingness to bring it up independently of being asked by their HCP. At a health systems level, structural barriers include time constraints, competing priorities, insufficient health workforce, and sociocultural barriers that limit patient accessibility to health services. In addition, economic barriers to care limit access and affordability, particularly for quality-of-life conditions, which may be deprioritized in the context of acute or life-threatening conditions. , In recognition of the numbers of women with UI who do not seek or receive care, a 2018 systematic review explored the utility of proactive UI screening for women, and a subsequent Women's Preventive Services Initiative recommendation was issued, endorsing annual screening for UI for women, ages 18+, using validated survey tools. By integrating current evidence and synthesizing UI screening and treatment recommendations, this infographic was developed with aims (1) to improve patient and provider knowledge of this health condition and (2) to facilitate shared decision-making about treatment options, in line with professional guidelines . Infographics convey important, often complex health information in a visually appealing display that may be quickly and easily understood. The American College of Obstetricians and Gynecologists (ACOG) endorses the use of decision aids, including infographics, citing evidence that these tools support patient-centered care and aid in counseling. , The infographic presented in this study may be shared widely across professional organizations and health systems and utilized by general practitioners providing routine and/or well-woman care, as well as women's health or pelvic medicine specialists. A brief description and evidence to support the components of this infographic are summarized. Whether through conventional “problem visit” care-seeking or implementation of standardized screening, women require preliminary evaluation of their UI. Screening may be facilitated using validated questionnaires, such as the 3-Incontinence Questions survey (3IQ), the Bladder Control Self-Assessment Questionnaire (B-SAQ), or the Michigan Incontinence Symptom Index (MISI). , Evaluation entails a thorough subjective history, symptom evaluation, and risk factor assessment and may include the use of validated surveys. The Urogenital Distress Inventory-6 (UDI-6) and the International Consultation on Incontinence Questionnaire-Short Form for Urinary Incontinence (ICIQSF-UI) are common in research and clinical practice. , Screening and symptom questionnaires may be used to determine presence of UI, suggest a provisional diagnosis of UI subtype or a more complicated cause, and/or to assess symptom severity and degree of bother. Physical examination, postvoid residual, and urinalysis are standard components. Potential causes of transient UI correspond to the acronym DIAPPERS (Delirium, Infection—urinary, Atrophic urethritis/vaginitis, Pharmaceuticals, Psychologic disorders, Excessive urine output, Restricted mobility, Stool impaction) and should be identified and addressed. Complicated UI due to congenital, neurological, or metabolic conditions, fistula, urinary retention, prolapse, or prior pelvic surgeries require specialist or subspecialist evaluation and care. , Once these causes have been ruled out, it is important to differentiate among the major types of UI, including stress urinary incontinence (SUI), urgency urinary incontinence (UUI), or mixed urinary incontinence (MUI) . Based on this diagnosis initial treatment recommendations can be made. There is broad international and multidisciplinary agreement on most components of the UI care pathway, including adopting a stepwise approach. , Universal consensus for first-line care for SUI, UUI, and MUI includes pelvic floor muscle training (PFMT) and other behavioral modifications such as bladder training, dietary changes, and/or fluid titration. PFMT is defined as “exercises for improving PFM strength, endurance, power and/or relaxation.” Level I evidence supports PFMT effectiveness and describes this intervention as most effective when performed under the supervision of a skilled HCP (supervised PFMT/sPFMT) for a period of at least 12 weeks. Recent publications highlight personal and structural barriers to implementing sPFMT for all women with UI. Gross limitations in the health workforce leave too few HCPs with the skills to provide sPFMT, compared with the numbers of women who require it. , New care models are being tested and proposed to build capacity within the broader health system to care for women with UI. These include group-based PFMT, unsupervised PFMT, and the use of mobile technologies. Components of first-line care are considered minimal or no risk and may also play a role in multimodal therapy, implemented alongside advanced interventions. Vaginal estrogen is recommended when vaginal atrophy is present with urinary symptoms. , , Beyond this and the first-line care described earlier, remaining treatment interventions for UI address either SUI or UUI (or the respective component of MUI). Although pessaries may be helpful for women with SUI, there are no FDA-approved medications for SUI available in the United States. , , Of those medications approved for UUI, anticholinergics are most often prescribed; however, HCPs must exercise caution when prescribing, in light of mounting evidence of increased dementia risk associated with chronic use. Beta-agonists are approved for urinary urgency and UUI, and are becoming more commonly prescribed due to fewer side effects, and thus greater tolerance and adherence among patients. , Limitations in utilization of beta-agonists may be financial in nature, as they are more costly to patients and payers compared with anticholinergics. , Advanced therapies for UI are often carried out at the specialist or subspecialist level although referral may be warranted at any time before this phase of care. For SUI, advanced therapies may include periurethral bulking agents and numerous surgical options. , , For UUI, these include intravesical Botox and in-office or implantable peripheral and sacral neuromodulation. , , Recent guidance indicates that remote care ( i.e., telemedicine) may be an appropriate vehicle for UI screening, initial evaluation, and the implementation of first-line care. , In this context, physical examination, postvoid residual, and urinalysis may be deferred according to HCP interpretation of other evaluation components. The exception to the use of telemedicine to implement first-line care is the indication for additional in-person evaluation and/or referral. In all contexts—telemedicine or in-person care—evaluation and discussion of a patient's desire for treatment and their response to a given treatment should occur and be followed by shared decision-making about any next steps in care. An example of next steps is implementation of additional testing, such as urodynamics and/or cystoscopy, after first-line care that did not yield sufficient symptom improvement. This infographic synthesizes the literature and society recommendations in a visual format. Important factors preceding and concurrent with the patient–provider interaction are depicted, and the stepwise treatment pathway that may unfold over time is clearly illustrated. It may be useful for HCPs who want to engage in shared decision-making with their female patients, and readers are encouraged to print and share the infographic as a useful tool in patient education and clinical practice. Future study will examine use of this infographic in various settings to assess its impact on patient knowledge and certainty about UI treatment and HCP perceptions of its role in patient counseling and decision-making. Supplemental data |
Proteomic and phosphoproteomic analyses reveal that TORC1 is reactivated by pheromone signaling during sexual reproduction in fission yeast | 94577201-42b7-49cd-9a21-11dd52701c1e | 11750111 | Biochemistry[mh] | All organisms are subject to variations in nutrient availability and face periods of starvation, to which they have developed adaptations ranging from arrest in a quiescent state to elaborate differentiation strategies. Differentiation responses to nitrogen (N) starvation include for instance encystation/sporulation of many protists and fungi, the transition to multicellularity in the slime mold Dictyostelium discoideum , the formation of appressoria for plant invasion in fungal pathogens, or the induction of sexual differentiation in several fungi and green algae . The presence of nitrogen is detected by signaling pathways that direct metabolic and growth responses, of which the mechanistic target of rapamycin (TOR) kinase forms a major hub. TOR kinase is part of 2 distinct complexes (TORC), of which TORC1 is a critical regulator of growth and metabolism . TORC1 supports anabolic growth by promoting protein, nucleotide, and lipid synthesis, and represses the catabolic autophagy pathway. Thus, TORC1 inactivation upon N starvation stops growth, but also de-represses autophagy, promoting the recycling of cellular material, which allows for specific transcriptional and translational programs supporting cell differentiation. Indeed, in many organisms, TORC1 inactivation, for instance through inhibition by its specific inhibitor rapamycin, is sufficient to produce an N starvation-like response. The fission yeast Schizosaccharomyces pombe is a great model to study cell differentiation in response to N starvation, as this triggers either entry into quiescence or, when compatible mates are present, sexual differentiation. Sexual differentiation depends on the master transcriptional regulator Ste11, whose function is repressed in rich medium by a host of signaling pathways, including protein kinase A, CDK1, and TORC1 . Some of this regulation is known in molecular detail, with for instance the upstream Ste11 transcription factors Rst2 and Fkh2 directly phosphorylated by PKA and CDK1, respectively . TORC1 plays a prominent role. Indeed, conditional inactivation of Tor2, the essential catalytic component of TORC1, mimics starvation responses and leads to sexual reproduction in rich media when compatible mates are present . Inactivation of other TORC1 subunits and activators, such as Rheb or RAG GTPases, similarly leads to starvation and sexual differentiation responses upon inactivation , while TORC1 hyperactivation or loss of negative regulators represses mating . Several additional kinases serve a positive role to induce Ste11 function and sexual differentiation, including the stress-MAPK cascade, AMPK, and TORC2 . These pathways are strongly interlinked. For example, AMPK contributes to TORC1 inhibition upon nitrogen stress ; TORC1 and TORC2 function in a signaling relay to promote sexual differentiation, where inactivation of TORC1 relieves the inhibition of the phosphatase PP2A on the main TORC2 kinase effector Gad8 , and their function converge on downstream targets such as the SAGA transcription factor . The choice between quiescence and sexual differentiation depends on the presence of compatible mates, which signal through secreted, diffusible pheromones. Two mating types, P ( h+ ) and M ( h- ), each secrete their own pheromone and detect the presence of the partner pheromone by a cognate G-protein-coupled receptor (GPCR), which elicits the activation of the same Ras-MAPK cascade in both cell types . Engagement of pheromone-MAPK signaling promotes the expression of Ste11, which enhances the expression of all pheromone-responsive genes, forming a positive feedback loop that locks cells in sexual differentiation. Pheromones are secreted from an initially mobile polarity patch at the cell surface and are interpreted directionally for partner cells to pair . Enhanced pheromone perception stabilizes the patch, leading to the growth of a cell projection (called shmoo), bringing partner cells in contact. Fusion of the 2 cells to form the diploid zygote then depends on the formation of an actin fusion focus assembled by the formin Fus1, which serves to concentrate the secretion of hydrolases that digest the cell wall at the contact site . The timing of fusion focus assembly and stabilization is critical to ensure that cell wall digestion is coupled to pair formation. Fusion timing is also controlled by pheromone-MAPK signaling playing a proximal role at the fusion focus, where all components of the signaling cascade accumulate . Finally, cell–cell fusion, closely followed by karyogamy, yields the diploid zygote, which immediately suppresses mating and enters meiosis. The zygotic fate is imposed by both transcriptional and post-transcriptional regulations involving the master meiotic regulators Mei3 and Mei2 , and culminates in the formation of four stress-resistant spores. While many kinases (and some phosphatases) have well-established roles in promoting sexual differentiation and cell fusion, we still have limited knowledge about their substrates. Which proteins are (de-)phosphorylated to drive differentiation and the morphogenetic changes for cell projection and cell fusion? To fill this knowledge gap, we have developed protocols to synchronize cell populations during mating and conducted time-course phosphoproteomics analyses to unveil the phosphorylation changes that occur dynamically as cells differentiate and then fuse together to form the zygote. We report on the unexpected finding that pheromone signaling leads to TORC1 re-activation in conditions of N starvation and that this is necessary for efficient mating and spore formation.
Synchronization of cell fusion by optogenetics We aimed to describe the protein phosphorylation changes occurring during sexual differentiation and cell fusion in the fission yeast Schizosaccharomyces pombe . One significant issue for population-based phosphoproteomics analysis is the lack of synchrony in the differentiation process of fission yeast cells. Inspired by the tools used to synchronize the cell cycle, and for which time-course phosphoproteomics on synchronized populations have been powerful in deciphering the order of events , we first developed a means to synchronize cells pre-fusion and release them synchronously into the fusion process. The Fus1 formin, which assembles the fusion focus, is a critical regulator of cell fusion, as fus1Δ cells are completely fusion-deficient, arresting as differentiated, paired cells. Because Fus1 is only expressed upon sexual differentiation and is solely required for cell–cell fusion , a conditional fus1 allele would in principle allow to specifically block cells pre-fusion in restrictive conditions and synchronously promote fusion upon reactivation. Because mating is inefficient at high temperatures, we could not use temperature as the conditional change. Instead, we designed a conditional fus1 allele ( fus1 opto ) based on an optogenetic design, where Fus1 N- and C-termini (Fus1N and Fus1C) are expressed as distinct polypeptides respectively linked to the A . thaliana photosensitive CIBN and CRY2 binding partners . Because Fus1N lacks actin-assembly activity and Fus1C lacks localization and condensation functionalities , the 2 separate halves are predicted to be nonfunctional in the dark, leading to arrest of cell pairs pre-fusion. Blue light illumination, which promotes CRY2-CIBN binding, should reunite the 2 halves, restoring a functional Fus1, allowing synchronous entry into the fusion process. Indeed, when fus1 opto cells were starved in the dark for 6 h, they formed pre-fusion pairs, in which the CRY2-Fus1C moiety was diffusely localized . Upon blue light illumination, CRY2-Fus1C immediately formed a bright focus at the site of cell–cell contact leading to cell fusion, as seen by entry of the P-cell expressed mTagBFP2 fluorophore into the M-cell. The median time between illumination and fusion was 20 min . Importantly, when fus1 opto cells were mated on plates kept in light conditions over 24 h, they formed zygotes as efficiently as WT cells , indicating that the allele is functional in permissive conditions. As expected, expressing CRY2-Fus1C alone did not support cell fusion, but we were surprised to observe that about 20% of Fus1N-CIBN and dark-kept fus1 opto cells successfully fused. These observations suggest a fusogenic activity of Fus1N, independent of actin assembly, which will be investigated elsewhere. As shown below, this weak fusogenic activity was not problematic in shorter mating reactions. Thus, the fus1 opto allele provides an efficient way to block cells pre-fusion and release them synchronously into fusion. In the experiment above, at the time of Fus1 opto activation, only about 5% of cells were engaged in mating (; see also , ammonium trace). In an attempt to maximize the fraction of pre-fusion cell pairs and synchronously release them all into fusion, we increased the time cells were mated in the dark. However, fusion efficiency dropped when cells were held in the dark for longer times , suggesting that cell pairs held in a pre-fusion state eventually lose fusion competence. Thus, the fus1 opto allele works well for early mating cells but requires faster sexual differentiation and cell pairing to synchronize the entire cell population. Conditions to increase mating speed and synchrony The G1 phase of the cell cycle is the only permissive phase for sexual differentiation . Fission yeast cells grown in presence of ammonium as main nitrogen source exhibit a very short G1 phase and spend most of their cell cycle in G2 phase. Therefore, in the population, only a small fraction of cells is in G1 phase. This provides one explanation for the slow, asynchronous mating process. Indeed, starved WT homothallic cells pre-grown in MSL-ammonium medium slowly accumulate zygotes over >20 h . We reasoned that increasing the fraction of G1 cells in the population would hasten the process. Entry into S-phase and thus the relative length of G1 and G2 phase can be modulated by nitrogen sources . We tested glutamate, phenylalanine, and proline as alternatives to ammonium. Pre-growing cells on medium containing these amino acids as main nitrogen source led to faster mating upon nitrogen starvation . Surprisingly, the fastest mating was observed with pre-growth on glutamate, a source slightly poorer than ammonium , but better than proline . The change was only in the rate of mating, not on the overall mating or fusion success . We thus used glutamate to hasten sexual differentiation in all subsequent experiments. As a second way to slow down S-phase entry and increase G1 phase, we deleted all non-essential cyclins ( cig1Δ cig2Δ crs1Δ puc1Δ rem1Δ , here named cycΔ5 ). These cells exhibit a delay in G1/S transition, resulting in an increased cell size and meiotic defects . However, these cells are highly mating competent, yielding similar pairing and fusion efficiencies as WT cells after 24 h of starvation . In addition, they mate substantially faster than WT cells. Like WT cells, cycΔ5 cells differentiated faster when pre-grown on glutamate than ammonium, with maximal mating reached about 6 h after starvation ( h+ and h- strains were pre-starved separately for 2 h before mixing; ). In these conditions, mating was remarkably synchronous, going from none to 60% of the population between 2 and 4 h post-mixing . Finally, we combined the cycΔ5 and fus1 opto alleles to create optimal synchronization conditions for mating and cell fusion. We conducted the following experiments (as well as all the phosphoproteomics preparations below) on agar plates, on which we observed even faster and more extensive mating than in liquid cultures, with 70% of cells forming pairs already 180 min after mixing . We note that in the cycΔ5 fast mating conditions, the Fus1N-CIBN weak fusogenic activity was not visible, as cycΔ5 fus1 opto cells held in the dark for up to 2 h 30 post-mixing were not observed to form zygotes . We probed for the optimal time to induce cell fusion by shifting the plate to the light for 30 min after varying amounts of time in the dark. Maximal zygote fraction (57.8% of the entire cell population) was seen after 180 min (150 min dark, 30 min light). In summary, we set up conditions for fast sexual differentiation and synchronous cell fusion in cycΔ5 fus1 opto cells, which allows to perform population-based time course studies. Phosphoproteomic time-course analyses of nitrogen starvation To describe the phosphorylation changes occurring during sexual differentiation, we collected protein extracts at 45 min interval of h- and h+ cells, mixed on agar plates lacking nitrogen, and kept in the dark to block cell–cell fusion (mating samples) . To discriminate the changes induced by pheromone signaling from those induced by starvation and transfer to solid medium, we also collected extracts of h- and h+ cells plated on distinct petri dishes and mixed only at the time of extract preparation (starvation samples). Using 10plex tandem mass tag (TMT), 3 biological replicates were then analyzed by proteomics and, after phospho-peptide enrichment, by phosphoproteomics. A total of 2,851 proteins (out of 5,133 coding genes in the S . pombe genome) and 10,828 phosphosites in 1,884 proteins were identified in at least 2 biological replicates. After data normalization to correct for batch (replicate) effect, principal component (PC) analysis revealed good clustering of replicates and stratification of starvation and mating data for both proteomic and phosphoproteomic data, with PC1 reflecting time and PC2 reflecting major differences between starvation and mating . All proteomic and phosphoproteomic changes described in this study are described in and Tables, as well as on an associated website for easy browsing ( https://pombephosphoproteomics.unige.ch/ ). Phosphorylation changes were highly reproducible across biological replicates (see website and for some examples). Even though the starvation data set does not probe for acute changes in the (phospho)proteome upon starvation, as strains were pre-starved in liquid conditions for 2 h before plating, control samples showed massive changes in the proteome and phospho-proteome, with at least 847 proteins and 3’100 phosphosites in about 1’000 proteins found to exhibit significant change in amounts over the 3 h time course ( and Figs and ). Because of the massive change in protein amounts, one concern in analyzing the phosphoproteome is that changes in phosphosite abundance may be a consequence of change in protein levels rather than a regulatory change. To probe for this possible confounding factor, we compared the slope of the changes in protein and phosphosite amounts over time. This analysis revealed that at least 78% of sites vary independently of protein concentration . Thus, even if a substantial fraction of phosphosites is concordant with protein abundance, global phosphoproteomic changes are not dominated by changes in protein levels . Gene ontology (GO) analysis showed that proteins involved in catabolic processes were enriched in the group whose levels increased, whereas proteins involved in biosynthetic processes, such as amino-acid biosynthesis, ribosomal, and translation functions, were enriched in the group whose levels decreased during starvation . This is in agreement with a general switch-off of de novo protein translation and the known re-use of existing cellular material, in particular through autophagy, upon nitrogen depletion . We indeed observed an enrichment in proteins involved in autophagy in the group whose phosphorylation increases during starvation . Interestingly, mitochondrial respiration functions also featured prominently in proteins increasing during starvation, perhaps reflecting the recently established link between starvation-induced autophagy and respiration . Significant changes in phosphorylation patterns included GO terms in autophagy, transmembrane transport, and cell cycle regulation, but also affected different classes of functions, mostly relating to the cytoskeleton and cell polarity . Comparison with prior phosphoproteomic studies revealed a small enrichment in targets of Orb6 and CDK1 kinases in phosphosites increasing during starvation ( , top). Strikingly, phosphosites decreasing during starvation were highly enriched in candidate TOR targets identified in 2 previous phosphoproteomic studies . While the attribution of sites to TORC1 or TORC2 has been difficult due to absence of specific TORC2 inhibitors, and differs in the 2 studies, a loss of TORC1-dependent phosphorylation is expected from the well-established inactivation of TORC1 upon starvation. Overall, our global analysis of proteomic and phosphoproteomic changes upon nitrogen starvation indicates a considerable re-wiring of cellular metabolism, reducing de novo protein translation and increasing nitrogen and carbon catabolism, likely imposed in part by reduction in TORC1 activity. While this forms a rich data set to explore the signaling consequences of nitrogen starvation, we did not investigate it further in this study. Instead, we used it as normalizing information to identify phosphosites changing specifically during mating. Phosphoproteomic time-course analyses of sexual differentiation To identify changes specific to mating, we looked for significant changes after subtraction of the starvation signal from the mating one. Specifically, we fit a linear model with the values ( i t m a t i n g - i t - 1 m a t i n g ) - ( i t s t a r v a t i o n - i t - 1 s t a r v a t i o n ) , where i t m a t i n g is the intensity of a given peptide in the mating sample at time t. Only 54 proteins showed significant changes in abundance (adjusted p -value <0.05; ), most of which increased. Consistent with the known positive feedback between pheromone signaling and transcriptional control during sexual differentiation , proteins with increased abundance were enriched in pheromone signaling proteins, but also in cell wall remodeling factors . We also identified a total of 272 phosphosites with significant changes along the mating time course. Of these, 116 changed only during mating and 156 also significantly changed upon starvation but changed differently in the presence of mating partners ( ; see also for examples of phosphorylation profiles during starvation, mating, and the corrected mating-starvation signal). Where proteomic information was available, we explicitly investigated which of these variations were independent of protein abundance, which showed that 79% to 93% of phosphorylation changes are independent of protein level changes . Thus, compared to nitrogen starvation, the presence of a mate induces relatively modest changes in the proteome and phospho-proteome, which are largely independent of each other. The phosphosites can be grouped in 2 clusters whose intensity increases or decreases monotonically during sexual differentiation . GO term enrichments were highly divergent in the 2 groups . Proteins involved in G1/S transition, including the MBF transcription complex (Cdc10, Res1, and Whi5), actin cytoskeleton organization, vesicle-mediated transport, cell polarity or localization to the cell tip were highly enriched in the group showing increased phosphorylation. Proteins involved in amino acid, peptide, and nucleobase transport were strongly enriched in the dephosphorylated group. In both groups, protein localization to the plasma membrane and the Golgi were overrepresented. These phosphorylation changes at the cell periphery are consistent with G1 arrest and reorganization of the polarity and actin machinery for shmoo growth and in preparation for cell–cell fusion. Examination of the sequence around phosphorylation sites showed overrepresentation of proline after the phospho-S/T residue, a motif typical of CDK and MAPK, and overrepresentation of basic residues before the phospho-S/T, characteristic of the group of AGC kinases ( , left) . The substrates of the pheromone-MAPK Spk1 await identification, but consistent with these observations, comparison with available phosphoproteome data sets showed a small enrichment in targets of CDK1 and very large enrichment in targets of the AGC group NDR/LATS-family Orb6 kinase ( , bottom and ). Orb6 is essential for mitotic growth and viability, and its activity is down-regulated upon nitrogen starvation . This suggests that Orb6 is re-activated upon pheromone signaling when its functions in cell polarization and vesicular trafficking described during mitotic growth are likely repurposed for shmoo growth. The enrichment for TOR targets associated with dephosphorylated peptides during nitrogen starvation was largely lost in our mating-specific data set, which discounts starvation signals, although surprisingly some enrichment remained significant with one of the 2 TOR target data sets (see below). We note that the overrepresentation of basic residues before the phospho-S/T remained after removal of all known Orb6 targets, suggesting activation of additional basic residues-directed kinases ( , right). Among sites whose phosphorylation increases, we identified several expected sites on proteins necessary for mating (colored in Figs and ), including in the pheromone MAPK-Spk1 active site (T199 and Y201) and on the C-terminal tails of pheromone receptors, likely to promote their internalization (T325 on Mam2 and S346, S349 on Map3) . Phosphorylation of the transcription factor Ste11 also significantly increased after 45 min, though surprisingly not at the previously mapped MAPK-Spk1 target sites , but at T173 shown to be phosphorylated by Pat1 kinase . Increase in Mei2 phosphorylation was also detected on 6 sites (S60, S171, S173, S420, S454, and T573), all different from the inhibitory Pat1 and TORC1 target sites , suggesting Mei2 is not only de-phosphorylated for mating induction but that other phosphorylation events also promote its activity. Unexpectedly, several proteins with specific function in meiosis showed significant increase in phosphorylation during mating (pre-fusion) . These phosphorylation events are not due to leakage fusion, as cells held in the dark showed no fusion in the phosphoproteomic screen conditions . These include Num1/Mcp5, the meiotic cortical dynein anchor, necessary for horsetail movement ; Rec10, a component of the meiotic linear element, functionally similar to the synaptonemal complex, on sites S347 and S529 previously identified in diploid cells , though with unclear functional significance ; the meiotic cohesin subunit Rec8 on sites S366 and S434, the first of which is phosphorylated by Casein kinase 1 to promote cleavage by separase ; the cohesin loading factor Mis4; and the RNA-binding protein Spo5, previously thought to be specifically expressed during meiosis . Spo5 has functions in sporulation and in promoting cyclin Cdc13 expression for meiosis II , and we found phosphorylation increase on sites necessary for timely degradation at meiosis II . Thus, expression and phosphorylation of (at least some) zygotic meiotic and sporulation factors occurs ahead of gamete fusion. Finally, we observed an unexpectedly large phosphorylation increase in Rps6 S235-S236 phosphorylation ( ; see below). Phosphoproteomic time-course analysis of cell–cell fusion To probe for specific phosphorylation changes over the fusion process, h- and h+ cells were mixed on agar plates lacking nitrogen and kept in the dark for 150 min. Protein extracts were collected upon illumination at 11 min interval (fusion samples; ). Because S . pombe does not encode light-sensing proteins, we did not specifically control for the light conditions. Data acquisition and analysis was performed as described above, but with 4 biological replicates. PC analysis showed decent clustering of replicate time points, with an interesting trajectory showing an inflexion at 22 min , corresponding to the median fusion time (see ). We identified 41 significant proteomic and 440 significant phosphoproteomic changes, 55 of which overlapped with sites shown to vary in the mating time course . As in the mating time course, most phosphorylation changes were independent of protein intensity changes . These phosphosites were grouped in 3 clusters: profiles in cluster 1 increased already over the first 11 to 22 min, those in cluster 2 increased only between 22 and 44 min, and those in cluster 3 decreased over the time course . Because 20 min is the median time from illumination to cell fusion (see ), the 2 classes of phosphorylation increase suggest that some sites are phosphorylated for cell fusion, whereas others are phosphorylated in the zygote in response to cell fusion perhaps to help terminate the process and prepare for meiosis and sporulation. Cluster 2 includes for instance Rep1/Rec16, which both promotes early meiotic gene expression and represses pheromone signaling genes , Spo5, Mis4, Rec10, and Rec8, which is further phosphorylated on multiple Casein kinase I and Polo kinase target sites . Both increasing clusters showed similar overrepresentation of actin-associated and cell tip-localized proteins . Enrichment for substrates of known kinases suggested a possible enrichment in TORC2 targets among phosphosites that increased in abundance, which may be consistent with its role in regulating the actin cytoskeleton . Among protein functions showing phosphorylation decrease, there was a strong overrepresentation of pheromone-MAPK cascade and polarity factors (Figs and ), suggesting that some factors phosphorylated for cell pair formation during mating become dephosphorylated during fusion. These include, for pheromone signaling, the MAPK Spk1, the pheromone receptor Map3, on the same sites that showed increase pre-fusion and the MAPKKK Byr2 ( , left). For polarity, we found many Orb6 targets as well as regulators and effectors of the small GTPase Cdc42 (the GEFs Scd1 and Gef1, scaffold Scd2, 2 Cdc42 GAPs and effector Shk2/Pak2 kinase) ( , right). Finally, we again identified an increase in Rps6 S235-S236 phosphorylation . Reactivation of TORC1 by the pheromone-MAPK pathway during nitrogen starvation Our mating and fusion phosphoproteomic data sets revealed strong phosphorylation of ribosomal protein S6 (Rps6), which is encoded by 2 paralogous genes ( rps601 and rps602 ), on the conserved residues S235-S236 (Figs and ). This phosphorylation event, which is conserved across eukaryotes, primarily depends on S6 kinases (S6K) downstream of TORC1 and is widely used as a marker of TORC1 activity . In fission yeast, Rps6 is phosphorylated during mitotic growth by the S6K-like kinase Psk1 downstream of TORC1 . However, upon nitrogen starvation, TORC1 activity is down-regulated, and down-regulation of TORC1 activity is necessary and sufficient for the initiation of sexual differentiation . Rps6 is also phosphorylated by the AGC family kinase Gad8 downstream of TORC2 , which positively regulates mating . We thus tested which TOR complex controls Rps6 phosphorylation during mating. We first verified the phosphoproteomic data by using phospho-specific antibodies against Rps6 S235-S236 phosphorylation in both cyc5Δ and WT cells. As expected, heterothallic h- strains showed low (but non-zero) Rps6 phosphorylation on S235-S236 after 6 h 30 (3 h for the cyc5Δ background) of nitrogen starvation. By contrast, homothallic h90 WT strains (which contain a mix of the 2 cell types and thus undergo sexual differentiation) or a mix of h+ and h- cycΔ5 cells showed high Rps6 phosphorylation at the same time points , confirming that Rps6 is phosphorylated during sexual differentiation also in WT cells. To probe which of the 2 TOR complexes functions upstream of Rps6 phosphorylation, we treated cells with Torin1, which inhibits both TOR complexes, and rapamycin, which specifically inhibits TORC1 . Drugs were added for 30 min on already differentiated cells. Rps6 phosphorylation was reduced in both conditions , suggesting dependence on TORC1 activity. Rps6 global levels were unchanged in these experiments. Because rapamycin has additional effects, including on Tor1, the main TORC2 kinase , and on the FKBP12 homologue Fkh1 during mating , we sought additional evidence to support the requirement for TORC1 activity. First, we found that the S6K-like Psk1 kinase, a direct substrate of TORC1 which in turn phosphorylates Rps6 during mitotic growth , was strongly phosphorylated on T415 in h90 but not in h- strains upon N-starvation, and that this phosphorylation was rapamycin dependent . 5 μm Torin1 was inefficient in the experiment shown in , but we confirmed that 25 μm Torin1 inhibited both Rps6 and Psk1 phosphorylation . By contrast, phosphorylation of the main TORC2 substrate Gad8 was largely unchanged in h90 strains and upon rapamycin treatment , confirming that rapamycin does not affect TORC2 activity. We note that Gad8 was destabilized upon Torin1 treatment . Second, the rapamycin-insensitive tor2 allele, tor2-S1837E , prevented loss of Rps6 and Psk1 phosphorylation upon rapamycin treatment, demonstrating that rapamycin inhibits these phosphorylation events by binding to the TORC1 kinase Tor2 . Interestingly, tor2-S1837E was also hypomorphic for Psk1 phosphorylation . We further confirmed that Psk1 phosphorylates Rps6 also during mating . We conclude that phosphorylation of Rps6 during mating depends on TORC1 signaling through Psk1 kinase. Sexual differentiation requires pheromone signaling through a GPCR-Ras-MAPK cascade functionally similar to the ERK cascade in mammalian cells. The cascade can be elicited in heterothallic cells by exposure to synthetic pheromone normally produced by partner cells. Nitrogen starvation of h- cells (lacking the P-factor protease Sxa2) led to acute loss of Psk1 and Rps6 phosphorylation over the first 20 to 40 min of starvation, followed by signal re-appearance after 4 to 8 h . This re-appearance of phospho-Rps6 and phospho-Psk1 after a few hours of starvation reflects the documented autophagy-mediated reactivation of TORC1, likely in response to autophagy-dependent release of intracellular nutrients . Addition of 1 μg/ml P-factor did not change the acute loss of phosphorylation upon starvation, but strongly boosted their re-appearance . We also observed similar dynamics when probing for phospho-Psk1 in a time course of mating cells . Thus, pheromone signaling promotes TORC1 re-activation. Pheromone-dependent reactivation of TORC1 does not require autophagy We probed for possible links between pheromone-dependent TORC1 re-activation and autophagy. In nitrogen-free medium, autophagy-deficient mutants are unable to initiate sexual reproduction due to their inability to recycle proteins to basic building blocks . However, they are fertile if induced to mate in low levels of nitrogen . We established that a shift from MSL + 15 mM to MSL + 0.5 (or 0.75) mM glutamate allowed both h- sxa2Δ (WT) and h- sxa2Δ atg1Δ mutants to form shmoos in response to 1 μg/ml P-factor . We first used these conditions to probe whether starvation-dependent induction of autophagy is modulated by pheromone signaling. As a readout, we used CFP-Atg8 cleavage, as CFP-Atg8 is delivered to vacuoles by the autophagic process, where only the CFP moiety is resistant to vacuolar proteases. In WT cells, CFP-Atg8 cleavage occurred upon transfer to low-glutamate conditions and increased over time in a similar manner whether cells were treated with MetOH or P-factor . This suggests that pheromone signaling does not induce autophagy. When we probed CFP-Atg8 cleavage in atg1Δ as control, we were surprised to observe significant residual levels of free CFP (Figs and ), as deletion of this upstream regulatory kinase is reported to fully block autophagy . We reasoned that this difference may be due to differences in media or nitrogen source, as we used MSL + glutamate, while earlier studies used EMM + ammonium. Indeed, whereas exchanging glutamate for ammonium during growth before starvation had no effect on the observed cleavage in MSL medium , using EMM medium abrogated cleavage , as previously reported. Thus, for currently unknown reasons, deletion of Atg1 kinase only partially prevents autophagy in MSL medium. To identify autophagy-deficient mutants in MSL medium, we screened through several atg mutants for those that would block CFP-Atg8 cleavage. This revealed that atg mutants exhibit a range of residual cleavage after 4 h starvation in MSL-N, with deletion of atg5 , atg7 , and atg18a being the most potent in blocking CFP release in these conditions . We thus chose, in addition to atg1Δ , atg5Δ , and atg18aΔ , which also responded to P-factor in our established conditions , to test for Psk1 phosphorylation. In WT ( sxa2Δ ), Psk1 phosphorylation loss was less acute upon transfer to low-glutamate than nitrogen-free medium, as expected. At late time points, Psk1 was further de-phosphorylated in DMSO-treated samples and was strongly re-phosphorylated in response to P-factor, as observed upon complete nitrogen starvation . In atg1Δ , atg5Δ , and atg18aΔ mutants, the dynamics of Psk1 phosphorylation was very similar: in all cases, the P-Psk1 signal initially dropped upon transfer to low glutamate (see 20’ and 40’ time points). In control conditions (MetOH), it then stayed low or was even further reduced at late time points (240’ and 480’). By contrast, in presence of P-factor, Psk1 was re-phosphorylated at these late time points, as happens in WT cells (Figs and ; compare the 240’ and 480’ time points for P-factor versus MetOH control). Western blots of CFP-Atg8 cleavage over the time course confirmed strong reduction of autophagy in the atg5Δ and atg18aΔ mutants . We conclude that pheromone-dependent TORC1 re-activation does not require the autophagy pathway. TORC1 activity plays a positive role in sexual reproduction We probed the possible function of Rps6 and its phosphorylation during sexual reproduction. Previous work had shown that rps601Δ and rps602Δ are synthetic lethal, but that neither gene alone, nor their phosphorylation, nor the Psk1 kinase, is essential for growth . We found that psk1Δ cells are fully competent for mating, and that non-phosphorylatable Rps6 ( rps601Δ rps602 SSAA ) is no less competent for mating than the control rps601Δ strain, indicating that Psk1 and Rps6 phosphorylation are not critical TORC1 substrates during mating, but merely act as good markers of its re-activation . Although rps601Δ cells had low mating efficiency, we noticed that rps601Δ and rps602Δ strains were fully sterile in strains auxotrophic for leucine . While the reasons for this synthetic effect are currently unclear, leucine is a potent direct activator of TORC1 activity in budding yeast and mammalian cells . S . pombe leu1- cells are rapamycin-sensitive and show fast TORC1 inactivation upon starvation , suggesting these cells also have reduced TORC1 activity . We indeed found Psk1 and Rps6 phosphorylation to be reduced during mating in leu1-32 compared to WT cells, indicating lower TORC1 re-activation . We had previously noted that leu1-32 cells exhibit fusion delays and defects . These cells also show strongly reduced mating efficiency . These correlative data suggest that TORC1 re-activation is required to promote efficient mating. To probe this point more stringently, we aimed to directly inhibit TORC1 during mating. TORC1 inhibition by rapamycin involves the formation of a trimeric complex between rapamycin, the TOR kinase and the FKBP12 prolyl isomerase, Fkh1 in S . pombe . Previous work showed that rapamycin inhibits mating in S . pombe , but the sterility of fkh1Δ cells makes it impossible to interpret whether TORC1 inactivation contributes to this inhibition . TORC1 can also be inhibited by caffeine or auxin . Addition of these compounds to WT mating cells led to dose-dependent inhibition of mating, consistent with TORC1 activity acting positively during sexual reproduction . We showed above that the rapamycin-insensitive allele tor2-S1837E is hypomorphic, as Psk1 phosphorylation is reduced in this strain. In contrast to tor2 temperature-sensitive mutants, which undergo mating in rich conditions at restrictive temperature , tor2-S1837E mutants do not spontaneously mate. By contrast, tor2-S1837E mutants were slow to agglutinate in liquid mating reactions and produced aberrant tetrads with many inviable spores . Similarly, tetrads produced by the tor2-ts10 allele shifted at 32°C to inactivate it only during sexual reproduction were aberrant and contained inviable spores, even if these spores were germinated at permissive temperature . Finally, we found that deletion of the non-essential TORC1 component tco89 , which exhibits reduced TORC1 activity for lifespan regulation and resistance to Torin1 and caffeine , caused a reduction in mating efficiency . We conclude that TORC1 re-activation by pheromone signaling is required for correct progression of sexual reproduction.
We aimed to describe the protein phosphorylation changes occurring during sexual differentiation and cell fusion in the fission yeast Schizosaccharomyces pombe . One significant issue for population-based phosphoproteomics analysis is the lack of synchrony in the differentiation process of fission yeast cells. Inspired by the tools used to synchronize the cell cycle, and for which time-course phosphoproteomics on synchronized populations have been powerful in deciphering the order of events , we first developed a means to synchronize cells pre-fusion and release them synchronously into the fusion process. The Fus1 formin, which assembles the fusion focus, is a critical regulator of cell fusion, as fus1Δ cells are completely fusion-deficient, arresting as differentiated, paired cells. Because Fus1 is only expressed upon sexual differentiation and is solely required for cell–cell fusion , a conditional fus1 allele would in principle allow to specifically block cells pre-fusion in restrictive conditions and synchronously promote fusion upon reactivation. Because mating is inefficient at high temperatures, we could not use temperature as the conditional change. Instead, we designed a conditional fus1 allele ( fus1 opto ) based on an optogenetic design, where Fus1 N- and C-termini (Fus1N and Fus1C) are expressed as distinct polypeptides respectively linked to the A . thaliana photosensitive CIBN and CRY2 binding partners . Because Fus1N lacks actin-assembly activity and Fus1C lacks localization and condensation functionalities , the 2 separate halves are predicted to be nonfunctional in the dark, leading to arrest of cell pairs pre-fusion. Blue light illumination, which promotes CRY2-CIBN binding, should reunite the 2 halves, restoring a functional Fus1, allowing synchronous entry into the fusion process. Indeed, when fus1 opto cells were starved in the dark for 6 h, they formed pre-fusion pairs, in which the CRY2-Fus1C moiety was diffusely localized . Upon blue light illumination, CRY2-Fus1C immediately formed a bright focus at the site of cell–cell contact leading to cell fusion, as seen by entry of the P-cell expressed mTagBFP2 fluorophore into the M-cell. The median time between illumination and fusion was 20 min . Importantly, when fus1 opto cells were mated on plates kept in light conditions over 24 h, they formed zygotes as efficiently as WT cells , indicating that the allele is functional in permissive conditions. As expected, expressing CRY2-Fus1C alone did not support cell fusion, but we were surprised to observe that about 20% of Fus1N-CIBN and dark-kept fus1 opto cells successfully fused. These observations suggest a fusogenic activity of Fus1N, independent of actin assembly, which will be investigated elsewhere. As shown below, this weak fusogenic activity was not problematic in shorter mating reactions. Thus, the fus1 opto allele provides an efficient way to block cells pre-fusion and release them synchronously into fusion. In the experiment above, at the time of Fus1 opto activation, only about 5% of cells were engaged in mating (; see also , ammonium trace). In an attempt to maximize the fraction of pre-fusion cell pairs and synchronously release them all into fusion, we increased the time cells were mated in the dark. However, fusion efficiency dropped when cells were held in the dark for longer times , suggesting that cell pairs held in a pre-fusion state eventually lose fusion competence. Thus, the fus1 opto allele works well for early mating cells but requires faster sexual differentiation and cell pairing to synchronize the entire cell population.
The G1 phase of the cell cycle is the only permissive phase for sexual differentiation . Fission yeast cells grown in presence of ammonium as main nitrogen source exhibit a very short G1 phase and spend most of their cell cycle in G2 phase. Therefore, in the population, only a small fraction of cells is in G1 phase. This provides one explanation for the slow, asynchronous mating process. Indeed, starved WT homothallic cells pre-grown in MSL-ammonium medium slowly accumulate zygotes over >20 h . We reasoned that increasing the fraction of G1 cells in the population would hasten the process. Entry into S-phase and thus the relative length of G1 and G2 phase can be modulated by nitrogen sources . We tested glutamate, phenylalanine, and proline as alternatives to ammonium. Pre-growing cells on medium containing these amino acids as main nitrogen source led to faster mating upon nitrogen starvation . Surprisingly, the fastest mating was observed with pre-growth on glutamate, a source slightly poorer than ammonium , but better than proline . The change was only in the rate of mating, not on the overall mating or fusion success . We thus used glutamate to hasten sexual differentiation in all subsequent experiments. As a second way to slow down S-phase entry and increase G1 phase, we deleted all non-essential cyclins ( cig1Δ cig2Δ crs1Δ puc1Δ rem1Δ , here named cycΔ5 ). These cells exhibit a delay in G1/S transition, resulting in an increased cell size and meiotic defects . However, these cells are highly mating competent, yielding similar pairing and fusion efficiencies as WT cells after 24 h of starvation . In addition, they mate substantially faster than WT cells. Like WT cells, cycΔ5 cells differentiated faster when pre-grown on glutamate than ammonium, with maximal mating reached about 6 h after starvation ( h+ and h- strains were pre-starved separately for 2 h before mixing; ). In these conditions, mating was remarkably synchronous, going from none to 60% of the population between 2 and 4 h post-mixing . Finally, we combined the cycΔ5 and fus1 opto alleles to create optimal synchronization conditions for mating and cell fusion. We conducted the following experiments (as well as all the phosphoproteomics preparations below) on agar plates, on which we observed even faster and more extensive mating than in liquid cultures, with 70% of cells forming pairs already 180 min after mixing . We note that in the cycΔ5 fast mating conditions, the Fus1N-CIBN weak fusogenic activity was not visible, as cycΔ5 fus1 opto cells held in the dark for up to 2 h 30 post-mixing were not observed to form zygotes . We probed for the optimal time to induce cell fusion by shifting the plate to the light for 30 min after varying amounts of time in the dark. Maximal zygote fraction (57.8% of the entire cell population) was seen after 180 min (150 min dark, 30 min light). In summary, we set up conditions for fast sexual differentiation and synchronous cell fusion in cycΔ5 fus1 opto cells, which allows to perform population-based time course studies.
To describe the phosphorylation changes occurring during sexual differentiation, we collected protein extracts at 45 min interval of h- and h+ cells, mixed on agar plates lacking nitrogen, and kept in the dark to block cell–cell fusion (mating samples) . To discriminate the changes induced by pheromone signaling from those induced by starvation and transfer to solid medium, we also collected extracts of h- and h+ cells plated on distinct petri dishes and mixed only at the time of extract preparation (starvation samples). Using 10plex tandem mass tag (TMT), 3 biological replicates were then analyzed by proteomics and, after phospho-peptide enrichment, by phosphoproteomics. A total of 2,851 proteins (out of 5,133 coding genes in the S . pombe genome) and 10,828 phosphosites in 1,884 proteins were identified in at least 2 biological replicates. After data normalization to correct for batch (replicate) effect, principal component (PC) analysis revealed good clustering of replicates and stratification of starvation and mating data for both proteomic and phosphoproteomic data, with PC1 reflecting time and PC2 reflecting major differences between starvation and mating . All proteomic and phosphoproteomic changes described in this study are described in and Tables, as well as on an associated website for easy browsing ( https://pombephosphoproteomics.unige.ch/ ). Phosphorylation changes were highly reproducible across biological replicates (see website and for some examples). Even though the starvation data set does not probe for acute changes in the (phospho)proteome upon starvation, as strains were pre-starved in liquid conditions for 2 h before plating, control samples showed massive changes in the proteome and phospho-proteome, with at least 847 proteins and 3’100 phosphosites in about 1’000 proteins found to exhibit significant change in amounts over the 3 h time course ( and Figs and ). Because of the massive change in protein amounts, one concern in analyzing the phosphoproteome is that changes in phosphosite abundance may be a consequence of change in protein levels rather than a regulatory change. To probe for this possible confounding factor, we compared the slope of the changes in protein and phosphosite amounts over time. This analysis revealed that at least 78% of sites vary independently of protein concentration . Thus, even if a substantial fraction of phosphosites is concordant with protein abundance, global phosphoproteomic changes are not dominated by changes in protein levels . Gene ontology (GO) analysis showed that proteins involved in catabolic processes were enriched in the group whose levels increased, whereas proteins involved in biosynthetic processes, such as amino-acid biosynthesis, ribosomal, and translation functions, were enriched in the group whose levels decreased during starvation . This is in agreement with a general switch-off of de novo protein translation and the known re-use of existing cellular material, in particular through autophagy, upon nitrogen depletion . We indeed observed an enrichment in proteins involved in autophagy in the group whose phosphorylation increases during starvation . Interestingly, mitochondrial respiration functions also featured prominently in proteins increasing during starvation, perhaps reflecting the recently established link between starvation-induced autophagy and respiration . Significant changes in phosphorylation patterns included GO terms in autophagy, transmembrane transport, and cell cycle regulation, but also affected different classes of functions, mostly relating to the cytoskeleton and cell polarity . Comparison with prior phosphoproteomic studies revealed a small enrichment in targets of Orb6 and CDK1 kinases in phosphosites increasing during starvation ( , top). Strikingly, phosphosites decreasing during starvation were highly enriched in candidate TOR targets identified in 2 previous phosphoproteomic studies . While the attribution of sites to TORC1 or TORC2 has been difficult due to absence of specific TORC2 inhibitors, and differs in the 2 studies, a loss of TORC1-dependent phosphorylation is expected from the well-established inactivation of TORC1 upon starvation. Overall, our global analysis of proteomic and phosphoproteomic changes upon nitrogen starvation indicates a considerable re-wiring of cellular metabolism, reducing de novo protein translation and increasing nitrogen and carbon catabolism, likely imposed in part by reduction in TORC1 activity. While this forms a rich data set to explore the signaling consequences of nitrogen starvation, we did not investigate it further in this study. Instead, we used it as normalizing information to identify phosphosites changing specifically during mating.
To identify changes specific to mating, we looked for significant changes after subtraction of the starvation signal from the mating one. Specifically, we fit a linear model with the values ( i t m a t i n g - i t - 1 m a t i n g ) - ( i t s t a r v a t i o n - i t - 1 s t a r v a t i o n ) , where i t m a t i n g is the intensity of a given peptide in the mating sample at time t. Only 54 proteins showed significant changes in abundance (adjusted p -value <0.05; ), most of which increased. Consistent with the known positive feedback between pheromone signaling and transcriptional control during sexual differentiation , proteins with increased abundance were enriched in pheromone signaling proteins, but also in cell wall remodeling factors . We also identified a total of 272 phosphosites with significant changes along the mating time course. Of these, 116 changed only during mating and 156 also significantly changed upon starvation but changed differently in the presence of mating partners ( ; see also for examples of phosphorylation profiles during starvation, mating, and the corrected mating-starvation signal). Where proteomic information was available, we explicitly investigated which of these variations were independent of protein abundance, which showed that 79% to 93% of phosphorylation changes are independent of protein level changes . Thus, compared to nitrogen starvation, the presence of a mate induces relatively modest changes in the proteome and phospho-proteome, which are largely independent of each other. The phosphosites can be grouped in 2 clusters whose intensity increases or decreases monotonically during sexual differentiation . GO term enrichments were highly divergent in the 2 groups . Proteins involved in G1/S transition, including the MBF transcription complex (Cdc10, Res1, and Whi5), actin cytoskeleton organization, vesicle-mediated transport, cell polarity or localization to the cell tip were highly enriched in the group showing increased phosphorylation. Proteins involved in amino acid, peptide, and nucleobase transport were strongly enriched in the dephosphorylated group. In both groups, protein localization to the plasma membrane and the Golgi were overrepresented. These phosphorylation changes at the cell periphery are consistent with G1 arrest and reorganization of the polarity and actin machinery for shmoo growth and in preparation for cell–cell fusion. Examination of the sequence around phosphorylation sites showed overrepresentation of proline after the phospho-S/T residue, a motif typical of CDK and MAPK, and overrepresentation of basic residues before the phospho-S/T, characteristic of the group of AGC kinases ( , left) . The substrates of the pheromone-MAPK Spk1 await identification, but consistent with these observations, comparison with available phosphoproteome data sets showed a small enrichment in targets of CDK1 and very large enrichment in targets of the AGC group NDR/LATS-family Orb6 kinase ( , bottom and ). Orb6 is essential for mitotic growth and viability, and its activity is down-regulated upon nitrogen starvation . This suggests that Orb6 is re-activated upon pheromone signaling when its functions in cell polarization and vesicular trafficking described during mitotic growth are likely repurposed for shmoo growth. The enrichment for TOR targets associated with dephosphorylated peptides during nitrogen starvation was largely lost in our mating-specific data set, which discounts starvation signals, although surprisingly some enrichment remained significant with one of the 2 TOR target data sets (see below). We note that the overrepresentation of basic residues before the phospho-S/T remained after removal of all known Orb6 targets, suggesting activation of additional basic residues-directed kinases ( , right). Among sites whose phosphorylation increases, we identified several expected sites on proteins necessary for mating (colored in Figs and ), including in the pheromone MAPK-Spk1 active site (T199 and Y201) and on the C-terminal tails of pheromone receptors, likely to promote their internalization (T325 on Mam2 and S346, S349 on Map3) . Phosphorylation of the transcription factor Ste11 also significantly increased after 45 min, though surprisingly not at the previously mapped MAPK-Spk1 target sites , but at T173 shown to be phosphorylated by Pat1 kinase . Increase in Mei2 phosphorylation was also detected on 6 sites (S60, S171, S173, S420, S454, and T573), all different from the inhibitory Pat1 and TORC1 target sites , suggesting Mei2 is not only de-phosphorylated for mating induction but that other phosphorylation events also promote its activity. Unexpectedly, several proteins with specific function in meiosis showed significant increase in phosphorylation during mating (pre-fusion) . These phosphorylation events are not due to leakage fusion, as cells held in the dark showed no fusion in the phosphoproteomic screen conditions . These include Num1/Mcp5, the meiotic cortical dynein anchor, necessary for horsetail movement ; Rec10, a component of the meiotic linear element, functionally similar to the synaptonemal complex, on sites S347 and S529 previously identified in diploid cells , though with unclear functional significance ; the meiotic cohesin subunit Rec8 on sites S366 and S434, the first of which is phosphorylated by Casein kinase 1 to promote cleavage by separase ; the cohesin loading factor Mis4; and the RNA-binding protein Spo5, previously thought to be specifically expressed during meiosis . Spo5 has functions in sporulation and in promoting cyclin Cdc13 expression for meiosis II , and we found phosphorylation increase on sites necessary for timely degradation at meiosis II . Thus, expression and phosphorylation of (at least some) zygotic meiotic and sporulation factors occurs ahead of gamete fusion. Finally, we observed an unexpectedly large phosphorylation increase in Rps6 S235-S236 phosphorylation ( ; see below).
To probe for specific phosphorylation changes over the fusion process, h- and h+ cells were mixed on agar plates lacking nitrogen and kept in the dark for 150 min. Protein extracts were collected upon illumination at 11 min interval (fusion samples; ). Because S . pombe does not encode light-sensing proteins, we did not specifically control for the light conditions. Data acquisition and analysis was performed as described above, but with 4 biological replicates. PC analysis showed decent clustering of replicate time points, with an interesting trajectory showing an inflexion at 22 min , corresponding to the median fusion time (see ). We identified 41 significant proteomic and 440 significant phosphoproteomic changes, 55 of which overlapped with sites shown to vary in the mating time course . As in the mating time course, most phosphorylation changes were independent of protein intensity changes . These phosphosites were grouped in 3 clusters: profiles in cluster 1 increased already over the first 11 to 22 min, those in cluster 2 increased only between 22 and 44 min, and those in cluster 3 decreased over the time course . Because 20 min is the median time from illumination to cell fusion (see ), the 2 classes of phosphorylation increase suggest that some sites are phosphorylated for cell fusion, whereas others are phosphorylated in the zygote in response to cell fusion perhaps to help terminate the process and prepare for meiosis and sporulation. Cluster 2 includes for instance Rep1/Rec16, which both promotes early meiotic gene expression and represses pheromone signaling genes , Spo5, Mis4, Rec10, and Rec8, which is further phosphorylated on multiple Casein kinase I and Polo kinase target sites . Both increasing clusters showed similar overrepresentation of actin-associated and cell tip-localized proteins . Enrichment for substrates of known kinases suggested a possible enrichment in TORC2 targets among phosphosites that increased in abundance, which may be consistent with its role in regulating the actin cytoskeleton . Among protein functions showing phosphorylation decrease, there was a strong overrepresentation of pheromone-MAPK cascade and polarity factors (Figs and ), suggesting that some factors phosphorylated for cell pair formation during mating become dephosphorylated during fusion. These include, for pheromone signaling, the MAPK Spk1, the pheromone receptor Map3, on the same sites that showed increase pre-fusion and the MAPKKK Byr2 ( , left). For polarity, we found many Orb6 targets as well as regulators and effectors of the small GTPase Cdc42 (the GEFs Scd1 and Gef1, scaffold Scd2, 2 Cdc42 GAPs and effector Shk2/Pak2 kinase) ( , right). Finally, we again identified an increase in Rps6 S235-S236 phosphorylation .
Our mating and fusion phosphoproteomic data sets revealed strong phosphorylation of ribosomal protein S6 (Rps6), which is encoded by 2 paralogous genes ( rps601 and rps602 ), on the conserved residues S235-S236 (Figs and ). This phosphorylation event, which is conserved across eukaryotes, primarily depends on S6 kinases (S6K) downstream of TORC1 and is widely used as a marker of TORC1 activity . In fission yeast, Rps6 is phosphorylated during mitotic growth by the S6K-like kinase Psk1 downstream of TORC1 . However, upon nitrogen starvation, TORC1 activity is down-regulated, and down-regulation of TORC1 activity is necessary and sufficient for the initiation of sexual differentiation . Rps6 is also phosphorylated by the AGC family kinase Gad8 downstream of TORC2 , which positively regulates mating . We thus tested which TOR complex controls Rps6 phosphorylation during mating. We first verified the phosphoproteomic data by using phospho-specific antibodies against Rps6 S235-S236 phosphorylation in both cyc5Δ and WT cells. As expected, heterothallic h- strains showed low (but non-zero) Rps6 phosphorylation on S235-S236 after 6 h 30 (3 h for the cyc5Δ background) of nitrogen starvation. By contrast, homothallic h90 WT strains (which contain a mix of the 2 cell types and thus undergo sexual differentiation) or a mix of h+ and h- cycΔ5 cells showed high Rps6 phosphorylation at the same time points , confirming that Rps6 is phosphorylated during sexual differentiation also in WT cells. To probe which of the 2 TOR complexes functions upstream of Rps6 phosphorylation, we treated cells with Torin1, which inhibits both TOR complexes, and rapamycin, which specifically inhibits TORC1 . Drugs were added for 30 min on already differentiated cells. Rps6 phosphorylation was reduced in both conditions , suggesting dependence on TORC1 activity. Rps6 global levels were unchanged in these experiments. Because rapamycin has additional effects, including on Tor1, the main TORC2 kinase , and on the FKBP12 homologue Fkh1 during mating , we sought additional evidence to support the requirement for TORC1 activity. First, we found that the S6K-like Psk1 kinase, a direct substrate of TORC1 which in turn phosphorylates Rps6 during mitotic growth , was strongly phosphorylated on T415 in h90 but not in h- strains upon N-starvation, and that this phosphorylation was rapamycin dependent . 5 μm Torin1 was inefficient in the experiment shown in , but we confirmed that 25 μm Torin1 inhibited both Rps6 and Psk1 phosphorylation . By contrast, phosphorylation of the main TORC2 substrate Gad8 was largely unchanged in h90 strains and upon rapamycin treatment , confirming that rapamycin does not affect TORC2 activity. We note that Gad8 was destabilized upon Torin1 treatment . Second, the rapamycin-insensitive tor2 allele, tor2-S1837E , prevented loss of Rps6 and Psk1 phosphorylation upon rapamycin treatment, demonstrating that rapamycin inhibits these phosphorylation events by binding to the TORC1 kinase Tor2 . Interestingly, tor2-S1837E was also hypomorphic for Psk1 phosphorylation . We further confirmed that Psk1 phosphorylates Rps6 also during mating . We conclude that phosphorylation of Rps6 during mating depends on TORC1 signaling through Psk1 kinase. Sexual differentiation requires pheromone signaling through a GPCR-Ras-MAPK cascade functionally similar to the ERK cascade in mammalian cells. The cascade can be elicited in heterothallic cells by exposure to synthetic pheromone normally produced by partner cells. Nitrogen starvation of h- cells (lacking the P-factor protease Sxa2) led to acute loss of Psk1 and Rps6 phosphorylation over the first 20 to 40 min of starvation, followed by signal re-appearance after 4 to 8 h . This re-appearance of phospho-Rps6 and phospho-Psk1 after a few hours of starvation reflects the documented autophagy-mediated reactivation of TORC1, likely in response to autophagy-dependent release of intracellular nutrients . Addition of 1 μg/ml P-factor did not change the acute loss of phosphorylation upon starvation, but strongly boosted their re-appearance . We also observed similar dynamics when probing for phospho-Psk1 in a time course of mating cells . Thus, pheromone signaling promotes TORC1 re-activation.
We probed for possible links between pheromone-dependent TORC1 re-activation and autophagy. In nitrogen-free medium, autophagy-deficient mutants are unable to initiate sexual reproduction due to their inability to recycle proteins to basic building blocks . However, they are fertile if induced to mate in low levels of nitrogen . We established that a shift from MSL + 15 mM to MSL + 0.5 (or 0.75) mM glutamate allowed both h- sxa2Δ (WT) and h- sxa2Δ atg1Δ mutants to form shmoos in response to 1 μg/ml P-factor . We first used these conditions to probe whether starvation-dependent induction of autophagy is modulated by pheromone signaling. As a readout, we used CFP-Atg8 cleavage, as CFP-Atg8 is delivered to vacuoles by the autophagic process, where only the CFP moiety is resistant to vacuolar proteases. In WT cells, CFP-Atg8 cleavage occurred upon transfer to low-glutamate conditions and increased over time in a similar manner whether cells were treated with MetOH or P-factor . This suggests that pheromone signaling does not induce autophagy. When we probed CFP-Atg8 cleavage in atg1Δ as control, we were surprised to observe significant residual levels of free CFP (Figs and ), as deletion of this upstream regulatory kinase is reported to fully block autophagy . We reasoned that this difference may be due to differences in media or nitrogen source, as we used MSL + glutamate, while earlier studies used EMM + ammonium. Indeed, whereas exchanging glutamate for ammonium during growth before starvation had no effect on the observed cleavage in MSL medium , using EMM medium abrogated cleavage , as previously reported. Thus, for currently unknown reasons, deletion of Atg1 kinase only partially prevents autophagy in MSL medium. To identify autophagy-deficient mutants in MSL medium, we screened through several atg mutants for those that would block CFP-Atg8 cleavage. This revealed that atg mutants exhibit a range of residual cleavage after 4 h starvation in MSL-N, with deletion of atg5 , atg7 , and atg18a being the most potent in blocking CFP release in these conditions . We thus chose, in addition to atg1Δ , atg5Δ , and atg18aΔ , which also responded to P-factor in our established conditions , to test for Psk1 phosphorylation. In WT ( sxa2Δ ), Psk1 phosphorylation loss was less acute upon transfer to low-glutamate than nitrogen-free medium, as expected. At late time points, Psk1 was further de-phosphorylated in DMSO-treated samples and was strongly re-phosphorylated in response to P-factor, as observed upon complete nitrogen starvation . In atg1Δ , atg5Δ , and atg18aΔ mutants, the dynamics of Psk1 phosphorylation was very similar: in all cases, the P-Psk1 signal initially dropped upon transfer to low glutamate (see 20’ and 40’ time points). In control conditions (MetOH), it then stayed low or was even further reduced at late time points (240’ and 480’). By contrast, in presence of P-factor, Psk1 was re-phosphorylated at these late time points, as happens in WT cells (Figs and ; compare the 240’ and 480’ time points for P-factor versus MetOH control). Western blots of CFP-Atg8 cleavage over the time course confirmed strong reduction of autophagy in the atg5Δ and atg18aΔ mutants . We conclude that pheromone-dependent TORC1 re-activation does not require the autophagy pathway.
We probed the possible function of Rps6 and its phosphorylation during sexual reproduction. Previous work had shown that rps601Δ and rps602Δ are synthetic lethal, but that neither gene alone, nor their phosphorylation, nor the Psk1 kinase, is essential for growth . We found that psk1Δ cells are fully competent for mating, and that non-phosphorylatable Rps6 ( rps601Δ rps602 SSAA ) is no less competent for mating than the control rps601Δ strain, indicating that Psk1 and Rps6 phosphorylation are not critical TORC1 substrates during mating, but merely act as good markers of its re-activation . Although rps601Δ cells had low mating efficiency, we noticed that rps601Δ and rps602Δ strains were fully sterile in strains auxotrophic for leucine . While the reasons for this synthetic effect are currently unclear, leucine is a potent direct activator of TORC1 activity in budding yeast and mammalian cells . S . pombe leu1- cells are rapamycin-sensitive and show fast TORC1 inactivation upon starvation , suggesting these cells also have reduced TORC1 activity . We indeed found Psk1 and Rps6 phosphorylation to be reduced during mating in leu1-32 compared to WT cells, indicating lower TORC1 re-activation . We had previously noted that leu1-32 cells exhibit fusion delays and defects . These cells also show strongly reduced mating efficiency . These correlative data suggest that TORC1 re-activation is required to promote efficient mating. To probe this point more stringently, we aimed to directly inhibit TORC1 during mating. TORC1 inhibition by rapamycin involves the formation of a trimeric complex between rapamycin, the TOR kinase and the FKBP12 prolyl isomerase, Fkh1 in S . pombe . Previous work showed that rapamycin inhibits mating in S . pombe , but the sterility of fkh1Δ cells makes it impossible to interpret whether TORC1 inactivation contributes to this inhibition . TORC1 can also be inhibited by caffeine or auxin . Addition of these compounds to WT mating cells led to dose-dependent inhibition of mating, consistent with TORC1 activity acting positively during sexual reproduction . We showed above that the rapamycin-insensitive allele tor2-S1837E is hypomorphic, as Psk1 phosphorylation is reduced in this strain. In contrast to tor2 temperature-sensitive mutants, which undergo mating in rich conditions at restrictive temperature , tor2-S1837E mutants do not spontaneously mate. By contrast, tor2-S1837E mutants were slow to agglutinate in liquid mating reactions and produced aberrant tetrads with many inviable spores . Similarly, tetrads produced by the tor2-ts10 allele shifted at 32°C to inactivate it only during sexual reproduction were aberrant and contained inviable spores, even if these spores were germinated at permissive temperature . Finally, we found that deletion of the non-essential TORC1 component tco89 , which exhibits reduced TORC1 activity for lifespan regulation and resistance to Torin1 and caffeine , caused a reduction in mating efficiency . We conclude that TORC1 re-activation by pheromone signaling is required for correct progression of sexual reproduction.
In this work, we performed large (phospho)proteomic studies of fission yeast sexual gamete differentiation and fusion, revealing the extensive remodeling of the (phospho)proteome during these developmental transitions. We present rich, high-confidence data sets, which document numerous phosphorylation changes likely to be driven by the activity of many kinases and phosphatases. These data sets constitute a valuable resource to understand how signaling promotes the physiology of starvation, mating, and cell fusion, which can be explored either in bulk ( and Tables) or by browsing through the associated website ( https://pombephosphoproteomics.unige.ch/ ). We have set up robust synchronization procedures to conduct population-based assays of mating and fusing cells, which will be of general use for other -omics analyses. It is however important to bear in mind the specific conditions allowing optimal synchrony. First, the cyc5Δ cells lacking all 5 non-essential cyclins ( cig1Δ cig2Δ crs1Δ puc1Δ rem1Δ ), which extends the differentiation-permissive G1 phase and increases cell length, may mask some of the (phospho-)proteomic changes governing the starvation- and/or pheromone-dependent cell cycle arrest of wild-type cells . Second, because of the use of the split fus1 opto allele for synchronization during cell–cell fusion, we should be careful about interpretation of phosphorylation events on Fus1 and accessory proteins, as these may not represent native changes at the fusion site. Nevertheless, the few phosphorylation events that were expected (for instance that of the MAPK Spk1 active site; ) or independently confirmed here (Rps6) validate the approach and suggest that identified phosphorylation events also occur in wild-type strains. Third, we have used glutamate in MSL medium, instead of ammonium for pre-growth of cells to hasten differentiation, and found evidence that medium composition (MSL versus EMM) may influence signaling outcome (see discussion on autophagy below). Our analyses of the (phospho)proteomic changes during sexual reproduction provide an interesting first global view, for which we highlight here only a few aspects. Nitrogen starvation causes massive signaling and metabolic changes, including reduction in de novo protein translation and increase in nitrogen and carbon catabolism, in part ruled by loss of TORC1 activity and increase in autophagy. Globally, these changes are very similar whether cells prepare for quiescence or differentiate sexually, as the presence of a mate modifies the starvation signaling landscape only in a relatively modest manner. The presence of a mate additionally induces signatures of G1 arrest, expression of pheromone signaling components, re-activation of growth control pathways (see below), and reorganization of the actin, polarity and cell wall machinery, likely for the morphogenetic events of shmoo outgrowth and cell fusion. A strong phospho-S/T-proline motif in phosphopeptides further suggests that many such phosphosites may be substrates of the pheromone-MAPK Spk1 though this awaits confirmation. Interestingly, some unexpected changes in differentiating haploid cells before fusion, such as expression and phosphorylation of several meiotic and sporulation factors, indicate that preparation of zygotic meiosis and sporulation already takes place in the gametes. Finally, the global analysis of (phospho)proteomic changes during cell fusion reveals an interesting inflexion at the median time of fusion, suggesting a very rapid qualitative change in signaling at the time of fusion pore opening. Thus, these data form a strong basis for future dissection of the signaling and fate changes underlying the haploid-to-diploid transition. Here, we have focused on the unexpected observation that Rps6 is highly phosphorylated during sexual differentiation and upon cell fusion. Reactivation of TORC1 by pheromone signaling Our findings demonstrate that Rps6 is phosphorylated by the S6K-like Psk1 downstream of TORC1. Indeed, not only are Rps6 S235-S236 phosphorylated by Psk1 during mating, but also Psk1 T415, which is a direct substrate of TORC1. We note that Psk1 T415 was not identified in our data sets due to absence of any peptide covering this site. Furthermore, these phosphorylation events are sensitive to rapamycin, which specifically inhibits TORC1, and this sensitivity is blocked by mutation of the TORC1 catalytic subunit Tor2. Thus, even though Rps6 is also substrate of Gad8 downstream of TORC2 , and TORC2 is essential for fertility , in wild-type cells, the critical activity change that leads to Rps6 phosphorylation during sexual reproduction is that of TORC1. In wild-type cells, our data also suggest that the activity of TORC2, although it is essential for fertility , may not increase during sexual reproduction, as observed from the observation that Gad8 phosphorylation does not increase. We note however that pheromone-dependent phosphorylation of Rps6 (but not Psk1) is still observed in the hypomorphic tor2-S1837E mutant, suggesting contribution of other kinases than Psk1 to Rps6 phosphorylation, perhaps as cellular adaptation to constitutive reduction of TORC1 activity. At first glance, the finding that TORC1 activity increases during mating is surprising. Indeed, TORC1 is well known to monitor nutrient availability and be inactivated upon starvation. Furthermore, many lines of evidence have shown that TORC1 inactivation is both necessary and sufficient to initiate sexual differentiation in S . pombe . For example, temperature-sensitive mutants of tor2 lead to sexual differentiation in rich media at restrictive temperature, while mutants with hyperactive TORC1 display sterility . Our data do not put this prior evidence in question but show that subsequent efficient progression through sexual reproduction coincides with, and requires, TORC1 reactivation. We present several lines of evidence showing that low TORC1 activity yields low mating efficiency. Mating success was reduced in a dose-dependent manner by caffeine or auxin, molecules demonstrated to inhibit TORC1 ; mating was reduced in absence of the non-essential TORC1 subunit Tco89, whose extended lifespan and sensitivity to Torin1 and caffeine indicate low TORC1 activity ; sexual agglutination was reduced in the hypomorphic tor2-S1837E allele, though we did not detect reproducible reduction in overall mating success; finally, mating was also strongly reduced in leu1-32 mutants, which we show here have reduced TORC1 activity during mating. This latter observation agrees with prior evidence of low TORC1 activity in leucine auxotrophs . Notably, leu1- but not WT strains are sensitive to rapamycin and exhibit fast onset of autophagy upon leucine depletion , suggesting that auxotrophs are on the verge of starvation and TORC1 inactivation. In S . cerevisiae and mammalian cells, leucine activates TORC1 either directly through the Leucyl-tRNA synthetase or indirectly through the Gtr1/2–RAGulator complex . In S . pombe , the levels of precursor tRNA, including that for leucine, function upstream of TORC1 and are down-regulated upon starvation , but the specific sensing pathway is unknown. We now link low TORC1 activity in leu1-32 with poor mating and inefficient cell–cell fusion ( and ). This link is further supported by the previous work on the Ppk32 kinase, deletion of which exhibits increased TORC1 activity specifically in the leu1-32 mutant background . During sexual differentiation, these cells reactivate TORC1 earlier than WT cells (a reactivation attributed to activation of autophagy in this earlier work), which restores better mating efficiency. Together, these data support the view that TORC1 activity is required for efficient mating progression. The activity of TORC1 is likely required throughout the sexual reproduction process. Indeed, both tor2-S1837E and tor2-ts10 mutants showed low spore viability. The conditional, temperature-sensitive tor2-ts10 allele is particularly informative, as its inactivation at restrictive temperature induces sexual differentiation , but sexual reproduction at this restrictive temperature yields a large fraction of inviable spores, even when these are returned to permissive temperature before germination. A role for TORC1 activity during meiosis and sporulation is also supported by the observation of meiotic defects in mutants of the Raptor-like TORC1 subunit Mip1 . In fact, it makes sense that morphogenetic processes such as conjugation, cell fusion, and spore formation should require the anabolic-promoting function of TORC1. Thus, even though TORC1 activity is lost to initiate sexual differentiation, it must then be restored both in gametes and likely in the zygote to sustain sexual development. An interesting question for the future is whether sexual development solely requires a quantitative increase in TORC1 activity or whether TORC1 signaling during mating is qualitatively different. Unfortunately, there are only few known bona fide sites in S . pombe besides Psk1 T415 . Several TORC1 target sites have been identified on Mei2 , for which we recovered traces for S39, but only in 2 replicates, with a tendency to increase during mating. TORC1 was proposed to phosphorylate Atg13 , but specific residues are unknown. Among sites identified in at least one of the 2 previous TORC1 phosphoproteomes , sites on eIF2β (Tif212), eIF4γ (Tif471), Osh2, Rga3, Rng2, Ppk30, Bbc1, Pom1, Trs130, and a few uncharacterized proteins also increase in our mating data set, suggesting phosphorylation of Psk1 and Rps6 are not the only consequences of TORC1 reactivation. We also see increase in phosphorylation on Oca2 and Cek1, orthologues of S . cerevisiae Npr1 and Rim15, phosphorylated downstream of TORC1 in this organism . By contrast, the main phosphorylation sites on Maf1, S63, which retards its electrophoretic migration and has been proposed as bona fide TORC1 substrate , does not show significant change in any of our data sets. A second open question is the mechanism of TORC1 reactivation by pheromone signaling. In response to nitrogen starvation, TORC1 inactivation drives autophagy, which in turn promotes TORC1 reactivation likely by increasing intracellular amino acid levels . We have shown that pheromone signaling does not boost autophagy, as estimated from the CTP-Atg8 cleavage assay. This result also agrees with prior data showing that ectopic activation of the sexual differentiation pathway in rich conditions does not promote autophagy . Furthermore, even if dynamics are not identical, autophagy-deficient cells exposed to P-factor show TORC1 re-activation as do WT cells, demonstrating that pheromone-dependent reactivation of TORC1 does not require the autophagic process per se. However, because addition of low levels of nitrogen is essential for autophagy mutants to differentiate sexually, and thus to be able to respond to pheromones, we cannot know whether TORC1 re-activation by pheromones requires the availability of small amounts of nitrogen, be it delivered exogenously or by the autophagic process. We can nevertheless conclude that the dynamics of TORC1 re-activation does not follow that of nitrogen availability. As a side note, the unexpected finding that atg1Δ mutants show some level (and atg12Δ wild-type–like levels) of CFP-Atg8 cleavage in the MSL medium suggests that autophagy is regulated by the environment. The Atg1 kinase complex sits at the top of a hierarchical autophagy pathway and integrates various inputs, including regulation by TOR kinase . While it is widely considered essential for induction of autophagy, previous work has shown that the Atg1 mammalian homologues (ULK1/2) are not absolutely required for basal autophagy . EMM (in which S . pombe atg1Δ cells are fully autophagy-deficient) and MSL are defined synthetic media with slightly different amounts of salts and ions and few additional differences. Our finding should invite examination of the role of medium composition on autophagy and more generally on cell physiology. All our experiments, except where indicated, were conducted upon starvation from glutamate-containing medium. The activation of TORC1 by pheromone signaling is reminiscent of the activation of mTORC1 by growth factor in mammalian cells. For example, the mammalian Ras-ERK pathway promotes mTORC1 activation by phosphorylation of the Tuberin Tsc2, a GAP for the Rheb GTPase TORC1 activator . Homologues of Rheb GTPase and its GAP exist and control TORC1 activity in S . pombe cells . It will be exciting to investigate whether the mechanism of TORC1 re-activation during mating is similar. Reactivation of growth promoting pathways by pheromone in nitrogen-starved cells While we focused our attention on the reactivation of TORC1, our phosphoproteomic dataset forms a rich resource to examine the signaling changes imposed by pheromone sensing. For example, the data set provides evidence that the NDR/LATS-family Orb6 kinase is also re-activated by pheromone signaling. In animal cells, the LATS1/2 kinases are central elements of the Hippo pathway, involved in tissue growth control, and NDR1/2 kinases have been involved in multiple forms of cancer . In fission yeast, the essential Orb6 kinase promotes polarized growth by antagonizing translational repression, notably of Ras GTPase activators, and by phosphorylating regulators of Cdc42 GTPase and the secretory system . During N-starvation, Orb6 activity is down-regulated to extend lifespan . Our data show an impressive 15-fold enrichment in Orb6 substrates among sites with phosphorylation increases during sexual differentiation relative to starvation-only conditions. Thus, like TORC1 and despite continued starvation, Orb6 is reactivated for mating, where it may promote growth of the mating projection. Thus, the 2 major growth-promoting pathways—TORC1 and Orb6—are both reactivated for sexual reproduction. It will be exciting to dissect the mechanisms by which pheromones act as growth factors.
Our findings demonstrate that Rps6 is phosphorylated by the S6K-like Psk1 downstream of TORC1. Indeed, not only are Rps6 S235-S236 phosphorylated by Psk1 during mating, but also Psk1 T415, which is a direct substrate of TORC1. We note that Psk1 T415 was not identified in our data sets due to absence of any peptide covering this site. Furthermore, these phosphorylation events are sensitive to rapamycin, which specifically inhibits TORC1, and this sensitivity is blocked by mutation of the TORC1 catalytic subunit Tor2. Thus, even though Rps6 is also substrate of Gad8 downstream of TORC2 , and TORC2 is essential for fertility , in wild-type cells, the critical activity change that leads to Rps6 phosphorylation during sexual reproduction is that of TORC1. In wild-type cells, our data also suggest that the activity of TORC2, although it is essential for fertility , may not increase during sexual reproduction, as observed from the observation that Gad8 phosphorylation does not increase. We note however that pheromone-dependent phosphorylation of Rps6 (but not Psk1) is still observed in the hypomorphic tor2-S1837E mutant, suggesting contribution of other kinases than Psk1 to Rps6 phosphorylation, perhaps as cellular adaptation to constitutive reduction of TORC1 activity. At first glance, the finding that TORC1 activity increases during mating is surprising. Indeed, TORC1 is well known to monitor nutrient availability and be inactivated upon starvation. Furthermore, many lines of evidence have shown that TORC1 inactivation is both necessary and sufficient to initiate sexual differentiation in S . pombe . For example, temperature-sensitive mutants of tor2 lead to sexual differentiation in rich media at restrictive temperature, while mutants with hyperactive TORC1 display sterility . Our data do not put this prior evidence in question but show that subsequent efficient progression through sexual reproduction coincides with, and requires, TORC1 reactivation. We present several lines of evidence showing that low TORC1 activity yields low mating efficiency. Mating success was reduced in a dose-dependent manner by caffeine or auxin, molecules demonstrated to inhibit TORC1 ; mating was reduced in absence of the non-essential TORC1 subunit Tco89, whose extended lifespan and sensitivity to Torin1 and caffeine indicate low TORC1 activity ; sexual agglutination was reduced in the hypomorphic tor2-S1837E allele, though we did not detect reproducible reduction in overall mating success; finally, mating was also strongly reduced in leu1-32 mutants, which we show here have reduced TORC1 activity during mating. This latter observation agrees with prior evidence of low TORC1 activity in leucine auxotrophs . Notably, leu1- but not WT strains are sensitive to rapamycin and exhibit fast onset of autophagy upon leucine depletion , suggesting that auxotrophs are on the verge of starvation and TORC1 inactivation. In S . cerevisiae and mammalian cells, leucine activates TORC1 either directly through the Leucyl-tRNA synthetase or indirectly through the Gtr1/2–RAGulator complex . In S . pombe , the levels of precursor tRNA, including that for leucine, function upstream of TORC1 and are down-regulated upon starvation , but the specific sensing pathway is unknown. We now link low TORC1 activity in leu1-32 with poor mating and inefficient cell–cell fusion ( and ). This link is further supported by the previous work on the Ppk32 kinase, deletion of which exhibits increased TORC1 activity specifically in the leu1-32 mutant background . During sexual differentiation, these cells reactivate TORC1 earlier than WT cells (a reactivation attributed to activation of autophagy in this earlier work), which restores better mating efficiency. Together, these data support the view that TORC1 activity is required for efficient mating progression. The activity of TORC1 is likely required throughout the sexual reproduction process. Indeed, both tor2-S1837E and tor2-ts10 mutants showed low spore viability. The conditional, temperature-sensitive tor2-ts10 allele is particularly informative, as its inactivation at restrictive temperature induces sexual differentiation , but sexual reproduction at this restrictive temperature yields a large fraction of inviable spores, even when these are returned to permissive temperature before germination. A role for TORC1 activity during meiosis and sporulation is also supported by the observation of meiotic defects in mutants of the Raptor-like TORC1 subunit Mip1 . In fact, it makes sense that morphogenetic processes such as conjugation, cell fusion, and spore formation should require the anabolic-promoting function of TORC1. Thus, even though TORC1 activity is lost to initiate sexual differentiation, it must then be restored both in gametes and likely in the zygote to sustain sexual development. An interesting question for the future is whether sexual development solely requires a quantitative increase in TORC1 activity or whether TORC1 signaling during mating is qualitatively different. Unfortunately, there are only few known bona fide sites in S . pombe besides Psk1 T415 . Several TORC1 target sites have been identified on Mei2 , for which we recovered traces for S39, but only in 2 replicates, with a tendency to increase during mating. TORC1 was proposed to phosphorylate Atg13 , but specific residues are unknown. Among sites identified in at least one of the 2 previous TORC1 phosphoproteomes , sites on eIF2β (Tif212), eIF4γ (Tif471), Osh2, Rga3, Rng2, Ppk30, Bbc1, Pom1, Trs130, and a few uncharacterized proteins also increase in our mating data set, suggesting phosphorylation of Psk1 and Rps6 are not the only consequences of TORC1 reactivation. We also see increase in phosphorylation on Oca2 and Cek1, orthologues of S . cerevisiae Npr1 and Rim15, phosphorylated downstream of TORC1 in this organism . By contrast, the main phosphorylation sites on Maf1, S63, which retards its electrophoretic migration and has been proposed as bona fide TORC1 substrate , does not show significant change in any of our data sets. A second open question is the mechanism of TORC1 reactivation by pheromone signaling. In response to nitrogen starvation, TORC1 inactivation drives autophagy, which in turn promotes TORC1 reactivation likely by increasing intracellular amino acid levels . We have shown that pheromone signaling does not boost autophagy, as estimated from the CTP-Atg8 cleavage assay. This result also agrees with prior data showing that ectopic activation of the sexual differentiation pathway in rich conditions does not promote autophagy . Furthermore, even if dynamics are not identical, autophagy-deficient cells exposed to P-factor show TORC1 re-activation as do WT cells, demonstrating that pheromone-dependent reactivation of TORC1 does not require the autophagic process per se. However, because addition of low levels of nitrogen is essential for autophagy mutants to differentiate sexually, and thus to be able to respond to pheromones, we cannot know whether TORC1 re-activation by pheromones requires the availability of small amounts of nitrogen, be it delivered exogenously or by the autophagic process. We can nevertheless conclude that the dynamics of TORC1 re-activation does not follow that of nitrogen availability. As a side note, the unexpected finding that atg1Δ mutants show some level (and atg12Δ wild-type–like levels) of CFP-Atg8 cleavage in the MSL medium suggests that autophagy is regulated by the environment. The Atg1 kinase complex sits at the top of a hierarchical autophagy pathway and integrates various inputs, including regulation by TOR kinase . While it is widely considered essential for induction of autophagy, previous work has shown that the Atg1 mammalian homologues (ULK1/2) are not absolutely required for basal autophagy . EMM (in which S . pombe atg1Δ cells are fully autophagy-deficient) and MSL are defined synthetic media with slightly different amounts of salts and ions and few additional differences. Our finding should invite examination of the role of medium composition on autophagy and more generally on cell physiology. All our experiments, except where indicated, were conducted upon starvation from glutamate-containing medium. The activation of TORC1 by pheromone signaling is reminiscent of the activation of mTORC1 by growth factor in mammalian cells. For example, the mammalian Ras-ERK pathway promotes mTORC1 activation by phosphorylation of the Tuberin Tsc2, a GAP for the Rheb GTPase TORC1 activator . Homologues of Rheb GTPase and its GAP exist and control TORC1 activity in S . pombe cells . It will be exciting to investigate whether the mechanism of TORC1 re-activation during mating is similar.
While we focused our attention on the reactivation of TORC1, our phosphoproteomic dataset forms a rich resource to examine the signaling changes imposed by pheromone sensing. For example, the data set provides evidence that the NDR/LATS-family Orb6 kinase is also re-activated by pheromone signaling. In animal cells, the LATS1/2 kinases are central elements of the Hippo pathway, involved in tissue growth control, and NDR1/2 kinases have been involved in multiple forms of cancer . In fission yeast, the essential Orb6 kinase promotes polarized growth by antagonizing translational repression, notably of Ras GTPase activators, and by phosphorylating regulators of Cdc42 GTPase and the secretory system . During N-starvation, Orb6 activity is down-regulated to extend lifespan . Our data show an impressive 15-fold enrichment in Orb6 substrates among sites with phosphorylation increases during sexual differentiation relative to starvation-only conditions. Thus, like TORC1 and despite continued starvation, Orb6 is reactivated for mating, where it may promote growth of the mating projection. Thus, the 2 major growth-promoting pathways—TORC1 and Orb6—are both reactivated for sexual reproduction. It will be exciting to dissect the mechanisms by which pheromones act as growth factors.
Yeast strains Strains were constructed using standard genetic manipulation of S . pombe either by tetrad dissection or transformation and can be found in . Plasmids generated for this study are listed in . To construct the optogenetic system, a homothallic ( h90 ) strain, auxotroph for uracil containing a myo52-tdtomato allele at the endogenous locus was transformed by PCR-based gene targeting using oligos osm933 and osm1670 on plasmid pSM693 to replace the fus1 ORF with an hphMX cassette and generate ySM4131. For homothallic fus1 opto cells, strain ySM4131 was first transformed with pSM2470 (3′region_fus1-5′region_fus1-fus1N 1-793 -CIBN-3′UTR_fus1-term fus1 -kanMX-pFA6a) linearized with BstZ17I to generate ySM4132, and then with pSM2475 (pUra4 AfeI -pfus1-cry2PHR-fus1C 796-1372 -sfGFP-term nmt1 ) linearized with AfeI to generate ySM4134. ySM4131 was also transformed with pSM2475 to generate a strain containing only the Cry2PHR-Fus1C-sfGFP fragment (ySM4133). To determine the timing of fusion in , a mTagBFP2 cytosolic marker under the map3 P-cell–specific promoter was introduced by transforming ySM4134 with pAV0761 linearized with SpeI to obtain ySM4130. For cycΔ5 fus1 opto , h- and h+ cycΔ5 strains containing the pfus1 :Cry2PHR- fus1C -sfGFP construct at the uracil locus were obtained through multiple crosses between yAV2050, yAV2038 , and ySM4130. These strains were then transformed with pSM3227 (3′region_fus1-5′region_fus1-fus1N 1-793 -CIBN-3′UTR_fus1-term fus1 -hphMX-pFA6a) linearized with BstZ17I to get the h- and h+ cycΔ5 fus1 opto strains ySM4135 and ySM4136, respectively. We used heterothallic cycΔ5 strains to control entry into sexual differentiation, because we observed homothallic strains to mate when reaching higher densities even on nitrogen-rich media. The h+ cycΔ5 fus1 opto strain (ySM4136) was further transformed with pSM3295 (pAde6 PmeI -pact1-mcherry-term tdh1 -patMX) linearized with RsrII to yield ySM4137. Colonies were selected on Edinburg Minimal Medium (EMM) supplemented with 400 μg/ml of glufosinate-ammonium (CatNo. G002P01G, Cluzeau Info Labo, Sainte-Foy-La-Grande, France) as previously described . The Rps6 prototroph mutants were obtained by crossing strain ANO229 to WT lab strains to generate ySM4138, ySM4139, and ySM4145. The heterothallic h- tor2 S1837E strain (ySM4063) is strain TA1397 from . The homothallic h90 tor2 S1837E strain (ySM4140) was obtained by crossing ySM4063 with a homothallic WT strain. The homothallic h90 psk1Δ deletion strains (ySM4141 and ySM4142) were obtained by transforming the homothallic wild-type strain ySM1396 with pSM3571 linearized with AfeI. The homothallic h90 psk1Δ leu1-32 strain (ySM4148) was obtained by transforming ySM4146 with pSM3571 linearized with AfeI. The h- sxa2Δ prototroph strain (ySM4143) was obtained by a marker switch on the strain S16 G10 from the Bioneer library to replace the kanMX6 cassette with an hphMX6 cassette, followed by crosses. The tor2-ts10 used in the study were obtained by crossing strain JV306 (ySM1591; h- tor2ts-10 ade6-M216 leu1-32 ) from to WT homothallic strains, giving ySM4144. Deletion of rps602 and rps601 in h90 leu1-32 strains were obtained by transforming ySM4147 with pSM3297 (3′region_rps602-5′region_rps602-kanMX-pFA6a) and pSM3298 (3′region_rps601-5′region_rps601-hphMX-pFA6a) linearized with AfeI to obtain ySM4149, ySM4150 and ySM4151, ySM4152, respectively. Deletion of tco89 was obtained by transforming ySM1396 with pSM3364 (3′UTR_tco89-5′UTR_tco89-hphMX-pFA6a) linearized with AfeI to generate ySM4157 . The heterothallic h- CFP-atg8 : leu1+ sxa2Δ :: natMX6 deleted for atg1 (ySM4157), atg5 (ySM4174), and atg18a (ySM4175) strains used to test the effect of autophagy on TORC1 reactivation were obtained by transforming DY11900 (ySM4158), DY4021 (ySM4170), and DY4031 (ySM4172) (Sun and colleagues) respectively, with pSM3274 (pFA6a -3′UTR_sxa2-5′UTR_sxa2-natMX) linearized with StuI. A heterothallic h- CFP-atg8 : leu1+ strain was isolated from a cross between DY11900 and a WT strain, auxotroph for leucine and then transformed with pSM3274 linearized with StuI to obtain ySM4156. Time-course phosphoproteomic experiments For time-course phosphoproteomic experiments, h- cycΔ5 fus1 opto and h+ cycΔ5 fus1 opto cells were grown and starved in a room kept in the dark and manipulation were done with the help of a red LED bulb (Osram). Unless specified, cells were always kept in the dark until they were quenched in 10% W/V TCA during the first step of protein extraction. For the time course during sexual differentiation, precultures of h- cycΔ5 fus1 opto and h+ cycΔ5 fus1 opto were grown to late log-phase in 40 ml, diluted to 400 ml of MSL + 15 mM glutamate, and grown to an OD 600nm = 0.5–0.9 at 30°C in the dark. For both cultures, a volume corresponding to 300 [OD 600nm ] was collected and washed separately 3 times with 20 ml of MSL-N. Centrifugation speed was set to 1,000 g for 2′ during the washes. Each strain was then resuspended in 200 ml of MSL-N and allowed to starve for 2 h at 30°C in the dark, and 100 ml of each strain were then mixed, cells were centrifuged at 1,000 g for 2′, the pellet was then resuspended in 2 ml of fresh MSL-N and 20 μl spots of the cell slurry were pipetted onto MSA-N plates at a density of 3 [OD 600nm ]/spot. We noted that spotting too many cells on the same plate drastically reduced mating efficiency and for that reason, only 5 spots were spotted on the same plate. We used an amount equivalent to 60 [OD 600nm ] per time point. The remaining 100 ml of culture of each heterothallic strain was centrifuged at 1,000 g for 2′ and the pellet resuspended in 1 ml of MSL-N. Both pellets were then spotted as 20 μl spots onto separate MSA-N at a density of 3 [OD 600nm ]/spot. We used an amount equivalent to 30 [OD 600nm ] per strain and time point. Plates were then incubated at 30°C in the dark until sample collection excepts the plates for the time point at 0′ which were directly used for protein extraction. For the phosphoproteomic time course during cell–cell fusion, pre-cultures of h- cycΔ5 fus1 opto and h+ cycΔ5 fus1 opto were grown to late log-phase in 30 ml, diluted to 300 ml of MSL + glutamate and grown to an OD 600nm = 0.5–0.9 at 30°C in the dark. For both cultures, an amount equivalent of 150 [OD 600nm ] of cells was collected and spun down at 1,000 g for 2′. The pellet was washed 3 times in 20 ml of MSL-N and re-suspended in 100 ml of MSL-N and cells were incubated for 2 h at 30°C in the dark. Cells were then mixed together and centrifuged at 1,000 g for 2′. The pellet was then resuspended in 2 ml of fresh MSL-N and spotted as 20 μl spot onto MSA-N plates (5 spot per plate). A total of 30 [OD 600nm ] were used per time point. Plates were incubated at 30°C in dark conditions for 2 h 30′ and then shifted to light condition until sample collection. For the time point at 0 min, the plates were directly used for protein extraction without being exposed to light. Protein extraction and digestion Extract protocol was adapted from . Yeast cells were collected from plates using 5 ml of 10% w/v ice-cold TCA. Cells were then spun down at 1,000 g for 2′ at 4°C and supernatant was discarded. Cells were washed first in 5 ml of acetone (cooled down at −20°C) and then in 1 ml of lysis buffer (50 mM ammonium bicarbonate, 10 mM DTT, 5% SDS). Pellets were resuspended in 400 μl of lysis buffer. Acid-washed glass beads (Sigma; G8772) were added to the samples and cells were lysed using a FastPrep-24 5G bead beating grinder (6 times shaking at 100 V for 30”, 30” break between runs). Samples were then centrifuged at max speed for 5′ at 4°C, and supernatant was recovered as protein fraction. Samples were immediately snap-frozen. For processing, samples were thawed rapidly and heated at 95°C for 10 min with shaking to lyse cells. Protein concentration was determined using the tryptophane fluorescence method . Trypsin digestion of 120 μg of protein material per sample was carried out according to the S-TRAP (Protifi, Farmingdale, New York) method as described . Briefly, after heating at 95°C to denature and reduce disulfides, cysteines were alkylated by reaction with 30 mM (final) chloroacetamide for 1 h in the dark at RT. An aliquot of 12% phosphoric acid was added to lower pH to 3.0, followed by dilution with 4 volumes of S-TRAP loading buffer (100 mM (final concentration) triethylammonium bicarbonate buffer (TEAB) (pH 8.0), in 90% MeOH). The obtained mixture was passed by centrifugation on S-TRAP Mini cartridges, which were then washed 3 times with 600 μl of loading buffer. Digestion was started by adding to the cartridges 20 μg of Trypsin (Promega) in 125 μl of 50 mM TEAB (pH 8) and was carried out for 2 h at 47°C without shaking. Digested peptides were eluted by centrifugation, followed by further elution of the cartridge with 80 μl of 50 mM TEAB, then 80 μl of 0.2 formic acid and 80 μl of 50% acetonitrile (each time at 3,000 × g for 1 min). All eluates were pooled, and samples were dried by evaporation. TMT labeling and cleanup Dried peptide mixtures were resuspended in 45 μl of 50 mM Hepes (pH 8.3) and reacted for 1 h at room temperature with 0.4 mg of TMT 10-plex reagents (Thermo Fisher Scientific, 90110) dissolved in 27.5 μl of pure acetonitrile. The reaction was quenched by adding 7 μl of 5% (v/w) hydroxylamine and incubating for 15 min at RT. Individual TMT labeled samples were pooled. The mixture obtained was acidified with TFA, frozen and the volume reduced to 1/10 of the initial by evaporation. After adding 9 volumes of aqueous 0.1% TFA, peptides were desalted on a C18 SepPak 1 cc 50 mg cartridge (Waters, #WAT054955). An aliquot of 10% of the eluate was dried and analyzed by MS as described below to assess labeling efficiency (which globally was found to be higher than 98.5%) and derive ratios of total protein content for normalization. Phosphopeptide enrichment The remaining of the TMT mix (approx. 1.0 mg) was dried, re-dissolved, and processed for phosphopeptide enrichment by IMAC using the High-Select Fe-NTA Phosphopeptide Enrichment Kit (Thermo Fisher Scientific, A32992) according to instructions from the manufacturer. The eluate from the IMAC cartridge was dried and resuspended in 70 μl of 2% MeCN, 0.05% TFA for LC-MS/MS analysis. Liquid chromatography-tandem mass spectrometry Tryptic peptide mixtures were injected on an Ultimate RSLC 3000 nanoHPLC system (Dionex, Sunnyvale, California, USA) interfaced to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, Bremen, Germany). Peptides were loaded onto a trapping microcolumn Acclaim PepMap100 C18 (20 mm × 100 μm ID, 5 μm, 100 Å, Thermo Scientific) before separation on a reversed-phase custom packed nanocolumn (75 μm ID × 40 cm, 1.8 μm particles, Reprosil Pur, Dr. Maisch). A flowrate of 0.25 μl/min was used with a gradient from 4% to 76% acetonitrile in 0.1% formic acid (total method time: 140 min). Both total proteome and phospho-enriched samples were injected multiple times with different methods and either with a normal nanospray ion source or a Field Asymmetric Waveform Ion Mobility Spectrometry interface (FAIMS pro, Thermo Fisher Scientific). For analyses with the normal nanoLC interface, a data-dependent acquisition method controlled by Xcalibur 4.2 software (Thermo Fisher Scientific) was used that optimized the number of precursors selected (“top speed”) of charge 2+ to 5+ while maintaining a fixed scan cycle of 1.5 s. The precursor isolation window used was 0.7 Th. Peptides were fragmented by higher energy collision dissociation (HCD) with a normalized energy of 37% or 40% using 2 separate methods (2 serial injections). MS2 scans were done at a resolution of 50’000 in the Orbitrap cell to resolve 10-plex TMT reporter ions. The m/z of fragmented precursors was then dynamically excluded from selection during 60 s. For analyses with the FAIMSpro interface, data-dependent acquisition methods controlled by Xcalibur 4.2 software (Thermo Fisher Scientific) were used, that alternated between 2 compensation voltages (CV) to acquire 2 survey scans within each cycle. Three methods were used, with the following CV pairs: −40/−60V, −50/−70V, −55/−65V. Following each survey scan at each CV, a “top speed” acquisition was performed to accumulate a maximum of MS2 spectra while keeping a maximum total cycle time of 1.0 s. MS2 scans were acquired with a normalized collision energy of 37%. All other parameters for MS2 spectra acquisition were the same as for the methods without FAIMS separation. Raw MS data analysis Raw MS files obtained with FAIMS ion separation were split into independent files relative to each CV using the software Freestyle 1.6.90.0 (Thermo Fisher Scientific). All tandem MS data were processed by the MaxQuant software (version 1.6.14.0) incorporating the Andromeda search engine . The S . pombe UNIPROT reference proteome (RefProt) sequence database of March 3, 2021 was used (5,141 sequences), supplemented with sequences of common contaminants. Trypsin (cleavage at Lys, Arg) was used as the enzyme definition, allowing 2 missed cleavages. Carbamidomethylation of cysteine was specified as a fixed modification. N-terminal acetylation of protein, oxidation of methionine and phosphorylation on Ser, Thr, and Tyr were specified as variable modifications. All identifications were filtered at 1% FDR at both the peptide and protein levels with default MaxQuant parameters. The isobaric match between runs functions of MaxQuant was used . For comparison of TMT runs, ratios were automatically calculated by MaxQuant as a function of the reference channels . MaxQuant data were further processed with Perseus software (version 1.6.15.0) for the filtering, log2-transformation and normalization of values. Medians of TMT ratios across samples obtained from the total protein measurements were used to correct ratios for phosphoproteomics data to account for differences in total loaded material. Statistical analysis Prior to statistical analysis, to correct for technical variation between replicates, for all phosphosites and protein identified, the median of all time point values collected for a given replicate was subtracted from individual values. In the time-course phosphoproteomic study during sexual differentiation, all 10 values (the 5 values from time points during starvation and 5 values from time points during sexual differentiation) were used to calculate the median. In the time-course phosphoproteomic study during cell–cell fusion, all 5 time point values were used. This method gave good stratification of starvation and mating data in principal component analysis (Figs and ). Sites that were absent in more than 1 biological replicate were discarded: in the time course during sexual differentiation, we only kept sites with no more than 10 missing values out of 30 total possible values (10,828 sites from 1,884 distinct proteins were retained); in the time course during cell–cell fusion, we only kept sites with no more than 5 missing values out of 15 total possible values (11,979 sites from 2,065 distinct proteins). In the raw data, a few sites have exactly the same numeric values across all measurements, because they originate from sites with multiple phosphorylation possibilities. To not bias the statistical multiple correction, redundancy was removed and only one of each value was kept for the statistical analysis ( n = 9,951 for the sexual differentiation, n = 11,286 for the cell–cell fusion). Duplicate sites were added to the results a posteriori. The normalized data was used to fit a linear model with the R Package “limma” to identify sites behaving differently during the time course using the eBayes function with default parameters. The following contrasts were used: For starvation: Ctrl_45–0 = (Ctrl.45min-Ctrl.0min) Ctrl_90–45 = (Ctrl.90min-Ctrl.45min) Ctrl_135–90 = (Ctrl.135min-Ctrl.90min) Ctrl_180–135 = (Ctrl.180min-Ctrl.135min) For mating: Diff_45–0 = (Mating.45min-Mating.0min)-(Ctrl.45min-Ctrl.0min) Diff_90–45 = (Mating.90min-Mating.45min)-(Ctrl.90min-Ctrl.45min) Diff_135–90 = (Mating.135min-Mating.90min)-(Ctrl.135min-Ctrl.90min) Diff_180–135 = (Mating.180min-Mating.135min)-(Ctrl.180min-Ctrl.135min) Subtraction of starvation signal to mating signal was used to eliminate sites that vary with a similar slope in the starvation and mating samples. Phosphosites that vary in both conditions but in a different manner (either same direction but different slopes, or increase versus decrease) will be retained . We note that the strategy may slightly underestimate the true extent of changes during mating, because possible addition of noise may increase variability between replicates and reduce statistical significance in a few cases. A total of 262 significant sites (Benjamini–Hochberg adjusted p -value ≤0.05) were identified as significantly different between the control and mating time course. Sites with the same original measurements as the significant ones that were excluded because they were redundant (multiple possibilities of phosphorylation in the fragment detected by the mass spectrometer) were added back, leading to a total of 272 candidate sites. The parameters for the linear model used in the cell–cell fusion time course were: Diff_11–0 = (Fusion.11min-Fusion.0min) Diff_22–11 = (Fusion.22min-Fusion.11min) Diff_33–22 = (Fusion.33min-Fusion.22min) Diff_44–33 = (Fusion.44min-Fusion.33min) A total of 428 significant sites (Benjamini–Hochberg adjusted p -value ≤0.001) were identified as significantly varying during the fusion time course. We used a lower adjusted p -value than for the sexual differentiation time course because this experiment had 4 biological replicates (3 for the differentiation time course). Sites with the same original measurements as the significant ones that were excluded because they were redundant (multiple possibilities of phosphorylation in the fragment detected by the mass spectrometer) were added back, leading to a total of 440 candidate sites. For proteomics analysis, MS2 spectra from unique peptides and razor peptides (assigned peptides to the protein with the highest Protein IDs among all possible proteins) were considered for protein quantification. Intensities at each time point are the sums of all individual peptide intensities belonging to a particular protein group. Unique and razor peptide intensities are used as default. For analysis of covariation, we first selected all the phosphosites varying significantly during starvation, sexual differentiation, and cell–cell fusion for which the protein had also been identified to change significantly in the proteomic data set. Then, log2 fold-changes values were used to fit a multiple linear regression model with interaction between a categorical variable (Proteomics versus Phosphoproteomics) and a continuous variable (time). The p -values for the interaction term were corrected for multiple testing using a Benjamini–Hochberg correction and phosphosites with p -values higher than 0.05 were considered as co-varying. If there was no significant change on protein levels, phosphosite changes were considered not to be co-varying. Gene ontology enrichment was done using PANTHER (Version 16.0/17.0). For the phosphoproteomic time course during mating, phosphosites significantly changing during starvation or mating (after accounting for changes during starvation) were first divided into 2 groups: increasing or decreasing phosphorylation. The list containing the Uniprot accession number of all the proteins for each group was analyzed with the PANTHER classification system . P -values of gene ontology were corrected for false discovery rates. Due to redundancy between some gene ontology terms, only a few significant terms are displayed. For kinase substrate enrichment, substrates identified in previous phosphoproteomic studies were used, keeping thresholds and criteria applied in these earlier studies. Specifically, Table S1 from was used to define Cdc2 substrate, table EV3 from was used for TORC1 and TORC2 substrates, S15 and S16 Tables from were used for TORC1 and TORC2 substrates, respectively, and from was used for Orb6 substrates. Fisher exact test was used to test for enrichment in kinase substrate during mating and starvation. For simplicity, only phosphosites detected on monophosphorylated phosphopeptides were considered for this analysis. For motif analysis, the sequences of all the phosphosites significantly increasing during mating were aligned from the amino acid at position -5 from the phosphoresidue to the amino acid at position +5. The logo was obtained using the R package “ggseqlogo” with the method parameter set to “prob” (probability). To identify new potential motifs, proline-directed sites and Orb6 substrates were removed, and the motif was plotted with the remaining phosphosites. Two-tailed two-sample t tests were used for statistical analysis on Figs , , and . The effect of the length of the incubation in the dark on the fusion efficiency of fus1 opto cells , and of Caffeine and Auxin treatment on wild-type cells during mating were analyzed by one-way ANOVA. Post hoc Tukey tests were performed and the p -values for comparisons between the condition without the drug and all other conditions are displayed. Strains, mediums, and growth conditions for microscopy Protocols described in were used with minor adaptation. Strains were always pre-cultured in 3 ml of MSL supplemented with appropriate amino acid and nitrogen source overnight at 30°C. Cultures were then diluted in the morning to OD 600nm = 0.2 and in the evening in 20 ml of medium at OD 600nm = 0.025–0.04. Strains used for experiments shown in Figs , were grown in MSL supplemented with 15.1 mM of ammonium sulfate (MSL+N) and all other experiments were done using MSL + 15.1 mM glutamate, unless stated otherwise on the figure panel. The next morning, 4.5 [OD 600nm ] of cells were washed 3 times in 1 ml of MSL-N (centrifugation steps: 1,000 g for 2’) and either spotted onto MSA-N plates (Figs , , ), incubated in liquid MSL-N (Figs , , and ) or incubated in liquid MSL-N before transfer to a MSL-N 2% agarose pad (Figs , , and ). Cells were generally grown at 30°C, except for data presented in , where cells were grown at 25°C. For quantifications in , caffeine and auxin (IAA; Indole-3-acetic acid), or DMSO were added when cultures were shifted to MSL-N. Caffeine (Cayman chemical, 14118) was freshly dissolved at 10 mM concentration in MSL-N liquid medium and diluted to reach indicated concentrations; IAA (Merck, 1.00353.0010) was dissolved in DMSO and added to cultures from a 1,000× concentrated stock. For quantifications in Figs , , heterothallic h+ and h- cells were starved separately for 2 h in MSL-N liquid at 30°C prior to cell mixing. The time point zero corresponds to the time the 2 mating types are resuspended together in MSL-N liquid or on MSA-N plate . Except where specified, cells were imaged after 24 h. For imaging from a pad (Figs , , and ), 100 μl of cells were taken after 3 to 4 h of incubation in liquid MSL-N if cells were grown in MSL + ammonium sulfate , 0 to 1 h of incubation in liquid MSL-N if cells were grown in MSL + glutamate (Figs and ) and centrifuged at 1,000 g for 2′. Approximately 90 μl of supernatant was removed and cells were resuspended in the remaining medium. A drop of 1 μl of the cell slurry was spotted onto MSL-N + 2% agarose pad between a slide and a coverslip, sealed with VALAP (Vaseline:Lanolin:Paraffin, 1:1:1), essentially as described . The MSL + glutamate, MSL + proline, and MSL + phenylalanine media used for experiment in were prepared similarly to the MSL+N medium with the replacement of the 15.1 mM of ammonium sulfate with 15.1 mM of L-Glutamic acid monosodium monohydrate (Fluka, 49621), L-Phenylalanine (Sigma-Aldrich, P2126), or L-Proline (Sigma-Aldrich, P8865). Spore viability assay To assess spore viability , h90 strains were spotted on MSA-N plates and incubated for 2 days at 25°C or 32°C. For , cells were resuspended in H 2 O and spread on YE plates. Tetrads were selected by using a tetrad micro dissector microscope (MSM400, Singer Instruments) to place 1 whole tetrad at each position. Plates were incubated for 5 days at 25°C, before taking images. For , cells were resuspended in H 2 O containing glusulase (from 100× stock, PerkinElmer, NEE154001EA) and incubated over night at 30°C. Spores were counted using a Neubauer counting chamber and 200, 100, and 50 spores were spread on duplicate YE plates. Plates were incubated for 4 days at 25°C, before counting colonies. We noted a systematic error in spore counting where an average of 192 colonies germinated for 100 counted WT spores at 25°C. We assumed this was due to imprecision in the volume of the counting chamber and corrected all values by 1.92. The graph therefore represents the “% corrected spore viability”. Microscopy and quantification Images shown and used for quantifications in Figs , , , were obtained using a DeltaVision platform (Applied Precision) composed of a customized inverted microscope (IX-71; Olympus), a UPlan Apochromat 100×/1.4 NA oil objective, a camera (4.2Mpx PrimeBSI sCMOS camera; Photometrics), and a color combined unit illuminator (Insight SSI 7; Social Science Insights). Images were acquired using softWoRx v4.1.2 software (Applied Precision). For optogenetic experiment, a 550 nm longpass filter (Thorlabs, Newton, New Jersey, USA) was installed on the condenser to allow for cell visualization without photoactivation before imaging. For , cells were imaged every 4′, with 150 ms exposure time in the mcherry channel, 15 ms exposure in the GFP channel, and 80 ms exposure in the DAPI channel. The images were first taken in the mcherry channel (575 nm laser excitation) to record an event before photo-activation. Images taken to optimize conditions for synchronous mating were acquired with a Leica DMI4000B microscope equipped with a standard mercury lamp and a sCMOS Pco edge 5.5 camera (Figs , and ). Images were acquired using micromanager. Bright field images were used to quantify mating efficiency. Fusion efficiency was quantified in the red channel (excitation filter (546/12), dichroic mirror (560), emission filter (605/75)) from a mix of non-fluorescent h- cycΔ5 cells and h+ cycΔ5 cells expressing mcherry under the p act1 promoter. Pairs were considered fused if the mcherry fluorescence were present in both cells of a mating pair. The mating efficiency (% mating) represents the % of cells engaged in mating (cell pairs or fused zygotes) relative to the total number of cells: m a t i n g e f f i c i e n c y = 2 ( m a t i n g p a i r s + z y g o t e s ) i n d i v i d u a l c e l l s + 2 ( m a t i n g p a i r s + z y g o t e s ) X 100 The fusion efficiency (% fusion) represents the % of fused pairs (zygotes) relative to cells engaged in mating (cell pairs and zygotes): f u s i o n e f f i c i e n c y = f u s e d p a i r s m a t i n g p a i r s x 100 In , we also show the % of zygotes relative to the total number of cells: % f u s i o n = 2 ( z y g o t e s ) i n d i v i d u a l c e l l s + 2 ( m a t i n g p a i r s + z y g o t e s ) X 100 Experimental conditions and protein extraction for western blot Pre-cultures in 10 ml of MSL + 15 mM glutamate were incubated overnight at 30°C on a rotative platform (180 rpm), diluted the next morning to [O.D.] 600nm = 0.2, in the evening to [O.D.] 600nm = 0.02–0.04 in 100 ml of MSL + 15 mM glutamate for homothallic strains and 50 ml of MSL + 15 mM glutamate for heterothallic strains. The morning of the third day, 60 [O.D.] 600nm of cells were washed 3 times in 10 ml of MSL-N, resuspended in 40 ml MSL-N, and incubated at 30°C for 6 to 7 h before protein extraction. cycΔ5 cells were only incubated 3 h in MSL-N as sexual differentiation is rapid in this mutant. For experiment with Torin1 and Rapamycin (Figs , and ), strains were incubated in MSL-N for 6 h 15 and 40 ml of cells were then treated with either 40 μl of Torin1 (5 mM stock in DMSO; final concentration 5 μm) (LC Laboratories; T-7887), 40 μl of Rapamycin (300 μg/ml stock in DMSO; final concentration 300 ng/ml) (LC Laboratories; R-5000), or 40 μl of DMSO and incubated for an additional 30 min at 30°C. We found that higher doses of Torin1 were required for loss of Rps6 and Psk1 phosphorylation during mating . The heterothallic sxa2Δ time course shown in in the presence of pheromones was performed as followed: heterothallic h- sxa2Δ cells were grown in 200 ml of MSL + 15.1 mM glutamate overnight at 30°C, diluted the next morning to [O.D.] 600nm = 0.2, diluted to [O.D.] 600nm = 0.02–0.04 in the evening in 1 L of MSL + 15.1 mM glutamate and incubated at 30°C. In the morning of the third day, 60 OD₆₀₀ of the culture were first spun down at 1,000 g for 2′ and resuspended in 40 ml of MSL+15.1 mM glutamate prior to protein extraction. This represents time point 0. An additional 480 [O.D.] 600nm (60 [O.D.] 600nm per time point and condition) of cells were collected, washed 3 times in 10 ml MSL-N, and resuspended in 320 ml MSL-N. The culture was then split in 2: one culture was treated with 160 μl of P-factor (Schafer-N) (from a 1 mg/ml stock in MetOH), the second culture was treated with the same volume of MetOH. The cultures were incubated at 30°C on a rotative platform shaking at 180 rpm; 40 ml of each culture were taken at each indicated time point for protein extraction. The time-course experiments on heterothallic h- sxa2Δ WT and autophagy-deficient mutants (Figs , ) were performed similarly to the experiment shown in , with 2 exceptions: (1) MSL-N was replaced by MSL + 0.5 mM glutamate (for WT and the atg1Δ mutant) or MSL + 0.75 mM glutamate (for WT and the atg5Δ and atg18aΔ mutants); (2) instead of resuspending the cells in MSL-N medium, 60 [O.D] 600nm of cells were plated onto MSA + 0.5 mM glutamate (for the atg1Δ experiment) or MSA + 0.75 mM glutamate (for the atg5Δ and atg18aΔ experiments), supplemented with 1 μg/ml of P-factor or the equivalent volume of MetOH. The choice of MSA over MSL is due to the observation that autophagy-deficient mutants did not respond to pheromone in the liquid media. We also observed that plating 60 [O.D] 600nm of cells on thick agar plates (made with 40 ml of medium) promoted a better response to P-factor in the autophagy-deficient strains compared to using plates made with 20 ml of medium. This may be due to the increased overall amount of glutamate available for shmoo formation. For the CFP-Atg8 cleavage assays (Figs , ), cells were grown in indicated medium (MSL or EMM with indicated nitrogen source) overnight at 30°C, diluted the next morning to [O.D.] 600nm = 0.2, diluted to [O.D.] 600nm = 0.02–0.04 in the evening, and incubated at 30°C. In the morning of the third day, 60 [O.D.] 600nm per sample of each strain were washed 3 times in 10 ml nitrogen-free medium, resuspended in 40 ml of the same medium–N, and incubated at 30°C on a rotative platform set to 180 rpm for indicated time prior to protein extraction. For protein extraction, cultures were quenched by adding 100% w/v ice-cold TCA to a final concentration of 10%, centrifuged at 4°C at 1,000 g for 2′ and the supernatant was removed. The pellet was washed first with 5 ml of −20°C acetone, then with 1 ml of western blot buffer (2% SDS, 5% Glycerol, 50 mM Tris-HCl, 0.2 M EDTA + cOmplete protease inhibitor cocktail tablet (Roche, 11697498001) + PhosSTOP (Roche, 4906837001)). Pellet was resuspended in 400 μl of western blot buffer. Acid-washed glass beads (Sigma, G8772) were added to the samples and cells were lysed using a FastPrep-24 5G bead beating grinder (6 times shaking at 100 V for 30”, 30” break between runs). Samples were then centrifuged at max speed for 5′ at 4°C, and supernatant was recovered as protein fraction. Samples were immediately snap-frozen. Protein samples were incubated in NuPAGE LDS sample buffer and denatured 10′ at 65°C; 1 μl of β-mercaptoethanol was added per 20 μl of sample and samples were incubated 10′. Samples were then either stored at −20°C or used directly for SDS-PAGE. Western blotting A total of about 75 ng of proteins were loaded on 4% to 20% acrylamide gels (GenScript, M00655) and run at 120 V for 90′ in commercial Tris-MOPS-SDS running buffer (GenScript, M00138). Transfer was done on a nitrocellulose membrane (Sigma, GE1060001) in homemade transfer buffer (50 mM tris base, 38 mM Glycine, 1% SDS) supplemented with 20% EtOH at 100 V for 90′. Membranes were blocked 1 h in TBST + 5% milk and incubated overnight in TBST + 5% milk + primary antibodies. Phospho-Akt substrate Rabbit monoclonal antibodies (1:1,000, Cell Signaling Technology, 9614) were used to detect phospho-Rps6. Rps6 levels were detected using anti-rps6 rabbit monoclonal antibodies (1:1,000, Abcam, ab40820). Psk1 phosphorylation was detected using phospho-P70 S6K mouse monoclonal antibodies (1:1,000, Cell Signaling Technology; 9206). Psk1 levels were detected with anti-Psk1 (1:5,000) antibodies . Membranes were washed 3 times in TBST for 15′ per wash and incubated 1 h in HRP-conjugated anti-rabbit (Promega, W4011) or anti-mouse (Promega, W402B) secondary antibodies (1:3,000) in TBST+ 5% Milk. Membranes were washed 3 additional times in TBST for 15′ per wash. Membranes were then covered with a mix of 1 ml of Reagent A and 1 ml of Reagent B from a Pierce ECL western blotting substrate kit (Thermo Fisher, 32106). Excess liquid was removed, and membranes were placed in transparent plastic sheets. Chemiluminescence was revealed using an Amersham imager 680 (GE Healthcare Life Sciences). Uncropped western blots are shown in .
Strains were constructed using standard genetic manipulation of S . pombe either by tetrad dissection or transformation and can be found in . Plasmids generated for this study are listed in . To construct the optogenetic system, a homothallic ( h90 ) strain, auxotroph for uracil containing a myo52-tdtomato allele at the endogenous locus was transformed by PCR-based gene targeting using oligos osm933 and osm1670 on plasmid pSM693 to replace the fus1 ORF with an hphMX cassette and generate ySM4131. For homothallic fus1 opto cells, strain ySM4131 was first transformed with pSM2470 (3′region_fus1-5′region_fus1-fus1N 1-793 -CIBN-3′UTR_fus1-term fus1 -kanMX-pFA6a) linearized with BstZ17I to generate ySM4132, and then with pSM2475 (pUra4 AfeI -pfus1-cry2PHR-fus1C 796-1372 -sfGFP-term nmt1 ) linearized with AfeI to generate ySM4134. ySM4131 was also transformed with pSM2475 to generate a strain containing only the Cry2PHR-Fus1C-sfGFP fragment (ySM4133). To determine the timing of fusion in , a mTagBFP2 cytosolic marker under the map3 P-cell–specific promoter was introduced by transforming ySM4134 with pAV0761 linearized with SpeI to obtain ySM4130. For cycΔ5 fus1 opto , h- and h+ cycΔ5 strains containing the pfus1 :Cry2PHR- fus1C -sfGFP construct at the uracil locus were obtained through multiple crosses between yAV2050, yAV2038 , and ySM4130. These strains were then transformed with pSM3227 (3′region_fus1-5′region_fus1-fus1N 1-793 -CIBN-3′UTR_fus1-term fus1 -hphMX-pFA6a) linearized with BstZ17I to get the h- and h+ cycΔ5 fus1 opto strains ySM4135 and ySM4136, respectively. We used heterothallic cycΔ5 strains to control entry into sexual differentiation, because we observed homothallic strains to mate when reaching higher densities even on nitrogen-rich media. The h+ cycΔ5 fus1 opto strain (ySM4136) was further transformed with pSM3295 (pAde6 PmeI -pact1-mcherry-term tdh1 -patMX) linearized with RsrII to yield ySM4137. Colonies were selected on Edinburg Minimal Medium (EMM) supplemented with 400 μg/ml of glufosinate-ammonium (CatNo. G002P01G, Cluzeau Info Labo, Sainte-Foy-La-Grande, France) as previously described . The Rps6 prototroph mutants were obtained by crossing strain ANO229 to WT lab strains to generate ySM4138, ySM4139, and ySM4145. The heterothallic h- tor2 S1837E strain (ySM4063) is strain TA1397 from . The homothallic h90 tor2 S1837E strain (ySM4140) was obtained by crossing ySM4063 with a homothallic WT strain. The homothallic h90 psk1Δ deletion strains (ySM4141 and ySM4142) were obtained by transforming the homothallic wild-type strain ySM1396 with pSM3571 linearized with AfeI. The homothallic h90 psk1Δ leu1-32 strain (ySM4148) was obtained by transforming ySM4146 with pSM3571 linearized with AfeI. The h- sxa2Δ prototroph strain (ySM4143) was obtained by a marker switch on the strain S16 G10 from the Bioneer library to replace the kanMX6 cassette with an hphMX6 cassette, followed by crosses. The tor2-ts10 used in the study were obtained by crossing strain JV306 (ySM1591; h- tor2ts-10 ade6-M216 leu1-32 ) from to WT homothallic strains, giving ySM4144. Deletion of rps602 and rps601 in h90 leu1-32 strains were obtained by transforming ySM4147 with pSM3297 (3′region_rps602-5′region_rps602-kanMX-pFA6a) and pSM3298 (3′region_rps601-5′region_rps601-hphMX-pFA6a) linearized with AfeI to obtain ySM4149, ySM4150 and ySM4151, ySM4152, respectively. Deletion of tco89 was obtained by transforming ySM1396 with pSM3364 (3′UTR_tco89-5′UTR_tco89-hphMX-pFA6a) linearized with AfeI to generate ySM4157 . The heterothallic h- CFP-atg8 : leu1+ sxa2Δ :: natMX6 deleted for atg1 (ySM4157), atg5 (ySM4174), and atg18a (ySM4175) strains used to test the effect of autophagy on TORC1 reactivation were obtained by transforming DY11900 (ySM4158), DY4021 (ySM4170), and DY4031 (ySM4172) (Sun and colleagues) respectively, with pSM3274 (pFA6a -3′UTR_sxa2-5′UTR_sxa2-natMX) linearized with StuI. A heterothallic h- CFP-atg8 : leu1+ strain was isolated from a cross between DY11900 and a WT strain, auxotroph for leucine and then transformed with pSM3274 linearized with StuI to obtain ySM4156.
For time-course phosphoproteomic experiments, h- cycΔ5 fus1 opto and h+ cycΔ5 fus1 opto cells were grown and starved in a room kept in the dark and manipulation were done with the help of a red LED bulb (Osram). Unless specified, cells were always kept in the dark until they were quenched in 10% W/V TCA during the first step of protein extraction. For the time course during sexual differentiation, precultures of h- cycΔ5 fus1 opto and h+ cycΔ5 fus1 opto were grown to late log-phase in 40 ml, diluted to 400 ml of MSL + 15 mM glutamate, and grown to an OD 600nm = 0.5–0.9 at 30°C in the dark. For both cultures, a volume corresponding to 300 [OD 600nm ] was collected and washed separately 3 times with 20 ml of MSL-N. Centrifugation speed was set to 1,000 g for 2′ during the washes. Each strain was then resuspended in 200 ml of MSL-N and allowed to starve for 2 h at 30°C in the dark, and 100 ml of each strain were then mixed, cells were centrifuged at 1,000 g for 2′, the pellet was then resuspended in 2 ml of fresh MSL-N and 20 μl spots of the cell slurry were pipetted onto MSA-N plates at a density of 3 [OD 600nm ]/spot. We noted that spotting too many cells on the same plate drastically reduced mating efficiency and for that reason, only 5 spots were spotted on the same plate. We used an amount equivalent to 60 [OD 600nm ] per time point. The remaining 100 ml of culture of each heterothallic strain was centrifuged at 1,000 g for 2′ and the pellet resuspended in 1 ml of MSL-N. Both pellets were then spotted as 20 μl spots onto separate MSA-N at a density of 3 [OD 600nm ]/spot. We used an amount equivalent to 30 [OD 600nm ] per strain and time point. Plates were then incubated at 30°C in the dark until sample collection excepts the plates for the time point at 0′ which were directly used for protein extraction. For the phosphoproteomic time course during cell–cell fusion, pre-cultures of h- cycΔ5 fus1 opto and h+ cycΔ5 fus1 opto were grown to late log-phase in 30 ml, diluted to 300 ml of MSL + glutamate and grown to an OD 600nm = 0.5–0.9 at 30°C in the dark. For both cultures, an amount equivalent of 150 [OD 600nm ] of cells was collected and spun down at 1,000 g for 2′. The pellet was washed 3 times in 20 ml of MSL-N and re-suspended in 100 ml of MSL-N and cells were incubated for 2 h at 30°C in the dark. Cells were then mixed together and centrifuged at 1,000 g for 2′. The pellet was then resuspended in 2 ml of fresh MSL-N and spotted as 20 μl spot onto MSA-N plates (5 spot per plate). A total of 30 [OD 600nm ] were used per time point. Plates were incubated at 30°C in dark conditions for 2 h 30′ and then shifted to light condition until sample collection. For the time point at 0 min, the plates were directly used for protein extraction without being exposed to light.
Extract protocol was adapted from . Yeast cells were collected from plates using 5 ml of 10% w/v ice-cold TCA. Cells were then spun down at 1,000 g for 2′ at 4°C and supernatant was discarded. Cells were washed first in 5 ml of acetone (cooled down at −20°C) and then in 1 ml of lysis buffer (50 mM ammonium bicarbonate, 10 mM DTT, 5% SDS). Pellets were resuspended in 400 μl of lysis buffer. Acid-washed glass beads (Sigma; G8772) were added to the samples and cells were lysed using a FastPrep-24 5G bead beating grinder (6 times shaking at 100 V for 30”, 30” break between runs). Samples were then centrifuged at max speed for 5′ at 4°C, and supernatant was recovered as protein fraction. Samples were immediately snap-frozen. For processing, samples were thawed rapidly and heated at 95°C for 10 min with shaking to lyse cells. Protein concentration was determined using the tryptophane fluorescence method . Trypsin digestion of 120 μg of protein material per sample was carried out according to the S-TRAP (Protifi, Farmingdale, New York) method as described . Briefly, after heating at 95°C to denature and reduce disulfides, cysteines were alkylated by reaction with 30 mM (final) chloroacetamide for 1 h in the dark at RT. An aliquot of 12% phosphoric acid was added to lower pH to 3.0, followed by dilution with 4 volumes of S-TRAP loading buffer (100 mM (final concentration) triethylammonium bicarbonate buffer (TEAB) (pH 8.0), in 90% MeOH). The obtained mixture was passed by centrifugation on S-TRAP Mini cartridges, which were then washed 3 times with 600 μl of loading buffer. Digestion was started by adding to the cartridges 20 μg of Trypsin (Promega) in 125 μl of 50 mM TEAB (pH 8) and was carried out for 2 h at 47°C without shaking. Digested peptides were eluted by centrifugation, followed by further elution of the cartridge with 80 μl of 50 mM TEAB, then 80 μl of 0.2 formic acid and 80 μl of 50% acetonitrile (each time at 3,000 × g for 1 min). All eluates were pooled, and samples were dried by evaporation.
Dried peptide mixtures were resuspended in 45 μl of 50 mM Hepes (pH 8.3) and reacted for 1 h at room temperature with 0.4 mg of TMT 10-plex reagents (Thermo Fisher Scientific, 90110) dissolved in 27.5 μl of pure acetonitrile. The reaction was quenched by adding 7 μl of 5% (v/w) hydroxylamine and incubating for 15 min at RT. Individual TMT labeled samples were pooled. The mixture obtained was acidified with TFA, frozen and the volume reduced to 1/10 of the initial by evaporation. After adding 9 volumes of aqueous 0.1% TFA, peptides were desalted on a C18 SepPak 1 cc 50 mg cartridge (Waters, #WAT054955). An aliquot of 10% of the eluate was dried and analyzed by MS as described below to assess labeling efficiency (which globally was found to be higher than 98.5%) and derive ratios of total protein content for normalization.
The remaining of the TMT mix (approx. 1.0 mg) was dried, re-dissolved, and processed for phosphopeptide enrichment by IMAC using the High-Select Fe-NTA Phosphopeptide Enrichment Kit (Thermo Fisher Scientific, A32992) according to instructions from the manufacturer. The eluate from the IMAC cartridge was dried and resuspended in 70 μl of 2% MeCN, 0.05% TFA for LC-MS/MS analysis.
Tryptic peptide mixtures were injected on an Ultimate RSLC 3000 nanoHPLC system (Dionex, Sunnyvale, California, USA) interfaced to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, Bremen, Germany). Peptides were loaded onto a trapping microcolumn Acclaim PepMap100 C18 (20 mm × 100 μm ID, 5 μm, 100 Å, Thermo Scientific) before separation on a reversed-phase custom packed nanocolumn (75 μm ID × 40 cm, 1.8 μm particles, Reprosil Pur, Dr. Maisch). A flowrate of 0.25 μl/min was used with a gradient from 4% to 76% acetonitrile in 0.1% formic acid (total method time: 140 min). Both total proteome and phospho-enriched samples were injected multiple times with different methods and either with a normal nanospray ion source or a Field Asymmetric Waveform Ion Mobility Spectrometry interface (FAIMS pro, Thermo Fisher Scientific). For analyses with the normal nanoLC interface, a data-dependent acquisition method controlled by Xcalibur 4.2 software (Thermo Fisher Scientific) was used that optimized the number of precursors selected (“top speed”) of charge 2+ to 5+ while maintaining a fixed scan cycle of 1.5 s. The precursor isolation window used was 0.7 Th. Peptides were fragmented by higher energy collision dissociation (HCD) with a normalized energy of 37% or 40% using 2 separate methods (2 serial injections). MS2 scans were done at a resolution of 50’000 in the Orbitrap cell to resolve 10-plex TMT reporter ions. The m/z of fragmented precursors was then dynamically excluded from selection during 60 s. For analyses with the FAIMSpro interface, data-dependent acquisition methods controlled by Xcalibur 4.2 software (Thermo Fisher Scientific) were used, that alternated between 2 compensation voltages (CV) to acquire 2 survey scans within each cycle. Three methods were used, with the following CV pairs: −40/−60V, −50/−70V, −55/−65V. Following each survey scan at each CV, a “top speed” acquisition was performed to accumulate a maximum of MS2 spectra while keeping a maximum total cycle time of 1.0 s. MS2 scans were acquired with a normalized collision energy of 37%. All other parameters for MS2 spectra acquisition were the same as for the methods without FAIMS separation.
Raw MS files obtained with FAIMS ion separation were split into independent files relative to each CV using the software Freestyle 1.6.90.0 (Thermo Fisher Scientific). All tandem MS data were processed by the MaxQuant software (version 1.6.14.0) incorporating the Andromeda search engine . The S . pombe UNIPROT reference proteome (RefProt) sequence database of March 3, 2021 was used (5,141 sequences), supplemented with sequences of common contaminants. Trypsin (cleavage at Lys, Arg) was used as the enzyme definition, allowing 2 missed cleavages. Carbamidomethylation of cysteine was specified as a fixed modification. N-terminal acetylation of protein, oxidation of methionine and phosphorylation on Ser, Thr, and Tyr were specified as variable modifications. All identifications were filtered at 1% FDR at both the peptide and protein levels with default MaxQuant parameters. The isobaric match between runs functions of MaxQuant was used . For comparison of TMT runs, ratios were automatically calculated by MaxQuant as a function of the reference channels . MaxQuant data were further processed with Perseus software (version 1.6.15.0) for the filtering, log2-transformation and normalization of values. Medians of TMT ratios across samples obtained from the total protein measurements were used to correct ratios for phosphoproteomics data to account for differences in total loaded material.
Prior to statistical analysis, to correct for technical variation between replicates, for all phosphosites and protein identified, the median of all time point values collected for a given replicate was subtracted from individual values. In the time-course phosphoproteomic study during sexual differentiation, all 10 values (the 5 values from time points during starvation and 5 values from time points during sexual differentiation) were used to calculate the median. In the time-course phosphoproteomic study during cell–cell fusion, all 5 time point values were used. This method gave good stratification of starvation and mating data in principal component analysis (Figs and ). Sites that were absent in more than 1 biological replicate were discarded: in the time course during sexual differentiation, we only kept sites with no more than 10 missing values out of 30 total possible values (10,828 sites from 1,884 distinct proteins were retained); in the time course during cell–cell fusion, we only kept sites with no more than 5 missing values out of 15 total possible values (11,979 sites from 2,065 distinct proteins). In the raw data, a few sites have exactly the same numeric values across all measurements, because they originate from sites with multiple phosphorylation possibilities. To not bias the statistical multiple correction, redundancy was removed and only one of each value was kept for the statistical analysis ( n = 9,951 for the sexual differentiation, n = 11,286 for the cell–cell fusion). Duplicate sites were added to the results a posteriori. The normalized data was used to fit a linear model with the R Package “limma” to identify sites behaving differently during the time course using the eBayes function with default parameters. The following contrasts were used: For starvation: Ctrl_45–0 = (Ctrl.45min-Ctrl.0min) Ctrl_90–45 = (Ctrl.90min-Ctrl.45min) Ctrl_135–90 = (Ctrl.135min-Ctrl.90min) Ctrl_180–135 = (Ctrl.180min-Ctrl.135min) For mating: Diff_45–0 = (Mating.45min-Mating.0min)-(Ctrl.45min-Ctrl.0min) Diff_90–45 = (Mating.90min-Mating.45min)-(Ctrl.90min-Ctrl.45min) Diff_135–90 = (Mating.135min-Mating.90min)-(Ctrl.135min-Ctrl.90min) Diff_180–135 = (Mating.180min-Mating.135min)-(Ctrl.180min-Ctrl.135min) Subtraction of starvation signal to mating signal was used to eliminate sites that vary with a similar slope in the starvation and mating samples. Phosphosites that vary in both conditions but in a different manner (either same direction but different slopes, or increase versus decrease) will be retained . We note that the strategy may slightly underestimate the true extent of changes during mating, because possible addition of noise may increase variability between replicates and reduce statistical significance in a few cases. A total of 262 significant sites (Benjamini–Hochberg adjusted p -value ≤0.05) were identified as significantly different between the control and mating time course. Sites with the same original measurements as the significant ones that were excluded because they were redundant (multiple possibilities of phosphorylation in the fragment detected by the mass spectrometer) were added back, leading to a total of 272 candidate sites. The parameters for the linear model used in the cell–cell fusion time course were: Diff_11–0 = (Fusion.11min-Fusion.0min) Diff_22–11 = (Fusion.22min-Fusion.11min) Diff_33–22 = (Fusion.33min-Fusion.22min) Diff_44–33 = (Fusion.44min-Fusion.33min) A total of 428 significant sites (Benjamini–Hochberg adjusted p -value ≤0.001) were identified as significantly varying during the fusion time course. We used a lower adjusted p -value than for the sexual differentiation time course because this experiment had 4 biological replicates (3 for the differentiation time course). Sites with the same original measurements as the significant ones that were excluded because they were redundant (multiple possibilities of phosphorylation in the fragment detected by the mass spectrometer) were added back, leading to a total of 440 candidate sites. For proteomics analysis, MS2 spectra from unique peptides and razor peptides (assigned peptides to the protein with the highest Protein IDs among all possible proteins) were considered for protein quantification. Intensities at each time point are the sums of all individual peptide intensities belonging to a particular protein group. Unique and razor peptide intensities are used as default. For analysis of covariation, we first selected all the phosphosites varying significantly during starvation, sexual differentiation, and cell–cell fusion for which the protein had also been identified to change significantly in the proteomic data set. Then, log2 fold-changes values were used to fit a multiple linear regression model with interaction between a categorical variable (Proteomics versus Phosphoproteomics) and a continuous variable (time). The p -values for the interaction term were corrected for multiple testing using a Benjamini–Hochberg correction and phosphosites with p -values higher than 0.05 were considered as co-varying. If there was no significant change on protein levels, phosphosite changes were considered not to be co-varying. Gene ontology enrichment was done using PANTHER (Version 16.0/17.0). For the phosphoproteomic time course during mating, phosphosites significantly changing during starvation or mating (after accounting for changes during starvation) were first divided into 2 groups: increasing or decreasing phosphorylation. The list containing the Uniprot accession number of all the proteins for each group was analyzed with the PANTHER classification system . P -values of gene ontology were corrected for false discovery rates. Due to redundancy between some gene ontology terms, only a few significant terms are displayed. For kinase substrate enrichment, substrates identified in previous phosphoproteomic studies were used, keeping thresholds and criteria applied in these earlier studies. Specifically, Table S1 from was used to define Cdc2 substrate, table EV3 from was used for TORC1 and TORC2 substrates, S15 and S16 Tables from were used for TORC1 and TORC2 substrates, respectively, and from was used for Orb6 substrates. Fisher exact test was used to test for enrichment in kinase substrate during mating and starvation. For simplicity, only phosphosites detected on monophosphorylated phosphopeptides were considered for this analysis. For motif analysis, the sequences of all the phosphosites significantly increasing during mating were aligned from the amino acid at position -5 from the phosphoresidue to the amino acid at position +5. The logo was obtained using the R package “ggseqlogo” with the method parameter set to “prob” (probability). To identify new potential motifs, proline-directed sites and Orb6 substrates were removed, and the motif was plotted with the remaining phosphosites. Two-tailed two-sample t tests were used for statistical analysis on Figs , , and . The effect of the length of the incubation in the dark on the fusion efficiency of fus1 opto cells , and of Caffeine and Auxin treatment on wild-type cells during mating were analyzed by one-way ANOVA. Post hoc Tukey tests were performed and the p -values for comparisons between the condition without the drug and all other conditions are displayed.
Protocols described in were used with minor adaptation. Strains were always pre-cultured in 3 ml of MSL supplemented with appropriate amino acid and nitrogen source overnight at 30°C. Cultures were then diluted in the morning to OD 600nm = 0.2 and in the evening in 20 ml of medium at OD 600nm = 0.025–0.04. Strains used for experiments shown in Figs , were grown in MSL supplemented with 15.1 mM of ammonium sulfate (MSL+N) and all other experiments were done using MSL + 15.1 mM glutamate, unless stated otherwise on the figure panel. The next morning, 4.5 [OD 600nm ] of cells were washed 3 times in 1 ml of MSL-N (centrifugation steps: 1,000 g for 2’) and either spotted onto MSA-N plates (Figs , , ), incubated in liquid MSL-N (Figs , , and ) or incubated in liquid MSL-N before transfer to a MSL-N 2% agarose pad (Figs , , and ). Cells were generally grown at 30°C, except for data presented in , where cells were grown at 25°C. For quantifications in , caffeine and auxin (IAA; Indole-3-acetic acid), or DMSO were added when cultures were shifted to MSL-N. Caffeine (Cayman chemical, 14118) was freshly dissolved at 10 mM concentration in MSL-N liquid medium and diluted to reach indicated concentrations; IAA (Merck, 1.00353.0010) was dissolved in DMSO and added to cultures from a 1,000× concentrated stock. For quantifications in Figs , , heterothallic h+ and h- cells were starved separately for 2 h in MSL-N liquid at 30°C prior to cell mixing. The time point zero corresponds to the time the 2 mating types are resuspended together in MSL-N liquid or on MSA-N plate . Except where specified, cells were imaged after 24 h. For imaging from a pad (Figs , , and ), 100 μl of cells were taken after 3 to 4 h of incubation in liquid MSL-N if cells were grown in MSL + ammonium sulfate , 0 to 1 h of incubation in liquid MSL-N if cells were grown in MSL + glutamate (Figs and ) and centrifuged at 1,000 g for 2′. Approximately 90 μl of supernatant was removed and cells were resuspended in the remaining medium. A drop of 1 μl of the cell slurry was spotted onto MSL-N + 2% agarose pad between a slide and a coverslip, sealed with VALAP (Vaseline:Lanolin:Paraffin, 1:1:1), essentially as described . The MSL + glutamate, MSL + proline, and MSL + phenylalanine media used for experiment in were prepared similarly to the MSL+N medium with the replacement of the 15.1 mM of ammonium sulfate with 15.1 mM of L-Glutamic acid monosodium monohydrate (Fluka, 49621), L-Phenylalanine (Sigma-Aldrich, P2126), or L-Proline (Sigma-Aldrich, P8865).
To assess spore viability , h90 strains were spotted on MSA-N plates and incubated for 2 days at 25°C or 32°C. For , cells were resuspended in H 2 O and spread on YE plates. Tetrads were selected by using a tetrad micro dissector microscope (MSM400, Singer Instruments) to place 1 whole tetrad at each position. Plates were incubated for 5 days at 25°C, before taking images. For , cells were resuspended in H 2 O containing glusulase (from 100× stock, PerkinElmer, NEE154001EA) and incubated over night at 30°C. Spores were counted using a Neubauer counting chamber and 200, 100, and 50 spores were spread on duplicate YE plates. Plates were incubated for 4 days at 25°C, before counting colonies. We noted a systematic error in spore counting where an average of 192 colonies germinated for 100 counted WT spores at 25°C. We assumed this was due to imprecision in the volume of the counting chamber and corrected all values by 1.92. The graph therefore represents the “% corrected spore viability”.
Images shown and used for quantifications in Figs , , , were obtained using a DeltaVision platform (Applied Precision) composed of a customized inverted microscope (IX-71; Olympus), a UPlan Apochromat 100×/1.4 NA oil objective, a camera (4.2Mpx PrimeBSI sCMOS camera; Photometrics), and a color combined unit illuminator (Insight SSI 7; Social Science Insights). Images were acquired using softWoRx v4.1.2 software (Applied Precision). For optogenetic experiment, a 550 nm longpass filter (Thorlabs, Newton, New Jersey, USA) was installed on the condenser to allow for cell visualization without photoactivation before imaging. For , cells were imaged every 4′, with 150 ms exposure time in the mcherry channel, 15 ms exposure in the GFP channel, and 80 ms exposure in the DAPI channel. The images were first taken in the mcherry channel (575 nm laser excitation) to record an event before photo-activation. Images taken to optimize conditions for synchronous mating were acquired with a Leica DMI4000B microscope equipped with a standard mercury lamp and a sCMOS Pco edge 5.5 camera (Figs , and ). Images were acquired using micromanager. Bright field images were used to quantify mating efficiency. Fusion efficiency was quantified in the red channel (excitation filter (546/12), dichroic mirror (560), emission filter (605/75)) from a mix of non-fluorescent h- cycΔ5 cells and h+ cycΔ5 cells expressing mcherry under the p act1 promoter. Pairs were considered fused if the mcherry fluorescence were present in both cells of a mating pair. The mating efficiency (% mating) represents the % of cells engaged in mating (cell pairs or fused zygotes) relative to the total number of cells: m a t i n g e f f i c i e n c y = 2 ( m a t i n g p a i r s + z y g o t e s ) i n d i v i d u a l c e l l s + 2 ( m a t i n g p a i r s + z y g o t e s ) X 100 The fusion efficiency (% fusion) represents the % of fused pairs (zygotes) relative to cells engaged in mating (cell pairs and zygotes): f u s i o n e f f i c i e n c y = f u s e d p a i r s m a t i n g p a i r s x 100 In , we also show the % of zygotes relative to the total number of cells: % f u s i o n = 2 ( z y g o t e s ) i n d i v i d u a l c e l l s + 2 ( m a t i n g p a i r s + z y g o t e s ) X 100
Pre-cultures in 10 ml of MSL + 15 mM glutamate were incubated overnight at 30°C on a rotative platform (180 rpm), diluted the next morning to [O.D.] 600nm = 0.2, in the evening to [O.D.] 600nm = 0.02–0.04 in 100 ml of MSL + 15 mM glutamate for homothallic strains and 50 ml of MSL + 15 mM glutamate for heterothallic strains. The morning of the third day, 60 [O.D.] 600nm of cells were washed 3 times in 10 ml of MSL-N, resuspended in 40 ml MSL-N, and incubated at 30°C for 6 to 7 h before protein extraction. cycΔ5 cells were only incubated 3 h in MSL-N as sexual differentiation is rapid in this mutant. For experiment with Torin1 and Rapamycin (Figs , and ), strains were incubated in MSL-N for 6 h 15 and 40 ml of cells were then treated with either 40 μl of Torin1 (5 mM stock in DMSO; final concentration 5 μm) (LC Laboratories; T-7887), 40 μl of Rapamycin (300 μg/ml stock in DMSO; final concentration 300 ng/ml) (LC Laboratories; R-5000), or 40 μl of DMSO and incubated for an additional 30 min at 30°C. We found that higher doses of Torin1 were required for loss of Rps6 and Psk1 phosphorylation during mating . The heterothallic sxa2Δ time course shown in in the presence of pheromones was performed as followed: heterothallic h- sxa2Δ cells were grown in 200 ml of MSL + 15.1 mM glutamate overnight at 30°C, diluted the next morning to [O.D.] 600nm = 0.2, diluted to [O.D.] 600nm = 0.02–0.04 in the evening in 1 L of MSL + 15.1 mM glutamate and incubated at 30°C. In the morning of the third day, 60 OD₆₀₀ of the culture were first spun down at 1,000 g for 2′ and resuspended in 40 ml of MSL+15.1 mM glutamate prior to protein extraction. This represents time point 0. An additional 480 [O.D.] 600nm (60 [O.D.] 600nm per time point and condition) of cells were collected, washed 3 times in 10 ml MSL-N, and resuspended in 320 ml MSL-N. The culture was then split in 2: one culture was treated with 160 μl of P-factor (Schafer-N) (from a 1 mg/ml stock in MetOH), the second culture was treated with the same volume of MetOH. The cultures were incubated at 30°C on a rotative platform shaking at 180 rpm; 40 ml of each culture were taken at each indicated time point for protein extraction. The time-course experiments on heterothallic h- sxa2Δ WT and autophagy-deficient mutants (Figs , ) were performed similarly to the experiment shown in , with 2 exceptions: (1) MSL-N was replaced by MSL + 0.5 mM glutamate (for WT and the atg1Δ mutant) or MSL + 0.75 mM glutamate (for WT and the atg5Δ and atg18aΔ mutants); (2) instead of resuspending the cells in MSL-N medium, 60 [O.D] 600nm of cells were plated onto MSA + 0.5 mM glutamate (for the atg1Δ experiment) or MSA + 0.75 mM glutamate (for the atg5Δ and atg18aΔ experiments), supplemented with 1 μg/ml of P-factor or the equivalent volume of MetOH. The choice of MSA over MSL is due to the observation that autophagy-deficient mutants did not respond to pheromone in the liquid media. We also observed that plating 60 [O.D] 600nm of cells on thick agar plates (made with 40 ml of medium) promoted a better response to P-factor in the autophagy-deficient strains compared to using plates made with 20 ml of medium. This may be due to the increased overall amount of glutamate available for shmoo formation. For the CFP-Atg8 cleavage assays (Figs , ), cells were grown in indicated medium (MSL or EMM with indicated nitrogen source) overnight at 30°C, diluted the next morning to [O.D.] 600nm = 0.2, diluted to [O.D.] 600nm = 0.02–0.04 in the evening, and incubated at 30°C. In the morning of the third day, 60 [O.D.] 600nm per sample of each strain were washed 3 times in 10 ml nitrogen-free medium, resuspended in 40 ml of the same medium–N, and incubated at 30°C on a rotative platform set to 180 rpm for indicated time prior to protein extraction. For protein extraction, cultures were quenched by adding 100% w/v ice-cold TCA to a final concentration of 10%, centrifuged at 4°C at 1,000 g for 2′ and the supernatant was removed. The pellet was washed first with 5 ml of −20°C acetone, then with 1 ml of western blot buffer (2% SDS, 5% Glycerol, 50 mM Tris-HCl, 0.2 M EDTA + cOmplete protease inhibitor cocktail tablet (Roche, 11697498001) + PhosSTOP (Roche, 4906837001)). Pellet was resuspended in 400 μl of western blot buffer. Acid-washed glass beads (Sigma, G8772) were added to the samples and cells were lysed using a FastPrep-24 5G bead beating grinder (6 times shaking at 100 V for 30”, 30” break between runs). Samples were then centrifuged at max speed for 5′ at 4°C, and supernatant was recovered as protein fraction. Samples were immediately snap-frozen. Protein samples were incubated in NuPAGE LDS sample buffer and denatured 10′ at 65°C; 1 μl of β-mercaptoethanol was added per 20 μl of sample and samples were incubated 10′. Samples were then either stored at −20°C or used directly for SDS-PAGE.
A total of about 75 ng of proteins were loaded on 4% to 20% acrylamide gels (GenScript, M00655) and run at 120 V for 90′ in commercial Tris-MOPS-SDS running buffer (GenScript, M00138). Transfer was done on a nitrocellulose membrane (Sigma, GE1060001) in homemade transfer buffer (50 mM tris base, 38 mM Glycine, 1% SDS) supplemented with 20% EtOH at 100 V for 90′. Membranes were blocked 1 h in TBST + 5% milk and incubated overnight in TBST + 5% milk + primary antibodies. Phospho-Akt substrate Rabbit monoclonal antibodies (1:1,000, Cell Signaling Technology, 9614) were used to detect phospho-Rps6. Rps6 levels were detected using anti-rps6 rabbit monoclonal antibodies (1:1,000, Abcam, ab40820). Psk1 phosphorylation was detected using phospho-P70 S6K mouse monoclonal antibodies (1:1,000, Cell Signaling Technology; 9206). Psk1 levels were detected with anti-Psk1 (1:5,000) antibodies . Membranes were washed 3 times in TBST for 15′ per wash and incubated 1 h in HRP-conjugated anti-rabbit (Promega, W4011) or anti-mouse (Promega, W402B) secondary antibodies (1:3,000) in TBST+ 5% Milk. Membranes were washed 3 additional times in TBST for 15′ per wash. Membranes were then covered with a mix of 1 ml of Reagent A and 1 ml of Reagent B from a Pierce ECL western blotting substrate kit (Thermo Fisher, 32106). Excess liquid was removed, and membranes were placed in transparent plastic sheets. Chemiluminescence was revealed using an Amersham imager 680 (GE Healthcare Life Sciences). Uncropped western blots are shown in .
S1 Fig cycΔ5 strains mate efficiently and Fus1N fusogenic activity is negligible at short time frames. ( A, B ) Mating efficiency (% of cell pairs among all cells; A) and fusion efficiency (% of paired cells that fused; B) of WT or cycΔ5 cells after 24 h on plates lacking nitrogen. The cells were pre-grown on ammonium or glutamate-containing medium. ( C ) Fusion efficiency (% of paired cells that fused) of h90 fus1 opto cells starved and kept in the dark for 24 h (same data as on ) and of h+ x h- cycΔ5 fus1 opto cells individually starved for 2 h and then mixed and kept in the dark for a further 2 h 30. The latter h+ x h- cycΔ5 fus1 opto conditions are identical to those used in the (phospho)proteomic time course, corresponding to time 150 min in . As these cells do not fuse, this confirms that all phosphorylation changes observed in the mating time course occur before gamete fusion. T test p -value is indicated. The underlying data can be found in . (PDF) S2 Fig Reproducibility of the data sets and expected phosphorylation sites. ( A ) Example profiles of phosphosites in the 3 biological replicates of the starvation/mating time course and the 4 biological replicates of the cell–cell fusion time course. Starvation (cells plated on distinct plates), mating (h+ and h- partners plated together), and starvation-corrected mating values (mating—starvation) are shown in the first 3 columns. The cell–cell fusion time course is shown in the last column. The first 3 examples (Spk1, Rps601, and Rec10) show examples of sites that do not vary significantly upon starvation but change during mating and cell–cell fusion. The next example (Ght1) shows a site that varies during starvation similarly whether cells are starved separately or in presence of their mating partner, resulting in absence of change in the starvation-corrected (mating–starvation) values, indicating no specific change during mating. The last 2 examples show sites that vary significantly both during starvation and mating, but with distinct dynamics. In the Mei2 example, variation occurs in opposite direction. In the Fur4 example, the slope of the change is distinct. Red line indicates average; black lines indicate individual replicates. Pale graphs indicate absence of significant change. The underlying data can be found in and Tables. ( B ) Short list of expected and identified phosphorylation increases during mating. References are to the expected sites. (PDF) S3 Fig Analysis of proteomic changes during N-starvation. Changes in the proteome of heterothallic cells in a time course of nitrogen starvation starting at t0 = plating of cells on MSL-N plates 2 h after transfer to liquid MSL-N. ( A ) Heatmap of the significant changes in the levels of 847 proteins during nitrogen starvation, showing 2 major clusters of proteins whose level increase (1) or decrease (2). The underlying data can be found in . ( B ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins whose level increases during nitrogen starvation. ( C ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins whose level decreases during nitrogen starvation. Significance levels were assessed by Fisher’s exact test and corrected for false discovery rate. (PDF) S4 Fig Analysis of phosphoproteomic changes during N-starvation. Changes in the phosphoproteome of heterothallic cells in a time course of nitrogen starvation starting at t0 = plating of cells on MSL-N plates 2 h after transfer to liquid MSL-N. ( A ) Heatmap of the significant changes in 3,100 phosphosites during nitrogen starvation, showing 2 major clusters of sites whose phosphorylation increase (1) or decrease (2). Corresponding protein dynamics is shown on the right with missing values in gray. Regression analysis estimated that about 78% of phosphosite level changes are independent of the changes in protein levels (see ). The underlying data can be found in . ( B ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins containing one or several sites showing phosphorylation increase during nitrogen starvation. ( C ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins containing one or several sites showing phosphorylation decrease during nitrogen starvation. Significance levels were assessed by Fisher’s exact test and corrected for false discovery rate. (PDF) S5 Fig Analysis of proteomic and phosphoproteomic changes during sexual differentiation. Changes in the proteome and phosphoproteome of mating cells in a time course starting at t0 = cell mixing on MSL-N plates in the dark. The cells were pre-grown separately in liquid MSL-N for 2 h. All Data are corrected for changes happening during starvation in absence of a mating partner. ( A ) Heatmap of the significant changes in the levels of 54 proteins during mating, showing 2 major clusters of proteins whose level increase (1) or decrease (2). The underlying data can be found in . ( B ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components in proteins whose level increases during mating. No significant enrichment was found for the few proteins whose levels decrease. ( C ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins containing one or several sites showing phosphorylation increase during nitrogen starvation. ( D ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins containing one or several sites showing phosphorylation decrease during nitrogen starvation. Significance levels were assessed by Fisher’s exact test and corrected for false discovery rate. ( E ) Analysis of the protein sequence surrounding serine and threonine residues phosphorylated during mating. The height of the amino acid one-letter code represents the frequency. The logo on the left was made using all 192 increasing phosphosites. For the one on the left, all proline-directed sites and Orb6 substrates were removed, leaving 122 phosphosites. (PDF) S6 Fig Analysis of proteomic and phosphoproteomic changes during cell–cell fusion. Changes in the proteome and phosphoproteome in a time course starting at t0 = light exposure. h- and h+ cycΔ5 fus1 opto cells were pre-grown separately in liquid MSL-N for 2 h, mixed on plates for 150 in the dark before illumination. ( A ) Heatmap of the significant changes in the levels of 41 proteins during mating, showing 2 major clusters of proteins whose level increase (1) or decrease (2). Note that cluster 1 can be separated in 2 subclusters with levels increase at 11 or 33 min. The underlying data can be found in . ( B ) Significant fold enrichment in GO annotations for cellular components of proteins whose level decreases during mating. No significant enrichment was found for the few proteins whose levels increase. ( C ) Significant fold enrichment in GO annotations for biological processes and cellular components of proteins containing one or several sites showing phosphorylation increase during cell–cell fusion. ( D ) Significant fold enrichment in GO annotations for biological processes, molecular functions, and cellular components of proteins containing one or several sites showing phosphorylation decrease during nitrogen starvation. Significance levels were assessed by Fisher’s exact test and corrected for false discovery rate. (PDF) S7 Fig Additional western blots and controls. ( A ) Phospho-Rps6 and phospho-Psk1 levels in WT heterothallic ( h- ) and homothallic ( h90 ) cells starved for 6 h 15 and then treated with Torin1 (5 or 25 μm) or DMSO (0) for 30 min. ( B ) CFP-Atg8 cleavage in WT and atg1Δ mutants after 4 h and 8 h of nitrogen starvation in MSL. The cells were pre-grown in MSL + glutamate (top) or MSL + ammonium (bottom). ( C ) Time course of phospho-Psk1 in h- sxa2Δ atg18aΔ cells transferred to MSL + 0.75 mg/ml glutamate in the presence of P-factor (1 μg/ml) or MetOH at T = 0 min. CFP-Atg8 cleavage is shown in the bottom blot. ( D ) Time course of CFP-Atg8 cleavage in h- sxa2Δ cells, which are otherwise WT (top), atg1Δ (middle), or atg5Δ (bottom), transferred to MSL + 0.75 mg/ml (or 0.5 mg/ml in case of atg1Δ ) glutamate in the presence of P-factor (1 μg/ml) or MetOH at T = 0 min. These are the same extracts as probed for phospho-Psk1 in . Uncropped western blots available in . (PDF) S1 Table (Phospho)proteomics data for all proteins and sites with significant change. All proteins and phosphosites whose levels are significantly changing during starvation, mating, or fusion time course. The data are presented as 6 distinct tabs in.xls format, presenting proteomics and phosphoproteomics over the starvation time course, proteomics and phosphoproteomics over the mating time course, and proteomics and phosphoproteomics over the fusion time course. Log2-transformed, median-corrected values are shown for all, except for the mating time course where the starvation values have been subtracted. Lighter shading indicates data from individual replicates; darker shading shows average values. An additional first tab presents a detailed legend. (XLSX) S2 Table Complete (phospho)proteomics data. All proteins and phosphosites identified in during starvation, mating, or fusion time course. Data presented in is also included. The data are presented as 4 distinct tabs in.xls format, presenting proteomics and phosphoproteomics over the starvation and mating time course, and proteomics and phosphoproteomics over the fusion time course. Log2-transformed, median-corrected values are shown for all. Lighter shading indicates data from individual replicates; darker shading shows average values. In the starvation-mating time course, different color shadings highlight the starvation data, the uncorrected mating data and the starvation-subtracted mating data. An additional first tab presents a detailed legend. (XLSX) S3 Table Strains used in this study. (PDF) S4 Table Plasmids used in this study. (PDF) S1 Raw Images Uncropped western blots. For each figure panel, the panel is shown for reference. Each of the full-size, uncropped western blots is shown by itself and overlaid with the ladder. The relevant protein is indicated with an arrowhead. Asterisks mark lanes not included in the final figures. (PDF) S1 Data Data set. (XLSX)
|
Reverse impact of chordae tendineae structural changes on its biomechanical properties as a part of pathogenesis in canine myxomatous mitral valve disease | 6ee133e6-a222-4466-9255-b301670f0412 | 11854349 | Cardiovascular System[mh] | Myxomatous mitral valve disease (MMVD) is a major problem in small-breed dogs and is representing a significant cause of morbidity and mortality. It accounts for approximately 75% of all heart diseases in canine cardiac patients . The mitral valve apparatus consists of a mitral annulus, septal and parietal leaflets, chordae tendineae (CT) and papillary muscles . CT are columnar structures that are connected to the mitral and tricuspid valves and modulate the transmission of strain to valve leaflets. They play a key role in the proper functioning of the valvular apparatus, contributing to valve integrity and facilitating coordinated leaflet movement during the cardiac cycle . The pathogenesis of degenerative mitral valve disease involves multiple factors, so the role of CT in the progression of MMVD is becoming increasingly important in veterinary cardiology . Myxomatous mitral valve disease is often accompanied by thickened chordae, which become elongated over time . This disease severely affects the mechanical properties of the CT of the mitral valve thereby resulting in regurgitation . Structural remodelling of the valve affected by MMVD is associated with characteristic histopathological changes in its structures including valve leaflets and CT. Changes of valve leaflets include expansion of the extracellular matrix by glycosaminoglycans and proteoglycans, changes in the interstitial cells of the valves, and weakening or loss of the collagen-laden fibrous layer. Additionally, fibrosis within the CT occurs . Histologically, normal CT exhibit a complex composition, comprising collagen, elastin, and glycosaminoglycans, imparting resilience and elasticity to these tendon-like structures . The arrangement of collagen fibers confers tensile strength, while elastin fibers contribute to flexibility, allowing the CT to withstand mechanical stress and deformation throughout the cardiac cycle. Replacement of normal tissue with pathological, proteoglycan-rich tissue results in loss of the mechanical properties of the CT . CT can be divided into various categories, according to: 1) the origin from the papillary muscle, 2) the attachment to the mitral valve leaflet or part of valvular apparatus, 3) the distribution on the valvular leaflet, 4) the branching pattern, and 5) gross structure . True CT attach to the apical part of the papillary muscle and one can distinguish commissural, cusp or cleft chordae among them . The cusp CT can be further divided into first-, second- and third-order CT [ – ]. First-order CT, also called marginal or primary CT, attach to the free edges of the mitral valve leaflets. Their role is to protect the valve from regurgitation and leaflet prolapse. The second-order CT, also called secondary, basal, ventricular, or strut CT, are thicker and more flexible and are responsible for valvular-ventricular interactions. Third-order CT connect the septal wall and second-order chordae and are rare in dogs [ – ]. During the course of myxomatous mitral valve disease, the phenomenon of CT rupture can occur . This is a condition that can lead to acute cardiogenic pulmonary oedema and is potentially life-threatening . There is much research on the biomechanical properties of CT in humans and other species, such as pigs and sheep, but there is still a lack of information on the influence of the degeneration of CT on their mechanical properties in dogs [ – ]. The mechanical changes in the leaflets, while significant, are not as drastic as the mechanical changes in CT . However, this phenomenon has not been studied in dogs. Understanding the morphological, biomechanical and molecular aspects of these structures provides the basis for elucidating the pathophysiological mechanisms of the disease. In the present study, the biomechanical parameters of the CT of healthy dogs and those affected by the degenerative process were determined. Understanding these properties may help in assessing the risk of CT rupture phenomena in the future.
Sample collection Mitral valves ( n = 54) were prepared from the hearts of dogs subjected for a necropsy examination at the Department of Pathology, including dogs diagnosed and treated at the Cardiology Unit. The material was collected between January 2021 and March 2023. The dogs were euthanised or died naturally. Due to the prevalence of the disease in old dogs, the inclusion criterium was animal’s age > 7 years old. The exclusion criteria were the presence of a congenital heart defect, a disease other than degenerative mitral valve disease (e.g., dilated cardiomyopathy) or the presence of a neoplastic lesion within the heart. Each dog underwent a post mortem examination with the collection of the heart in a routine manner; the organ was then dissected and the mitral valve with CTs and their origin on papillary muscles was collected from each dog. Only true cusp chordae were collected and were subsequently divided to first- and second-order CTs (Fig. ). The CTs were obtained from both the anterior and posterior leaflets of the mitral valve. Biomechanical analysis Between four and six 1st order (marginal) and 2nd order (ventricular) CTs were taken from each mitral valve. Single CTs were excluded due to technical issues during mechanical examination (slipping out of the grips). As a result, at least three 1st and 2nd order CTs were examined in each of the included dogs. In order to guarantee tissue hydration status and biomechanical properties, after dissection samples were put into fluid for perfusion and washing of organs referred for transplantation (Biolasol solution, Biochefa, Sosnowiec, Poland) and then kept at reduced temperature (4 °C) until examination, as described previously . During the biomechanical study, uni-axial static tensile testing of the CT was performed on a Zwick/Roell EPZ 005 testing machine (ZwickRoell GmbH & Co. KG, Ulm, Germany) (Fig. A). Before the test, individual CT were excised from the mitral valve, leaving a fragment of the papillary muscle and valve leaflet so that the CT could be fixed in the clamps. Prior to the test, all CTs were parameterized by taking digital images, allowing the structure of the specimen, which could be damaged when using manual measuring tools such as a calliper, to remain unchanged. To provide precise calibration, values were obtained for the CT dimensions, i.e., cross-sectional area and length (Fig. B). To reduce the risk of slipping of the tissue material from the grip, the ends were additionally inserted in a high-gradation grid and then placed in flat, knurled grips to obtain the most realistic results (Fig. C). The specimens were stretched at a velocity of 5 mm/min . The measurements were performed using Zeiss AxioVisionRel software version 4.8. Force–displacement charts were obtained, which were then converted into charts allowing the determination of biomechanical parameters: tensile strength and destructive strain (Fig. ). For each dog the mean tensile strength and destructive strain were calculated. During the dissection of the heart, one first-order CT for each valve leaflet was collected and immersed in 7% buffered formalin solution for fixation . After 24-h of fixation the specimens were embedded in paraffin blocks and cut into 6 μm sections in a standard manner. After being placed on microscope slides, the specimens were stained with haematoxylin–eosin (HE) and picro Sirius red stain and underwent a microscopic examination. The examination was performed using a Leica DM500 microscope coupled with a Leica ICC50W camera (Leica Microsystems, KAWA.SKA, Poland). The specimens were evaluated at 100 × magnification. Due to the lack of clear criteria of CT degeneration severity in the literature, a 4-grade scale was developed (Table , Fig. ). To develop the scale, all samples were pre-evaluated simultaneously by the main author and a supervisor with > 10 years of experience in cardiac pathology (IJZ) to identify the main features of CT degeneration. After developing the scale, the degeneration grade of each specimen was assessed separately by two authors (JG and IJZ) in a blinded-manner basing on HE and picro Sirius red stains. The specimens where grade differed between investigators, were re-evaluated by both authors simultaneously and a final grade was assessed. Figure A and B show grade “0” CT in dogs as evident by normally arranged fiber structure, stained by haematoxylin–eosin and Sirius red, respectively. Figure C and D show grade “1” CT in dogs with corrugation of fiber structure and spaces between fibers with diameter smaller than fiber thickness in less than 50% of CT, stained by haematoxylin–eosin and Sirius red, respectively. Figure E and F show grade “2” CT in dogs as evident by corrugation of fiber structure and spaces between fibers with diameter smaller than fiber thickness in more than 50% or spaces between fibers with diameter larger than fiber thickness in less than 30% of CT, stained by haematoxylin–eosin and Sirius red, respectively. Figure G and H show grade “3” CT in dogs, characterized by corrugation of structure with spaces larger than fiber thickness in more than 30% of CT, chaotic arrangement of fibers and branching structure, stained by haematoxylin–eosin and Sirius red, respectively. Magnification 100x. Statistical analysis The results of CT biomechanical strength were analysed in relation to the results of the histopathological grading of CT degeneration. Statistical analysis was performed using the STATISTICA 13.3 software, TIBCO Software Inc; University of Zielona Góra licence agreement 2022–2024 and Past4 software (Natural History Museum; University of Oslo). Data normality was tested using Shapiro–Wilk test, all data are presented as the median (interquartile range); due to non-normal distribution of data, the comparisons between groups were performed using either Mann–Whitney U analysis (for comparisons between sexes and between 1st and 2nd order CT) or Kruskal–Wallis analysis with Dunn post-hoc test (for comparison between various grade CT); correlation between variables was tested using Spearman’s correlation analysis; statistical significance was set for p ≤ 0.05.
Mitral valves ( n = 54) were prepared from the hearts of dogs subjected for a necropsy examination at the Department of Pathology, including dogs diagnosed and treated at the Cardiology Unit. The material was collected between January 2021 and March 2023. The dogs were euthanised or died naturally. Due to the prevalence of the disease in old dogs, the inclusion criterium was animal’s age > 7 years old. The exclusion criteria were the presence of a congenital heart defect, a disease other than degenerative mitral valve disease (e.g., dilated cardiomyopathy) or the presence of a neoplastic lesion within the heart. Each dog underwent a post mortem examination with the collection of the heart in a routine manner; the organ was then dissected and the mitral valve with CTs and their origin on papillary muscles was collected from each dog. Only true cusp chordae were collected and were subsequently divided to first- and second-order CTs (Fig. ). The CTs were obtained from both the anterior and posterior leaflets of the mitral valve.
Between four and six 1st order (marginal) and 2nd order (ventricular) CTs were taken from each mitral valve. Single CTs were excluded due to technical issues during mechanical examination (slipping out of the grips). As a result, at least three 1st and 2nd order CTs were examined in each of the included dogs. In order to guarantee tissue hydration status and biomechanical properties, after dissection samples were put into fluid for perfusion and washing of organs referred for transplantation (Biolasol solution, Biochefa, Sosnowiec, Poland) and then kept at reduced temperature (4 °C) until examination, as described previously . During the biomechanical study, uni-axial static tensile testing of the CT was performed on a Zwick/Roell EPZ 005 testing machine (ZwickRoell GmbH & Co. KG, Ulm, Germany) (Fig. A). Before the test, individual CT were excised from the mitral valve, leaving a fragment of the papillary muscle and valve leaflet so that the CT could be fixed in the clamps. Prior to the test, all CTs were parameterized by taking digital images, allowing the structure of the specimen, which could be damaged when using manual measuring tools such as a calliper, to remain unchanged. To provide precise calibration, values were obtained for the CT dimensions, i.e., cross-sectional area and length (Fig. B). To reduce the risk of slipping of the tissue material from the grip, the ends were additionally inserted in a high-gradation grid and then placed in flat, knurled grips to obtain the most realistic results (Fig. C). The specimens were stretched at a velocity of 5 mm/min . The measurements were performed using Zeiss AxioVisionRel software version 4.8. Force–displacement charts were obtained, which were then converted into charts allowing the determination of biomechanical parameters: tensile strength and destructive strain (Fig. ). For each dog the mean tensile strength and destructive strain were calculated. During the dissection of the heart, one first-order CT for each valve leaflet was collected and immersed in 7% buffered formalin solution for fixation . After 24-h of fixation the specimens were embedded in paraffin blocks and cut into 6 μm sections in a standard manner. After being placed on microscope slides, the specimens were stained with haematoxylin–eosin (HE) and picro Sirius red stain and underwent a microscopic examination. The examination was performed using a Leica DM500 microscope coupled with a Leica ICC50W camera (Leica Microsystems, KAWA.SKA, Poland). The specimens were evaluated at 100 × magnification. Due to the lack of clear criteria of CT degeneration severity in the literature, a 4-grade scale was developed (Table , Fig. ). To develop the scale, all samples were pre-evaluated simultaneously by the main author and a supervisor with > 10 years of experience in cardiac pathology (IJZ) to identify the main features of CT degeneration. After developing the scale, the degeneration grade of each specimen was assessed separately by two authors (JG and IJZ) in a blinded-manner basing on HE and picro Sirius red stains. The specimens where grade differed between investigators, were re-evaluated by both authors simultaneously and a final grade was assessed. Figure A and B show grade “0” CT in dogs as evident by normally arranged fiber structure, stained by haematoxylin–eosin and Sirius red, respectively. Figure C and D show grade “1” CT in dogs with corrugation of fiber structure and spaces between fibers with diameter smaller than fiber thickness in less than 50% of CT, stained by haematoxylin–eosin and Sirius red, respectively. Figure E and F show grade “2” CT in dogs as evident by corrugation of fiber structure and spaces between fibers with diameter smaller than fiber thickness in more than 50% or spaces between fibers with diameter larger than fiber thickness in less than 30% of CT, stained by haematoxylin–eosin and Sirius red, respectively. Figure G and H show grade “3” CT in dogs, characterized by corrugation of structure with spaces larger than fiber thickness in more than 30% of CT, chaotic arrangement of fibers and branching structure, stained by haematoxylin–eosin and Sirius red, respectively. Magnification 100x.
The results of CT biomechanical strength were analysed in relation to the results of the histopathological grading of CT degeneration. Statistical analysis was performed using the STATISTICA 13.3 software, TIBCO Software Inc; University of Zielona Góra licence agreement 2022–2024 and Past4 software (Natural History Museum; University of Oslo). Data normality was tested using Shapiro–Wilk test, all data are presented as the median (interquartile range); due to non-normal distribution of data, the comparisons between groups were performed using either Mann–Whitney U analysis (for comparisons between sexes and between 1st and 2nd order CT) or Kruskal–Wallis analysis with Dunn post-hoc test (for comparison between various grade CT); correlation between variables was tested using Spearman’s correlation analysis; statistical significance was set for p ≤ 0.05.
Animals and samples There were 54 dogs in the group, including mixed-breed dogs ( n = 28), Labradors, American Staffordshire ( n = 4 each), German Shepherds ( n = 3), Cocker Spaniels, French Bulldogs and Yorkshire Terriers ( n = 2 each), as well as Shih Tzu, a Polish Greyhound, a Weimaraner, a Maltese, a Husky, a Collie Sheepdog, a Dachshund, a Medium Schnauzer and a Cavalier King Charles Spaniel ( n = 1 each). Twenty-eight dogs were males (51,85%), and twenty-six were females (48,15%). The median body weight was 15 kg (range 2.5 to 38 kg) and the mean age was 11.3 years (range 8 to 15 years old). Fifty-four hearts were initially collected. In six hearts the histopathological examination did not allow a proper analysis (the CT size was too small to perform a reliable grading) and further correlation with biomechanical tests, and in six additional hearts the biomechanical examination failed (CTs were too short to be caught in a grip). Finally, specimens obtained from 42 hearts were analysed in regard to biomechanical-structural correlation. Histopathological grading Based on histopathological classification, 10 dogs were classified as 0 grade lesions, 18 dogs as 1 grade, 11 dogs as 2 grade and three dogs as 3 grade. Biomechanical analysis The tensile strengths and destructive strain value showed no difference regarding animal’s sex ( p > 0.05; Mann–Whitney U analysis) and no correlation with animal’s age or weight ( p > 0.05; Spearman correlation analysis). The mean tensile strengths together with mean destructive strain for each grade of CT degeneration are summarized in Table . Based on Table , it can be noted that the significant differences in tensile strength occur between CTs assessed as healthy (grade 0) and CTs with mild (grade 1; p = 0.001) and moderate (grade 2; p = 0.02) changes. Moreover, a negative correlation was noted between the CT grade and mean tensile strength ( p = 0.03; r = −0.33), suggesting that with the progression of the degenerative process, the elasticity of CT decreases, causing them to rupture more quickly. This is also evident in the lower values of the mean destructive strain at the incidence of CT degeneration. Considering the location of the CT in the valvular apparatus, a significant difference in the strength of the CT was noted between 1st and 2nd order CTs (Fig. ). The mean tensile strength of the 2nd order CT (ventricular) was up to twice that of the 1st order CT (marginal). The comparison of mean values of tensile strength and destructive strain between 1st and 2nd order CT. Mann–Whitney U analysis; p- values are presented on the figure; significant difference is marked with asterisk; results presented as median, interquartile range (box) and range (whiskers). Due to the differences noted between 1st- and 2nd-order CTs, further analysis was performed separately for each category. The results of tensile strength and destructive strain for 1st and 2nd order CTs are presented in Tables and . To further evaluate the relationship between the CT degeneration and biomechanical results, the correlation analysis was performed. The 1st order CT showed no correlation between the histopathological changes grade and tensile strength ( p > 0.05; Spearman correlation analysis) but showed a negative correlation between the histopathological grade and mean destructive strain ( p = 0.01; r = −0.43; Spearman correlation analysis). At the same time the 2nd order CT showed a negative correlation between the histopathological grade and mean tensile strength values ( p = 0.008, r = −0.47; Spearman correlation analysis) but showed no correlation between the histopathological grade and destructive strain values ( p > 0.05; Spearman correlation analysis).
There were 54 dogs in the group, including mixed-breed dogs ( n = 28), Labradors, American Staffordshire ( n = 4 each), German Shepherds ( n = 3), Cocker Spaniels, French Bulldogs and Yorkshire Terriers ( n = 2 each), as well as Shih Tzu, a Polish Greyhound, a Weimaraner, a Maltese, a Husky, a Collie Sheepdog, a Dachshund, a Medium Schnauzer and a Cavalier King Charles Spaniel ( n = 1 each). Twenty-eight dogs were males (51,85%), and twenty-six were females (48,15%). The median body weight was 15 kg (range 2.5 to 38 kg) and the mean age was 11.3 years (range 8 to 15 years old). Fifty-four hearts were initially collected. In six hearts the histopathological examination did not allow a proper analysis (the CT size was too small to perform a reliable grading) and further correlation with biomechanical tests, and in six additional hearts the biomechanical examination failed (CTs were too short to be caught in a grip). Finally, specimens obtained from 42 hearts were analysed in regard to biomechanical-structural correlation.
Based on histopathological classification, 10 dogs were classified as 0 grade lesions, 18 dogs as 1 grade, 11 dogs as 2 grade and three dogs as 3 grade.
The tensile strengths and destructive strain value showed no difference regarding animal’s sex ( p > 0.05; Mann–Whitney U analysis) and no correlation with animal’s age or weight ( p > 0.05; Spearman correlation analysis). The mean tensile strengths together with mean destructive strain for each grade of CT degeneration are summarized in Table . Based on Table , it can be noted that the significant differences in tensile strength occur between CTs assessed as healthy (grade 0) and CTs with mild (grade 1; p = 0.001) and moderate (grade 2; p = 0.02) changes. Moreover, a negative correlation was noted between the CT grade and mean tensile strength ( p = 0.03; r = −0.33), suggesting that with the progression of the degenerative process, the elasticity of CT decreases, causing them to rupture more quickly. This is also evident in the lower values of the mean destructive strain at the incidence of CT degeneration. Considering the location of the CT in the valvular apparatus, a significant difference in the strength of the CT was noted between 1st and 2nd order CTs (Fig. ). The mean tensile strength of the 2nd order CT (ventricular) was up to twice that of the 1st order CT (marginal). The comparison of mean values of tensile strength and destructive strain between 1st and 2nd order CT. Mann–Whitney U analysis; p- values are presented on the figure; significant difference is marked with asterisk; results presented as median, interquartile range (box) and range (whiskers). Due to the differences noted between 1st- and 2nd-order CTs, further analysis was performed separately for each category. The results of tensile strength and destructive strain for 1st and 2nd order CTs are presented in Tables and . To further evaluate the relationship between the CT degeneration and biomechanical results, the correlation analysis was performed. The 1st order CT showed no correlation between the histopathological changes grade and tensile strength ( p > 0.05; Spearman correlation analysis) but showed a negative correlation between the histopathological grade and mean destructive strain ( p = 0.01; r = −0.43; Spearman correlation analysis). At the same time the 2nd order CT showed a negative correlation between the histopathological grade and mean tensile strength values ( p = 0.008, r = −0.47; Spearman correlation analysis) but showed no correlation between the histopathological grade and destructive strain values ( p > 0.05; Spearman correlation analysis).
Incidence and progression of degenerative process in CT significantly impacts their biomechanical properties. The opening and closing motion of the valve is impaired, and valve leaflets may prolapse as a result of varying degrees of degenerative changes caused by MMVD . As the literature indicates, not all affected valves appear to prolapse. CT rupture is an expected outcome in patients having advanced MMVD and almost always results in valve prolapse and acute severe mitral regurgitation, which is a serious medical complication . Echocardiographic studies often reveal that some of the remodelled valve tissue stretches through the annulus into the left atrium during contraction . What still remains inconclusive is how to best assess the intensity and clinical significance of valve pathology and the degree of valve regurgitation, and how these factors vary by breed, stage of disease and aging process . In our study, a relationship was noted between the degree of degeneration and the mechanical properties of the CT in the mitral valve apparatus. With the progression of CT degeneration, uneven fibres structure, presence of free spaces between the fibres and undulating fibre structure, the value of tensile strength and deformability of the tissue decreased, therefore causing the CT to rupture more quickly. The biomechanical studies on explanted tissues requires proper handling of the specimens. Tissue samples are most often secured in saline solution, simulated body fluid (SBF) or phosphate buffered saline (PBS) , which can falsify the results and reduce the mechanical properties of soft tissues as a result of the degradation of these . Type of fluid (Biolasol) used in our study guarantees the nutrition and maintenance of optimal tissue hydration, without changing the mechanical properties from the time the CT were taken until the time of the test, which allows for real results . Looking at the data in Tables and , one can observe a tendency for the strength of CT to decrease with the progression of structural changes with differences in tensile strength (for both 1st and 2nd order CT) observed between 0 and 1 grade, and 0 and 2 grade changes, and with differences in destructive strain (for 1st order CT) observed between 0 and 2, and 1 and 2 grade changes. In the course of histopathological degeneration, the tissue resistance is affected (in grade 1). Less significant differences between severely changed CT (grade 3) and other groups were most probably caused by a small number of CT classified as grade 3 ( n = 3). The corrugation of the fibers in CT affected by grade 1, 2 and 3 of the disease indicates a loss of elastic properties and permanent deformation of the CT, which undergoes mechanical wear as a result of the biomechanical work of opening and closing the valve. It should be noted that the strength characteristics of the CT indicate their hyperelastic model, which is noticeable in many soft tissues. At the same time, in biomechanical terms, the CT acts only in the direction of tension and relaxation, which, in the context of the corrugation of collagen fibers, indicates the loss of elastic properties, through permanent deformation and lack of adequate tension of the CT during movement. This has the effect of lowering the tension of the valve leaflets and on valve leakage. Furthermore, statistically significant differences were observed in mean values of tensile strength between 1st and 2nd order CT, with higher reported values for 2nd order CT. This is a characteristic related to the function of 2nd order CTs, as they are responsible for valvular-ventricular interactions and are larger, thicker and more elastic . There are limited reports of biomechanical studies on the chordae tendineae of human, ovine and porcine models. One of the earliest studies of the strength of human-derived CT showed that the CT were less extensible with increasing strain rate, that larger cross-sectional area corresponded to lower tensile strength, and that the CT fractured at 21.4% strain . In our case, it was noted that the larger the cross-sectional area of the CT, the higher the value of the force needed to rupture it, and we obtained a similar order of magnitude for the rupture strain, specifically for grade 2 and 3 histopathological changes. In the case of a comparison between the CT of the human and ovine models, it was noted that the human ones were significantly stiffer than the corresponding ovine CT and showed significantly lower rupture strains . No comparison with canine CTs has been performed so far. Another study on a human model shows that CT from myxomatous valves showed significantly lower elastic modulus (40.4 ± 10.2 versus 132 ± 15 MPa) and fracture stress (6.0 ± 0.6 MPa versus 25.7 ± 1.8 MPa) than healthy CT, but tensile and fracture strain were similar . However, in our study we did not determine Young’s modulus. In another study on human chordae tendineae, a quasi-static tensile test was performed to analyze how calcified marginal CT affect the biomechanics of the valve. It was observed that diseased CT were three to seven times more compliant than normal CT . In our study the tensile strength of the CTs decreased with degeneration progression, but one cannot compare the results to the abovementioned study due to different character of pathological changes within the CT. In the porcine model, the marginal CT fractured at 68% less load (in the context of tensile stress) and 28% less strain than the basal CT, which is a similar feature we observed in our research, where the corresponding values were 61% and 22%, respectively. It was also noted that the CT from the posterior leaflet fractured at 43% less load and 22% less strain than the CT from the anterior leaflet . The comparison of CTs from each leaflet was out of the scope of the current study. Another team examined mitral and tricuspid valve CT in pigs using a static tensile testing approach with strain markers and observed that: (I) strut CT were stiffer than marginal and basal ones; (II) basal CTs had greater extensibility than marginal CT which is analogous to our results; (III) mitral valve CT were stiffer than their tricuspid valve counterparts; and CT connecting to the tricuspid valve septal leaflets were more extensible than CT connecting to the other two tricuspid leaflets . In the current research we only examined mitral valve CTs. Based on the above-mentioned studies, it can be seen that the results of the analyses between animal species are different. Analyses on the biomechanical model of the canine CTs are not found in the literature, although this species shows naturally occurring disease. Studies on the biomechanics of the mitral valve in dogs, including consideration of strength analysis of the CT, were only developed in preliminary studies by our team . The main limitation of the study is that due to tissue damage during biomechanical examination, simultaneous histopathological and biomechanical evaluation of the same CT is not possible. This limitation is significant because of the large variation in the expression of disease within a single individual. As indicated in the literature , within the valve, the leaflets are not uniformly affected by MMVD and there are local degenerative changes, which may also influence the uneven degeneration and strength loss among of both basal and marginal CT in a single individual. In addition, echocardiographic information was not available for most dogs, which did not allow a comparison of MMVD stage according to ACVIM classification , which could have contributed a lot of clinical information. Therefore, the analysis was based on the relationship between CT degeneration and biomechanical performance without analysing the impact of MMVD on the results. The next limitations were combined to technical properties of biochemical examination: some of the CTs were too short to be caught in the grips, while others slipped out from the grips during the examination and had to be removed from the group.
The results of biomechanical and histopathological studies confirm our thesis about correlation: the more advanced the structural changes, the weaker the biomechanical properties as expressed in terms of strength, deformation and strain. First- and second-order CTs show various resistance to stretch, suggesting their different role in preserving valvular function. The degenerative process, affecting mainly the mitral valve, can also be observed in CTs leading to changes in their biomechanical properties. Future studies should be directed towards correlating the histopathological and reverse biomechanical changes of the CT to the MMVD progression, clinical symptoms and echocardiographic picture. This, in turn, may be useful in assessing the risk of a CT rupture episode.
|
The effects of microglia on tauopathy progression can be quantified using Nexopathy | 611940d0-f61e-4b87-9abf-5a0d98aeb4be | 9729195 | Pathology[mh] | Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that exhibits extracellular amyloid- [12pt]{minimal} $$$$ β plaques and intracellular aggregates of hyperphosphorylated tau as its main neuropathological hallmarks. The regional location and density of tau tangles correlate more with the degree of clinical symptoms than those of amyloid- [12pt]{minimal} $$$$ β plaques , . The stronger association between tau and regional atrophy, hypometabolism, and cognitive decline in AD , makes it imperative to shift the long-standing focus from amyloid- [12pt]{minimal} $$$$ β to tau and to further investigate modulators of tau accumulation and spread. Although tau mainly spreads via the brain’s anatomic connectome, the complete characterization of tauopathies should be based on the notion of Molecular Nexopathy , which states that divergent spatiotemporal patterns of different tauopathies may result from biochemical differences between tau species and the neural cells in the brain that interact with tau as well as the connectome-based mechanism of transsynaptic transmission . To unpack this, even though the key mediatory role in tau spread is played by the connectome, there are many other interactions between cells, circuits, proteopathic species, and the general neurological milieu in the neurodegenerative brain that remains outside of the connectome-centric view of these diseases. Evidence for this notion has become stronger over years suggesting the existence of important condition-altering differences in pathological tau species between degenerative conditions. For instance, tau displays differential 3R and 4R isoforms and hyperphosphorylation sites in different tauopathies. Gross tangle structure also differs across diseases, with AD tau forming neurofibrillary tangles and pre-helical formations, Pick bodies in Pick’s disease, and progressive supranuclear palsy, corticobasal degeneration, and frontotemporal dementia presenting more abundant tangles . Indeed, pathological tau transits with a different mechanism in a disorder-dependent manner , . Tau propagates in a stereotypical pattern in AD, starting from the entorhinal cortex to the hippocampus followed by the neocortex as described by Braak staging . Experiments involving the injection of synthetic tau fibrils in the mouse brain have demonstrated that tau can propagate from the injection site to distant but connected regions . Tauopathy progression is prion-like - templated and self-propagating , with pre-existing aggregates acting as seeds to induce further misfolding. Since tau primarily migrates along the white matter tracts of the brain, it is reasonable to model its spatiotemporal dynamics in a first-order approximation as a diffusive process between connected brain regions. Prior work by our group has quantified pathology progression in humans and animals using such an approach via a graph-theory based system of differential equations called the Network Diffusion Model (NDM) – . However, although the NDM framework considers the connectome graph as the central (sole) mediator of pathology spread along a networked representation of the brain, it fails to consider the impact of molecular nexopathic species on tau spread. Here, we attempt to capture this nexopathy paradigm using mathematical in silico modeling, which we refer to as Nexopathy in silico or “Nex is” . Nex is models the following biological processes: trans-neuronal transmission along the connectome (modeled using the NDM); prion-like amplification of tau over time; and the nexopathic effects of extra-connectomic molecular players that modulate these processes. We formulate from first principles the process by which neuronal and other cells, genes, and other modulators might interact with misfolded tau during its accumulation and transmission along neural connections. Although these principles are qualitatively well-established from preclinical in vivo and in vitro studies, the unique benefit of formulating them in a mathematical model is that the model then becomes a means for testing interesting hypotheses and becomes a testbed for novel hypothesis generation. The overall design of Nex is and its typical workflow is depicted in Fig. a–d. Important nexopathic effects are wielded by microglia , the brain’s resident immune cells, and tau dynamics are well known to be influenced by the action of microglia often found in the vicinity of pathological aggregates. Recent studies of postmortem AD brains have revealed that microglia co-localize with tau and activate themselves after internalizing tau aggregates . Tau-containing neurons can also activate adjacent microglia causing them to form junctions with other neurons as conduits to transfer tau and propagate pathology . Using mouse models of rapid tau propagation, Asai et al. and others have demonstrated that microglia markedly contribute to the clearance, recruitment, and spread of pathology. Microglial activation and neuroinflammation accompany tau tangle formation and importantly, parallel AD clinical symptoms . Specifically, activated microglia express the triggering receptor expressed on myeloid cells 2 (TREM2). Trem2 is a well-known AD risk-gene , with loss-of-function mutations (such as R47H and R62H) increasing the susceptibility to late-onset AD , and being related to the rate of cognitive decline . In vivo, Trem2 over-expression in pathological conditions is related to the recruitment of microglia to pathological aggregates. Although, studies assessing the relationship between Trem2 deletion and tau pathology and neurodegeneration have yielded variable results , empirical clinical and preclinical data suggest that the modulation of tau propagation by the action of microglia , , is germane to the study of progressive tauopathies. Owing to the microglial contributions, quantitative estimates based solely on connectivity and regional pathology levels, e.g. those from our prior NDM , cannot recapitulate all aspects of pathology spread patterns in AD. Therefore here, we apply our novel Nex is framework to interrogate the modulatory effect of microglia on tauopathy progression, by introducing microglial signatures within the Nex is model (Fig. e). We explore microglial influence under two potential modes, whereby microglia alter: (1) the rate of intra-regional tau accumulation and (2) the rate of inter-regional tau spread. These effects are not mutually exclusive but have different biological and mathematical implications. We first apply the non-microglia-influenced Nex is model of tau aggregation and spread (i.e. Nex is :global) characterized solely by tau accumulation/clearance rate in addition to its diffusion along the connectome. This model is further extended into a microglia-based Nex is :microglia model to incorporate their effect on tauopathy progression. We apply these Nex is frameworks to data from mouse models of tauopathy using a 426-region mouse connectome . By utilizing microglial homeostasis genes (first principal component of P2ry12 , Cx3cr1 , Fcrls , Olfml3 , Hexb , Siglech , Sox5 , and Jun ) and an activated microglial signature ( Trem2 ) gene , we show that the inclusion of microglia in the model quantitatively improves its predictive power, with Trem2 inclusion largely outperforming that of microglial homeostasis genes. We developed an optimization algorithm to infer model parameters. Results confirm and quantify reduction of tau burden (accumulation) by Trem2 and increase in tau transmissibility to regions ipsi- and contralateral to the hippocampal seed area. Given the strain-specific nature of tau pathology, we verified that the Nex is : Trem2 model parameters for datasets DS6 and DS9 also fit the pathology from the corresponding 1:10 dilution studies from Kaufman et al. . We then explored Apoe , another putative modulator of tau propagation, and found that it does not contribute significantly to tau progression in the current datasets used. Nex is :microglia was also applied to some other microglial genes ( Sorl1 , Cd33 , Clu , Sirpa , and Aif1 ) regulating phagocytosis and neuroinflammation in AD progression , . Overall, the Nex is :microglia framework provides a more complete model of tau propagation. This study brings together empirical mouse tauopathy models , the mouse mesoscale connectome , and gene expression atlas to achieve the first quantitative study of microglial involvement in tauopathy propagation developed and validated at the whole-brain level. Association between spatiotemporal tau pathology and microglial spatial distribution Correlations of tau pathology spread in sample dataset DS9 (Fig. a) with microglial homeostasis genes (Fig. b, c) as well as with Trem2 (Fig. b, d) were significant, although modest. These data form a solid basis on which to undertake a more detailed and model-based exploration of the effects of microglia on tauopathy progression. Workflow of model implementation and validation To implement and validate the Nex is models we chose whole-brain analysis approach with both the structural connectome and regional tau defined at the same parcellation scheme consisting of 426 regions (213 each hemisphere), each of which serves as a node of the connectome graph. Fig. a–c schematizes the network representation of the mouse brain for which we extract the connectivity matrix C . This connectome is the central (sole) mediator of tau spread in NDM (Fig. d). For Nex is , the analytical framework is augmented to account for the influence of tau accumulation/clearance rate (Nex is :global) followed by the influence of microglia (Nex is :microglia) (Fig. e). Since homeostasis genes and Trem2 have distinct functional roles and display distinct baseline distributions (Fig. b), we chose to model them separately with two models following the Nex is :microglia framework: Nex is :homeostasis and Nex is : Trem2 . We validated the models at the whole-brain level on mouse tauopathy datasets DS4, DS6, and DS9 from Kaufman et al. . Instead of pre-selecting certain brain regions and their tau levels as is usually done in such studies, we used all regional tau data, wherever available, in a completely unbiased fashion. Model parameterization depends on tau strain The Nex is :microglia models include the following parameters: global accumulation/clearance rate of pathology, [12pt]{minimal} $$$$ α ; global diffusivity rate constant denoting the diffusion of pathology between regions, [12pt]{minimal} $$$$ β ; accumulation/clearance of pathology as modulated by microglia, p ; and pathology diffusion rate as modulated by microglia, b . Nex is :global is the non-microglial model with [12pt]{minimal} $$p = b = 0$$ p = b = 0 . The tau strains of the selected datasets were derived from mutually distinct sources: DS4 mice were injected with human AD-derived tau isolates, whereas DS6 and DS9 mice had tau derived from a P301S mouse and recombinant tau fibrils, respectively . These pre-characterized datasets were made available by the Kaufman et al. study where they demonstrated distinct tau bio-activity marked by varying capacities to form and spread tau aggregates (seeding and spreading potential), widespread induction of pathology, and capacity to induce microgliosis. Figure illustrates the distinct spatiotemporal patterns of tau pathology in DS4, DS6, and DS9 captured in silico . We implemented a robust strategy for parameter estimation using a 100 iteration bootstrapping approach on 80% resampled data, allowing us to acquire 95% confidence intervals around the parameter means and to assess reproducibility. Table summarizes the inferred parameters and quality-of-fit metrics. The results of parameter estimation using the bootstrapping approach (Fig. ) include the mean and 95% confidence intervals for the model fit, global, and microglial parameters for Nex is :global and Nex is : Trem2 of DS4, DS6, and DS9. Non-zero p and b values demonstrate microglial influence on tau accumulation and spread above and beyond Nex is :global. When Trem2 microglial effects were included, DS6 and DS9 indicated enhanced clearance ( p < 0) but increased inter-regional spread of tau ( b > 0). Homeostatic microglial effects, however, led to lower inter-regional spread in DS6 and DS9 ( b < 0 for DS6 and DS9; Table ). DS4 demonstrated increased tau accumulation and spread due to Trem2 compared to homeostatic microglia (Table ). Inclusion of microglia improves the predictive power of Nex is The NDM performed weakly compared to Nex is models for all datasets (Table ). Nex is :global performed weakly for DS4 ( [12pt]{minimal} $$R^2$$ R 2 = 0.17) but well for DS6 and DS9 ( [12pt]{minimal} $$R^2$$ R 2 = 0.46 and 0.44, respectively), successfully recapitulating the spatiotemporal profile of empirical tau pathology. This suggests that passive network diffusion is insufficient and additional aggregation processes enhance the model’s predictive power. The inclusion of microglial gene expression demonstrated higher predictive power ( [12pt]{minimal} $$R^2$$ R 2 ) compared to Nex is :global. The improvement was statistically significant (per Fisher’s-R-to-Z transform p value) for DS6 and DS9. Trem2 imparted significantly higher predictive power compared to microglial homeostasis genes. It is important to note that across all datasets, AIC and BIC decreased upon the inclusion of microglia, indicating that the better fits achieved by Nex is :microglia were not simply due to the introduction of additional model parameters. Instead, these results reflect reduction in uncertainty in the global parameters, [12pt]{minimal} $$$$ α and [12pt]{minimal} $$$$ β , as indicated by their narrower 95% confidence intervals. The eight chosen microglial homeostasis genes are used in the model as their first principal component. The accumulation and spread parameters for these Nex is :microglia models were either negative, non-significant, or much lower than those obtained for Nex is : Trem2 (Table ). Further, we consider the datasets DS6_110 and DS9_110, which use the same injected tau strains as DS6 and DS9, respectively, but at 1:10 dilution . We expected that the tau-strain-specific parameters [12pt]{minimal} $$$$ α , [12pt]{minimal} $$$$ β , b , and p should be consistent between these experiments, but the seed rescale parameter, [12pt]{minimal} $$$$ γ , should be lower. We found that by applying Nex is : Trem2 on DS6_110 and DS9_110 after fixing the former parameters to their respective values for DS6 and DS9 and fitting only for [12pt]{minimal} $$$$ γ , the model provides strong fits to the data with lower [12pt]{minimal} $$$$ γ values, as anticipated (Table ). Effect of Trem2 on tau burden and spread Since the fitting results highlight Trem2 as the more significant modulator compared to homeostatic genes, the remainder of our results will focus on it. Figure shows the results for DS9 in detail; they are very similar for DS6 (Fig. ). For dataset DS4 (Fig. ), while the Nex is models were significantly better than the passive NDM, improvement from Nex is :global to Nex is :microglia was not statistically significant (Table ). Both Nex is -modeled distributions differ significantly from the NDM-modeled distribution. While both Nex is -modeled distributions look similar visually, there are clear differences between them that begin to become apparent at later time-points, especially in striatal and entorhinal areas. While both produce excellent recapitulation of empirical tau burden at all time points, the Nex is : Trem2 model produces by far the better fits (Fig. f, g). Model fits in all cases improve over time as may be expected from a ramifying pathological process. The Nex is : Trem2 fits incurred an increase in the microglia-dependent spread parameter, b (Table ); hence Trem2 appears to allow more pathology to efflux from the hippocampal seed regions at early time-points (Fig. d). We next assessed, visually, the effect that microglia are imparting on the spread of tau within the model. Flow-maps depicting major pathology spreading directions were constructed using Brainframe, our in-house Matlab-based function. Figure highlights differences in tau burden and spread as modeled by Nex is :global and Nex is : Trem2 for DS9. A prevalence of magenta arrows (Fig. b) indicates that Trem2 allowed more widespread propagation of tau (also reflected in the higher b values in Table for DS9). Although modeled tau propagated from the hippocampal seed area to connected regions, Trem2 appeared to mediate more spread of tau to the striatal, entorhinal, pallidal, and contralateral hippocampal regions. Similarly, a prevalence of green spheres in the striatal, pallidal, and hippocampal areas indicates that these regions evidence higher tau burden due to Trem2 than would otherwise be expected. Tau burden was unaffected by the presence of Trem2 in the parasubiculum (red sphere at all time points). Changes in Trem2 expression and the subsequent response of tau Given the effect of Trem2 on model performance and its quantified effects on tau transmissibility ( b ), we hypothesized that there would be a dose-dependent effect of Trem2 on tau, where higher Trem2 expression would lead to lower tau levels. In silico simulation of changes in baseline Trem2 levels at the global brain level, followed by application of the Nex is : Trem2 model, indicated a unidirectional trend in tau. Higher Trem2 expression resulted in a reduction of tau in five sample regions of the mouse brain (Fig. a). Specifically, there appeared to be a preponderance of an early effect of Trem2 in the hippocampus and the entorhinal cortex. Nex is :microglia models applied to additional candidate genes implicated in tauopathies We investigated Apoe , a strongly implicated AD-risk gene – expressed both in microglia and astrocytes . Nex is : Apoe did not outperform Nex is :global in any of the datasets (Table , Fig. ). Since Trem2 and Apoe also work in concert in AD , we conducted a combined Nex is :microglia analysis including them both, but results for all three datasets matched those produced by Nex is : Trem2 alone (Table ). We then applied Nex is :microglia to some candidate genes involved in phagocytosis and inflammation ( Sorl1 , Cd33 , Clu , Sirpa , and Aif1 ). The accumulation ( p ) and spread ( b ) parameters obtained for these genes were either negative or insignificant, agreeing with their role in containing and eliminating pathological entities. Correlations of tau pathology spread in sample dataset DS9 (Fig. a) with microglial homeostasis genes (Fig. b, c) as well as with Trem2 (Fig. b, d) were significant, although modest. These data form a solid basis on which to undertake a more detailed and model-based exploration of the effects of microglia on tauopathy progression. To implement and validate the Nex is models we chose whole-brain analysis approach with both the structural connectome and regional tau defined at the same parcellation scheme consisting of 426 regions (213 each hemisphere), each of which serves as a node of the connectome graph. Fig. a–c schematizes the network representation of the mouse brain for which we extract the connectivity matrix C . This connectome is the central (sole) mediator of tau spread in NDM (Fig. d). For Nex is , the analytical framework is augmented to account for the influence of tau accumulation/clearance rate (Nex is :global) followed by the influence of microglia (Nex is :microglia) (Fig. e). Since homeostasis genes and Trem2 have distinct functional roles and display distinct baseline distributions (Fig. b), we chose to model them separately with two models following the Nex is :microglia framework: Nex is :homeostasis and Nex is : Trem2 . We validated the models at the whole-brain level on mouse tauopathy datasets DS4, DS6, and DS9 from Kaufman et al. . Instead of pre-selecting certain brain regions and their tau levels as is usually done in such studies, we used all regional tau data, wherever available, in a completely unbiased fashion. The Nex is :microglia models include the following parameters: global accumulation/clearance rate of pathology, [12pt]{minimal} $$$$ α ; global diffusivity rate constant denoting the diffusion of pathology between regions, [12pt]{minimal} $$$$ β ; accumulation/clearance of pathology as modulated by microglia, p ; and pathology diffusion rate as modulated by microglia, b . Nex is :global is the non-microglial model with [12pt]{minimal} $$p = b = 0$$ p = b = 0 . The tau strains of the selected datasets were derived from mutually distinct sources: DS4 mice were injected with human AD-derived tau isolates, whereas DS6 and DS9 mice had tau derived from a P301S mouse and recombinant tau fibrils, respectively . These pre-characterized datasets were made available by the Kaufman et al. study where they demonstrated distinct tau bio-activity marked by varying capacities to form and spread tau aggregates (seeding and spreading potential), widespread induction of pathology, and capacity to induce microgliosis. Figure illustrates the distinct spatiotemporal patterns of tau pathology in DS4, DS6, and DS9 captured in silico . We implemented a robust strategy for parameter estimation using a 100 iteration bootstrapping approach on 80% resampled data, allowing us to acquire 95% confidence intervals around the parameter means and to assess reproducibility. Table summarizes the inferred parameters and quality-of-fit metrics. The results of parameter estimation using the bootstrapping approach (Fig. ) include the mean and 95% confidence intervals for the model fit, global, and microglial parameters for Nex is :global and Nex is : Trem2 of DS4, DS6, and DS9. Non-zero p and b values demonstrate microglial influence on tau accumulation and spread above and beyond Nex is :global. When Trem2 microglial effects were included, DS6 and DS9 indicated enhanced clearance ( p < 0) but increased inter-regional spread of tau ( b > 0). Homeostatic microglial effects, however, led to lower inter-regional spread in DS6 and DS9 ( b < 0 for DS6 and DS9; Table ). DS4 demonstrated increased tau accumulation and spread due to Trem2 compared to homeostatic microglia (Table ). is The NDM performed weakly compared to Nex is models for all datasets (Table ). Nex is :global performed weakly for DS4 ( [12pt]{minimal} $$R^2$$ R 2 = 0.17) but well for DS6 and DS9 ( [12pt]{minimal} $$R^2$$ R 2 = 0.46 and 0.44, respectively), successfully recapitulating the spatiotemporal profile of empirical tau pathology. This suggests that passive network diffusion is insufficient and additional aggregation processes enhance the model’s predictive power. The inclusion of microglial gene expression demonstrated higher predictive power ( [12pt]{minimal} $$R^2$$ R 2 ) compared to Nex is :global. The improvement was statistically significant (per Fisher’s-R-to-Z transform p value) for DS6 and DS9. Trem2 imparted significantly higher predictive power compared to microglial homeostasis genes. It is important to note that across all datasets, AIC and BIC decreased upon the inclusion of microglia, indicating that the better fits achieved by Nex is :microglia were not simply due to the introduction of additional model parameters. Instead, these results reflect reduction in uncertainty in the global parameters, [12pt]{minimal} $$$$ α and [12pt]{minimal} $$$$ β , as indicated by their narrower 95% confidence intervals. The eight chosen microglial homeostasis genes are used in the model as their first principal component. The accumulation and spread parameters for these Nex is :microglia models were either negative, non-significant, or much lower than those obtained for Nex is : Trem2 (Table ). Further, we consider the datasets DS6_110 and DS9_110, which use the same injected tau strains as DS6 and DS9, respectively, but at 1:10 dilution . We expected that the tau-strain-specific parameters [12pt]{minimal} $$$$ α , [12pt]{minimal} $$$$ β , b , and p should be consistent between these experiments, but the seed rescale parameter, [12pt]{minimal} $$$$ γ , should be lower. We found that by applying Nex is : Trem2 on DS6_110 and DS9_110 after fixing the former parameters to their respective values for DS6 and DS9 and fitting only for [12pt]{minimal} $$$$ γ , the model provides strong fits to the data with lower [12pt]{minimal} $$$$ γ values, as anticipated (Table ). Trem2 on tau burden and spread Since the fitting results highlight Trem2 as the more significant modulator compared to homeostatic genes, the remainder of our results will focus on it. Figure shows the results for DS9 in detail; they are very similar for DS6 (Fig. ). For dataset DS4 (Fig. ), while the Nex is models were significantly better than the passive NDM, improvement from Nex is :global to Nex is :microglia was not statistically significant (Table ). Both Nex is -modeled distributions differ significantly from the NDM-modeled distribution. While both Nex is -modeled distributions look similar visually, there are clear differences between them that begin to become apparent at later time-points, especially in striatal and entorhinal areas. While both produce excellent recapitulation of empirical tau burden at all time points, the Nex is : Trem2 model produces by far the better fits (Fig. f, g). Model fits in all cases improve over time as may be expected from a ramifying pathological process. The Nex is : Trem2 fits incurred an increase in the microglia-dependent spread parameter, b (Table ); hence Trem2 appears to allow more pathology to efflux from the hippocampal seed regions at early time-points (Fig. d). We next assessed, visually, the effect that microglia are imparting on the spread of tau within the model. Flow-maps depicting major pathology spreading directions were constructed using Brainframe, our in-house Matlab-based function. Figure highlights differences in tau burden and spread as modeled by Nex is :global and Nex is : Trem2 for DS9. A prevalence of magenta arrows (Fig. b) indicates that Trem2 allowed more widespread propagation of tau (also reflected in the higher b values in Table for DS9). Although modeled tau propagated from the hippocampal seed area to connected regions, Trem2 appeared to mediate more spread of tau to the striatal, entorhinal, pallidal, and contralateral hippocampal regions. Similarly, a prevalence of green spheres in the striatal, pallidal, and hippocampal areas indicates that these regions evidence higher tau burden due to Trem2 than would otherwise be expected. Tau burden was unaffected by the presence of Trem2 in the parasubiculum (red sphere at all time points). Trem2 expression and the subsequent response of tau Given the effect of Trem2 on model performance and its quantified effects on tau transmissibility ( b ), we hypothesized that there would be a dose-dependent effect of Trem2 on tau, where higher Trem2 expression would lead to lower tau levels. In silico simulation of changes in baseline Trem2 levels at the global brain level, followed by application of the Nex is : Trem2 model, indicated a unidirectional trend in tau. Higher Trem2 expression resulted in a reduction of tau in five sample regions of the mouse brain (Fig. a). Specifically, there appeared to be a preponderance of an early effect of Trem2 in the hippocampus and the entorhinal cortex. is :microglia models applied to additional candidate genes implicated in tauopathies We investigated Apoe , a strongly implicated AD-risk gene – expressed both in microglia and astrocytes . Nex is : Apoe did not outperform Nex is :global in any of the datasets (Table , Fig. ). Since Trem2 and Apoe also work in concert in AD , we conducted a combined Nex is :microglia analysis including them both, but results for all three datasets matched those produced by Nex is : Trem2 alone (Table ). We then applied Nex is :microglia to some candidate genes involved in phagocytosis and inflammation ( Sorl1 , Cd33 , Clu , Sirpa , and Aif1 ). The accumulation ( p ) and spread ( b ) parameters obtained for these genes were either negative or insignificant, agreeing with their role in containing and eliminating pathological entities. The spread of hyperphosphorylated tau in AD occurs along the brain’s connectome and is also influenced by the surrounding molecular players such as certain cell types. The presented modeling effort was inspired by and expands upon the “molecular nexopathy” paradigm in that a central role is noted for network-mediated spread, but also includes other sources of protein aggregation and mediators of trans-neuronal spread, especially microglia. Microglia co-localize with tau inclusions, taking up and releasing bioactive seed-competent tau, and thus emerging as key mediators of AD pathology propagation along with connectome-mediated spread , . Phagocytic microglia, instead of degrading tau, may package it into exosomes and form aberrant junctions with surrounding neurons through which tau might be transmitted between cells , . While prior studies – have successfully quantified key aspects of pathology progression along the connectome, they do not constitute a true Nexopathy model, since they did not account for the influence of non-neuronal modulators. Thus, here we presented a mathematical model called Nex is , that attempts to incorporate diverse effects including misfolded protein seeding, aggregation, connectome-mediated spread, as well as the modulation of all the above processes by extra-connectomic agents residing within the brain’s neurological milieu. Nex is augments prior quantitative models (NDM) by including microglial modulation via two key processes: local tau accumulation and its enhanced trans-neuronal transmission. Our results quantify how the inclusion of microglia improves the predictive power of the model and reduces the uncertainty in the accumulation and diffusion parameter estimates (see AIC and narrower confidence intervals around model parameters, Table ). The full model recapitulates empirical data with [12pt]{minimal} $$R^2$$ R 2 values up to 0.60. The NDM, on the other hand, cannot capture the extent of the spatiotemporal pathology because it fails to capture scale. Thus even though the individual Pearson r values per time-point are significant, the overall model fit ( [12pt]{minimal} $$R^2$$ R 2 ) is very poor with the NDM. The Nex is -modeled spatiotemporal progression of tau closely matches the Braak staging scheme , with pathology progressing from the hippocampal injection site to the contralateral hippocampus and entorhinal cortex, followed by the neocortical, striatal, and thalamic areas. We acknowledge that the data used here were from a seeded mouse model, which may have biased the starting point of pathology to be the hippocampus. Future models applied to unseeded mouse models of AD would help since they do demonstrate tau spatiotemporal behavior as dictated by Braak staging . In addition, the data are from a model of primary tauopathy, and hence cannot necessarily be characterized in the same way as AD tauopathy. This is because there is insufficient information on staging of primary tauopathies , and the classical pattern of tau spread to the frontal regions might be due to the influence of amyloid- [12pt]{minimal} $$$$ β in AD. Microglial modulation was strongly dependent on the tau strain used in the specific dataset. Datasets DS6 and DS9 have medium to high seeding capacity and are characterized by tau from a P301S tauopathy mouse and recombinant fibrils, respectively. For these datasets, Trem2 was associated with decreased tau accumulation rate ( p ) and increased spread ( b ) to regions ipsi- and contralateral to the hippocampal seed area (Table ). Moreover, Trem2 distribution (Fig. b) did not spatially correlate with regions with higher Trem2 -mediated tau burden (Fig. d), suggesting that microglial effects are not strictly local. Instead, as shown by differences in fitted model parameters, Trem2 appears to mobilize tau increasing its network transmissibility. We hypothesize that Trem2 renders tau more mobile, thus increasing its brain-wide spread but reducing its accumulation rate. Although not possible to model here, the effect of Trem2 on tau mobilization may be occurring via the action of exosomes . Trem2 -mediated high tau efflux with slower accumulation may reflect clearance and possible mitigation of pathology. Mouse studies have indeed demonstrated Trem2 -mediated decrease in tau accumulation in early disease stages , , Trem2 -mediated neuroprotection , as well as Trem2 loss resulting in enhanced tau seeding and exacerbated pathology . Microglia can internalize and release seed-competent tau, but also potentially degrade its seeding activity . Our model parameters here suggest that the decreased accumulation rate ( p ) may support the empirically observed reduced seeding activity , whilst still maintaining tau spreading potential ( b ). Intriguingly, our results posit this effect of microglia to be specific to Trem2 , and not to homeostatic microglial genes.We schematically summarize the relationship between Trem2 and tau as modeled by Nex is and as observed in these datasets, in Fig. . Neither Nex is framework recapitulates one of the datasets, DS4, particularly well (Table ). DS4, characterized by the human AD-brain derived tau isolate, demonstrated elevated rates of both tau accumulation ( p ) and spread ( b ) in the presence of either Trem2 or homeostatic microglia. However, neither Nex is :homeostasis nor Nex is : Trem2 was significantly better than Nex is :global. As observed by Kaufman et al. , DS4 was a low-tangle tau strain with lower seeding capacity than DS6 and DS9. This may explain its comparatively weaker Nex is :global model fit, with a failure of any additional improvements with the inclusion of microglial genes. We also speculate that since DS4 was characterized by amyloid-comorbid tau from a human-AD brain it evinced higher values of accumulation and spread rates. AD derived amyloid-comorbid tau can acquire more pathogenic conformations leading to its aggregation and accumulation at a faster rate . Our results thus point to the importance of quantifying the heterogeneous effects that tau has on neuropathology progression, importantly because tau strain diversity based on differences in bioactivity (seeding and spreading) may explain the myriad of AD clinical outcomes . Trem2 is upregulated in microglia during disease progression, and in fact a lack of Trem2 worsens tau pathology . Similar observations have been made by Zhu et al. where Trem2 deletion caused enhanced spread of tau pathology from the injection site to other connected regions of a Trem2 knock-out mouse model. We showed here that Nex is is capable of confirming the histopathologically observed mechanistic relationship between Trem2 and tau (Fig. a). Trem2 expression levels lower and higher than baseline simulated its deletion (or knock-out) and over-expression, respectively. Nex is : Trem2 set up using varying scales of Trem2 expression could provide quantifiable trends of its inverse relationship with tau as quantified in five sample regions of the mouse brain. The baseline expression levels of Trem2 (Fig. a inset) in these regions is in agreement with the pattern observed in murine and human studies , with the hippocampus and entorhinal cortex evidencing the highest levels followed by the neocortical and striatal regions. Apoe is a canonical risk-gene implicated in AD, and along with being a strong influencer of amyloid pathology, it is also suggested to influence microglial function and tau pathology . We thus explored the effects of Apoe via Nex is :microglia model alone and in conjunction with Trem2 as these two genes appear to act in concert . Compared to Nex is : Trem2 , the lack of improvement in Nex is : Apoe and comparable performance by Nex is : Trem2 : Apoe can be related to the fact that Apoe acts downstream of Trem2 (in disease conditions). Therefore, the key mediatory role manifested by Trem2 can be expected to encompass the contribution of Apoe . In conclusion, our in silico Nex is :microglia model quantifies the contribution of microglia to tau neuropathology progression. It recapitulates the characteristics of observed tau pathology progression, providing distinct results specific to the tau species employed in the mouse models we studied. Although prior bench studies have suggested a modulatory role for microglia, the present study is the first to demonstrate this quantitatively on the whole mouse brain, and for specific microglial genes and specific tau conformations. In particular, our results provide quantitative support to the observations made by Asai et al. in that, microglia aid increased propagation of tau across brain regions. The Nex is :microglia framework can greatly aid experimentalists in further hypothesis generation, validation, and quantification. It may be readily adapted to interrogate the effects of other putative modulators of pathology progression such as astrocytes and oligodendrocytes. The datasets used monitored the levels of exogenous tau at 4, 8, and 12 weeks post hippocampal seeding. Additional time-points may enable a more thorough understanding of the spatiotemporal behavior of tau progression. The current monitoring over time was also not longitudinal in the real sense since measurements did not track the same animal. Instead, tau progression was observed in groups of tau-injected mice sacrificed at different time-points post-injection. Further, the gene expression data used as microglial signatures were from a healthy mouse brain and thus can infer only about the effects of baseline microglial density on pathology. However, interestingly, a recent study indeed supports the effect of baseline microglia, rather than the brain’s anatomic network, on tau propagation . We were also restricted to using the coronal series from the AGEA due to its superior spatial coverage relative to the sagittal series, but with significantly fewer quantified genes. As a result, not all microglial genes of interest could be assessed with Nex is . Next, Trem2 assumes opposing roles (protective vs. detrimental) in AD pathology depending on disease-stage. Such disease-stage dependent dual roles of Trem2 have been observed with respect to amyloid [12pt]{minimal} $$$$ β plaque pathology . Our current in silico investigation of the role of Trem2 in neuropathology progression is restricted to tau and is conducted on data acquired within three months of exogenous tau seeding in mice. Future goals would be to apply this approach to data collected at extended time-points post tau-seeding (depending on data availability), as well as to include amyloid pathology in the model since amyloid [12pt]{minimal} $$$$ β has been speculated to potentiate microglial influence on tauopathy propagation . A possibility that the seed region where tau was initially injected may confer specific molecular vulnerability to tau, remains outside the scope of the current study design. Our Nex is approach itself is fully capable of testing for site-specific effects but paucity of experimental data to date makes this challenging. We also note that the nexopathy paradigm enumerates a far more diverse set of potential mechanisms by which molecular dysfunction might interact with the neural architecture than is possible to explore in a single study. Epigenetic or post-translational modifications due to variable causes will produce gene expression in diseased groups that is different from healthy gene expression used here. However, the primary goal of uncovering the molecular versus network correlates of AD topography may be largely unaffected by these additional factors, since they are either unlikely to have spatial gradients (e.g. age, environment) or have non-AD-like spatial patterning (oxidative/vascular stress). Finally, the current modeling attempts were restricted to data from mouse models of primary tauopathy. There are two main reasons why we decided to start with mouse models: (1) the datasets compared here are from mice having the same genetic background thus minimizing any effects of genetic heterogeneity on model results, and (2) since mice are exogenously injected with pathology in specific brain regions, pathology progression from a common seed location can be modeled and compared. Pathology seeding data from humans are not possible, and despite the stereotyped progression of AD tau pathology, there is evidence that significant heterogeneity exists between patients at disease inception . Another advantage of seeded mouse models is that the diverse effects of different strains of the pathological species can be compared. That being said, the true translational aspect of Nex is will be appreciated by its application to human data. This will also be important for inter-species comparison of the glial response to pathology, since transcriptional signatures of glia in AD pathology can be markedly different between mouse AD models and specimens from human AD patients . Nevertheless, despite this difference, the fact that Trem2 impacts microglia function is a common theme in mouse and human AD . We are currently in the process of curating and inspecting human data to which we can apply Nex is . Study design Empirical spatiotemporal tau pathology data were obtained from a study on PS19 mice expressing 4R1N P301S with human tau, exogenously injected with distinct strains of pathological tau . Regional levels of tau burden were extracted from heat maps displaying semi-quantified pathology anatomically following the approach from Mezias et al. . The Kaufman datasets used in this study are DS4 (containing tau isolate derived from the human AD-brain), DS6 (contained tau injectate from a P301S tauopathy mouse), and DS9 (characterized by recombinant tau fibrils). Our 426-region mouse connectome is from the Allen Brain Atlas (ABA) , with 213 region bi-hemispheric symmetry. The microglial baseline spatial gene expression maps employed as signatures of baseline homeostatic and phagocytic microglia are from the ABA gene expression atlas (AGEA) of healthy 56-day old male C57BL/6J mice . Note that these microglial signatures represent quiescent rather than activated states and may not be considered disease-activated microglia (DAM). Moreover we selectively use only the genes from the coronal AGEA and exclude those from the sagittal atlas because the spatial coverage of gene expression is higher in the former. Thus, the final gene list that we referred to has a subset containing 3855 genes from a total of approximately 26000 genes in the AGEA (sagittal + coronal). Regional homeostatic microglial abundance was defined as the first principal component of the expression levels of P2ry12 , Cx3cr1 , Fcrls , Olfml3 , Hexb , Siglech , Sox5 , and Jun , . We separately tested the influence of Trem2 , Cd33 , and Apoe , activated microglial surface-markers and regulators of microglia-mediated phagocytosis strongly implicated in AD pathogenesis . Cd33 ’s influence (Table ) was negligible compared to Trem2 given its smaller effect size compared to Trem2 . Nexopathy in silico Model (Nex is ) The Nex is model is derived from the Network Diffusion Model (NDM) —a mathematical model that describes pathology spread as a diffusive process between connected brain regions. Let [12pt]{minimal} $$c_{12}$$ c 12 denote the inter-region connection strength between two brain regions (1,2) as inferred from tractography data, let and [12pt]{minimal} $$x_{1}(t)$$ x 1 ( t ) and [12pt]{minimal} $$x_{2}(t)$$ x 2 ( t ) be the pathology concentration at two brain regions at time t as inferred from semi-quantitative immunohistochemistry analysis from the specific reference study. Following the NDM , we impose first-order diffusive dynamics: 1 [12pt]{minimal} $$ }{dt}= c_{12} (x_{2}-x_{1}), $$ d x 1 dt = β c 12 ( x 2 - x 1 ) , where [12pt]{minimal} $$$$ β is the global diffusivity rate that controls how fast pathology can flow from the more concentrated region to the less concentrated one independent of any other modulators. Nex is : global adds another parameter to the NDM, [12pt]{minimal} $$ x_{1}$$ α x 1 , a linear accumulation or clearance term depending on its sign. 2 [12pt]{minimal} $$ }{dt}= c_{12} (x_{2}-x_{1}) + x_{1}, $$ d x 1 dt = β c 12 ( x 2 - x 1 ) + α x 1 , This pair-wise relationship can be extended to the entire brain connectivity network using vector [12pt]{minimal} $$ { {R}}^{n}$$ x → ∈ R n over all n ROIs, to give [12pt]{minimal} $$}{dt} = - L $$ d x → dt = - β L x → , where L is the graph Laplacian matrix (when [12pt]{minimal} $$ = 0$$ α = 0 ) . Nex is :microglia The Nex is :microglia framework incorporates the potential effects that different modulators, for instance cell-types such as microglia, may have on neuropathology progression, particularly in the context of AD and other tauopathies. It has two additional cell-type specific parameters p and b , which indicate the effect of the chosen cell type, u , on pathology accumulation/clearance and spread, respectively. Letting [12pt]{minimal} $$u_{1}$$ u 1 and [12pt]{minimal} $$u_{2}$$ u 2 indicate the local gene expression in regions 1 and 2, the extended pairwise relationship in becomes 3 [12pt]{minimal} $$ }{dt} = c_{12} [(1 + u_{2}b)x_{2} - (1 + u_{1}b)x_{1} ] + x_{1} + u_{1} p x_{1}. $$ d x 1 dt = β c 12 [ ( 1 + u 2 b ) x 2 - ( 1 + u 1 b ) x 1 ] + α x 1 + u 1 p x 1 . Equation reduces to NDM equation if [12pt]{minimal} $$b = p = = 0$$ b = p = α = 0 , and can similarly be extended brain-wide via pathology [12pt]{minimal} $$$$ x → and gene [12pt]{minimal} $$$$ u → vectors to [12pt]{minimal} $$}{dt} = $$ d x → dt = [ Λ ( u → ) - β L ~ ( u → ) ] x → , where [12pt]{minimal} $$$$ Λ is a diagonal matrix and [12pt]{minimal} $${}$$ L ~ is a new Laplacian, both now depending on gene expression u . See the for a detailed explanation. For convenience of reading, in the remainder of the document the model name ’Nex is :microglia’ will be replaced by Nex is :homeostasis in the case of homeostatic microglia or Nex is :( microglial gene name ) in the case of a particular microglial gene being investigated. Seed scaling and model initialization Both NDM and Nex is require a seed-rescale parameter [12pt]{minimal} $$$$ γ , which denotes the scaling factor for initial pathology at the site of injection of pathology (tau in the current case). This parameter needs to be quantified since, although we know the site of the injection (hippocampus for instance), we do not know the amount of pathology being injected. Thus, the pathology at t = 0 is represented as a binary vector multiplied by [12pt]{minimal} $$$$ γ and is used to initialize the model. Robust strategy for parameter estimation We implemented a two-step, robust strategy for estimating parameters for the NDM and Nex is models. First, we fit the spread parameter ( [12pt]{minimal} $$$$ β ) for the NDM, followed by the accumulation and spread parameters ( [12pt]{minimal} $$$$ α and [12pt]{minimal} $$$$ β ) for Nex is :global via grid search and random initialization. This was then followed by 100 iterations of bootstrapping with 80 [12pt]{minimal} $$\%$$ % re-sampling. Next, for the Nex is :microglia model, we initialized the values of [12pt]{minimal} $$$$ α and [12pt]{minimal} $$$$ β to those obtained from Nex is :global while also fitting (without initialization) the microglia-mediated accumulation and spread parameters ( p and b ). Anchoring the parameters shared by both Nex is frameworks within a reasonable range from each in the optimization routines prevents the algorithm from reaching local minima that yield better fits but have overall biologically-implausible parameter values. This also allowed us to assess reproducibility by acquiring the 95% confidence intervals around the mean estimates of the model fit as well as global and microglial parameters (simulations in Fig. for Nex is :global and Nex is : Trem2 ). Importantly, model parameters were fit using all available quantified regions and all time points; prior formulations only allowed fitting to single time points , . Statistics and reproducibility The statistical significance of the differences in model fits between the NDM and the Nex is frameworks was determined by carrying out Fisher’s-R-to-Z transformation. The p value of significance was obtained by the Student’s t test on the Fisher-transformed Z scores. The model fits presented in Table were determined by averaging the parameters across 100 iterations of bootstrapping with 80% re-sampling of the regions quantified by Kaufman et al. . When fitting the Nex is :microglia models we used the mean and 95% confidence intervals determined for the Nex is :global parameters [12pt]{minimal} $$$$ α , [12pt]{minimal} $$$$ β , and [12pt]{minimal} $$$$ γ for each dataset as the initial guesses and bounds for those parameters, respectively; this helps to minimize the risk of finding degenerate solutions. Please see for details on parameter fitting. The [12pt]{minimal} $$R^2$$ R 2 values of the full model are within the range of the bootstrapped iterations of the 80% re-sampled data (Fig. c), thus confirming reproducibility. All analyses were conducted in Matlab R2020b. Simulation of changes in Trem2 expression to assess response of tau Significant improvement in model fit upon inclusion of Trem2 warranted further in silico quantification of tau response to simulated changes in Trem2 expression levels. Baseline Trem2 levels were scaled by factors of two above and below the baseline level and tau was quantified in the hippocampus, entorhinal cortex, neocortex, striatum, and amygdala. The scaling was performed by multiplying the microglial gene (here, Trem2 ) vector by powers of 2 (from [12pt]{minimal} $$2^{-3}$$ 2 - 3 to [12pt]{minimal} $$2^4$$ 2 4 ) before entering it into the model. Ethics approval and consent to participate All data were obtained from publicly available datasets and published papers. No human subjects, live animals, patient or animal derived tissue, cell lines, or biological material were directly used in this research and thus, no consent or ethics approval was required. Empirical spatiotemporal tau pathology data were obtained from a study on PS19 mice expressing 4R1N P301S with human tau, exogenously injected with distinct strains of pathological tau . Regional levels of tau burden were extracted from heat maps displaying semi-quantified pathology anatomically following the approach from Mezias et al. . The Kaufman datasets used in this study are DS4 (containing tau isolate derived from the human AD-brain), DS6 (contained tau injectate from a P301S tauopathy mouse), and DS9 (characterized by recombinant tau fibrils). Our 426-region mouse connectome is from the Allen Brain Atlas (ABA) , with 213 region bi-hemispheric symmetry. The microglial baseline spatial gene expression maps employed as signatures of baseline homeostatic and phagocytic microglia are from the ABA gene expression atlas (AGEA) of healthy 56-day old male C57BL/6J mice . Note that these microglial signatures represent quiescent rather than activated states and may not be considered disease-activated microglia (DAM). Moreover we selectively use only the genes from the coronal AGEA and exclude those from the sagittal atlas because the spatial coverage of gene expression is higher in the former. Thus, the final gene list that we referred to has a subset containing 3855 genes from a total of approximately 26000 genes in the AGEA (sagittal + coronal). Regional homeostatic microglial abundance was defined as the first principal component of the expression levels of P2ry12 , Cx3cr1 , Fcrls , Olfml3 , Hexb , Siglech , Sox5 , and Jun , . We separately tested the influence of Trem2 , Cd33 , and Apoe , activated microglial surface-markers and regulators of microglia-mediated phagocytosis strongly implicated in AD pathogenesis . Cd33 ’s influence (Table ) was negligible compared to Trem2 given its smaller effect size compared to Trem2 . in silico Model (Nex is ) The Nex is model is derived from the Network Diffusion Model (NDM) —a mathematical model that describes pathology spread as a diffusive process between connected brain regions. Let [12pt]{minimal} $$c_{12}$$ c 12 denote the inter-region connection strength between two brain regions (1,2) as inferred from tractography data, let and [12pt]{minimal} $$x_{1}(t)$$ x 1 ( t ) and [12pt]{minimal} $$x_{2}(t)$$ x 2 ( t ) be the pathology concentration at two brain regions at time t as inferred from semi-quantitative immunohistochemistry analysis from the specific reference study. Following the NDM , we impose first-order diffusive dynamics: 1 [12pt]{minimal} $$ }{dt}= c_{12} (x_{2}-x_{1}), $$ d x 1 dt = β c 12 ( x 2 - x 1 ) , where [12pt]{minimal} $$$$ β is the global diffusivity rate that controls how fast pathology can flow from the more concentrated region to the less concentrated one independent of any other modulators. Nex is : global adds another parameter to the NDM, [12pt]{minimal} $$ x_{1}$$ α x 1 , a linear accumulation or clearance term depending on its sign. 2 [12pt]{minimal} $$ }{dt}= c_{12} (x_{2}-x_{1}) + x_{1}, $$ d x 1 dt = β c 12 ( x 2 - x 1 ) + α x 1 , This pair-wise relationship can be extended to the entire brain connectivity network using vector [12pt]{minimal} $$ { {R}}^{n}$$ x → ∈ R n over all n ROIs, to give [12pt]{minimal} $$}{dt} = - L $$ d x → dt = - β L x → , where L is the graph Laplacian matrix (when [12pt]{minimal} $$ = 0$$ α = 0 ) . is :microglia The Nex is :microglia framework incorporates the potential effects that different modulators, for instance cell-types such as microglia, may have on neuropathology progression, particularly in the context of AD and other tauopathies. It has two additional cell-type specific parameters p and b , which indicate the effect of the chosen cell type, u , on pathology accumulation/clearance and spread, respectively. Letting [12pt]{minimal} $$u_{1}$$ u 1 and [12pt]{minimal} $$u_{2}$$ u 2 indicate the local gene expression in regions 1 and 2, the extended pairwise relationship in becomes 3 [12pt]{minimal} $$ }{dt} = c_{12} [(1 + u_{2}b)x_{2} - (1 + u_{1}b)x_{1} ] + x_{1} + u_{1} p x_{1}. $$ d x 1 dt = β c 12 [ ( 1 + u 2 b ) x 2 - ( 1 + u 1 b ) x 1 ] + α x 1 + u 1 p x 1 . Equation reduces to NDM equation if [12pt]{minimal} $$b = p = = 0$$ b = p = α = 0 , and can similarly be extended brain-wide via pathology [12pt]{minimal} $$$$ x → and gene [12pt]{minimal} $$$$ u → vectors to [12pt]{minimal} $$}{dt} = $$ d x → dt = [ Λ ( u → ) - β L ~ ( u → ) ] x → , where [12pt]{minimal} $$$$ Λ is a diagonal matrix and [12pt]{minimal} $${}$$ L ~ is a new Laplacian, both now depending on gene expression u . See the for a detailed explanation. For convenience of reading, in the remainder of the document the model name ’Nex is :microglia’ will be replaced by Nex is :homeostasis in the case of homeostatic microglia or Nex is :( microglial gene name ) in the case of a particular microglial gene being investigated. Both NDM and Nex is require a seed-rescale parameter [12pt]{minimal} $$$$ γ , which denotes the scaling factor for initial pathology at the site of injection of pathology (tau in the current case). This parameter needs to be quantified since, although we know the site of the injection (hippocampus for instance), we do not know the amount of pathology being injected. Thus, the pathology at t = 0 is represented as a binary vector multiplied by [12pt]{minimal} $$$$ γ and is used to initialize the model. We implemented a two-step, robust strategy for estimating parameters for the NDM and Nex is models. First, we fit the spread parameter ( [12pt]{minimal} $$$$ β ) for the NDM, followed by the accumulation and spread parameters ( [12pt]{minimal} $$$$ α and [12pt]{minimal} $$$$ β ) for Nex is :global via grid search and random initialization. This was then followed by 100 iterations of bootstrapping with 80 [12pt]{minimal} $$\%$$ % re-sampling. Next, for the Nex is :microglia model, we initialized the values of [12pt]{minimal} $$$$ α and [12pt]{minimal} $$$$ β to those obtained from Nex is :global while also fitting (without initialization) the microglia-mediated accumulation and spread parameters ( p and b ). Anchoring the parameters shared by both Nex is frameworks within a reasonable range from each in the optimization routines prevents the algorithm from reaching local minima that yield better fits but have overall biologically-implausible parameter values. This also allowed us to assess reproducibility by acquiring the 95% confidence intervals around the mean estimates of the model fit as well as global and microglial parameters (simulations in Fig. for Nex is :global and Nex is : Trem2 ). Importantly, model parameters were fit using all available quantified regions and all time points; prior formulations only allowed fitting to single time points , . The statistical significance of the differences in model fits between the NDM and the Nex is frameworks was determined by carrying out Fisher’s-R-to-Z transformation. The p value of significance was obtained by the Student’s t test on the Fisher-transformed Z scores. The model fits presented in Table were determined by averaging the parameters across 100 iterations of bootstrapping with 80% re-sampling of the regions quantified by Kaufman et al. . When fitting the Nex is :microglia models we used the mean and 95% confidence intervals determined for the Nex is :global parameters [12pt]{minimal} $$$$ α , [12pt]{minimal} $$$$ β , and [12pt]{minimal} $$$$ γ for each dataset as the initial guesses and bounds for those parameters, respectively; this helps to minimize the risk of finding degenerate solutions. Please see for details on parameter fitting. The [12pt]{minimal} $$R^2$$ R 2 values of the full model are within the range of the bootstrapped iterations of the 80% re-sampled data (Fig. c), thus confirming reproducibility. All analyses were conducted in Matlab R2020b. Trem2 expression to assess response of tau Significant improvement in model fit upon inclusion of Trem2 warranted further in silico quantification of tau response to simulated changes in Trem2 expression levels. Baseline Trem2 levels were scaled by factors of two above and below the baseline level and tau was quantified in the hippocampus, entorhinal cortex, neocortex, striatum, and amygdala. The scaling was performed by multiplying the microglial gene (here, Trem2 ) vector by powers of 2 (from [12pt]{minimal} $$2^{-3}$$ 2 - 3 to [12pt]{minimal} $$2^4$$ 2 4 ) before entering it into the model. All data were obtained from publicly available datasets and published papers. No human subjects, live animals, patient or animal derived tissue, cell lines, or biological material were directly used in this research and thus, no consent or ethics approval was required. Supplementary Information. |
Evaluation of the efficiency of smear layer removal during endodontic treatment using scanning electron microscopy: an in vitro study | 96ecdbd8-c84b-4b1e-bb0e-37bbe3bf9c3c | 11776335 | Dentistry[mh] | The rapid development in dentistry allows for effective tooth-preserving manipulations. The number of successful cases of endodontic treatment has increased significantly due to improved mechanical treatment of root canal (RC), which makes it possible to excise infected dentin and obtain optimal RC geometry, accumulation of knowledge of root canal treatment (RCT) and improvement of obturation techniques, which reduce the risk of insufficient and excessive filling of root canal . During the preparation (instrumentation) of root canals (RC) with manual or rotary instruments, a smear layer (SL) is formed, which may remain in the root canal sealing the dentinal tubules, which decrease the success of endodontic treatment. During the treatment of necrotic forms of pulpitis and periodontitis, the endodontist must not only widen the root canal for adequate obturation and disinfection, but also ensure maximum treatment of the dentinal tubules adjacent to the RC wall with temporary and permanent filling materials . The SL formed during the treatment must be removed, so great importance should be related to the irrigation protocol used in endodontic treatment . The use of modern irrigants allows for effective control of the endodontic microbiota and allows for the removal of up to 92% of microorganisms from the RC system during irrigation with sodium hypochlorite (NaOCL) . One of the most frequently used irrigants is NaOCL (1–5%). It effectively combats the RC microbiota and dissolves soft tissue residues . Its antiseptic activity has a wide spectrum of action against E. Faecalis, P. Endodontalis, St. Intermedius. However, the smear layer (SL) reduces its antiseptic activity . The duration of the process causes a formation of biofilms on the walls of the root canal and a penetration of microflora into the dentinal tubules up to 200 μm . Therefore, an important aspect in endodontic treatment is the removal of smear layer after mechanical treatment. This allows NaOCL (surface tension 75 dyne/cm) to better penetrate the dentinal tubules . The smear layer can be removed by physical and chemical methods. For physical action, laser action is used, which effectively removes SL and microorganisms. However, it requires equipment and specialized skills of the doctor . For the chemical method of removing the smear layer in the root canal, ethylenediaminetetraacetic acetate (EDTA) 17–20% is most often used . EDTA has low antiseptic activity, so it is used in combination with NaOCL 1–5% . This combination of irrigants affects the microbiota (NaOCL 1–5%) and root canal debris (EDTA 17–20%). This widely used protocol has proven its effectiveness and has good long-term clinical results. However, it has disadvantages, as according to existing studies, the use of EDTA 17–20% is ineffective for removing the smear layer in the apical third of the root canal, which can cause the development of a relapse of chronic apical periodontitis . To improve the quality of root canal treatment with EDTA solution, additional activation is necessary, for this purpose ultrasonic or laser activation is used . Combinations of these irrigants have several disadvantages, such as the need to use several syringes with different solutions and intermediate rinsing with water, since no precipitates are formed upon contact of irrigants, but gaseous chlorine is released . Also, EDTA 17–20% damages peritubular dentin upon prolonged contact with dentin . All the above disadvantages of EDTA are the reason for further search for irrigants for SL removal. Recently, solutions based on etidronic acid have become very popular. Etidronic acid is a soft and biocompatible chelating agent, a distinctive feature of which is the ability to dilute etidronate directly in NaOCL . An additional advantage is the release of root dentin growth factors after treatment with etidronate . The most common concentration is 9%; due to the instability of the solution, it must be prepared immediately before use. To prepare the irrigant, 0.9 g of etidronate is dissolved in 10 ml of NaOCL or distilled water and used within an hour . The etidronate solution in hypochlorite has a mild chelating effect on the root canal walls, removes debris at all stages of treatment, does not damage the structure of dentin and does not affect its hardness . Ultrasonic activation of the solution also improves the quality of SL removal . The aim of the study was to compare the efficiency of removing smear layer and debris in root canals using different irrigation protocols in vitro.
The ethics committee at medical institute of peoples’ Friendship University of Russia approved this study (Protocol 21 at 19.10.2023). This in vitro study was conducted to evaluate three irrigation protocols on removing smear layer and debris from the root canal walls. For the study, 30 intact teeth with straight root canals, extracted for periodontal indications (18 incisors, 3 canines, 9 premolars) were selected. Depending on the protocol used, the teeth were divided into 3 groups, 10 in each. Informed consent was obtained from all subjects, whose teeth were used after extraction. Group 1 (protocol 1) 10 teeth (6 incisors, 1 canine and 3 premolars) were mechanically processed with S-flexi instruments (Geosoft Endoline, Russia) till apical master file 35 and irrigated according to the protocol with 3% NaOCL (2 ml of solution) alternately with 17% EDTA (2 ml of solution) for 30 s. Each solution was activated by an EQ-S sonic endoactivator (Meta Biomed, South Korea) for 30 s after each instrument, the final irrigation was carried out with distilled water. Group 2 (protocol 2) 10 teeth (6 incisors, 1 canine and 3 premolars) were mechanically processed with S-flexi instruments till apical master file 35 and the root canal was irrigated according to protocol 2 using 3% NaOCL (2 ml of solution) alternately with 9% etidronate solution diluted with distilled water (2 ml of solution) for 30 s. Each solution was activated by the EQ-S sonic endoactivator for 30 s after each instrument, the final irrigation was performed with distilled water. Group 3 (protocol 3) 10 teeth (6 incisors, 1 canine and 3 premolars) were mechanically processed with S-flexi instruments till master file 35 and canals were irrigated according to Protocol 3 using a 9% solution of etidronate diluted in 3% NaOCL (2 ml of solution) after each instrument for 30 s, with activation by an EQ-S endoactivator for 30 s, final irrigation was performed with distilled water. Upon completion of sample preparation, the root canal was dried with absorbent paper points and the root was cut longitudinally with a diamond disk and divided into three parts (apical, middle and coronal). The dried samples were prepared for electron scanning microscopy and covered with a layer of platinum. The prepared samples were examined on a LEO-1430 VP scanning electron microscope (Carl Zeiss, Germany). In the experiment, the resolution was about 40 nm. The quality of smear layer removal was assessed by two independent experts using photographs obtained after scanning electron microscopy (SEM) using a scoring system from 1 to 5 proposed by Hulsmann et al. . The following values served as criteria: Score 1: No smear layer, dentinal tubules visualized. Score 2: Small amount of smear layer covering the root canal, many dentinal tubules are visualized. Score 3: Smear layer and debris covering the root canal walls and a few dentinal tubules are visualized. Score 4: The surface of root canal covered completely with smear layer; no dentinal tubules visualized. Score 5: Heavy smear layer and debris covered the root canal surface. Statistical analysis was performed using IBM SPSS Statistics Version 22 for Windows (IBM ® , NY, USA). The nonparametric Mann-Whitney and Kruskal Wallis tests were used for the comparison between the irrigation protocols. Statistical significance was set at 0.05.
10 teeth (6 incisors, 1 canine and 3 premolars) were mechanically processed with S-flexi instruments (Geosoft Endoline, Russia) till apical master file 35 and irrigated according to the protocol with 3% NaOCL (2 ml of solution) alternately with 17% EDTA (2 ml of solution) for 30 s. Each solution was activated by an EQ-S sonic endoactivator (Meta Biomed, South Korea) for 30 s after each instrument, the final irrigation was carried out with distilled water.
10 teeth (6 incisors, 1 canine and 3 premolars) were mechanically processed with S-flexi instruments till apical master file 35 and the root canal was irrigated according to protocol 2 using 3% NaOCL (2 ml of solution) alternately with 9% etidronate solution diluted with distilled water (2 ml of solution) for 30 s. Each solution was activated by the EQ-S sonic endoactivator for 30 s after each instrument, the final irrigation was performed with distilled water.
10 teeth (6 incisors, 1 canine and 3 premolars) were mechanically processed with S-flexi instruments till master file 35 and canals were irrigated according to Protocol 3 using a 9% solution of etidronate diluted in 3% NaOCL (2 ml of solution) after each instrument for 30 s, with activation by an EQ-S endoactivator for 30 s, final irrigation was performed with distilled water. Upon completion of sample preparation, the root canal was dried with absorbent paper points and the root was cut longitudinally with a diamond disk and divided into three parts (apical, middle and coronal). The dried samples were prepared for electron scanning microscopy and covered with a layer of platinum. The prepared samples were examined on a LEO-1430 VP scanning electron microscope (Carl Zeiss, Germany). In the experiment, the resolution was about 40 nm. The quality of smear layer removal was assessed by two independent experts using photographs obtained after scanning electron microscopy (SEM) using a scoring system from 1 to 5 proposed by Hulsmann et al. . The following values served as criteria: Score 1: No smear layer, dentinal tubules visualized. Score 2: Small amount of smear layer covering the root canal, many dentinal tubules are visualized. Score 3: Smear layer and debris covering the root canal walls and a few dentinal tubules are visualized. Score 4: The surface of root canal covered completely with smear layer; no dentinal tubules visualized. Score 5: Heavy smear layer and debris covered the root canal surface. Statistical analysis was performed using IBM SPSS Statistics Version 22 for Windows (IBM ® , NY, USA). The nonparametric Mann-Whitney and Kruskal Wallis tests were used for the comparison between the irrigation protocols. Statistical significance was set at 0.05.
No smear layer, dentinal tubules visualized.
Small amount of smear layer covering the root canal, many dentinal tubules are visualized.
Smear layer and debris covering the root canal walls and a few dentinal tubules are visualized.
The surface of root canal covered completely with smear layer; no dentinal tubules visualized.
Heavy smear layer and debris covered the root canal surface. Statistical analysis was performed using IBM SPSS Statistics Version 22 for Windows (IBM ® , NY, USA). The nonparametric Mann-Whitney and Kruskal Wallis tests were used for the comparison between the irrigation protocols. Statistical significance was set at 0.05.
The SEM results showed that in the first group, when using protocol 1 (3% NaOCL and 17% EDTA), the smallest number of open dentinal tubules was determined. The results are presented in Table . When using a combination of 3% NaOCL and 17% EDTA, SL removal in the coronal part was more pronounced than in the middle and apical parts (Fig. , -A, -B and -C). When analyzing the second group, it was revealed that the wall of the root canal in the coronal and middle parts was characterized by an almost complete absence of smear layer, well-open dentinal tubules, and clean peritubular dentin, with the absence of chemical erosions. The apical third of the root canal was covered with a thinner smear layer compared to the first irrigation protocol. Data on the preservation of the smear layer in the apical part of the canal were shown in Fig. ( -A, -B, -C). Samples of the third group Fig. ( -A, -B, -C) showed results like the samples of the second group. Treatment with etidronate in hypochlorite quite effectively removed the SL in all thirds of the root canal, open dentinal tubules with clean peritubular dentin were visualized in SEM photographs. Comparison of the results of scanning electron microscopy at the apical part in all three protocols revealed a lesser effect of 17% EDTA solution on the smear layer. At the same time, no significant difference was found between treatment with 9% etidronate solution and treatment with 9% etidronate solution in 3% Sodium hypochlorite. According to the results of scanning electron microscopy of 10 samples in each group (30 samples in total), it was revealed that the maximum number of open dentinal tubules was observed in the second and third groups. When compared the smear layer removal among protocols in coronal, middle and apical thirds of the root canal, Kruskal Wallis test showed statistical differences P = 0.01, 0.00 and 0.03, respectively. And when analyzing the smear layer removal differences between the protocols in the coronal, middle and apical third of the root canal, Mann-Whitney criterion demonstrated that Protocols 2 and 3 outperformed the first protocol in different parts ( P ≤ 0.05) (Table ). As a result of the statistical analysis, it can be concluded that protocols 2 and 3 have statistically significant differences from the control protocol 1. The use of irrigation protocol 3 (Etidronate 9% in 3% NaOCL) shows high efficiency in the apical part in 40% of cases (assigned a score of 1 point). When using the other protocols (9% etidronate in physiological solution + 3% NaOCL), a result of 1 score was obtained in 20% of cases and 2 score in 40% of cases (insignificant smear layer, single dentinal tubules are visualized). And in protocol 1 (EDTA 17% + 3% NaOCL), the distribution of points was only within 4–5 (abundant smear layer, dentinal tubules are not visualized). The data was presented in Fig. .
Elimination of microorganisms in the root canal system during treatment is an important task in endodontic treatment . This is why removing the smear layer from the root canal walls is so important since microbiota may remain in the thickness of the debris. The use of etidronate is a promising direction for further research. A distinctive feature is the lack of hypochlorite neutralization . Based on the results of this study, it can be assumed that treatment with a 17% EDTA solution is less effective in removing the smear layer. According to Yang. G. et al. 2008, the apical part of the root canal most often remains untreated when irrigated with 17% EDTA . Also, 17% EDTA neutralizes the alkaline environment of 3% NaOCL, reducing its antiseptic activity. Additionally, the literature indicates the erosive effect of 17% EDTA on peritubular dentin . In combination with an obviously better removal of the smear layer in the apical part, etidronate becomes preferable in the endodontic treatment . Similar results were obtained in a study conducted by Aoun. C. et al., 2023, when comparing the efficiency of SL removal in groups using 17% EDTA and groups using etidronate, the best result was obtained after etidronate irrigation . The protocol used etidronate diluted in 3% NaOCL is the most convenient for clinical use. The use of a solution that simultaneously has the properties of an antiseptic and a chelating agent optimizes the workflow. Additionally, the risk of losing the working length due to obturation of part of the canal with sawdust is reduced . Many authors obtained data similar to ours when studying etidronate in endodontic treatment. However, in their studies, they used etidronate as a finishing irrigant and investigated the effect of sound and ultrasound activation on the effectiveness of the irrigant . However, many studies evaluated the etidronate as finial irrigation solution, but not constant or sequential, after each endodontic instrument, as proposed in our study, specifically for the prevention of debris accumulation in the isthmuses and dentinal tubes . An important factor is patients’ safety during treatment. That is why, in addition to efficiency, it is necessary to consider the safety of irrigants. Given the high activity of 5.25% NaOCL, we propose using etidronate dilution in 3% NaOCL to prevent complications and maintain high bactericidal activity . Thus, the use of etidronate can be an alternative to irrigation with EDTA due to the high efficiency of SL removal and optimization of the irrigation process. The use of etidronate as a chelating solution allows for almost complete removal of the smear layer along the entire length of the root canal, which increases the efficiency of irrigation.
The smear layer formed during mechanical treatment of the root canal was removed during irrigation with solutions from all the stated protocols. Using etidronate solution protocol increases the efficiency of smear layer removal during root canal treatment by 4 times compared to EDTA.
|
Multimorbidity, health Literacy, and quality of life among older adults in an urban slum in India: a community-based cross-sectional study | 105cc79f-c381-4310-b4bf-d5f5dcf8d71b | 11234527 | Health Literacy[mh] | Multimorbidity, defined as the coexistence of two or more chronic conditions in one person, is increasingly common among older adults globally . The prevalence of multimorbidity increases with age and is greater in low- and middle-income countries than in high-income countries . A multi-country population-based study stretching across low-, middle-, and high-income countries found that the prevalence of multimorbidity increases with age . However, there was heterogeneity in the estimates based on setting, participant age group, and the number and type of chronic conditions included. In India, the burden of multimorbidity is expected to rise dramatically due to the rapidly aging population coupled with the epidemiological transition from communicable to noncommunicable diseases . Studies from both urban and rural parts of India have shown a high prevalence of multimorbidity among older adults, ranging from 55 to 83% . The most common chronic disase included hypertension, diabetes, heart disease, chronic respiratory conditions, musculoskeletal disorders, and mental health conditions. With the accumulation of multiple chronic conditions, older adults are at increased risk of adverse health outcomes, including declines in physical and cognitive functioning, poor quality of life, and increased healthcare utilization . Multimorbidity has been associated with lower health-related quality of life across different populations . An impaired quality of life leads to a loss of independence, social isolation, and greater demands on family members as caregivers . In India, there are wide urban-rural and socioeconomic disparities in access to healthcare and social support systems for older adults with multimorbidity. Those living in urban slums are especially vulnerable due to poverty, substandard housing, lack of infrastructure, and barriers to healthcare access in these informal settlements . The challenges of managing multimorbidity are greater for slum dwellers because of high out-of-pocket expenditures for health services and medications . Despite the growing size of this vulnerable population, there is limited community-based data on the burden and impact of multimorbidity among older adults in urban slums in India. Most related studies have been conducted in community or hospital settings, with an underrepresentation of urban slum populations. Given their deprived living conditions and lack of social protection for healthcare, older slum dwellers likely experience a disproportionately greater burden of multimorbidity and related adverse consequences. There is a need for representative data on the prevalence of multimorbidity and its relationship with health-related quality of life in this urban slum population. This approach can help identify high-risk groups and modifiable factors to inform targeted interventions and appropriate health services for multimorbidity management. Health literacy, defined as the degree to which individuals can obtain, process, and understand basic health information needed to make appropriate decisions, is an important factor in the prevention and management of chronic diseases. However, there is limited data on the association between health literacy and multimorbidity burden, especially among vulnerable populations like urban slum dwellers in India. Therefore, the present study aimed to (i) assess the prevalence of multimorbidity among older adults living in an urban slum, (ii) examine the association of multimorbidity with health literacy and quality of life in this population, (iii) identify high-risk groups based on sociodemographic factors. We hypothesized that multimorbidity would be highly prevalent in this population and associated with poor health literacy and quality of life. This is the first study, to our knowledge, that aims to assess health literacy related to multimorbidity among urban slum dwellers aged ≥ 65 years.
Study design and setting This was a community-based cross-sectional study conducted in an urban slum located in Gujarat between April 2023- Dec2023. This slum has a population of approximately 50,000 residing across 20 municipal wards. The residents belong predominantly to lower socioeconomic status groups and face challenges such as poverty, inadequate housing, poor sanitation, and limited access to healthcare service. Sample size calculation Considering a prevalence of 45% based on previous Indian studies , with 5% absolute precision and a design effect of 2 for the multistage sampling, the sample size was calculated using the formula: n = Z2*P(1-P)/d2 *design effect Where, Z = 1.96 at 95% confidence interval P = 45% = 0.45, d = 5% = 0.05 Design effect = 2 Plugging in the values: n = (1.96)2 × 0.45(1-0.45) / (0.05)2 × 2 n = 768, the final minimum sample size was estimated rounded up to be 800. The sampling technique employed was a multistage random sampling approach. In the first stage, four out of the approximately 20 municipal wards in the urban slum area were randomly selected using the lottery method. Subsequently, systematic random sampling was applied in each of the four chosen wards to select households. A rough sketch map was used to guide the process, with every 5th household included in the sample. In the third stage, within each selected household, eligible individuals (aged ≥ 65 years) were listed, and one older adult was randomly chosen using the lottery method. If the initially selected participant was not available at home after two visits, the next older adult from that household was approached. If no eligible individual was available in the selected household, the adjacent household was approached using the lottery method again. The eligibility criteria included age ≥ 65 years, residence in the selected households, and providing informed consent, while the exclusion criteria included inability to communicate, bedridden, or unwilling to participate. Data collection In this research endeavor, a meticulously designed pretested interviewer-administered questionnaire served as the primary tool for data collection. The questionnaire elicited sociodemographic information from participants, providing insights into the studied population. Additionally, participants self-reported chronic conditions, providing information on health issue prevalence. The chronic conditions elicited included hypertension, diabetes, heart disease, stroke, chronic respiratory diseases, musculoskeletal disorders, neurological disorders, mental health conditions, cancer, and others. The questionnaire tool used to collect this data was adapted from validated instruments used in previous Indian and global studies on multimorbidity. It was pretested in a pilot study for comprehensibility, validity, and reliability before use in this study. We have provided the final questionnaire as a supplementary file . The Short Form-12 (SF-12) assessed health-related quality of life . This widely recognized instrument evaluates physical and mental health. Furthermore, the 47-item Short Form Health Literacy Scale (HLS-SF-47) was administered to measure participants’ health literacy across different domains . Anthropometric data such as height, weight, and blood pressure were also recorded. Physical activity level: Assessed using the Global Physical Activity Questionnaire (GPAQ) . Smoking status: Categorized as a non-smoker, current smoker, or former smoker. Alcohol use: Drinking frequency and number of drinks per occasion. Dietary patterns: Assessed using a food frequency questionnaire . Social support/living arrangements: Measured using the Multidimensional Scale of Perceived Social Support (MSPSS) . Healthcare access: Based on the distance to the nearest health facility and reported barriers to healthcare. (Limited healthcare access: Defined based on the distance to the nearest health facility (> 5 km) and self-reported barriers to healthcare access including lack of transportation, inability to pay fees, and lack of social support to attend appointments. Access was categorized as adequate if the nearest facility was within 5 km and no barriers were reported, moderately accessible if the facility was > 5 km but no other barriers, and limited access if the facility was > 5 km and participants reported ≥ 1 barrier). To ensure high-quality data collection, all investigators and field workers were thoroughly trained in the study objectives, methodology, and use of the data collection tools. The training also emphasized building rapport, confidentiality, and ethical conduct throughout the data-gathering process. Data were checked regularly in the field for completeness and accuracy. Any unclear or missing responses were verified and corrected in the field itself by revisiting the concerned households. Table summarizes the key-dependent, independent, and covariate variables that were measured in the study, along with the tools used to assess each variable. The table categorizes the variables into dependent (multimorbidity), independent (health literacy, physical activity, social support, quality of life), and covariates including sociodemographic factors, health behaviors, and healthcare access. Data analysis The collected data were meticulously processed and analyzed using the Statistical Package for Social Sciences (SPSS version 26). Socioeconomic status was measured using the modified BG Prasad classification based on the consumer price index for the study year. This tool classifies SES into upper, upper middle, middle, lower middle, and lower categories based on the monthly per-capita income . The prevalent disease clusters were identified based on the chronic condition combinations reported by the study participants with multimorbidity. The three most frequently occurring combinations were categorized as the common multimorbidity clusters in this population. The SF-12 scale was analyzed by calculating the physical and mental component summary scores, which range from 0 to 100 with higher scores indicating better health-related quality of life. The physical and mental component scores were computed using standard scoring algorithms and compared between older adults with and without multimorbidity using the independent samples t-test. Descriptive statistics were used to summarize the prevalence of multimorbidity and the pattern of chronic disease clusters in the study population. Bivariate analyses using t-tests and chi-square tests were conducted to compare health-related quality of life between older adults with and without multimorbidity.” Multivariable logistic regression analysis was employed to identify factors associated with multimorbidity after adjusting for sociodemographic covariates. The model is represented by: Log(p/1 − p) = b0 + b1 × 1 + b2 × 2 + …. + bpXp. Where p is the probability of having multimorbidity, b0 is the constant, b1 to bp are regression coefficients, and X1 to Xp are explanatory variables including health literacy, physical activity, social support, and sociodemographic factors. The significance level was set at p < 0.05, ensuring a robust statistical threshold. Ethical consideration This study started after ethical clearance was obtained from the Institutional Ethics Committee. (REF No: 216/03/23). Written Informed consent was obtained first after the purpose of the study was explained, and participants were not obliged to answer any questions they did not like or were free to terminate the interview at any given time. Assurance was given that confidentiality concerning their information would be strictly maintained.
This was a community-based cross-sectional study conducted in an urban slum located in Gujarat between April 2023- Dec2023. This slum has a population of approximately 50,000 residing across 20 municipal wards. The residents belong predominantly to lower socioeconomic status groups and face challenges such as poverty, inadequate housing, poor sanitation, and limited access to healthcare service.
Considering a prevalence of 45% based on previous Indian studies , with 5% absolute precision and a design effect of 2 for the multistage sampling, the sample size was calculated using the formula: n = Z2*P(1-P)/d2 *design effect Where, Z = 1.96 at 95% confidence interval P = 45% = 0.45, d = 5% = 0.05 Design effect = 2 Plugging in the values: n = (1.96)2 × 0.45(1-0.45) / (0.05)2 × 2 n = 768, the final minimum sample size was estimated rounded up to be 800. The sampling technique employed was a multistage random sampling approach. In the first stage, four out of the approximately 20 municipal wards in the urban slum area were randomly selected using the lottery method. Subsequently, systematic random sampling was applied in each of the four chosen wards to select households. A rough sketch map was used to guide the process, with every 5th household included in the sample. In the third stage, within each selected household, eligible individuals (aged ≥ 65 years) were listed, and one older adult was randomly chosen using the lottery method. If the initially selected participant was not available at home after two visits, the next older adult from that household was approached. If no eligible individual was available in the selected household, the adjacent household was approached using the lottery method again. The eligibility criteria included age ≥ 65 years, residence in the selected households, and providing informed consent, while the exclusion criteria included inability to communicate, bedridden, or unwilling to participate.
In this research endeavor, a meticulously designed pretested interviewer-administered questionnaire served as the primary tool for data collection. The questionnaire elicited sociodemographic information from participants, providing insights into the studied population. Additionally, participants self-reported chronic conditions, providing information on health issue prevalence. The chronic conditions elicited included hypertension, diabetes, heart disease, stroke, chronic respiratory diseases, musculoskeletal disorders, neurological disorders, mental health conditions, cancer, and others. The questionnaire tool used to collect this data was adapted from validated instruments used in previous Indian and global studies on multimorbidity. It was pretested in a pilot study for comprehensibility, validity, and reliability before use in this study. We have provided the final questionnaire as a supplementary file . The Short Form-12 (SF-12) assessed health-related quality of life . This widely recognized instrument evaluates physical and mental health. Furthermore, the 47-item Short Form Health Literacy Scale (HLS-SF-47) was administered to measure participants’ health literacy across different domains . Anthropometric data such as height, weight, and blood pressure were also recorded. Physical activity level: Assessed using the Global Physical Activity Questionnaire (GPAQ) . Smoking status: Categorized as a non-smoker, current smoker, or former smoker. Alcohol use: Drinking frequency and number of drinks per occasion. Dietary patterns: Assessed using a food frequency questionnaire . Social support/living arrangements: Measured using the Multidimensional Scale of Perceived Social Support (MSPSS) . Healthcare access: Based on the distance to the nearest health facility and reported barriers to healthcare. (Limited healthcare access: Defined based on the distance to the nearest health facility (> 5 km) and self-reported barriers to healthcare access including lack of transportation, inability to pay fees, and lack of social support to attend appointments. Access was categorized as adequate if the nearest facility was within 5 km and no barriers were reported, moderately accessible if the facility was > 5 km but no other barriers, and limited access if the facility was > 5 km and participants reported ≥ 1 barrier). To ensure high-quality data collection, all investigators and field workers were thoroughly trained in the study objectives, methodology, and use of the data collection tools. The training also emphasized building rapport, confidentiality, and ethical conduct throughout the data-gathering process. Data were checked regularly in the field for completeness and accuracy. Any unclear or missing responses were verified and corrected in the field itself by revisiting the concerned households. Table summarizes the key-dependent, independent, and covariate variables that were measured in the study, along with the tools used to assess each variable. The table categorizes the variables into dependent (multimorbidity), independent (health literacy, physical activity, social support, quality of life), and covariates including sociodemographic factors, health behaviors, and healthcare access.
The collected data were meticulously processed and analyzed using the Statistical Package for Social Sciences (SPSS version 26). Socioeconomic status was measured using the modified BG Prasad classification based on the consumer price index for the study year. This tool classifies SES into upper, upper middle, middle, lower middle, and lower categories based on the monthly per-capita income . The prevalent disease clusters were identified based on the chronic condition combinations reported by the study participants with multimorbidity. The three most frequently occurring combinations were categorized as the common multimorbidity clusters in this population. The SF-12 scale was analyzed by calculating the physical and mental component summary scores, which range from 0 to 100 with higher scores indicating better health-related quality of life. The physical and mental component scores were computed using standard scoring algorithms and compared between older adults with and without multimorbidity using the independent samples t-test. Descriptive statistics were used to summarize the prevalence of multimorbidity and the pattern of chronic disease clusters in the study population. Bivariate analyses using t-tests and chi-square tests were conducted to compare health-related quality of life between older adults with and without multimorbidity.” Multivariable logistic regression analysis was employed to identify factors associated with multimorbidity after adjusting for sociodemographic covariates. The model is represented by: Log(p/1 − p) = b0 + b1 × 1 + b2 × 2 + …. + bpXp. Where p is the probability of having multimorbidity, b0 is the constant, b1 to bp are regression coefficients, and X1 to Xp are explanatory variables including health literacy, physical activity, social support, and sociodemographic factors. The significance level was set at p < 0.05, ensuring a robust statistical threshold.
This study started after ethical clearance was obtained from the Institutional Ethics Committee. (REF No: 216/03/23). Written Informed consent was obtained first after the purpose of the study was explained, and participants were not obliged to answer any questions they did not like or were free to terminate the interview at any given time. Assurance was given that confidentiality concerning their information would be strictly maintained.
Table shows the sociodemographic characteristics of the 800 study participants. Frequencies and percentages are presented for the categories of age, sex, religion, marital status, education level, and socioeconomic status (SES). The prevalence of multimorbidity (defined as ≥ 2 chronic conditions) among the 800 study participants is presented in Table . Overall, 500 participants (62.5%) were found to have multimorbidity. Table presents the prevalent disease clusters among older adults with multimorbidity in the urban slum setting. The most common combination was hypertension paired with diabetes, affecting 32% (160 participants) of those with multimorbidity. The second most frequent cluster was hypertension combined with osteoarthritis, observed in 24% (120 participants) of the multimorbid group. Other notable disease combinations included diabetes with heart disease (16%), respiratory disease with heart disease (12%), and depression with osteoarthritis (8%). Less common pairings were diabetes with stroke (4%) and heart disease with cancer (2%). Interestingly, a small proportion (2%) of participants exhibited a triad of conditions: hypertension, diabetes, and osteoarthritis. These findings highlight the complex interplay of chronic conditions in this population, with cardiovascular and metabolic disorders frequently co-occurring. The prevalence of these specific disease clusters underscores the need for integrated care approaches that address multiple chronic conditions simultaneously in older adults residing in urban slums. Table presents the comparison of health-related quality of life, as measured by the SF-12 physical and mental component summary scores, and health literacy, as measured by the HLS-SF-47, between older adults with and without multimorbidity. The mean SF-12 physical component score was significantly lower for those with multimorbidity (39.7 ± 6.5) compared to those without multimorbidity (42.5 ± 5.2), with a p -value < 0.001. Similarly, the mean SF-12 mental component score was significantly lower for the multimorbidity group (45.3 ± 8.9) versus the non-multimorbidity group (49.2 ± 7.1), with p < 0.001. This indicates that the presence of multimorbidity was associated with poorer physical and mental quality of life. For health literacy, the multimorbidity group had a significantly lower mean HLS-SF-47 score (24.7 ± 6.2) than the non-multimorbidity group (32.1 ± 5.8), p < 0.001, showing that increased multimorbidity was correlated with lower health literacy. Additionally, the multimorbidity group had significantly lower social support, as measured by lower mean MSPSS scores (60.3 ± 13.5) compared to the non-multimorbidity group (71.5 ± 11.2), p < 0.001. Overall, Table highlights the significant negative impact of multimorbidity on health-related quality of life across physical, mental, and social health domains in this older adult urban slum population. Table shows the Pearson correlation between health literacy (HLS-SF-47) and quality-of-life scores on the SF-12 scale. Table shows the Multivariable logistic regression analysis was conducted to identify factors associated with multimorbidity. Older age (per year increase) was associated with greater odds of multimorbidity (AOR 1.05, 95% CI 1.02–1.09). Female gender (AOR 1.86, 95% CI 1.12–3.08), widowed status (AOR 2.05, 95% CI 1.15–3.65), no formal education (AOR 3.12, 95% CI 1.52–6.41), lower socioeconomic status (AOR 2.35, 95% CI 1.22–4.52), being a current smoker (AOR 2.35, 95% CI 1.67–3.46) or former smoker (AOR 2.15, 95% CI 1.59–4.23), physical inactivity (AOR 1.68, 95% CI 1.027–2.77), and lack of social support (AOR 1.57, 95% CI 1.01–2.45) also increased the likelihood of multimorbidity. For every 1 unit increase in the health literacy score, the odds of having multimorbidity decrease by 19% (AOR 0.81, 95% CI 0.78–0.91). Additionally, limited healthcare access was associated with higher odds of multimorbidity (AOR 2.49, 95% CI 1.88–4.27). In summary, the odds of multimorbidity were positively associated with older age, female sex, being widowed, lower levels of education, smoking, physical inactivity, lack of social support, lack of health literacy, and limited access to healthcare. The analysis highlights the impact of sociodemographic disparities on multimorbidity risk in the study population. Targeted interventions to address modifiable risk factors like health literacy and social support may help reduce the burden of multimorbidity among vulnerable older adults.
In this community-based cross-sectional study, we found a high prevalence of multimorbidity (≥ 2 chronic conditions) affecting more than 60% of older adults residing in urban slum areas. The most prevalent chronic disase were hypertension, diabetes, musculoskeletal disorders, respiratory diseases, and mental health issues. Multimorbidity was significantly associated with lower quality of life, with older adults reporting poorer physical and mental health on the SF-12 scale. Our findings on the high burden of multimorbidity align with those of previous studies in India, which reported a prevalence ranging from 55 to 65% among community-dwelling older adults in urban slums . The pattern of common chronic conditions observed in this urban slum population also conforms to the epidemiological transition underway in urban regions . With continuing demographic and lifestyle changes, India is facing escalating burdens of noncommunicable diseases manifesting as multimorbidity among its rapidly growing older adult population. The two most prevalent clusters were hypertension paired with diabetes (in 80 participants, 32%) and hypertension paired with osteoarthritis (in 60 participants, 24%). These patterns align with multimorbidity data from previous studies in India that also noted hypertension, diabetes, cardiovascular disease, and musculoskeletal disorders as the predominant co-occurring chronic conditions among older adults . The high prevalence of certain clusters emphasizes the need to strengthen the integrated screening and management of comorbid conditions such as diabetes and hypertension that tend to coexist and negatively impact outcomes. Tackling common modifiable risk factors and addressing disease combinations through a patient-centered approach can help reduce the burden of multimorbidity as the population ages. The strong inverse association between multimorbidity and quality of life is consistent with reports across diverse global settings . Managing multiple chronic conditions simultaneously has a detrimental additive effect on physical capacities, psychological well-being, social relationships, and independence in daily living. Multimorbidity also results in complex healthcare needs and polypharmacy, which older adults in resource-constrained slums are ill-equipped to handle. Their poor living conditions, limited access to health services, and lack of social protection exacerbate the challenges of multimorbidity. A key finding was the high prevalence of inadequate health literacy associated with multimorbidity. After adjusting for sociodemographic variables, the odds ratio of 0.81 indicates that for every 1 unit increase in the health literacy score, the odds of having multimorbidity decrease by 19%. This finding aligns with prior research showing that health literacy is an independent predictor of multimorbidity . Low health literacy can impede the self-management of chronic diseases, medication adherence, and utilization of preventive services . Enhancing health literacy through community education and capacity building may help reduce the risk and effects of multimorbidity among vulnerable older adult people. The Multivariable regression analysis showed that inadequate health literacy, lack of physical activity, and lack of social support were significantly associated with a higher likelihood of multimorbidity in this urban slum population. These findings align with prior studies demonstrating the role of health literacy and social determinants in multimorbidity risk. A systematic review found that low health literacy was associated with greater multimorbidity prevalence in several studies . Other research has linked social isolation and poorer social support with an increased number of chronic conditions among older adults . Finally, a cohort study in Brazil concluded that insufficient physical activity was predictive of developing multimorbidity over a 2-year follow-up period . Taken together, these modifiable factors related to health behaviors, capacities, and social environment appear to contribute significantly to the development of multimorbidity, even after accounting for sociodemographic characteristics. Our study provides representative data on the prevalence of multimorbidity and its impact on quality of life, specifically among older residents of an Indian urban slum - an underserved population often excluded from national health surveys. This highlights the disproportionate multimorbidity burden imposed by socioeconomically marginalized older adult groups dwelling in informal settlements. The much higher prevalence compared to rural counterparts implies urban slum conditions like concentrated poverty, lack of infrastructure, and constrained healthcare access could exacerbate the development of multiple chronic illnesses. The cumulative out-of-pocket expenditure for healthcare poses catastrophic financial risks to low-income slum households already struggling to meet daily necessities. Older adults with multimorbidity likely face prohibitive barriers in affording treatment and medications over long periods. Many are forced to choose between healthcare costs and other basic needs, which further worsens their disease prognosis and quality of life. Their social vulnerability and lack of financial protection mechanisms make it difficult to effectively manage complex healthcare needs arising from multimorbidity. Targeted interventions to alleviate the multimorbidity burden among the urban poor should address context-specific social determinants of health-spanning living conditions, access to health services, community awareness, and support systems. Health policies must recognize deprived urban groups and provide tailored financial risk protection alongside prevention and screening initiatives. Limitations of this study include the cross-sectional design, which restricts causal inference about the association between multimorbidity and quality of life. The reliance on self-reported diagnoses of chronic conditions could result in underreporting. We selected only one slum area, which may limit the generalizability of the findings to other urban slums that differ substantially in their demographic composition and health profiles. Nonetheless, the study provides novel insights into the vulnerability of older slum dwellers to multimorbidity and its adverse effects. Recommendations Integrated screening and management programs for multimorbidity should be implemented in urban slums targeting older adults. Affordable primary care and geriatric services need to be made accessible within slum settings. Public health policies and interventions must address social determinants such as education, financial security, and living conditions in slums. Family members and caregivers of older adults with multimorbidity require training and support. Community awareness of healthy lifestyles, preventive behaviors, and self-care should be created. Health literacy of older adults in urban slums should be improved through community education programs. Social support systems and financial protection mechanisms are needed for vulnerable older adult groups.
Integrated screening and management programs for multimorbidity should be implemented in urban slums targeting older adults. Affordable primary care and geriatric services need to be made accessible within slum settings. Public health policies and interventions must address social determinants such as education, financial security, and living conditions in slums. Family members and caregivers of older adults with multimorbidity require training and support. Community awareness of healthy lifestyles, preventive behaviors, and self-care should be created. Health literacy of older adults in urban slums should be improved through community education programs. Social support systems and financial protection mechanisms are needed for vulnerable older adult groups.
Multimorbidity among older adults in urban slums requires urgent policy attention and action. A multipronged strategy should focus on both preventive and management aspects, spanning health promotion, community-based screening, affordable primary care, geriatric services, and social assistance. Tackling socioeconomic deprivation alongside lifestyle risks and timely disease management can help reduce the multimorbidity burden and improve the quality of life among marginalized older adult people in urban India. The high prevalence of inadequate health literacy associated with multimorbidity suggests low health awareness and self-care capacities among urban older adult slum dwellers. Targeted interventions to improve health literacy through community outreach, patient education, simplified treatment guidelines, and capacity building of family caregivers are essential.
Below is the link to the electronic supplementary material. Supplementary Material 1
|
Enhanced Bacterial and Biofilm Adhesion Resistance of ALD Nano-TiO | 52c44608-0aad-456e-9f00-cd1b26db8cee | 11537173 | Dentistry[mh] | Dental implant restorations have emerged as the preferred therapeutic modality for rehabilitating missing teeth, wherein the dental abutment stands as a pivotal component crucial to the soft tissue sealing. Unfortunately, unlike the strong soft tissue sealing surrounding natural teeth, dental abutments exhibit significantly lower functional sealing performance. This discrepancy arises due to the fiber orientation being parallel to the abutment surface and the limited blood supply, predominantly stemming from thin connective tissue. Consequently, these conditions create a conducive environment for bacterial invasion and colonization on the surfaces of dental abutments. As biofilms migrate further, a cascade of tissue inflammation around the dental abutment is triggered. This process can result in localized mucositis and, ultimately, advance to peri-implantitis. In contemporary clinical practice, commonly utilized implant abutment materials comprise titanium (Ti) and zirconia (ZrO 2 ). In a meta-analysis encompassing 19 clinical randomized controlled studies over an average observation period of 36 months, Sanz-Martin et al observed a notable increase in mucosal inflammation reactions associated with Ti abutments compared to the ZrO 2 ones. The reasons for this discrepancy lie in the challenges posed by bacterial plaque adhesion on ZrO 2 surfaces. This underscores the direct link between bacterial adhesion on implant abutment surfaces and the choice of materials. However, compared to Ti, the relatively low mechanical strength limits the widespread clinical application of ZrO 2 . Thus, to mitigate the formation of biofilms and plaque accumulation, facilitating antibacterial ability on dental Ti abutments could prove beneficial. Given this context, nanometer titanium dioxide (nano-TiO 2 ) coatings exhibit broad-spectrum effectiveness against diverse microorganisms, encompassing Gram-positive and Gram-negative bacteria, among other types of microbes. Recent investigations have highlighted the ability of nano-TiO 2 coatings on dental implant materials to elicit an antibacterial effect on common bacteria, usually with a rough surface. It’s worth noting that, surface treatments such as sandblasting and acid etching are commonly employed in the manufacturing of dental titanium implants to enhance surface roughness, facilitating osteoblast adhesion and proliferation, thereby improving osseointegration. However, these treatments also increase bacterial adhesion. In contrast, dental Ti abutments require a relatively smooth surface to promote the adhesion of surrounding soft tissue. Consequently, machining and polishing are the preferred treatment for the emergence profile of dental Ti abutments at the manufacture, although it still offers limited antibacterial properties. There are two methods for preparing nano-TiO 2 coatings on dental Ti abutment materials. One is the anodic oxidation (AO) technology, a conventional and widespread coating preparation method, which has found extensive use in creating nano-TiO 2 coatings on surfaces of Ti-based materials like abutments (eg, Dentsply Astra Tech). This process involves utilizing a metal as the anode in conjunction with an electrolyte such as sulfuric acid or phosphoric acid. When an external electric current is applied, oxidation reactions take place on the metal surface, resulting in the formation of an oxide layer. Moreover, this process can alter the surface characteristics and properties of the metal, encompassing factors such as surface color, biocompatibility, and corrosion resistance. Studies indicated that nano-TiO 2 coatings produced through AO can enhance the hydrophilicity of Ti abutment surfaces and facilitate the formation of anatase crystal structure TiO 2 , consequently reducing early bacterial colonization on these surfaces. However, findings from Hall et al, in a randomized controlled clinical study, revealed that compared to untreated Ti abutments, AO-modified Ti abutment surfaces did not significantly diminish bacterial biofilm formation. Nonetheless, they did note an enhancement in soft tissue integration on the material’s surface. Consequently, the debate surrounding the antibacterial efficacy of AO-modified Ti abutments persists. Another technology for the preparation of nano-TiO 2 coatings on dental Ti abutment materials is atomic layer deposition (ALD) technology, which has been described by our previous study. It involves sequentially depositing materials in the form of atomic layers onto the surface of a substrate using alternating pulses of gas-phase precursors. This method can create uniform nano-scale coatings with controllable thickness and strong adhesion on micro or complex structures at relatively low temperatures, showcasing its technological superiority compared to the AO technology. Studies highlighted ALD’s capability in enhancing surface biocompatibility and curbing bacterial adhesion by modifying surface chemistry and nano structural morphology. This allows for precise control over surface nanostructure and surface energy. , , Based on these advantages, nano-TiO 2 coatings synthesized via ALD exhibit promising antibacterial effectiveness across various applications, including orthopedic implants and denture base materials. Our previous study also has indicated the ALD-TiO₂ coating with the thickness of 100 nm not only reduced bacterial adhesion but also exhibited favorable biocompatibility, corrosion resistance, and a warm yellow hue, enhancing the aesthetics of dental abutments. Thus, using ALD to create nano-TiO 2 coatings on dental Ti abutment surfaces holds significant clinical potential for imparting excellent antibacterial properties. However, whether the antibacterial effect of nano-TiO 2 coatings prepared by ALD is better than those produced by AO remains unknown. The interface between the abutment and soft tissue, serving as the primary barrier against bacterial intrusion, is influenced not only by material surface properties but also by the diversity of bacteria involved. Both Gram-positive and Gram-negative bacteria adhere and form biofilms through a multi-step process involving initial adhesion, irreversible attachment, maturation, and dispersion. , Initial bacterial attachment varies based on surface properties, while irreversible attachment is mediated by specific bacterial surface components. The irreversible attachment is facilitated by surface structures like MSCRAMMs in Gram-positive bacteria and pili in Gram-negative bacteria, while extracellular polymeric substances (EPS) form a protective matrix. Gram-negative bacteria rely on lipopolysaccharides (LPS), and Gram-positive bacteria on teichoic acids (TA), both of which provide surface charges that influence adhesion and biofilm formation. Quorum sensing (QS) regulates biofilm growth and dispersal, with Gram-positive bacteria using autoinducing peptides (AIP) and Gram-negative bacteria employing acyl-homoserine lactones (AHL). Furthermore, Gram-positive bacteria boast thicker peptidoglycan layers, granting them higher resistance to reactive oxygen species (ROS), while the lipid layers and lipopolysaccharides of Gram-negative bacteria are more susceptible to disruption via photocatalytic means. Additionally, the oral microbial community exhibits considerable diversity and complexity, engaging in interactions such as symbiosis, competition, and antagonism. Regarding the mixed-species models, Dorkhan et al found bacterial adhesion to anodized Ti surfaces involving early colonizers ( oral streptococci, viridans streptococci, Aggregatibacter actinomycetemcomitans , and Streptococcus sanguinis ) revealed reduced adhesion for mixed-species and single-species after 2 hours compared to commercially pure Ti. However, the reduction in mixed-species adhesion did not yield statistically significant differences. Similarly, Mouratidou et al established single-species bacterial biofilms ( Actinomyces naeslundii, Fusobacterium nucleatum , and Streptococcus gordonii ) alongside mixed-species biofilms to explore antibiotic sensitivity. Their findings highlighted that mixed-species biofilms displayed reduced antibiotic sensitivity compared to single-species biofilms. Consequently, developing models involving both single-species and mixed-species biofilms becomes indispensable in studying the antibacterial effectiveness of abutment materials, providing a more accurate comprehension of the intricate interplay between oral microbiota and diverse materials. Therefore, the present study aims to compare the surface properties of nano-TiO 2 coatings applied to dental Ti abutments, fabricated by ALD and AO techniques. This evaluation will be assessed in comparison with ZrO 2 and Ti surfaces. To further investigate the bacterial and biofilm adhesion resistance of these four surfaces, in vitro bacterial models were employed, encompassing single-species and mixed-species compositions of three peri-implantitis-associated bacteria, such as the Gram-negative obligate anaerobe P. gingivalis , and Gram-positive obligate anaerobe S. mutans , and S. aureus . The results of this investigation would establish the foundational knowledge regarding the performance of these materials as dental abutment materials in the oral cavity and offer recommendations for dental abutment materials to help prevent peri-implantitis. The hypotheses under investigation are as follows: (1) Distinct differences exist in surface properties among the four surfaces. (2) Discrepancies are observed in the bacterial adhesion of S. mutans, S. aureus , and P. gingivalis on these four surfaces. (3) Variances emerge in the adhesion of mixed-species bacteria to the four materials compared to the adhesion observed with single-species compositions.
This study did not involve humans or animals and did not require ethics approval. Preparation of Materials In this study, two dental materials, Ti (Ti6Al4V, Nissin, China) and ZrO 2, (Aidite, China) were utilized. Test samples were designed as disk-shaped specimens (10.0 mm in diameter, 2.0 mm in thickness) and they were fabricated by the computer-aided design and computer-aided manufacturing (CAD-CAM) mill. All specimens were wet-polished with silicon carbide sandpapers to grade #7000. Subsequently, they were sequentially cleaned and dried with acetone, ethanol, and deionized water using ultrasonication. That were Ti group and ZrO 2 group. And then, two methods were used to prepare the nano-TiO 2 coatings on Ti surface. The first method, ALD technology, was described as follows: the samples were placed in the ALD (TALD-06G, Jiaxingkeming, China), operating in thermal mode. The nano-TiO 2 coatings were then applied to the Ti substrates using Titanium tetrakis dimethylamide (TDMAT) and H 2 O as precursors, along with argon (99.999%) as the purge gas. Each ALD cycle in this study consisted of a 0.1-second TDMAT, a 25-second argon purge, a 0.02-second H 2 O purge, and another 25-second argon purge. The nano-TiO 2 coatings were grown at a temperature of 260°C, with a total of approximately 1625 cycles prepared. And then the nano-TiO 2 coatings with the thickness of 100 nm were produced. That were Ti+ALD group . Following the treatments, the samples were rinsed with deionized water and the hot air dried. , The second method employed in this study involved the utilization of AO technology. Electrochemical staining was conducted on Ti samples using a 1M phosphoric acid electrolytic solution within a temperature-regulated enclosure (DH1719A-5 model, Dahua, China). The Ti was affixed to the anode, while a 20.0 mm × 30.0 mm stainless steel plate served as the cathode, positioned at a distance of 2 cm from the anode. A constant voltage of 60 V was applied through a direct current stabilizing power supply (DH1719A-5 model) for a duration of 60 seconds to facilitate AO. The specific reaction process is outlined as follows: Upon application of an appropriate oxidation voltage, the anode sheds electrons, generating Ti 4+ ions according to formula (1). Simultaneously, O 2- ions and OH − ions manifest at the cathode as indicated in formula (2). These resulting ions migrate via the electric field and electrolyte transport. The interaction of Ti 4+ ions with O 2- ions leads to the formation of TiO 2 at the titanium/electrolyte interface, outlined in formula (3). That were Ti+AO group . Following the treatments, the samples were rinsed with deionized water and hot air dried. , (1) [12pt]{minimal}
[substack]{amsmath}
[mathscr]{eucal}
{linotext }{}
$${} - 4{{}^ - } {}{{}^{4 + }}$$ (2) [12pt]{minimal}
[substack]{amsmath}
[mathscr]{eucal}
{linotext }{}
$$2{}{{}^ - } + 2{{}^ - } 2{{}^{2 - }} + {{}_2} $$ (3) [12pt]{minimal}
[substack]{amsmath}
[mathscr]{eucal}
{linotext }{}
$${}{{}^{4 + }} + 2{{}^{2 - }} {}{{}_2}$$ Subsequently, all the samples from four groups were stored after sterilization with ethylene oxide gas. Surface Properties Chemical Analysis The chemical compositions of the randomly selected samples from Ti, Ti+AO and Ti+ALD groups were determined by the X-ray photoelectron spectroscopy (XPS) system (Thermo Scientific K-Alpha+, Thermo Fisher, USA). Crystalline Forms Analysis The crystalline forms of the randomly selected samples from Ti, Ti+AO, and Ti+ALD groups were performed using X-ray diffraction (XRD) (Miniflex 600, Rigaku, Japan) equipped with Cu Kα radiation. Surface Topography and Roughness Analysis The randomly selected samples from four groups were subjected to surface topography and roughness analysis. The scanning electron microscope (SEM) device (Sigma 300, Zeiss, Germany) was used for surface topography evaluation. The non-conductive ZrO 2 samples were gold-sputtered, and then the samples were placed in the SEM device at magnifications of × 5.00 K, and × 20.00 K. Atomic Force Microscope (AFM) (Dimension Icon, Bruker, USA) was used for roughness analysis. Hydrophilicity The hydrophilicity of the samples from each group ( n = 6) was assessed through water contact angle (WCA) measurements. Using an optical contact angle meter (SDC-100, Shengding, China), a 4 μL drop of distilled water was placed on the specimen’s surface. Final contact angle values were determined by averaging three measurements taken at different parts of the surfaces. Antibacterial Assays Single-Species Biofilm Formation S. aureus (ATCC 25,923), S. mutans (ATCC 25,175), and P. gingivalis (ATCC 33,277) were cultured as described in the previous study. Mixed-Species Biofilm Formation Single-species biofilm formation was conducted following the protocol as described in a previous study. The concentrations of S. aureus, S. mutans , and P. gingivalis were subsequently adjusted to 1×10 8 CFU/mL, with S. mutans and S. aureus further diluted to 1×10 3 CFU/mL. These three bacterial suspensions were combined in equal proportions. Subsequently, the samples were placed into 15 mL centrifuge tubes ( n = 6), and 200 μL of the aforementioned suspensions were added to each tube, supplemented with 5 mL of brain heart infusion (BHI) culture medium, constituting the mixed microbial solution for this study. These samples were then transferred to 24-well plates ( n = 6), where 100 μL of the same suspensions were added to each well, along with 1 mL of bacterial culture medium. The plates were subsequently incubated at 37°C for 24 hours. Colony‐forming Unit (CFU) Counts After a 24-hour incubation period, the samples ( n = 6) with single-species or mixed-species biofilms underwent two washes with PBS to eliminate non-adherent microorganisms, following which they were transferred to a tube containing 1 mL of PBS. Subsequently, the samples were vortexed for 1 minute to collect the biofilm. Culture of the microorganisms was performed using BHI agar plates. These agar plates were then anaerobically incubated at 37°C, varying in duration based on the microorganism: S. aureus and S. mutans for 2 days, P. gingivalis for 5–7 days, and mix-species for 4–7 days, to enumerate CFUs. Detection of Metabolic Activity of Biofilm After the incubation of biofilms, samples ( n = 6) underwent two washes with PBS to eliminate non-adherent bacteria before being transferred into a new 24-well plate. To assess the metabolic activity of biofilms on test samples, 3-(4, 5-dimethylthiazolyl-2)-2, 5-diphenyltetrazolium bromide (MTT kit, MCE, US) was utilized. Each well received 1 mL of 0.5 mg/mL MTT solution, submerging the test discs. Following a 2-hour incubation at 37°C in a micro-aerobic environment, the MTT solution was replaced with 1 mL of dimethyl sulfoxide (DMSO). After gently shaking the 24-well plate for 20 minutes, 200 μL of DMSO from each well was transferred into a 96-well plate, and the absorbance at 540 nm was measured using a spectrophotometer (Spectra Max iD3, Molecular Devices, US) to evaluate the biofilm’s metabolic activity. Biofilm Imaging Following incubation, the randomly selected samples with single-species or mixed-species 24-hour biofilms underwent two washes with phosphate buffer solution (PBS) to eliminate non-adherent cells. These samples were then immersed in glutaraldehyde solution (2.5%), dehydrated progressively using ethanol solutions (ranging from 30% to 100%), and subsequently desiccated. Upon desiccation, gold-sputtering was performed, and adherent microorganisms were visualized using SEM (Sigma 300, Zeiss, Germany). Statistical Analysis The experiments were conducted in triplicate and repeated on three separate occasions. Statistical analyses were carried out using SPSS software version 20.0. Outliers were identified using boxplot analysis. The Shapiro–Wilk test was performed to assess normality. For data that followed a normal distribution, one-way ANOVA was employed. In contrast, the Kruskal–Wallis test was used for non-normal data. Levene’s test was conducted to check for homogeneity of variances. In cases of heterogeneous variances, Welch’s ANOVA was applied. Post hoc analysis for one-way ANOVA involved Tukey’s multiple-comparison test for homogeneous data and Dunnett’s T3 test for cases with non-homogeneous variances. For the Kruskal–Wallis test, Dunn’s Test was utilized for post hoc comparisons. A significance level of α = 0.05 was applied to all tests.
In this study, two dental materials, Ti (Ti6Al4V, Nissin, China) and ZrO 2, (Aidite, China) were utilized. Test samples were designed as disk-shaped specimens (10.0 mm in diameter, 2.0 mm in thickness) and they were fabricated by the computer-aided design and computer-aided manufacturing (CAD-CAM) mill. All specimens were wet-polished with silicon carbide sandpapers to grade #7000. Subsequently, they were sequentially cleaned and dried with acetone, ethanol, and deionized water using ultrasonication. That were Ti group and ZrO 2 group. And then, two methods were used to prepare the nano-TiO 2 coatings on Ti surface. The first method, ALD technology, was described as follows: the samples were placed in the ALD (TALD-06G, Jiaxingkeming, China), operating in thermal mode. The nano-TiO 2 coatings were then applied to the Ti substrates using Titanium tetrakis dimethylamide (TDMAT) and H 2 O as precursors, along with argon (99.999%) as the purge gas. Each ALD cycle in this study consisted of a 0.1-second TDMAT, a 25-second argon purge, a 0.02-second H 2 O purge, and another 25-second argon purge. The nano-TiO 2 coatings were grown at a temperature of 260°C, with a total of approximately 1625 cycles prepared. And then the nano-TiO 2 coatings with the thickness of 100 nm were produced. That were Ti+ALD group . Following the treatments, the samples were rinsed with deionized water and the hot air dried. , The second method employed in this study involved the utilization of AO technology. Electrochemical staining was conducted on Ti samples using a 1M phosphoric acid electrolytic solution within a temperature-regulated enclosure (DH1719A-5 model, Dahua, China). The Ti was affixed to the anode, while a 20.0 mm × 30.0 mm stainless steel plate served as the cathode, positioned at a distance of 2 cm from the anode. A constant voltage of 60 V was applied through a direct current stabilizing power supply (DH1719A-5 model) for a duration of 60 seconds to facilitate AO. The specific reaction process is outlined as follows: Upon application of an appropriate oxidation voltage, the anode sheds electrons, generating Ti 4+ ions according to formula (1). Simultaneously, O 2- ions and OH − ions manifest at the cathode as indicated in formula (2). These resulting ions migrate via the electric field and electrolyte transport. The interaction of Ti 4+ ions with O 2- ions leads to the formation of TiO 2 at the titanium/electrolyte interface, outlined in formula (3). That were Ti+AO group . Following the treatments, the samples were rinsed with deionized water and hot air dried. , (1) [12pt]{minimal}
[substack]{amsmath}
[mathscr]{eucal}
{linotext }{}
$${} - 4{{}^ - } {}{{}^{4 + }}$$ (2) [12pt]{minimal}
[substack]{amsmath}
[mathscr]{eucal}
{linotext }{}
$$2{}{{}^ - } + 2{{}^ - } 2{{}^{2 - }} + {{}_2} $$ (3) [12pt]{minimal}
[substack]{amsmath}
[mathscr]{eucal}
{linotext }{}
$${}{{}^{4 + }} + 2{{}^{2 - }} {}{{}_2}$$ Subsequently, all the samples from four groups were stored after sterilization with ethylene oxide gas.
Chemical Analysis The chemical compositions of the randomly selected samples from Ti, Ti+AO and Ti+ALD groups were determined by the X-ray photoelectron spectroscopy (XPS) system (Thermo Scientific K-Alpha+, Thermo Fisher, USA). Crystalline Forms Analysis The crystalline forms of the randomly selected samples from Ti, Ti+AO, and Ti+ALD groups were performed using X-ray diffraction (XRD) (Miniflex 600, Rigaku, Japan) equipped with Cu Kα radiation. Surface Topography and Roughness Analysis The randomly selected samples from four groups were subjected to surface topography and roughness analysis. The scanning electron microscope (SEM) device (Sigma 300, Zeiss, Germany) was used for surface topography evaluation. The non-conductive ZrO 2 samples were gold-sputtered, and then the samples were placed in the SEM device at magnifications of × 5.00 K, and × 20.00 K. Atomic Force Microscope (AFM) (Dimension Icon, Bruker, USA) was used for roughness analysis. Hydrophilicity The hydrophilicity of the samples from each group ( n = 6) was assessed through water contact angle (WCA) measurements. Using an optical contact angle meter (SDC-100, Shengding, China), a 4 μL drop of distilled water was placed on the specimen’s surface. Final contact angle values were determined by averaging three measurements taken at different parts of the surfaces.
The chemical compositions of the randomly selected samples from Ti, Ti+AO and Ti+ALD groups were determined by the X-ray photoelectron spectroscopy (XPS) system (Thermo Scientific K-Alpha+, Thermo Fisher, USA).
The crystalline forms of the randomly selected samples from Ti, Ti+AO, and Ti+ALD groups were performed using X-ray diffraction (XRD) (Miniflex 600, Rigaku, Japan) equipped with Cu Kα radiation.
The randomly selected samples from four groups were subjected to surface topography and roughness analysis. The scanning electron microscope (SEM) device (Sigma 300, Zeiss, Germany) was used for surface topography evaluation. The non-conductive ZrO 2 samples were gold-sputtered, and then the samples were placed in the SEM device at magnifications of × 5.00 K, and × 20.00 K. Atomic Force Microscope (AFM) (Dimension Icon, Bruker, USA) was used for roughness analysis.
The hydrophilicity of the samples from each group ( n = 6) was assessed through water contact angle (WCA) measurements. Using an optical contact angle meter (SDC-100, Shengding, China), a 4 μL drop of distilled water was placed on the specimen’s surface. Final contact angle values were determined by averaging three measurements taken at different parts of the surfaces.
Single-Species Biofilm Formation S. aureus (ATCC 25,923), S. mutans (ATCC 25,175), and P. gingivalis (ATCC 33,277) were cultured as described in the previous study. Mixed-Species Biofilm Formation Single-species biofilm formation was conducted following the protocol as described in a previous study. The concentrations of S. aureus, S. mutans , and P. gingivalis were subsequently adjusted to 1×10 8 CFU/mL, with S. mutans and S. aureus further diluted to 1×10 3 CFU/mL. These three bacterial suspensions were combined in equal proportions. Subsequently, the samples were placed into 15 mL centrifuge tubes ( n = 6), and 200 μL of the aforementioned suspensions were added to each tube, supplemented with 5 mL of brain heart infusion (BHI) culture medium, constituting the mixed microbial solution for this study. These samples were then transferred to 24-well plates ( n = 6), where 100 μL of the same suspensions were added to each well, along with 1 mL of bacterial culture medium. The plates were subsequently incubated at 37°C for 24 hours. Colony‐forming Unit (CFU) Counts After a 24-hour incubation period, the samples ( n = 6) with single-species or mixed-species biofilms underwent two washes with PBS to eliminate non-adherent microorganisms, following which they were transferred to a tube containing 1 mL of PBS. Subsequently, the samples were vortexed for 1 minute to collect the biofilm. Culture of the microorganisms was performed using BHI agar plates. These agar plates were then anaerobically incubated at 37°C, varying in duration based on the microorganism: S. aureus and S. mutans for 2 days, P. gingivalis for 5–7 days, and mix-species for 4–7 days, to enumerate CFUs. Detection of Metabolic Activity of Biofilm After the incubation of biofilms, samples ( n = 6) underwent two washes with PBS to eliminate non-adherent bacteria before being transferred into a new 24-well plate. To assess the metabolic activity of biofilms on test samples, 3-(4, 5-dimethylthiazolyl-2)-2, 5-diphenyltetrazolium bromide (MTT kit, MCE, US) was utilized. Each well received 1 mL of 0.5 mg/mL MTT solution, submerging the test discs. Following a 2-hour incubation at 37°C in a micro-aerobic environment, the MTT solution was replaced with 1 mL of dimethyl sulfoxide (DMSO). After gently shaking the 24-well plate for 20 minutes, 200 μL of DMSO from each well was transferred into a 96-well plate, and the absorbance at 540 nm was measured using a spectrophotometer (Spectra Max iD3, Molecular Devices, US) to evaluate the biofilm’s metabolic activity. Biofilm Imaging Following incubation, the randomly selected samples with single-species or mixed-species 24-hour biofilms underwent two washes with phosphate buffer solution (PBS) to eliminate non-adherent cells. These samples were then immersed in glutaraldehyde solution (2.5%), dehydrated progressively using ethanol solutions (ranging from 30% to 100%), and subsequently desiccated. Upon desiccation, gold-sputtering was performed, and adherent microorganisms were visualized using SEM (Sigma 300, Zeiss, Germany).
S. aureus (ATCC 25,923), S. mutans (ATCC 25,175), and P. gingivalis (ATCC 33,277) were cultured as described in the previous study.
Single-species biofilm formation was conducted following the protocol as described in a previous study. The concentrations of S. aureus, S. mutans , and P. gingivalis were subsequently adjusted to 1×10 8 CFU/mL, with S. mutans and S. aureus further diluted to 1×10 3 CFU/mL. These three bacterial suspensions were combined in equal proportions. Subsequently, the samples were placed into 15 mL centrifuge tubes ( n = 6), and 200 μL of the aforementioned suspensions were added to each tube, supplemented with 5 mL of brain heart infusion (BHI) culture medium, constituting the mixed microbial solution for this study. These samples were then transferred to 24-well plates ( n = 6), where 100 μL of the same suspensions were added to each well, along with 1 mL of bacterial culture medium. The plates were subsequently incubated at 37°C for 24 hours.
After a 24-hour incubation period, the samples ( n = 6) with single-species or mixed-species biofilms underwent two washes with PBS to eliminate non-adherent microorganisms, following which they were transferred to a tube containing 1 mL of PBS. Subsequently, the samples were vortexed for 1 minute to collect the biofilm. Culture of the microorganisms was performed using BHI agar plates. These agar plates were then anaerobically incubated at 37°C, varying in duration based on the microorganism: S. aureus and S. mutans for 2 days, P. gingivalis for 5–7 days, and mix-species for 4–7 days, to enumerate CFUs.
After the incubation of biofilms, samples ( n = 6) underwent two washes with PBS to eliminate non-adherent bacteria before being transferred into a new 24-well plate. To assess the metabolic activity of biofilms on test samples, 3-(4, 5-dimethylthiazolyl-2)-2, 5-diphenyltetrazolium bromide (MTT kit, MCE, US) was utilized. Each well received 1 mL of 0.5 mg/mL MTT solution, submerging the test discs. Following a 2-hour incubation at 37°C in a micro-aerobic environment, the MTT solution was replaced with 1 mL of dimethyl sulfoxide (DMSO). After gently shaking the 24-well plate for 20 minutes, 200 μL of DMSO from each well was transferred into a 96-well plate, and the absorbance at 540 nm was measured using a spectrophotometer (Spectra Max iD3, Molecular Devices, US) to evaluate the biofilm’s metabolic activity.
Following incubation, the randomly selected samples with single-species or mixed-species 24-hour biofilms underwent two washes with phosphate buffer solution (PBS) to eliminate non-adherent cells. These samples were then immersed in glutaraldehyde solution (2.5%), dehydrated progressively using ethanol solutions (ranging from 30% to 100%), and subsequently desiccated. Upon desiccation, gold-sputtering was performed, and adherent microorganisms were visualized using SEM (Sigma 300, Zeiss, Germany).
The experiments were conducted in triplicate and repeated on three separate occasions. Statistical analyses were carried out using SPSS software version 20.0. Outliers were identified using boxplot analysis. The Shapiro–Wilk test was performed to assess normality. For data that followed a normal distribution, one-way ANOVA was employed. In contrast, the Kruskal–Wallis test was used for non-normal data. Levene’s test was conducted to check for homogeneity of variances. In cases of heterogeneous variances, Welch’s ANOVA was applied. Post hoc analysis for one-way ANOVA involved Tukey’s multiple-comparison test for homogeneous data and Dunnett’s T3 test for cases with non-homogeneous variances. For the Kruskal–Wallis test, Dunn’s Test was utilized for post hoc comparisons. A significance level of α = 0.05 was applied to all tests.
Surface Properties As shown in , the color of both Ti+AO and Ti+ALD samples changed to yellow hue, which is conducive to peri‑implant soft tissue color of Ti abutment. Chemical Analysis To determine the elemental composition and chemical bonds of the nano-TiO 2 coatings prepared by AO and ALD on Ti abutment surfaces, XPS analysis was performed on the Ti, Ti+AO, and Ti+ALD samples. and displayed the full XPS spectrum of the Ti, Ti+AO, Ti+ALD groups, which all had three distinct peaks attributed to Ti, oxygen (O), and carbon (C) elements. Specifically, in the Ti+ALD group, the relative atomic content of C and Ti elements was the highest, accounting for 29.65 at. % and 21.34 at. %, respectively, with trace amounts of Nitrogen (N) detected. In the Ti+AO group, the relative atomic content of O element was the highest, at 51.21 at. %. The high-resolution Ti2p spectra in showed double peaks at approximately 464.2 eV and 458.5 eV, corresponding to Ti 4+ 2p 1/2 and Ti 4+ 2p 3/2 , indicating the presence of Ti 4+ . These peaks were consistent with the reference spectra for TiO 2 , confirming that the coatings consisted of TiO 2 . Additionally, a broad shoulder peak at 453.8 eV in the Ti group represented the elementary substance Ti, which was not detected in the Ti+AO and Ti+ALD groups, indicating that Ti was successfully oxidized during the AO and ALD processes. Furthermore, the highest relative atomic content of Ti 4+ (3.04 at.%) was observed in the Ti+ALD group, indicating the highest level of TiO 2 . This finding also corresponded to the notably heightened atomic concentration of Ti element (21.34 at.%) in the Ti+ALD group. The high-resolution O1s spectra in showed two peaks at approximately 529.9 eV and 531.7 eV, corresponding to Ti-O and Ti-OH bonds, respectively. Specifically, ALD-TiO 2 exhibited the highest relative atomic content of Ti-O at 5.75 at.%, while Ti-OH showed the lowest presence (1.56 at.%) in this group. Crystalline Forms Analysis To determine the crystalline structure of nano-TiO2 coatings prepared by AO and ALD on Ti abutment surfaces, XRD analysis was performed on the Ti, Ti+AO, and Ti+ALD samples, as shown in . The diffraction peaks for the α-phase (100), (002), (101), and (102) planes and the β-phase (110) were observed in all groups. Notably, the Ti+ALD group showed a strong A (101) diffraction peak at 2θ=25.4°, presenting the formation of anatase TiO 2 phase compared to the other groups. Surface Topography and Roughness Analysis The surface topography and roughness of Ti, Ti+AO, Ti+ALD, and ZrO 2 groups were examined using SEM and AFM, illustrated in and . SEM analysis revealed distinctive features: The Ti group displayed non-uniform surface structures with noticeable tripping and flake formations. Conversely, the Ti+AO group exhibited a smoother surface with reduced and shallower scratches alongside various nano-sized pores on the oxide film. The ALD-TiO 2 coatings appeared denser than the AO-TiO 2 and non-coated surfaces, uniformly covering the substrate with TiO 2 nanoparticles, resulting in a visibly smoother surface. Similarly, the ZrO 2 group displayed a uniformly smooth substrate surface with shallow groove-like scratches, notably smoother than the other groups. AFM images mirrored these trends. The Ti sample displayed ridge-uplift and valley depression, whereas the AO-TiO 2 coated surfaces showed subtle variations with pore-like structures. In contrast, the ALD-TiO 2 coated samples presented the sharp and spikelike nanostructures. The ZrO 2 surface exhibited the smoothest surface (SEM image), featuring densely distributed point-like particles with scattered needle-like protrusions (AFM image). Additionally, and presented the roughness values. Compared to the Ti group, the surface roughness (Ra) of the Ti+AO, Ti+ALD, and ZrO2 groups decreased from (30.07 ± 5.06) nm to (28.40 ± 2.01) nm, (14.47 ± 4.39) nm, and (7.33 ± 0.87) nm, respectively. Both Ti and Ti+AO groups were significantly different with Ti+ALD ( p < 0.01) and ZrO 2 ( p < 0.001) groups, respectively. However, the differences between the Ti group and the Ti+AO group, as well as between the Ti+ALD group and the ZrO 2 group, were not statistically significant ( p > 0.05). The value of RMS mirrored these trends. Hydrophilicity The water contact angles (WCA) of the ZrO 2 (72.45 ± 4.64)°, Ti+ALD (62.91 ± 3.71)°, and Ti+AO (67.75 ± 0.45)° were found to be higher than that of the Ti group (52.70 ± 0.86)° , all the difference were significant ( p < 0.01), indicating that the presence of loaded nano-TiO 2 reduced the hydrophilicity of the Ti substrates, but they are still hydrophilicity materials (WCA < 90°). Furthermore, there were statistical differences between Ti+ALD and ZrO 2 groups ( p < 0.05). Antibacterial Capability of Single-Species Colony‐forming Unit (CFU) Counts In , significant differences in CFU results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with S. mutans ( p < 0.01), S. aureus ( p < 0.001), and P. gingivalis ( p < 0.01), respectively. The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Then, the evaluation of the antibacterial effect was determined by comparing the bacteria adhesion of bacteria to the Ti+ALD and ZrO 2 groups with that of Ti group. The result showed the lowest observed antimicrobial effect was about 50% ( S. aureus and P. gingivalis ), while the highest one was about 80% ( S. mutans ). Detection of Metabolic Activity of Biofilm Similar with the result of CFUs, the significant differences in MTT results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with S. mutans ( p < 0.01), S. aureus ( p < 0.05), and P. gingivalis ( p < 0.05), respectively. The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Biofilm Imaging Generally, the SEM images at low magnification (× 5.00K) exhibited a lower abundance of three bacterial species in Ti+ALD and ZrO 2 groups. After 24 h co-culturing with S. mutans , the bacteria of Ti group and Ti+AO group were deposited on the material surface, particularly in irregular areas such as scratches and gullies, and formed chain-like patterns. High magnification SEM (× 20.00K) showed that individual bacteria consisted of multiple “capsule” structures in series, with smooth surfaces, full shapes, and intact envelopes. After 24 h co-culturing with S. aureus , the Ti and Ti+AO groups displayed bacteria clustered on the material surface in a “grape-like” distribution. In contrast, the Ti+ALD and ZrO 2 groups showed scattered spherical bacteria on the specimen surfaces. High magnification SEM (×20.00K) revealed spherical bacteria, full in shape and with complete envelopes in all groups. After 24 h co-culturing with P. gingivalis , the Ti and Ti+AO groups exhibited densely dispersed bacteria on the specimen surface. However, the Ti+ALD and ZrO 2 groups showed scattered spherical or rod-shaped bacteria. High magnification SEM (×20.00K) illustrated that individual bacteria were spherical or rod-shaped, with rough surfaces, full shapes, and complete envelopes. Antibacterial Capability of Mixed-Species Colony‐forming Unit (CFU) Counts In , significant differences ( p < 0.05) in CFU results were observed between the ZrO 2 and Ti+ALD groups compared to Ti group after co-culturing with mixed-species consisting of S. aureus, S. mutans , and P. gingivalis . The comparable bacterial adhesion ( p > 0.05) was observed between Ti+AO and Ti groups, while significant differences were found among ZrO 2 and Ti+ALD groups ( p < 0.01), the antibacterial effect was evaluated by comparing bacterial adhesion in the Ti+ALD and ZrO 2 groups with that in the Ti group. The result showed the observed antimicrobial effect was approximately 75% for ALD-TiO 2 and ZrO 2 surfaces. Detection of Metabolic Activity of Biofilm Similar with the result of CFUs, the significant differences ( p < 0.05) in MTT results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with mixed-species that consist of S. aureus, S. mutans , and P. gingivalis . The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Biofilm Imaging Low magnification SEM images (×5.00K) exhibited that bacteria in the Ti and Ti+AO groups aggregated on the material surface, especially in irregular areas, while the Ti+ALD and ZrO 2 groups had a lower abundance of all three bacterial species. High magnification SEM (×20.00K) revealed specific characteristics of S. aureus, S. mutans , and P. gingivalis. S. aureus appeared round and larger than the other two bacteria, with a smooth surface, scattered or arranged in “grape-like” clusters. S. mutans consisted of several “capsule” structures arranged in chains or singly, with smooth surfaces and full shapes. P. gingivalis was bulbous or rod-shaped, with a rough surface, full form, and complete envelope. These three bacteria were interleaved and arranged on the specimen surface.
As shown in , the color of both Ti+AO and Ti+ALD samples changed to yellow hue, which is conducive to peri‑implant soft tissue color of Ti abutment. Chemical Analysis To determine the elemental composition and chemical bonds of the nano-TiO 2 coatings prepared by AO and ALD on Ti abutment surfaces, XPS analysis was performed on the Ti, Ti+AO, and Ti+ALD samples. and displayed the full XPS spectrum of the Ti, Ti+AO, Ti+ALD groups, which all had three distinct peaks attributed to Ti, oxygen (O), and carbon (C) elements. Specifically, in the Ti+ALD group, the relative atomic content of C and Ti elements was the highest, accounting for 29.65 at. % and 21.34 at. %, respectively, with trace amounts of Nitrogen (N) detected. In the Ti+AO group, the relative atomic content of O element was the highest, at 51.21 at. %. The high-resolution Ti2p spectra in showed double peaks at approximately 464.2 eV and 458.5 eV, corresponding to Ti 4+ 2p 1/2 and Ti 4+ 2p 3/2 , indicating the presence of Ti 4+ . These peaks were consistent with the reference spectra for TiO 2 , confirming that the coatings consisted of TiO 2 . Additionally, a broad shoulder peak at 453.8 eV in the Ti group represented the elementary substance Ti, which was not detected in the Ti+AO and Ti+ALD groups, indicating that Ti was successfully oxidized during the AO and ALD processes. Furthermore, the highest relative atomic content of Ti 4+ (3.04 at.%) was observed in the Ti+ALD group, indicating the highest level of TiO 2 . This finding also corresponded to the notably heightened atomic concentration of Ti element (21.34 at.%) in the Ti+ALD group. The high-resolution O1s spectra in showed two peaks at approximately 529.9 eV and 531.7 eV, corresponding to Ti-O and Ti-OH bonds, respectively. Specifically, ALD-TiO 2 exhibited the highest relative atomic content of Ti-O at 5.75 at.%, while Ti-OH showed the lowest presence (1.56 at.%) in this group. Crystalline Forms Analysis To determine the crystalline structure of nano-TiO2 coatings prepared by AO and ALD on Ti abutment surfaces, XRD analysis was performed on the Ti, Ti+AO, and Ti+ALD samples, as shown in . The diffraction peaks for the α-phase (100), (002), (101), and (102) planes and the β-phase (110) were observed in all groups. Notably, the Ti+ALD group showed a strong A (101) diffraction peak at 2θ=25.4°, presenting the formation of anatase TiO 2 phase compared to the other groups. Surface Topography and Roughness Analysis The surface topography and roughness of Ti, Ti+AO, Ti+ALD, and ZrO 2 groups were examined using SEM and AFM, illustrated in and . SEM analysis revealed distinctive features: The Ti group displayed non-uniform surface structures with noticeable tripping and flake formations. Conversely, the Ti+AO group exhibited a smoother surface with reduced and shallower scratches alongside various nano-sized pores on the oxide film. The ALD-TiO 2 coatings appeared denser than the AO-TiO 2 and non-coated surfaces, uniformly covering the substrate with TiO 2 nanoparticles, resulting in a visibly smoother surface. Similarly, the ZrO 2 group displayed a uniformly smooth substrate surface with shallow groove-like scratches, notably smoother than the other groups. AFM images mirrored these trends. The Ti sample displayed ridge-uplift and valley depression, whereas the AO-TiO 2 coated surfaces showed subtle variations with pore-like structures. In contrast, the ALD-TiO 2 coated samples presented the sharp and spikelike nanostructures. The ZrO 2 surface exhibited the smoothest surface (SEM image), featuring densely distributed point-like particles with scattered needle-like protrusions (AFM image). Additionally, and presented the roughness values. Compared to the Ti group, the surface roughness (Ra) of the Ti+AO, Ti+ALD, and ZrO2 groups decreased from (30.07 ± 5.06) nm to (28.40 ± 2.01) nm, (14.47 ± 4.39) nm, and (7.33 ± 0.87) nm, respectively. Both Ti and Ti+AO groups were significantly different with Ti+ALD ( p < 0.01) and ZrO 2 ( p < 0.001) groups, respectively. However, the differences between the Ti group and the Ti+AO group, as well as between the Ti+ALD group and the ZrO 2 group, were not statistically significant ( p > 0.05). The value of RMS mirrored these trends. Hydrophilicity The water contact angles (WCA) of the ZrO 2 (72.45 ± 4.64)°, Ti+ALD (62.91 ± 3.71)°, and Ti+AO (67.75 ± 0.45)° were found to be higher than that of the Ti group (52.70 ± 0.86)° , all the difference were significant ( p < 0.01), indicating that the presence of loaded nano-TiO 2 reduced the hydrophilicity of the Ti substrates, but they are still hydrophilicity materials (WCA < 90°). Furthermore, there were statistical differences between Ti+ALD and ZrO 2 groups ( p < 0.05).
To determine the elemental composition and chemical bonds of the nano-TiO 2 coatings prepared by AO and ALD on Ti abutment surfaces, XPS analysis was performed on the Ti, Ti+AO, and Ti+ALD samples. and displayed the full XPS spectrum of the Ti, Ti+AO, Ti+ALD groups, which all had three distinct peaks attributed to Ti, oxygen (O), and carbon (C) elements. Specifically, in the Ti+ALD group, the relative atomic content of C and Ti elements was the highest, accounting for 29.65 at. % and 21.34 at. %, respectively, with trace amounts of Nitrogen (N) detected. In the Ti+AO group, the relative atomic content of O element was the highest, at 51.21 at. %. The high-resolution Ti2p spectra in showed double peaks at approximately 464.2 eV and 458.5 eV, corresponding to Ti 4+ 2p 1/2 and Ti 4+ 2p 3/2 , indicating the presence of Ti 4+ . These peaks were consistent with the reference spectra for TiO 2 , confirming that the coatings consisted of TiO 2 . Additionally, a broad shoulder peak at 453.8 eV in the Ti group represented the elementary substance Ti, which was not detected in the Ti+AO and Ti+ALD groups, indicating that Ti was successfully oxidized during the AO and ALD processes. Furthermore, the highest relative atomic content of Ti 4+ (3.04 at.%) was observed in the Ti+ALD group, indicating the highest level of TiO 2 . This finding also corresponded to the notably heightened atomic concentration of Ti element (21.34 at.%) in the Ti+ALD group. The high-resolution O1s spectra in showed two peaks at approximately 529.9 eV and 531.7 eV, corresponding to Ti-O and Ti-OH bonds, respectively. Specifically, ALD-TiO 2 exhibited the highest relative atomic content of Ti-O at 5.75 at.%, while Ti-OH showed the lowest presence (1.56 at.%) in this group.
To determine the crystalline structure of nano-TiO2 coatings prepared by AO and ALD on Ti abutment surfaces, XRD analysis was performed on the Ti, Ti+AO, and Ti+ALD samples, as shown in . The diffraction peaks for the α-phase (100), (002), (101), and (102) planes and the β-phase (110) were observed in all groups. Notably, the Ti+ALD group showed a strong A (101) diffraction peak at 2θ=25.4°, presenting the formation of anatase TiO 2 phase compared to the other groups.
The surface topography and roughness of Ti, Ti+AO, Ti+ALD, and ZrO 2 groups were examined using SEM and AFM, illustrated in and . SEM analysis revealed distinctive features: The Ti group displayed non-uniform surface structures with noticeable tripping and flake formations. Conversely, the Ti+AO group exhibited a smoother surface with reduced and shallower scratches alongside various nano-sized pores on the oxide film. The ALD-TiO 2 coatings appeared denser than the AO-TiO 2 and non-coated surfaces, uniformly covering the substrate with TiO 2 nanoparticles, resulting in a visibly smoother surface. Similarly, the ZrO 2 group displayed a uniformly smooth substrate surface with shallow groove-like scratches, notably smoother than the other groups. AFM images mirrored these trends. The Ti sample displayed ridge-uplift and valley depression, whereas the AO-TiO 2 coated surfaces showed subtle variations with pore-like structures. In contrast, the ALD-TiO 2 coated samples presented the sharp and spikelike nanostructures. The ZrO 2 surface exhibited the smoothest surface (SEM image), featuring densely distributed point-like particles with scattered needle-like protrusions (AFM image). Additionally, and presented the roughness values. Compared to the Ti group, the surface roughness (Ra) of the Ti+AO, Ti+ALD, and ZrO2 groups decreased from (30.07 ± 5.06) nm to (28.40 ± 2.01) nm, (14.47 ± 4.39) nm, and (7.33 ± 0.87) nm, respectively. Both Ti and Ti+AO groups were significantly different with Ti+ALD ( p < 0.01) and ZrO 2 ( p < 0.001) groups, respectively. However, the differences between the Ti group and the Ti+AO group, as well as between the Ti+ALD group and the ZrO 2 group, were not statistically significant ( p > 0.05). The value of RMS mirrored these trends.
The water contact angles (WCA) of the ZrO 2 (72.45 ± 4.64)°, Ti+ALD (62.91 ± 3.71)°, and Ti+AO (67.75 ± 0.45)° were found to be higher than that of the Ti group (52.70 ± 0.86)° , all the difference were significant ( p < 0.01), indicating that the presence of loaded nano-TiO 2 reduced the hydrophilicity of the Ti substrates, but they are still hydrophilicity materials (WCA < 90°). Furthermore, there were statistical differences between Ti+ALD and ZrO 2 groups ( p < 0.05).
Colony‐forming Unit (CFU) Counts In , significant differences in CFU results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with S. mutans ( p < 0.01), S. aureus ( p < 0.001), and P. gingivalis ( p < 0.01), respectively. The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Then, the evaluation of the antibacterial effect was determined by comparing the bacteria adhesion of bacteria to the Ti+ALD and ZrO 2 groups with that of Ti group. The result showed the lowest observed antimicrobial effect was about 50% ( S. aureus and P. gingivalis ), while the highest one was about 80% ( S. mutans ). Detection of Metabolic Activity of Biofilm Similar with the result of CFUs, the significant differences in MTT results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with S. mutans ( p < 0.01), S. aureus ( p < 0.05), and P. gingivalis ( p < 0.05), respectively. The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Biofilm Imaging Generally, the SEM images at low magnification (× 5.00K) exhibited a lower abundance of three bacterial species in Ti+ALD and ZrO 2 groups. After 24 h co-culturing with S. mutans , the bacteria of Ti group and Ti+AO group were deposited on the material surface, particularly in irregular areas such as scratches and gullies, and formed chain-like patterns. High magnification SEM (× 20.00K) showed that individual bacteria consisted of multiple “capsule” structures in series, with smooth surfaces, full shapes, and intact envelopes. After 24 h co-culturing with S. aureus , the Ti and Ti+AO groups displayed bacteria clustered on the material surface in a “grape-like” distribution. In contrast, the Ti+ALD and ZrO 2 groups showed scattered spherical bacteria on the specimen surfaces. High magnification SEM (×20.00K) revealed spherical bacteria, full in shape and with complete envelopes in all groups. After 24 h co-culturing with P. gingivalis , the Ti and Ti+AO groups exhibited densely dispersed bacteria on the specimen surface. However, the Ti+ALD and ZrO 2 groups showed scattered spherical or rod-shaped bacteria. High magnification SEM (×20.00K) illustrated that individual bacteria were spherical or rod-shaped, with rough surfaces, full shapes, and complete envelopes.
In , significant differences in CFU results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with S. mutans ( p < 0.01), S. aureus ( p < 0.001), and P. gingivalis ( p < 0.01), respectively. The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Then, the evaluation of the antibacterial effect was determined by comparing the bacteria adhesion of bacteria to the Ti+ALD and ZrO 2 groups with that of Ti group. The result showed the lowest observed antimicrobial effect was about 50% ( S. aureus and P. gingivalis ), while the highest one was about 80% ( S. mutans ).
Similar with the result of CFUs, the significant differences in MTT results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with S. mutans ( p < 0.01), S. aureus ( p < 0.05), and P. gingivalis ( p < 0.05), respectively. The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups.
Generally, the SEM images at low magnification (× 5.00K) exhibited a lower abundance of three bacterial species in Ti+ALD and ZrO 2 groups. After 24 h co-culturing with S. mutans , the bacteria of Ti group and Ti+AO group were deposited on the material surface, particularly in irregular areas such as scratches and gullies, and formed chain-like patterns. High magnification SEM (× 20.00K) showed that individual bacteria consisted of multiple “capsule” structures in series, with smooth surfaces, full shapes, and intact envelopes. After 24 h co-culturing with S. aureus , the Ti and Ti+AO groups displayed bacteria clustered on the material surface in a “grape-like” distribution. In contrast, the Ti+ALD and ZrO 2 groups showed scattered spherical bacteria on the specimen surfaces. High magnification SEM (×20.00K) revealed spherical bacteria, full in shape and with complete envelopes in all groups. After 24 h co-culturing with P. gingivalis , the Ti and Ti+AO groups exhibited densely dispersed bacteria on the specimen surface. However, the Ti+ALD and ZrO 2 groups showed scattered spherical or rod-shaped bacteria. High magnification SEM (×20.00K) illustrated that individual bacteria were spherical or rod-shaped, with rough surfaces, full shapes, and complete envelopes.
Colony‐forming Unit (CFU) Counts In , significant differences ( p < 0.05) in CFU results were observed between the ZrO 2 and Ti+ALD groups compared to Ti group after co-culturing with mixed-species consisting of S. aureus, S. mutans , and P. gingivalis . The comparable bacterial adhesion ( p > 0.05) was observed between Ti+AO and Ti groups, while significant differences were found among ZrO 2 and Ti+ALD groups ( p < 0.01), the antibacterial effect was evaluated by comparing bacterial adhesion in the Ti+ALD and ZrO 2 groups with that in the Ti group. The result showed the observed antimicrobial effect was approximately 75% for ALD-TiO 2 and ZrO 2 surfaces. Detection of Metabolic Activity of Biofilm Similar with the result of CFUs, the significant differences ( p < 0.05) in MTT results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with mixed-species that consist of S. aureus, S. mutans , and P. gingivalis . The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups. Biofilm Imaging Low magnification SEM images (×5.00K) exhibited that bacteria in the Ti and Ti+AO groups aggregated on the material surface, especially in irregular areas, while the Ti+ALD and ZrO 2 groups had a lower abundance of all three bacterial species. High magnification SEM (×20.00K) revealed specific characteristics of S. aureus, S. mutans , and P. gingivalis. S. aureus appeared round and larger than the other two bacteria, with a smooth surface, scattered or arranged in “grape-like” clusters. S. mutans consisted of several “capsule” structures arranged in chains or singly, with smooth surfaces and full shapes. P. gingivalis was bulbous or rod-shaped, with a rough surface, full form, and complete envelope. These three bacteria were interleaved and arranged on the specimen surface.
In , significant differences ( p < 0.05) in CFU results were observed between the ZrO 2 and Ti+ALD groups compared to Ti group after co-culturing with mixed-species consisting of S. aureus, S. mutans , and P. gingivalis . The comparable bacterial adhesion ( p > 0.05) was observed between Ti+AO and Ti groups, while significant differences were found among ZrO 2 and Ti+ALD groups ( p < 0.01), the antibacterial effect was evaluated by comparing bacterial adhesion in the Ti+ALD and ZrO 2 groups with that in the Ti group. The result showed the observed antimicrobial effect was approximately 75% for ALD-TiO 2 and ZrO 2 surfaces.
Similar with the result of CFUs, the significant differences ( p < 0.05) in MTT results were observed between the ZrO 2 and Ti+ALD groups compared to the Ti+AO and Ti groups after co-culturing with mixed-species that consist of S. aureus, S. mutans , and P. gingivalis . The comparable bacterial adhesion ( p > 0.05) was observed among ZrO 2 and Ti+ALD groups, as well as Ti+AO and Ti groups.
Low magnification SEM images (×5.00K) exhibited that bacteria in the Ti and Ti+AO groups aggregated on the material surface, especially in irregular areas, while the Ti+ALD and ZrO 2 groups had a lower abundance of all three bacterial species. High magnification SEM (×20.00K) revealed specific characteristics of S. aureus, S. mutans , and P. gingivalis. S. aureus appeared round and larger than the other two bacteria, with a smooth surface, scattered or arranged in “grape-like” clusters. S. mutans consisted of several “capsule” structures arranged in chains or singly, with smooth surfaces and full shapes. P. gingivalis was bulbous or rod-shaped, with a rough surface, full form, and complete envelope. These three bacteria were interleaved and arranged on the specimen surface.
The results indicated a statistically significant difference in surface properties and bacterial adhesion of S. mutans, S. aureus , and P. gingivalis among the Ti, AO-TiO 2 , ALD-TiO 2 , and ZrO 2 surfaces. Interestingly, the adhesion of mixed-species bacteria resembled that of single-species bacteria. Consequently, we could accept the first and second null hypotheses while rejecting the third null hypothesis. Chemical analysis showed that, in the Ti+ALD group, the relative atomic content of C and Ti elements is the highest, with trace amounts of Nitrogen (N) detected. The presence of C element may originate not only from airborne carbon contamination but also residues of C and N elements from the byproduct generated during the ALD reaction of TDMAT with H 2 O, which also explained the source of the detected N element. In the Ti+AO group, the relative atomic content of O element is the highest. This may be attributed to its association with Silicon (Si), Phosphorus (P), and N elements, capable of bonding with O. The Si element, akin to that in the Ti group, might stem from residual Si after polishing with silicon carbide paper on the Ti substrate. It could also originate from the AO process of TiO 2 preparation, explaining the absence of Si element residue in the Ti+ALD group. Prior studies, such as Wang et al, corroborate these findings, noting minor Si and P elements on anodized Ti substrates, suggesting the incorporation of electrolytes into the film-forming process. Additionally, Kern et al indicated the reproducible existence of P in the corresponding oxides during anodization experiments with phosphate-containing electrolytes, where phosphate ions comprised a significant portion of the total film thickness. Phosphate, being a fundamental component of living organisms with good biocompatibility, justifies its selection as an electrolyte in the present study. The results of Crystalline forms analysis indicated that the anatase TiO 2 phase was only found in the Ti+ALD group. This observation is in line with previous studies that reported the formation of the anatase TiO 2 phase through ALD. , Fenoglio et al found anatase TiO 2 can generate ROS in ambient light conditions (even in darkness) and possesses certain antibacterial properties. Key factors determining the phase structure of nano TiO 2 coatings via ALD technology include precursor types and growth temperatures. Firstly, precursor selection significantly influences the ALD temperature range. Ti alkoxides and amides, commonly used for TiO₂ films, decompose below 300°C and produce non-toxic by-products. , Accordingly, the present study employed TDMAT as the precursor at temperatures below 300°C. A linear increase in layer thickness with ALD cycles has been observed between 50°C and 300°C. In contrast, Chung et al used the heteroleptic precursor, allowing ALD at temperatures up to 400°C. While halides like TiCl₄ also support growth above 400°C, their by-products (eg, HCl) pose corrosion risks. Growth temperature also affects film crystallinity. , , Both Reiners et al and Chung et al indicated that higher growth temperature (TDMAT, 250°C and heteroleptic, 236°C) promoted the anatase-phase crystallization, while the lower growth temperature showed amorphous with smooth surfaces. Similarly, Liu et al used TDMAT and H 2 O precursors via ALD on Ti implants at temperatures of 120°C, 160°C, and 190°C. Only at 190°C did anatase-phase crystallization occur. In the present study, the ALD reaction temperature was 260°C, exhibiting anatase-phase nano TiO 2 on Ti+ALD samples, consistent with prior researches. However, Ti+AO group lacked anatase diffraction peaks, possibly due to the lower AO voltage (60 V) used herein. Brunello et al used 8 V and 2.2 A for AO modification on Ti alloys, finding similar bacterial adhesion to mechanically processed Ti. In contrast, Giordano et al used 100 V and 120 V for AO, showing anatase diffraction peaks at both voltages, especially intense at higher voltage, indicating electrochemical-induced anatase inhibition against bacterial colonization. As illustrated in the results of surface topography and roughness analysis , compared to Ti group, Ti+AO group exhibited a smoother surface with reduced and shallower scratches alongside various nano-sized pores on the oxide film. This alteration may be linked to gas emission during AO, consistent with findings from previous studies. The roughness of the Ti+ALD group decreased significantly as a result of particle deposition on the surface, which partially filled the valley concavities. Furthermore, the surface roughness of the Ti-based substrate decreased after AO and ALD treatments, yet it still did not match the smoothness of the ZrO 2 group, consistent with the observations from the four groups of SEM morphology and AFM surface topography. Nevertheless, the SEM image of ZrO 2 surface exhibited the smoothest surface, while AFM image of that featured densely distributed point-like particles with scattered needle-like protrusions, these were in accordance with other studies. The reason for the different surface topography observed in AFM and SEM images is that AFM and SEM differ significantly in terms of imaging mechanisms, resolution, and dimensionality. These two high-resolution surface investigations are complementary techniques that provides a more complete representation of a surface when used together than if each were the only technique available. Particularly, the AFM and SEM image of Ti+ALD samples exhibited a smooth and spikelike nanostructure. This contrasted with the roughen surfaces observed in other studies, , likely also due to differences in production parameters of ALD process, such as precursor type, growth temperature, substrate and coating thickness. The influence of precursor type on the growth temperature has been previously discussed. As Chung et al noted, ALD growth temperature significantly affected the film’s microstructure, and roughness. Their films grown at 182°C were amorphous with smooth surfaces, while those above 236°C showed the roughening surfaces. At temperatures above 300°C, only very fine grains were apparent. Liu et al also found a remarkable nanorough surfaces at 120°C, 160°C and 190°C. In the present study, the growth temperature of 260°C was applied, which was aligns with the 263°C reported by Chung et al, yet differences in surface morphology persisted. Substrate choice is another factor influencing crystallization behavior. , The distinct microstructures we observed likely result from the differing substrates used for TiO₂ coating. Additionally, coating thickness plays a key role in determining surface morphology. , , Thicker films tend to crystallize, with particle size and crystallization rate increasing with thickness. Beyond a threshold, fewer active O-sites reduce the crystallization rate. Building on our previous findings, the 100 nm ALD-TiO₂ coating not only reduced bacterial adhesion but also exhibited favorable biocompatibility, corrosion resistance, and a warm yellow hue, enhancing the aesthetics of dental abutments. To achieve similar performance and avoid the corrosion risks associated with TiCl₄ at high temperatures, we changed the precursor to TDMAT and conducted this ALD process in the present study. It can be found that the similar color, surface properties and antibacterial performance were observed. The decreased hydrophilicity was found in Ti+AO, Ti+ALD and ZrO 2 groups, these phenomena could stem from the relatively smooth surface of the ALD-TiO 2 and ZrO 2 , while they might also be attributed to the interaction between surface roughness and hydroxyl density. Further research is needed to explore the extent to which these factors influence water contact angles, as well as the potential involvement of other factors. Despite the diversity within oral microbial communities, the formation of plaque and the development of biofilms are continuous and dynamic processes. Streptococcus , a Gram-positive, facultative anaerobic bacteria with a short doubling time, has been established as an early colonizer on surfaces of teeth and dental implants. , The late colonizers, represented by slow-growing obligate anaerobes like P. gingivalis , gradually integrate into the oral biofilm, , coinciding with the decrease of early colonization bacteria. Hence, this study selected three bacteria: S. mutans and S. aureus , both Gram-positive early colonizers, and P. gingivalis , a Gram-negative late colonizer and obligate anaerobe. The evaluation of the antibacterial effect was determined by comparing the adhesion of bacteria to the Ti group with that of the other three groups. This study investigated the antibacterial properties of ZrO 2 and Ti, as well as ALD/AO-TiO 2 on Ti substrates via CFUs and MTT . The ZrO 2 and Ti+ALD groups showed comparable antibacterial properties, while Ti+AO group showed no antibacterial effect. Specifically, the lowest observed antimicrobial effect was about 50% ( S. aureus and P. gingivalis ), while the highest one was approximately 80% ( S. mutans ). The higher bactericidal efficacy against Gram-positive S. mutans (80%) compared to Gram-negative P. gingivalis (50%) was surprising, given the general belief that Gram-positive species are more resistant to mechanical rupture. This discrepancy could be attributed to their different modes of cell division, with S. aureus forming clusters and S. mutans forming chains. Consequently, S. mutans may encounter difficulty dividing laterally across nanostructured surfaces, exposing them to a larger surface area and potentially causing higher levels of membrane stress. The surface properties of materials play a crucial role in bacterial adhesion and growth, influenced by various factors such as surface roughness, wettability, charge, surface topography, and chemical composition. , The surface characteristics of the ZrO 2 group exhibited extreme smoothness, uniformity, and absence of pronounced grooves, and show the largest contact angle, indicating the poorest hydrophilicity. This may explain, in part, the reduced bacterial adhesion observed in the ZrO 2 group, aligning with findings from other in vitro and in vivo studies. Another research suggested that ZrO 2 is an amphoteric metal, potentially displaying a positive charge or being neutral. In contrast, the metal Ti surface has an isoelectric point of around 6.0 and carries a negative charge. Bacterial adhesion, facilitated by calcium ions, tends to occur more on negatively charged Ti surfaces, resulting in a localized positive charge. While most bacteria remain negatively charged, which promotes bacterial adhesion on the Ti surface. This elucidated the higher bacterial adhesion observed on Ti surfaces, contributing to the significantly lower bacterial adhesion in the ZrO 2 group within this study. However, bacterial adhesion is a complex process, demanding further investigation into the relative impact of these factors and potential synergistic effects. On another note, Ti-based substrate materials treated with AO and ALD exhibited an increase in water contact angle, indicating decreased hydrophilicity. Intriguingly, only the Ti+ALD group demonstrated antibacterial properties comparable to the ZrO 2 group. There are several explanations for this phenomenon. Firstly, ALD-TiO 2 effectively concealing deep and extensive scratches and fissures on Ti substrate surfaces. Moreover, the smoother and more hydrophobic surfaces potentially reduced bacteria adherence, , subsequently inhibiting bacterial capacity to withstand shear forces and proliferate on material surfaces. Secondly, ALD, as a nano-coating technique, leads to densely clustered nanoparticles on TiO 2 coatings. Prior research has highlighted that this augmentation in surface nanostructures diminishes bacterial adhesion by reducing bacterial anchoring points and enhancing fibronectin adsorption. , Hayles et al also demonstrated that spiked titanium nanostructures could eliminate anaerobic dental pathogens in both single-species and dual-species. Similarly, Lorenzetti et al indicated that in TiO₂-coated samples, the nanocrystals reduced the spacing between microasperities, introducing nanoroughness. This decreased the contact area between bacteria and the surface, resulting in up to 50% less bacterial adhesion compared to untreated titanium. Thirdly, ALD technology facilitates the concurrent production of anatase TiO 2 with precise temperature regulation, thereby resulting in heightened antibacterial efficacy in comparison to the amorphous layers of TiO 2 coatings formed on the Ti and Ti+AO group’s surfaces. , Dorkhan et al also confirm that early colonizer adherence diminishes notably on surfaces abundant in anatase (ALD-TiO 2 ). Another reason for the relatively poorer antibacterial performance of the Ti+AO group might be its similarity in roughness to the Ti group. Additionally, SEM images indicated a more porous surface morphology of AO-TiO 2 , which further facilitates bacterial adhesion. , The SEM images exhibited a lower abundance of three bacteria in Ti+ALD and ZrO 2 groups. Moreover, all the bacteria formed thick membrane with geometric three-dimensional complexity, suggesting the absence of obvious bactericide mechanism of ALD-TiO 2 . This may be attributed to the limitations of nano-TiO 2 as a bactericide, which is in nanoparticle form and relies on photofunctionalization. Apart from material surface physicochemical properties, competitive interactions among bacteria within mixed-species environments also might impact the extent of bacterial adhesion. This behavior might stem from bacterial competition for nutrients and/or inhibition by other species, with varying inhibitory activities among different bacterial species. For instance, Li et al observed that S. mutans biofilm cells predominated in mixed cultures due to shorter generation times, thus having an advantage in competition with other bacterial species. Tu et al also confirmed the supernatants from S. mutans markedly suppressed the growth and biofilm formation of P. gingivalis . Moreover, metabolic interactions and signaling molecules between bacteria contribute to the maturity of biofilms. Multi-species biofilms generate extracellular polymeric substances (EPS) that consist of polysaccharides, proteins, and extracellular DNA, aggregating them favorably in three-dimensional spatial arrangements. To better simulate the oral environment, considering bacterial proliferation time and nutrient competition, a mixed in vitro bacterial model comprising three species was constructed in the present study. Initial concentrations in the mixed model were set as 1×10 3 CFU/mL for S. mutans and S. aureus , and 1×10 8 CFU/mL for P. gingivalis , enabling the coexistence and reproduction of the three strains, which also aligns with the decline in the early colonization bacteria proportion when P. gingivalis participates in biofilm formation. Interestingly, SEM observations and MTT assay results indicated that bacterial adhesion on the surfaces of the four groups correlated with the trend observed for single-species adhesion. The only variation noted was in the CFU results for the mixed-species cultures, where the Ti+ALD group had a significantly lower CFU count than the ZrO 2 group ( p <0.01). However, the MTT assay showed similar metabolic activity between the two groups. This discrepancy arises from differences in the assay principles. The CFU assay measures the number of viable cells capable of colony formation, while the MTT assay assesses metabolic activity by measuring the reduction of MTT to formazan crystals, indicating cell viability. Therefore, a sample with high metabolic activity (reflected in MTT results) may not have a corresponding high CFU count, especially if some cells are alive but not replicating. Additionally, interactions in mixed-species cultures can affect bacterial growth and metabolism. For example, some oral bacterial strains, such as S.mutans , can inhibit growth of P. gingivalis through various mechanisms, , which could lead to a lower CFU count but sustained metabolic activity. Although the CFU results showed that the Ti+ALD group had significantly lower bacterial counts than the ZrO 2 group, the values for them were notably low compared to Ti+AO and Ti groups. Based on the mixed-species MTT results and the CFU and MTT results for single species, the antibacterial efficacy of the ALD-TiO₂ coating and the ZrO₂ surface remains comparable. This suggested that, in a more complex environment containing multiple bacteria, the Ti+ALD and ZrO 2 groups, influenced by the physicochemical properties of the material surface, still exhibit the least bacterial adhesion compared to the Ti and Ti+AO groups, making them ideal choices for dental abutment materials. However, although this study’s inclusion of three peri-implantitis-related bacterial species adds complexity compared to other single-species in vitro studies, limitations in the complexity of the oral microbial environment mean it is still distant from the human oral microbiome. Further investigations will aim to optimize bacterial growth conditions, simulate the surrounding environment more accurately, including microbial diversity and relationships, providing potential laboratory-based evidence for the selection of clinical abutment materials. Furthermore, the perimucosal integration surrounding the abutment serves as the initial defense against pathogenic infiltration leading to peri-implantitis. Consequently, the surface characteristics of the abutment profoundly influence this soft tissue response. Thus, the effects of nano-TiO 2 coatings prepared by ALD and AO technologies on soft tissue integration also need further investigation.
With the limitation of this study, the following conclusions were drawn. Nano-TiO 2 coatings prepared by ALD rendered the surface of dental Ti abutment dense, flat, smooth, and less hydrophilia, featuring an anatase phase. This modification significantly reduced the adhesion of S. aureus, S. mutans , and P. gingivalis , by at least 50%, comparable to the effect observed on ZrO 2 surfaces. Nano-TiO 2 coatings prepared by AO made the surface of dental Ti abutment less hydrophilia, characterized by various nano-sized pores within the oxide film. However, this alteration did not impact the adhesion of S. aureus, S. mutans , and P. gingivalis . The adhesion behavior of mixed-species bacteria (consisting of the aforementioned three species) closely mirrored that of single-species bacteria. It confirmed that, akin to ZrO 2 abutments, nano-TiO 2 coatings prepared by ALD on Ti abutments exhibited strong potential for application in preventing peri-implantitis. In contrast, the antibacterial efficacy of AO-TiO 2 abutments was not obvious, resembling that of Ti abutments.
|
Racial and Ethnic Diversity Among Obstetrics and Gynecology, Surgical, and Nonsurgical Residents in the US From 2014 to 2019 | 4a8d503d-bc60-46ae-95a7-9d31ce1f4c12 | 8134986 | Gynaecology[mh] | Health inequities are prevalent throughout US society. Within obstetrics and gynecology (OBGYN), Native American or Alaskan Native and non-Hispanic Black women are 3- to 4-fold more likely to have a pregnancy-related death compared with non-Hispanic White women. While addressing these disparities requires multiple strategies, one approach is to increase diversity among health care practitioners, as patient-physician racial/ethnic concordance is associated with increased patient satisfaction and higher levels of trust. Previous research, such as a 2020 study by Nieblas-Bedolla et al, indicates that the OBGYN workforce includes more underrepresented physicians compared with other specialties. By examining current and recent trainees, we can assess recruitment efforts and estimate the future racial and ethnic diversity of the physician workforce. In this cross-sectional study, we examine the contemporary composition and trends in race and ethnicity among OBGYN, surgical, and nonsurgical residents. The University of California, Davis, institutional review board determined that this cross-sectional study was not human participants research; therefore, it was exempt from review and informed consent. We followed the Strengthening the Reporting of Observational Studies in Epidemiology ( STROBE ) reporting guideline. In this cross-sectional study, we abstracted deidentified publicly available data on the race and ethnicity of OBGYN, surgical, and nonsurgical residents from the JAMA Medical Education reports from 2014 to 2019, when multiracial first appeared as a racial category. We utilized the American College of Surgeons definition of a surgical specialty. We analyzed categorical data using χ 2 tests and, for each race and ethnicity, used logistic regression to estimate the dependent variable as the change in odds of a resident identifying as a given race or ethnicity, across year and specialty. Given the small number of residents, we combined Native Hawaiian or Pacific Islander with Native American or Alaskan Native into a single Native category for analysis. We analyzed Hispanic ethnicity separately from race. We used SAS statistical software version 9.4 (SAS Institute) for statistical analysis. P values were 2-sided, and statistical significance was set at P < .05. Data were analyzed from 2014 to 2019. A total of 520 116 US medical residents from 2014 to 2019 were included in this study. For each year, OBGYN, surgical, and nonsurgical residents most commonly identified as White (eg, 2014-2015: 58 098 residents [59.3%] ), followed by Asian (eg, 2014-2015: 26 010 residents [26.6%]) . Native American or Alaskan Native and Native Hawaiian or Pacific Islander residents were the least represented in all residency categories (eg, 2014-2015: 325 residents [0.3%]). The racial and ethnic composition of residents varied among OBGYN, surgical and nonsurgical specialties each year, with higher proportions of OBGYN residents who identified as Black (eg, 2014-2015: 514 residents [10.2%]; P < .001) or Hispanic (eg, 2014-2015: 481 residents [9.6%]; P < .001) compared with surgical (eg, 2014-2015: 872 Black residents [4.7%]; P < .001; 1299 Hispanic residents [7.0%]; P < .001) and nonsurgical specialties (eg, 2014-2015: 4341 Black residents [5.8%]; P < .001; 5675 Hispanic residents [7.6%]; P < .001). Among OBGYN residents, we noted a decrease in White (odds ratio [OR], 0.96; 95% CI, 0.94-0.98) and Black residents (OR, 0.93; 95% CI, 0.90-0.96) and an increase in those categorized as other or unknown race/ethnicity (OR, 1.26; 95% CI, 1.22-1.31) across the 5-year period . In surgical specialties, there were decreases in White (OR, 0.97; 95% CI, 0.97-0.98) and Black residents (OR, 0.97; 95% CI, 0.95-0.99) and increases in multiracial (OR, 1.04; 95% CI, 1.02-1.07) and other or unknown residents (OR, 1.14; 95% CI, 1.12-1.16). Lastly, among nonsurgical residents, there was a decrease in White (OR, 0.97; 95% CI, 0.96-0.97) and Asian residents (OR, 0.98; 95% CI, 0.97-0.98), while there was an increase in multiracial (OR, 1.07; 95% CI, 1.06-1.08), other or unknown (OR, 1.17; 95% CI, 1.16-1.18), and Hispanic residents (OR, 1.02; 95% CI, 1.02-1.03). The findings of this cross-sectional study suggest that while OBGYN residencies had higher proportions of Black and Hispanic residents compared with surgical and nonsurgical specialties, proportions of Black OBGYN residents decreased, while the proportion of Hispanic residents remained unchanged. Furthermore, our findings indicate that the diversity of OBGYN resident physicians still lagged behind the changing US demographic characteristics: non-Hispanic Black people represent 13% of the population, and Hispanic people represent 18% of the population, and both groups are projected to continue increasing in the coming years. As a diverse workforce is more likely to provide care for underserved populations and foster higher levels of trust, efforts to recruit and retain underrepresented individuals from racial/ethnic minority groups, such as Black, Hispanic, Native American or Alaskan Native, Native Hawaiian or Pacific Islander individuals, within OBGYN are critical as we strive to close gaps in health disparities. , This study has some limitations, including that the database was comprised of self-reported data. Additionally, the other or unknown race/ethnicity category could indicate missing data or that residents did not identify with the listed racial/ethnic categories, thereby contributing to the downward trend in other racial categories. |
Neuropathological microscopic features of abortions induced by | 84fd33f0-f643-4293-9129-50697c827a3b | 4260183 | Pathology[mh] | In recent years, with the importation of sheep from abroad, the prevalence of many diseases, especially abortion diseases, has increased in Iran. Surveys on abortion diseases in domestic sheep have been carried out, but most were restricted to brucellosis, campylobacteriosis, coxiellaburnetii, salmonellosis, leptospirosis , neosporosis , toxoplasmosis and other diseases . Nevertheless, determining the viral cause of abortion in ovine is obscure, but can be improved with the proper sampling and testing, good communication between veterinarians and diagnostic labs and awareness of the current disease situation in a certain area through authorities’ notification. Therefore, pathologists and field veterinarians who play a very significant role in diagnosis and control should be kept up to date regarding the spread of individual viruses into new geographic areas. On the other hand, despite the importance of fetal viral infections in both humans and animals, many questions regarding mechanisms of transplacental transmission, virus spread within the fetus and the consequences of infection for target cells and the fetus as a whole remain unanswered . Whereas, the pathways of virus infection of the fetus and potential protective mechanisms, notably exerted by the innate immune system, are poorly understood despite the fact that transplacental virus infections account for considerable mortality and morbidity in both animals and humans . Pathologic studies can help to confirm the clinical diagnosis and further the understanding of the disease pathogenesis and are very useful in outbreak investigations . Outbreaks of congenital abnormalities in fetal or neonatal ruminants have been related to exposure of pregnant dams to a number of viruses, including pestiviruses , bunyaviruses , flaviviruses and arboviruses , such as Bluetongue (BT), Border disease virus (BDV) Wesselsbron (WSL), Rift Valley fever (RVF), Cache valley virus (CVV) and Akabane viruses (AKV) . These abnormalities included stillbirths, mummified fetuses, defects of the central nervous system and musculoskeletal problems. Moreover, the most defects, such as hydranencephaly, hydroencephaly, porencephaly and arthrogryposis and cerebellar hypoplasia, are usually associated with infection with mentioned viruses. In parallel, porencephaly and cerebellar hypoplasia among other congenital anomalies were described in aborted or newborn calves to cows experimentally infected with Wesselsbron disease . One report of hydranencephaly and arthrogryposis in sheep infected with Wesselsbron disease and Rift Valley fever viruses was described by , and also, in Akabane disease, necropsy findings of the aborted fetuses are mainly reported in the brain and include microcephaly, hydrocephalus, porencephaly and hydranencephaly . On the other hand, based on experimental and clinical studies performed by researchers, the histological hallmarks of most viral infections in the CNS are neuronal degeneration, perivascular cuffing by inflammatory cells and glial reactivity. Neuronal injury is characterized by central chromatolysis and swelling that progresses to necrosis. The inflammatory reaction is typically non-suppurative and perivascular cuffs mainly consist of lymphocytes, with fewer plasma cells and macrophages, and proliferating vascular adventitial cells. Focal or diffuse microgliosis and formation of glial nodules are characteristic features of viral infections . According to these studies , viral encephalitis is usually part of a systemic infection rather than the agent having a predilection for neural tissue; however, some viruses are neurotropic and a few multiply within, and cause damage to, the nervous system. Nevertheless, most infections are haematogenous, but some viruses use the fast axoplasmic transport system in nerves to assist invasion . In parallel, RVF virus-induced nonsuppurative encephalitis has been reported in natural infections in human beings and in experimentally infected gerbils and certain strains of rats, but the pathologic characterization of the central nervous system (CNS) lesions has not been described in RVF virus-infected ruminants. In a study, Weiss reported viral encephalitis in two lambs born to ewes vaccinated , and in other study, Maar et al. described a case of nonsuppurativeencephalitis in a RVF patient . Another case with encephalitis and retinitis was described by Alrajhi et al. , in these patients, the histopathological lesions in brains were characterized by focal necroses associated with an infiltration of round cells, mostly lymphocytes and macrophages, and perivascular cuffing . The aim of the study was to the neuropathological diagnostic features of naturally occurring, a suspected viral infection in the aborted and stillbirth lambs in North of Iran.
Ethics statement, animals and area All experiments described in this study were performed in full accordance with the guidelines for animal studies released by the National Institute of Animal Health. The present study was carried out in the different area located in Mazandaran province in the north of Iran (Including the cities of Amol, Ghaemshahr, Neka and Larijan). The number of pregnant sheep in farms varied from 15 to 400. We visited ewes ranches with an abortion rate over 50% for the past 1 year (From February 2012 to March 2013). Whereas, more than 50% of the flocks had experienced abortions, stillbirths and deformities of newborn lambs, but the adult sheep were not affected. Moreover, the sheep flocks comprised mainly indigenous breeds, such as White Mountain Sheep, Brown Mountain Sheep, Zel Breed Sheep and Black headed Mutton. History of the outbreak and blood sampling A total of 650 aborted fetuses including 793 pregnant ewes were studied from 8 flocks at different area in the Mazandaran province during the period of 2012–2013. In some cases, the blood samples from sheep and aborted fetuses were randomly collected from four different locations around Mazandaran province. After coagulation, sera were separated by centrifugation and stored at -20°C until serological testing. But, the results were negative for the detection of Brucella spp., Listeria spp., Campylobacter spp., Mycoplasma spp and other infectious agents such as viral, fungal and parasitic. Clinical samples and tissue collection Following macroscopic examination, brain and spinal cord were removed from each fetus. However, the condition of some of the fetuses was such that not all tissues could be collected. Systematic necropsy was performed to collect tissues, and all gross findings were recorded at necropsy by the pathologist. Furthermore, not all tissues were available from each case because the studied abortions occurred under natural conditions, where predation or degree of autolysis resulted in the failure to submit all tissues. Histopathological analysis Tissues collected at necropsy were processed and embedded in paraffin after 48–72 hours of fixation in neutral-buffered 10% formalin. Tissue was sectioned at 5 μm, stained with hematoxylin and eosin, and examined for lesions by light microscopy. Where the brain was available, 14 different sections (from cerebral lobes to medulla oblongata, including cerebellum) together with cervical, thoracic, lumbar and sacral spinal cord segments were studied. Finally, unfortunately, according to the existing facilities at the university, we conducted a limited number of necropsies of aborted fetuses.
All experiments described in this study were performed in full accordance with the guidelines for animal studies released by the National Institute of Animal Health. The present study was carried out in the different area located in Mazandaran province in the north of Iran (Including the cities of Amol, Ghaemshahr, Neka and Larijan). The number of pregnant sheep in farms varied from 15 to 400. We visited ewes ranches with an abortion rate over 50% for the past 1 year (From February 2012 to March 2013). Whereas, more than 50% of the flocks had experienced abortions, stillbirths and deformities of newborn lambs, but the adult sheep were not affected. Moreover, the sheep flocks comprised mainly indigenous breeds, such as White Mountain Sheep, Brown Mountain Sheep, Zel Breed Sheep and Black headed Mutton.
A total of 650 aborted fetuses including 793 pregnant ewes were studied from 8 flocks at different area in the Mazandaran province during the period of 2012–2013. In some cases, the blood samples from sheep and aborted fetuses were randomly collected from four different locations around Mazandaran province. After coagulation, sera were separated by centrifugation and stored at -20°C until serological testing. But, the results were negative for the detection of Brucella spp., Listeria spp., Campylobacter spp., Mycoplasma spp and other infectious agents such as viral, fungal and parasitic.
Following macroscopic examination, brain and spinal cord were removed from each fetus. However, the condition of some of the fetuses was such that not all tissues could be collected. Systematic necropsy was performed to collect tissues, and all gross findings were recorded at necropsy by the pathologist. Furthermore, not all tissues were available from each case because the studied abortions occurred under natural conditions, where predation or degree of autolysis resulted in the failure to submit all tissues.
Tissues collected at necropsy were processed and embedded in paraffin after 48–72 hours of fixation in neutral-buffered 10% formalin. Tissue was sectioned at 5 μm, stained with hematoxylin and eosin, and examined for lesions by light microscopy. Where the brain was available, 14 different sections (from cerebral lobes to medulla oblongata, including cerebellum) together with cervical, thoracic, lumbar and sacral spinal cord segments were studied. Finally, unfortunately, according to the existing facilities at the university, we conducted a limited number of necropsies of aborted fetuses.
In the most cases, arthrogryposis was the most common musculoskeletal defects (Figure ). At postmortem examination significant gross changes were seen in the brains. Malformations of the brain included unilateral or bilateral internal hydrocephalus, characterized by dilated ventricles/or destruction of adjacent neuroparenchyma but still distinguishable gray and white matter; and hydranencephaly (Figure ), which characterized by segmental or complete loss of the cerebral cortex without discernable gray and white matter; and cerebellar hypoplasia was severe (Figure ), so that only the brain stem, including pons and medulla oblongata were distinguishable. Porencephalies of the brain stem and gray and white matter of the brain were found in some animals, so that the subcortical cavitations throughout both cerebral hemispheres were also noted (Figure ). No abnormalities were seen in other tissues. In our study, microscopic lesions are mostly confined to throughout the brain and the white and/or gray matter of the brain stem, particularly the pons and the medulla oblongata, and the spinal cord, but, in some cases, CNS lesions mainly identified in the cerebral hemispheres, periventricular areas, midbrain, cerebellum, brainstem and occasionally in the spinal cord. Furthermore, the distribution and severity of lesions in the brain varied among multifarious cases. Microscopically, inflammation in the CNS characterized by a lymphoplasmacytic , mainly perivascular, infiltration of the gray and white matter in most investigated brain areas. Perivascular cuffs ranged from multifarious layers of mononuclear cells. In parallel, encephalitic changes were detected in most aborted fetuses. In these cases, a mild to moderate non-suppurative encephalitis, characterized by foci of perivascular cuffing with mononuclear cells, predominantly lymphocytes, was observed (Figures and ). The perivascular cuffing was generally associated with a mild to moderate gliosis (focal or diffuse), these changes were most prominent in the cranial brain regions and occasionally the caudal brain regions (pons and medulla). On the other hand, glial nodules (Figure ) formed predominantly in the mesencephalon, thalamus, hippocampus, pons and medulla oblongata. Nevertheless, mild-to-moderate focal-to-multifocal gliosis associated with acute neuronal necrosis was observed for most cases (Figure ). Additionally, Virchow–Robin spaces were expanded by cuffs of lymphocytes and plasma cells, admixed with fewer histiocytes and neutrophils. Moreover, multifocal hemorrhages were seen in several cases. These lesions varied in age within each case and between cases. On the other hand, the most common lesions observed in all infected fetuses which studied were multiple small to large areas of microcavitation/or cyst (Figure ). These cavities , typically were most common in the cortex. In some cases, in the CNS of aborted fetuses, there was mild to moderate cavitation of the cortex and subcortical white matter (Figure ). Also, in others, severe cavitation was also observed in the white matter of the cerebellar and cerebellum and also in the adjacent gray matter and cervical spinal cord. However, mild cavitation was also observed in the pons, and lumbar spinal. In some areas, acute neuronal necrosis (Including the eosinophilic/or hypereosinophilic cytoplasm with nuclear pyknosis (ischemic cell changes) and central chromatolysis were mainly in the cortex, subcortical and brainstem regions (Figure ), and also, in some fetuses, cellular necrosis were observed in the thalamus and temporal cortex in aborted fetuses. In these lesions, degenerating neurons and necrosis can be found, but any nuclear or cytoplasmic inclusions are not observed. The most consistently involved regions are cerebral gray matter and brainstem, closely followed by leptomeninges, as well as, in particular, necrosis neurons were observed in the ventral horns of spinal cord (Figure ) and mainly were found in arthrogryposis fetuses. In the more severe cases, cerebellum is also involved. Whereas, in most fetuses, there were multiple and discrete foci of cortical necrosis, with loss of the neuropil, increased prominence of blood vessels, gliosis, and microcavitation. The affected blood vessels were lined by hypertrophied endothelial cells and sometimes perivascular edema (Figure ). The edema was severe in some areas and resulted in marked loosening of the neuropil, resulting in the formation of cavitations. These lesions had variable distribution among the cases and among the various sections of brain regions in each case. Briefly ,histopathologic findings in the brain and spinal cord included hyperemia, hemorrhage, non-suppurative encephalitis , mononuclear perivascular cuffing, multifocal gliosis, microcavitation,central chromatolysis (Figure ), neuronal degeneration and necrosis, perineuronal and perivascular edema and necrotic neurons in ventral horn gray matter of spinal cord were present in the all regions of the brain and spinal cord.
Evaluating the areas at risk for the introduction of a new pathogen is challenging. Nevertheless, given the possibility of severe consequences on public and animal health associated with the introduction of a pathogen such as viral infections, the veterinary experts require suitable information on where and how to target surveillance and preventive actions . Based on available data, few studies have been carried out to investigate neuropathological changes after viral infection in the aborted sheep fetuses and also, it is not known why these aborted fetuses demonstrated a different extent of viral infection. Therefore, in parallel, histopathology has been utilized as the gold standard for diagnosis of viral infection, it is well recognized that false-negative results can occur based on the uneven distribution of lesions, particularly in clinical biopsy . Previous studies have shown that the most fetal infections with viral causes result in persistent subclinical infection, fetal death, or defects such as cerebellar hypoplasia, hydranencephaly, internal hydrocephalus, microencephaly, and porencephaly. Therefore, these observations are similar to those described in our study. Nevertheless, a number of viruses, including pestiviruses , bunyaviruses , flaviviruses and arboviruses are as a teratogenic causes, such as RVF, WSL, CVV, AKV, BTV and pestiviruses like border disease virus (BDV) . All these virus infections show similar gross findings including cerebellar hypoplasia, por- or hydranencephaly and skeletal malformations like brachygnathia and arthrogryposis of in utero-infected neonates . Our results revealed that these malformations occurred in similar high percentages in aborted sheep fetuses with and without CNS inflammation. Cerebellar hypoplasia, porencephaly and hydranencephaly represented the most frequently detected malformations in aborted fetuses together with skeletal malformations like arthrogryposis. In addition to the gross lesions,porencephaly was also detected by light microscopy mainly in the cortex and subcortical white matter. In severe cases, the white matter of the cerebellar and cerebellum was also affected by formation of such cavities. In humans, the occurrence of multiple cysts in the brain due to a hypoxic-ischemic pathogenesis has been described. This entity is termed multicystic encephalopathy . The pathological changes associated with viral infection in ruminants seem to fit the description of multicystic encephalopathy. While a variety of exogenous and endogenous substances are capable of inducing an inflammatory response, a useful principle of neuropathology is that bacterial infections are associated with suppurative inflammation while viral infections are associated with nonsuppurative inflammation . Accordingly, the nonsuppurative encephalitis in the aborted ovine fetuses in the present study has the histological hallmarks of a viral infection of the central nervous system: neuronal degeneration and necrosis, reactivity of the glia, and perivascular cuffing with lymphocytes and histiocytes. Furthermore, studies indicated that the variations in the histopathological characteristics of the inflammatory response were detected between animals and anatomical sites and malacia was the most commonly seen feature, but infiltrative or vascular patterns, with malacia, were also detected. Most of these reports however, were based on experimental lesions resulting from the injection of virus into the CNS. Although the pathological investigations of viral encephalitis vary somewhat depending on the specific infectious agent and the immunologic status of the aborted fetuses, most viral infections of the CNS are characterized by a triad of findings including perivascular chronic inflammation, microglial nodules, and neuronal necrosis. Therefor, the mentioned cases are in agreement with our study, moreover, expressed lesions were observed in many samples of our study, and also, the distribution of these findings as well as the presence of characteristic intranuclear or intracytoplasmic viral inclusions can lead to a specific diagnosis in an appropriate clinical setting . Ancillary techniques, including immunohistochemistry (IHC), in-situ hybridization (ISH), or polymerase chain reaction (PCR) amplification, are useful in some settings. These cases are, in contrast, with our observations that not were detected intranuclear or intracytoplasmic viral inclusions in the aborted fetuses. In general, in most conducted studies with viral agents on the aborted ovine fetuses, histologic lesions consisted the focal/or multifocal nonsuppurative encephalitis together with the areas of necrosis and loss of the neuronal and motor neurons, cavitation, gliosis ,perivascular and perineural edema at various neuroanatomic sites of the brain and spinal, therefore, in parallel ,based on our study, the mentioned lesions are similar to those described . Based on these findings, the gross and histologic examination of the brains appears to be important, and viral evaluate may be useful in the postmortem investigation of fetuses with a history of clinical signs referable to the brain.
In conclusion, therefore, we believe that the histopathological pattern using detected in this study could be associated with either viral infection and or mainly by a Bunyavirus / or Flavivirus strains that extensively shares common lesions with rift valley fever, WSL and CVV. The true sources of these infections are not known, however, a link between the infected sheep and the condition described here could be suggested. Additional data on these cases are not available because much time has elapsed since it occurred. However, from the history and diagnostic findings on these cases, the etiologic role of Bunyavirus / or Flavivirus families are a plausible conclusion, thus making these cases the first well-documented evidence of the occurrence of these condition in Iran. And because of the known neurotropism and histologic description of non suppurative encephalitis in viral–infected fetuses, these agents were considered a possible etiologic agent. Finally, our study suggested that the aborted/and or infected sheep fetuses are fully susceptible to viral infections and may even develop neurological disease upon natural inoculation of mentioned pathogens. To our knowledge, these are the first direct evidences of the susceptibility to viral causes of aborted fetuses in the north of Iran.
|
Size-controllable synthesis of hydroxyapatite nanorods via fluorine modulation: applications in dental adhesives for enhanced enamel remineralization | dbdfc594-7846-44b5-b069-fbeddd2feca5 | 11806586 | Dentistry[mh] | Enamel demineralization is a common concern in clinical dental practice, particularly during orthodontic treatment. It may arise from various factors including inadequate nutrition, poor oral hygiene, and improper bonding techniques . Enamel demineralization leads to the susceptibility to tooth decay due to enamel fragility and reduced acid resistance . However, demineralization and remineralization keep a dynamic balance, which means HA crystals, once partially demineralized, can restore their original size under favorable conditions . Therefore, many mineralizing agents have been developed to facilitate remineralization. Current dental adhesives lack mineralizing components, and recent advancements have focused on incorporating nano-calcium phosphorus complex (nCaP) and other mineralization agents into adhesives. Nevertheless, limitations have been discovered in these innovations, such as short-term ion release, heterogeneous filler dispersity, and unsatisfactory remineralization performance . HAs have been widely used in bone regeneration and dentistry due to their excellent mechanical properties, biocompatibility, and mineralization capability . A biomimetic nano-hydroxyapatite mineralizing solution has been reported to enhance the microhardness values of the demineralized enamel surface . Fluoride plays a crucial role in preventing enamel dissolution and degradation by accelerating the formation of larger fluorapatite crystals . Fluoride is also known to promote remineralization and improve the acid resistance of tooth enamel, thereby preventing dental caries . HAs and their F-substituents are the primary inorganic components in teeth and bones . Partially F-substituted HAs have been reported to be more stable than pure hydroxyapatites , as fluorine ions integrate better into the lattice than -OH due to the higher net negative charge and charge symmetry . HA coatings with higher fluoride content exhibited higher apatite deposition efficiency in simulated body fluid, representing superior mineralization . FHAs have been used in various biological agents, including dental sealers , to enhance tooth remineralization. However, there is little research on their application in enamel adhesives to date. Previous studies primarily focused on the effect of fluorine content on enamel remineralization . However, the microscopic shape, size, and size distribution of the apatite crystals significantly affect their mechanical properties, chemical properties, and mineralization performance . The partially F-substituted HA with a plate-like microstructure was reported to be less effective than the bioactive glass (BAG) for remineralization, attributed to its irregular shape, excessive size, and significant agglomeration . The unique one-dimensional (1-D) structure of HA nanorods existing in natural teeth enhances both ductility and hardness . HA nanorods with a higher surface-to-volume ratio have more surface atomic active sites, leading to an increased release rate of specific ions . However, few studies have reported the effects of FHA nanorod size on enamel remineralization. To date, uniform hydroxyapatite nanorods with varying aspect ratios and surface properties have been prepared using hydrothermal and chemical precipitation methods . Most studies have used organic modifiers to achieve the size-controllable synthesis of nanorods . However, these approaches needed extensive washing and purification to suit biological applications, and could lead to non-uniform distributions in size and shape, which affect the reproducibility of the desired nanorod characteristics and impact mineralization . It was reported that the incorporation of F in the lattice of HA resulted in crystallites close in size to bone-like apatite and orientate along the a-axis rather than the c-axis . Consequently, modulating the size of FHA nanorods through the fluorine doping level may represent a promising strategy. In this study, uniform FHA nanorods of varying sizes were successfully synthesized by adjusting the fluorine doping level. For the first time, the synergistic effects of the fluoride content and FHA nanorod size on enamel remineralization were investigated by incorporating the nanorods into adhesives. Additionally, novel composite adhesives were developed, and their physical properties, SBS, as well as mineralization performance were evaluated. The null hypothesis of the study was that the composite adhesives containing FHA nanorods would not maintain adequate shear bond strength or enhance enamel remineralization. Sample size calculation The sample size was calculated using G*Power 3.1. The study aimed for a type I error (α) of 0.05 and a type II error (β) of 0.2. According to previous studies, the effect sizes in this study were estimated at 0.5 for the SBS test and the degree of conversion (DC %) test , 0.6 for the ion release test , and 0.8 for the mineralization test ; the sample sizes for each group were n = 12 for the SBS test, n = 5 for the DC % test, n = 3 for the ion release test, n = 6 for both the in vitro and in vivo EDS tests. For qualitatively analyzing the crystalline phase by XRD, n = 3 was needed for each experiment . Synthesis of FHA nanorods with different sizes The synthesis was based on a hydrothermal method . Quantity of precursors for synthesizing different groups of HAs were listed in the Supplementary Table . Briefly, octadecane-1-amine (2 g) was dissolved in oleic acid (16 mL) at 40–50 ℃, followed by the addition of ethanol (64 mL) at room temperature. Then this solution was mixed with Ca (NO 3 ) 2 .4H 2 O (2 g / 30 mL of distilled water), NaF (0.02 / 0.06 / 0.1 g / 20 mL of distilled water, the final products of each group labeled H1, H2, and H3, respectively), and Na 3 PO 4 .12H 2 O (2 g / 30 mL of distilled water) in order while stirring. The stirring was continued for 30 min. Next, the mixture was transferred to a 200 mL autoclave, followed by hydrothermal heating at 200 ℃ for 10 h before cooling naturally. The final products, which were collected at the bottom of the vessel, were washed with ethanol and distilled water, lyophilized, and sieved using a 400-mesh screen. Based on this process, the fluorine contents in the final products H1, H2, and H3 were estimated to be 2 wt%, 6 wt%, and 10 wt%, respectively. Characterization of FHA nanorods The microstructure of the FHA nanorods was characterized using TEM (JEM 2100 F, JEOL, Japan), and the element ratio was analyzed using EDS. ImageJ software was used to measure the length, diameter, and aspect ratio of the nanorods. The crystalline phase of the nanorods was determined using XRD (SmartLab SE, Rigaku, Japan) with Cu-Ka radiation working at 10 mA, 30 kV, and a scan speed of 10°/min in the 2θ range of 10°–80°. FTIR (Nicolet iS20, Thermo Fisher Scientific, USA) was used to detect the chemical compositions of the nanorods with the KBr pellet method. Data were collected at a resolution of 4 cm − 1 and wavenumbers ranging between 400 and 4000 cm − 1 . A zeta analyzer (Zetasizer Nano ZS90, Malvern, UK) was used to measure the zeta potential of FHA dispersed in distilled water. Preparation of composite adhesives An experimental adhesive resin matrix was prepared as follows. The analytically pure materials of the methyl propionic acid (MPA)-modified epoxy resin, methacrylate monomer, phosphate ester, and camphor quinone were procured from Shanghai National Pharmaceuticals. The MPA-modified epoxy resin, methacrylate monomer, and phosphate ester were mixed at a 30%: 45%: 25% weight ratio, along with camphor quinone (0.5 wt% of the above mixture), and stirred at 30 °C for 30 min. Next, this mixture was warmed to 60 °C and stirred for another 30 min. Finally, the mixture was allowed to cool naturally, resulting in the experimental adhesive labeled as A0. Based on a previous published procedure , the synthesized FHA nanorods were incorporated into the adhesive solution at a weight ratio of 10%, followed by ultrasonication for 30 min to allow uniform dispersion of the powders in the adhesive. The composite adhesives containing 10 wt% of H1, H2, and H3 HA powders were labeled AH1, AH2, and AH3, respectively. Characterization of composite adhesives The adhesive sample was cured and molded into a cylindrical shape (diameter: 4 mm, height: 2 mm), followed by cutting into pieces perpendicular to the exposed surface and drying. Then the cross-sections were sputter-coated with gold and observed via SEM (Regulus 8100, Hitachi, Japan) operating at 10 kV in the secondary electron mode. EDS was used to map the fluorine distribution to assess the distribution of FHA nanorods in the adhesive with an accelerating voltage of 20 kV. Colloidal stabilit y The FHA nanorods were dispersed in the experimental adhesive at a weight ratio of 10%, followed by ultrasonication for 10 min to obtain the specimens to be tested. A separation analyzer (MS20, Dataphysics, Germany) was used to measure the transmittance at evert height of the test tube using near-infrared light with a wavelength of 870 ± 30 nm and a scanning frequency of 10 s − 1 for 2 h. Evaluation of the light-curing property The tilt test was used to assess the changes in the state of the adhesives before and after curing . 3 mL of different composite adhesives were added to some glass bottles, standing for 30 min and irradiated with a light source (wavelength: 420–480 nm; intensity: 1200–2000 mw/cm 2 ; LY-A180, Liangya, China) from a distance of 2 cm above the center of the bottom of the bottle for 0 s, 5 s, 10 s, and 20 s respectively. Next, the bottles were tilted for 5 min to check the flow of the adhesives. The curing state was recorded using a digital camera (Z30 Double Kit, Nikon, Japan) at a distance of 40 cm, with an exposure time of 0.02 s and a resolution of 4096 × 2730 pixels. All the procedures were conducted in the darkness at 25 ℃, except for the use of a flash during photography. Degree of conversion (DC %) test A hollow cylindrical silicone with a diameter of 5 mm and a height of 1 mm was affixed onto a Mylar strip (Quantum mylar, 10 mm width, Medi-Dent, Australia) . This apparatus was then placed on a glass slide (Sail brand, China) with a thickness of 1 mm. Adhesive samples were filled into the silicone using a pipette. The slides were placed on a support, and the light-curing tip was fixed 2 cm above the center of the adhesive. The adhesives were photopolymerized for 5 s, 10 s, and 20 s, respectively. This test was conducted in the darkness. After that, uncured samples and samples with different curing times were determined by FTIR (Nicolet iS20, Thermo Fisher Scientific, USA) to assess the DC % . The formula for calculating the conversion degrees is as follows : [12pt]{minimal} $$\:DC\:=[1-\:({I}_{1}/{I}_{2})/\:({I}_{3}/{I}_{4})]\:\:\:100$$ I 1 : C = C peak intensity at 1638 cm − 1 (After curing); I 2 : C-C peak intensity at 1608 cm − 1 (After curing); I 3 : C = C peak intensity at 1638 cm − 1 (Before curing); I 4 : C-C peak intensity at 1608 cm − 1 (Before curing). Demineralized enamel specimen preparation Bovine incisors free of any cracks or lesions were obtained from cattle that were slaughtered in accordance with Chinese animal slaughter regulations at permitted markets. The bovine sacrifice was not related to this study and was conducted solely for the purposes of food processing. Their periodontal membranes and calculus were removed. A high-speed diamond disc was used to remove the tooth roots. The enamel surfaces were polished with 240-, 800-, and 1500-mesh SiC sandpaper under running water, followed by ultrasonication for 10 min to remove the smear layer. The crown specimens were stored in 0.2% thymol solution for further experiments . Demineralized models were prepared according to previously established methods . Briefly, each specimen was immersed in 37% phosphoric acid (10 ml) for 45 s, washed with distilled water three times, sonicated for 5 min, and then used for subsequent experiments. SBS test Forty-eight demineralized tooth crowns were randomly divided into four groups ( n = 12). Each crown was embedded in self-curing acrylic resin (New Century Dental Materials, Shanghai, China) perpendicular to the buccal axis of the crown to align the buccal surfaces with the applied forces during testing. The demineralized enamel surfaces were dried with an oil- and moisture-free air stream from a standard dental 3-way syringe (Waldent, Dentalstall, India) for 10 s until they became chalky , and then they were applied with different composite adhesives, followed by the placement of premolar orthodontic metal brackets (8216-20B, Shinye, China) with a base area of 9.24 mm 2 on the prepared surfaces, curing for 20 s. These samples with brackets were placed in distilled water at 37 °C for 24 h before testing for shear bond strength. The test was performed using a universal testing machine (EZ20, LLOYD, UK) with a displacement rate of 0.5 mm/min (Supplementary Fig. ). A previously described method was used to calculate the SBS . Failure mode analysis The failure modes were assessed using stereomicroscope loupe (S9E, Leica, Germany) under 40 × magnification. The samples were classified into the adhesive failure mode (failure occurring at the adhesive interface), the mixed failure mode (failure occurring at the adhesive interface, with adhesive residues left on the enamel surface), and the cohesive failure mode (failure occurring within the enamel, resulting in an undercut) . Degradation and release of ions The adhesive sample was cured and molded into a cylindrical shape (diameter: 1 cm; height: 1 mm). Then the samples were separately soaked in 2 mL of distilled water and were placed in a 37 °C shaker. The supernatant (1 mL) was removed on days 1, 5, 10, 15, and 30, respectively, and replaced by the same amount of fresh distilled water. A pH meter was used to determine the pH values of the supernatant. The ICP (720ES(OES), Agilent, the US) and IC (ICS-5000+, Thermo Fisher Scientific, the US) were used to determine the concentrations of Ca 2+ and F − , respectively. The weight loss ratio was calculated using the initial weight (w 0 ) and the final weight after incubation (w 1 ) of dried samples, according to the following formula: (w 1 -w 0 ) / w 0 %. In vitro mineralization performance test Different adhesives (200 µL) were added to each well of a 24-well culture plate and cured for 60 s. Next, 1 ml of the simulated oral fluid (SOF) was added to each well and was refreshed every 24 h. The chemical composition of the SOF included 1.13 mM CaCl₂•2 H₂O, 0.29 mM MgCl₂•6 H₂O, 8.38 mM KCl, 4.62 mM K₂HPO₄, and 2.40 mM KH₂PO₄. The pH was adjusted to 6.8 using Tris, and no precipitation was observed in the SOF throughout the experimental duration . The samples were placed in a 37 °C shaker to mimic the oral physiological environment. They were collected on day 7, followed by sonication in distilled water for 2 min and dried. Then they were gold coated. SEM (Regulus 8100, Hitachi, Japan) was used to analyze the morphology of mineral crystals operating at 10 kV. The mineralized sites were chemically characterized using EDS operating at 20 kV. The regenerated crystals on the surfaces of the AH1, AH2 and AH3 specimens were analyzed by XRD (SmartLab SE, Rigaku, Japan) with Cu-Ka radiation at 10 mA, 30 kV, and a scan speed of 10°/min in the 2θ range of 10–80. In vivo animal study To further validate the results of the in vitro experiments, in vivo experiments were conducted to evaluate the remineralization capabilities of different groups of adhesives in a realistic oral environment. The study received ethics approval from the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (China) (SH9H-2024-A1221-1), and all methods were conducted in accordance with relevant guidelines and regulations. Each tooth crown was cut into an enamel square disc (0.5 cm × 0.5 cm). A 1-mm-diameter hole was made using a drill for fixation. Twelve Sprague-Dawley rats (male, aged 8 weeks, 250–300 g) were randomly divided into four groups. The rats were purchased from Shanghai JieSiJie Laboratory Animal Company (Shanghai, China). Before the experiment, the rats were kept in a standard experimental environment for one week to acclimate to their surroundings. The rats were anesthetized with an intraperitoneal injection of Zoletil 50 (tilidine hydrochloride and midazolam hydrochloride, 80 mg/kg, Virbac, France). Two enamel samples applied with the adhesives were fixed adjacent to the two third molar teeth, respectively, using ligature wire to ensure their stability. Furthermore, the animals’ food supply was pulverized to prevent any potential harm to the samples. The incubation period was set at 7 days to ensure consistency with the in vitro experiments. After 7 days, all the rats were euthanized using carbon dioxide, and all the samples were collected. A low-speed dental burr was used to carefully remove adhesives on the sample surface. Next, the surfaces were polished using 1500-mesh SiC sandpaper under running water, followed by ultrasonication for 10 min to remove the smear layer. After drying the samples, the surfaces of the specimens were analyzed by XRD (SmartLab SE, Rigaku, Japan) with Cu-Ka radiation at 10 mA, 30 kV, and a scan speed of 10°/min in the 2θ range of 10°–80°. The surface micro-hardness was measured using a Vickers hardness testing machine (HXD-1000TMC, Taiming, China) with an applied load of 1000 gf and a dwell time of 10 s. Each sample was tested for five times and the average value was calculated. The surface micro-hardness of each sample before and after demineralization was also measured. Then their surfaces were sputter-coated with Au. SEM (Phenom Pharos G2, Phenom, the US) was used to observe the morphology operating at 5 kV. EDS was used to determine the chemical compositions of the remineralization sites operating at 15 kV. Finally, each enamel sample was cut into two halves perpendicular to the exposed surface to enable the observation of the cross sections via SEM as introduced above. Statistical analysis All statistical data were analyzed using SPSS software version 27.0. The normality of the data distribution was evaluated by Shapiro-Wilk test. The equality of variances was assessed by Levene test. The mean values of SBS, weight loss ratios, atomic percentages of Ca, and surface micro-hardness were compared using one-way ANOVA with post-hoc Tukey’s test, based on their variance equality. Data on the release of Ca 2+ and F − conformed to the criteria for normal distribution but failed in the homogeneity of variance. Then Welch’s analysis of variance with Tamhane’s test was used. The failure mode was shown in absolute frequency . The association between adhesive groups and failure mode was analyzed with the Fisher’s exact test, while the differencess of failure modes between different groups were analyzed using Kruskal-Wallis test. The level of statistical significance was set at α = 0.05. The sample size was calculated using G*Power 3.1. The study aimed for a type I error (α) of 0.05 and a type II error (β) of 0.2. According to previous studies, the effect sizes in this study were estimated at 0.5 for the SBS test and the degree of conversion (DC %) test , 0.6 for the ion release test , and 0.8 for the mineralization test ; the sample sizes for each group were n = 12 for the SBS test, n = 5 for the DC % test, n = 3 for the ion release test, n = 6 for both the in vitro and in vivo EDS tests. For qualitatively analyzing the crystalline phase by XRD, n = 3 was needed for each experiment . The synthesis was based on a hydrothermal method . Quantity of precursors for synthesizing different groups of HAs were listed in the Supplementary Table . Briefly, octadecane-1-amine (2 g) was dissolved in oleic acid (16 mL) at 40–50 ℃, followed by the addition of ethanol (64 mL) at room temperature. Then this solution was mixed with Ca (NO 3 ) 2 .4H 2 O (2 g / 30 mL of distilled water), NaF (0.02 / 0.06 / 0.1 g / 20 mL of distilled water, the final products of each group labeled H1, H2, and H3, respectively), and Na 3 PO 4 .12H 2 O (2 g / 30 mL of distilled water) in order while stirring. The stirring was continued for 30 min. Next, the mixture was transferred to a 200 mL autoclave, followed by hydrothermal heating at 200 ℃ for 10 h before cooling naturally. The final products, which were collected at the bottom of the vessel, were washed with ethanol and distilled water, lyophilized, and sieved using a 400-mesh screen. Based on this process, the fluorine contents in the final products H1, H2, and H3 were estimated to be 2 wt%, 6 wt%, and 10 wt%, respectively. The microstructure of the FHA nanorods was characterized using TEM (JEM 2100 F, JEOL, Japan), and the element ratio was analyzed using EDS. ImageJ software was used to measure the length, diameter, and aspect ratio of the nanorods. The crystalline phase of the nanorods was determined using XRD (SmartLab SE, Rigaku, Japan) with Cu-Ka radiation working at 10 mA, 30 kV, and a scan speed of 10°/min in the 2θ range of 10°–80°. FTIR (Nicolet iS20, Thermo Fisher Scientific, USA) was used to detect the chemical compositions of the nanorods with the KBr pellet method. Data were collected at a resolution of 4 cm − 1 and wavenumbers ranging between 400 and 4000 cm − 1 . A zeta analyzer (Zetasizer Nano ZS90, Malvern, UK) was used to measure the zeta potential of FHA dispersed in distilled water. An experimental adhesive resin matrix was prepared as follows. The analytically pure materials of the methyl propionic acid (MPA)-modified epoxy resin, methacrylate monomer, phosphate ester, and camphor quinone were procured from Shanghai National Pharmaceuticals. The MPA-modified epoxy resin, methacrylate monomer, and phosphate ester were mixed at a 30%: 45%: 25% weight ratio, along with camphor quinone (0.5 wt% of the above mixture), and stirred at 30 °C for 30 min. Next, this mixture was warmed to 60 °C and stirred for another 30 min. Finally, the mixture was allowed to cool naturally, resulting in the experimental adhesive labeled as A0. Based on a previous published procedure , the synthesized FHA nanorods were incorporated into the adhesive solution at a weight ratio of 10%, followed by ultrasonication for 30 min to allow uniform dispersion of the powders in the adhesive. The composite adhesives containing 10 wt% of H1, H2, and H3 HA powders were labeled AH1, AH2, and AH3, respectively. The adhesive sample was cured and molded into a cylindrical shape (diameter: 4 mm, height: 2 mm), followed by cutting into pieces perpendicular to the exposed surface and drying. Then the cross-sections were sputter-coated with gold and observed via SEM (Regulus 8100, Hitachi, Japan) operating at 10 kV in the secondary electron mode. EDS was used to map the fluorine distribution to assess the distribution of FHA nanorods in the adhesive with an accelerating voltage of 20 kV. y The FHA nanorods were dispersed in the experimental adhesive at a weight ratio of 10%, followed by ultrasonication for 10 min to obtain the specimens to be tested. A separation analyzer (MS20, Dataphysics, Germany) was used to measure the transmittance at evert height of the test tube using near-infrared light with a wavelength of 870 ± 30 nm and a scanning frequency of 10 s − 1 for 2 h. The tilt test was used to assess the changes in the state of the adhesives before and after curing . 3 mL of different composite adhesives were added to some glass bottles, standing for 30 min and irradiated with a light source (wavelength: 420–480 nm; intensity: 1200–2000 mw/cm 2 ; LY-A180, Liangya, China) from a distance of 2 cm above the center of the bottom of the bottle for 0 s, 5 s, 10 s, and 20 s respectively. Next, the bottles were tilted for 5 min to check the flow of the adhesives. The curing state was recorded using a digital camera (Z30 Double Kit, Nikon, Japan) at a distance of 40 cm, with an exposure time of 0.02 s and a resolution of 4096 × 2730 pixels. All the procedures were conducted in the darkness at 25 ℃, except for the use of a flash during photography. A hollow cylindrical silicone with a diameter of 5 mm and a height of 1 mm was affixed onto a Mylar strip (Quantum mylar, 10 mm width, Medi-Dent, Australia) . This apparatus was then placed on a glass slide (Sail brand, China) with a thickness of 1 mm. Adhesive samples were filled into the silicone using a pipette. The slides were placed on a support, and the light-curing tip was fixed 2 cm above the center of the adhesive. The adhesives were photopolymerized for 5 s, 10 s, and 20 s, respectively. This test was conducted in the darkness. After that, uncured samples and samples with different curing times were determined by FTIR (Nicolet iS20, Thermo Fisher Scientific, USA) to assess the DC % . The formula for calculating the conversion degrees is as follows : [12pt]{minimal} $$\:DC\:=[1-\:({I}_{1}/{I}_{2})/\:({I}_{3}/{I}_{4})]\:\:\:100$$ I 1 : C = C peak intensity at 1638 cm − 1 (After curing); I 2 : C-C peak intensity at 1608 cm − 1 (After curing); I 3 : C = C peak intensity at 1638 cm − 1 (Before curing); I 4 : C-C peak intensity at 1608 cm − 1 (Before curing). Bovine incisors free of any cracks or lesions were obtained from cattle that were slaughtered in accordance with Chinese animal slaughter regulations at permitted markets. The bovine sacrifice was not related to this study and was conducted solely for the purposes of food processing. Their periodontal membranes and calculus were removed. A high-speed diamond disc was used to remove the tooth roots. The enamel surfaces were polished with 240-, 800-, and 1500-mesh SiC sandpaper under running water, followed by ultrasonication for 10 min to remove the smear layer. The crown specimens were stored in 0.2% thymol solution for further experiments . Demineralized models were prepared according to previously established methods . Briefly, each specimen was immersed in 37% phosphoric acid (10 ml) for 45 s, washed with distilled water three times, sonicated for 5 min, and then used for subsequent experiments. Forty-eight demineralized tooth crowns were randomly divided into four groups ( n = 12). Each crown was embedded in self-curing acrylic resin (New Century Dental Materials, Shanghai, China) perpendicular to the buccal axis of the crown to align the buccal surfaces with the applied forces during testing. The demineralized enamel surfaces were dried with an oil- and moisture-free air stream from a standard dental 3-way syringe (Waldent, Dentalstall, India) for 10 s until they became chalky , and then they were applied with different composite adhesives, followed by the placement of premolar orthodontic metal brackets (8216-20B, Shinye, China) with a base area of 9.24 mm 2 on the prepared surfaces, curing for 20 s. These samples with brackets were placed in distilled water at 37 °C for 24 h before testing for shear bond strength. The test was performed using a universal testing machine (EZ20, LLOYD, UK) with a displacement rate of 0.5 mm/min (Supplementary Fig. ). A previously described method was used to calculate the SBS . The failure modes were assessed using stereomicroscope loupe (S9E, Leica, Germany) under 40 × magnification. The samples were classified into the adhesive failure mode (failure occurring at the adhesive interface), the mixed failure mode (failure occurring at the adhesive interface, with adhesive residues left on the enamel surface), and the cohesive failure mode (failure occurring within the enamel, resulting in an undercut) . The adhesive sample was cured and molded into a cylindrical shape (diameter: 1 cm; height: 1 mm). Then the samples were separately soaked in 2 mL of distilled water and were placed in a 37 °C shaker. The supernatant (1 mL) was removed on days 1, 5, 10, 15, and 30, respectively, and replaced by the same amount of fresh distilled water. A pH meter was used to determine the pH values of the supernatant. The ICP (720ES(OES), Agilent, the US) and IC (ICS-5000+, Thermo Fisher Scientific, the US) were used to determine the concentrations of Ca 2+ and F − , respectively. The weight loss ratio was calculated using the initial weight (w 0 ) and the final weight after incubation (w 1 ) of dried samples, according to the following formula: (w 1 -w 0 ) / w 0 %. Different adhesives (200 µL) were added to each well of a 24-well culture plate and cured for 60 s. Next, 1 ml of the simulated oral fluid (SOF) was added to each well and was refreshed every 24 h. The chemical composition of the SOF included 1.13 mM CaCl₂•2 H₂O, 0.29 mM MgCl₂•6 H₂O, 8.38 mM KCl, 4.62 mM K₂HPO₄, and 2.40 mM KH₂PO₄. The pH was adjusted to 6.8 using Tris, and no precipitation was observed in the SOF throughout the experimental duration . The samples were placed in a 37 °C shaker to mimic the oral physiological environment. They were collected on day 7, followed by sonication in distilled water for 2 min and dried. Then they were gold coated. SEM (Regulus 8100, Hitachi, Japan) was used to analyze the morphology of mineral crystals operating at 10 kV. The mineralized sites were chemically characterized using EDS operating at 20 kV. The regenerated crystals on the surfaces of the AH1, AH2 and AH3 specimens were analyzed by XRD (SmartLab SE, Rigaku, Japan) with Cu-Ka radiation at 10 mA, 30 kV, and a scan speed of 10°/min in the 2θ range of 10–80. To further validate the results of the in vitro experiments, in vivo experiments were conducted to evaluate the remineralization capabilities of different groups of adhesives in a realistic oral environment. The study received ethics approval from the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (China) (SH9H-2024-A1221-1), and all methods were conducted in accordance with relevant guidelines and regulations. Each tooth crown was cut into an enamel square disc (0.5 cm × 0.5 cm). A 1-mm-diameter hole was made using a drill for fixation. Twelve Sprague-Dawley rats (male, aged 8 weeks, 250–300 g) were randomly divided into four groups. The rats were purchased from Shanghai JieSiJie Laboratory Animal Company (Shanghai, China). Before the experiment, the rats were kept in a standard experimental environment for one week to acclimate to their surroundings. The rats were anesthetized with an intraperitoneal injection of Zoletil 50 (tilidine hydrochloride and midazolam hydrochloride, 80 mg/kg, Virbac, France). Two enamel samples applied with the adhesives were fixed adjacent to the two third molar teeth, respectively, using ligature wire to ensure their stability. Furthermore, the animals’ food supply was pulverized to prevent any potential harm to the samples. The incubation period was set at 7 days to ensure consistency with the in vitro experiments. After 7 days, all the rats were euthanized using carbon dioxide, and all the samples were collected. A low-speed dental burr was used to carefully remove adhesives on the sample surface. Next, the surfaces were polished using 1500-mesh SiC sandpaper under running water, followed by ultrasonication for 10 min to remove the smear layer. After drying the samples, the surfaces of the specimens were analyzed by XRD (SmartLab SE, Rigaku, Japan) with Cu-Ka radiation at 10 mA, 30 kV, and a scan speed of 10°/min in the 2θ range of 10°–80°. The surface micro-hardness was measured using a Vickers hardness testing machine (HXD-1000TMC, Taiming, China) with an applied load of 1000 gf and a dwell time of 10 s. Each sample was tested for five times and the average value was calculated. The surface micro-hardness of each sample before and after demineralization was also measured. Then their surfaces were sputter-coated with Au. SEM (Phenom Pharos G2, Phenom, the US) was used to observe the morphology operating at 5 kV. EDS was used to determine the chemical compositions of the remineralization sites operating at 15 kV. Finally, each enamel sample was cut into two halves perpendicular to the exposed surface to enable the observation of the cross sections via SEM as introduced above. All statistical data were analyzed using SPSS software version 27.0. The normality of the data distribution was evaluated by Shapiro-Wilk test. The equality of variances was assessed by Levene test. The mean values of SBS, weight loss ratios, atomic percentages of Ca, and surface micro-hardness were compared using one-way ANOVA with post-hoc Tukey’s test, based on their variance equality. Data on the release of Ca 2+ and F − conformed to the criteria for normal distribution but failed in the homogeneity of variance. Then Welch’s analysis of variance with Tamhane’s test was used. The failure mode was shown in absolute frequency . The association between adhesive groups and failure mode was analyzed with the Fisher’s exact test, while the differencess of failure modes between different groups were analyzed using Kruskal-Wallis test. The level of statistical significance was set at α = 0.05. Characterization of F-doped hydroxyapatite nanorods The TEM micrographs displayed that uniform nanorods with different sizes were synthesized (Fig. a). ImageJ Software was used to determine the size parameters (Fig. c, Supplementary Fig. ). The results revealed that the lengths of nanorods (H1, H2, and H3) were 505.31 ± 104.43 nm, 111.27 ± 22.89 nm, and 66.21 ± 12.68 nm, respectively, while the diameters were 22.28 ± 4.34 nm, 9.62 ± 1.71 nm, and 10.34 ± 1.56 nm ( n = 50, mean ± SD). The aspect ratios were 24.93 ± 1.25, 11.7 ± 2.78, and 6.6 ± 1.52 ( n = 50, mean ± SD). Therefore, as the fluorine doping level was increased, the length and aspect ratio of the FHA nanorods decreased. As shown in Fig. b, the EDS analysis revealed that the Ca/F atomic ratios of the final products (16.8, 5.9, and 3.7) were close to the theoretical Ca/F atomic ratios of the mixture of Ca(NO 3 ) 2 ·4H 2 O, NaF, and Na 3 PO 4 ·12H 2 O (17.8, 5.9, and 3.6). The P/F and Ca/P atomic ratios also yielded similar results, indicating that nearly all the main elements were converted into the final products. The XRD patterns of the nanorods corresponded to the hexagonal phase of hydroxyapatite [Ca 10 (PO 4 ) 6 (OH) 2 , JCPDF #01-074-0566] (Fig. d). With an increase in the fluorine doping level, all the peaks moved to higher degrees; this suggested that the substitution of F − for -OH caused the lattice volume to shrink. Meanwhile, the FTIR spectrum of the nanorods revealed the absorption bands of -OH, P-O, -CH 3 , and -CH 2 groups (Fig. e). With an increase in the fluorine doping level, the intensity of the absorption peak associated with -OH decreased sharply, while the other peaks remained almost unchanged, indicating the successful substitution of F − for -OH. SEM characterization of different groups of adhesives Figure a shows that the A0 group displayed a smooth surface without any fillers. In the AH1, AH2, and AH3 groups, FHA nanorods were uniformly distributed within the adhesive. The profiles of some nanorods in the AH2 and AH3 groups were unclear due to their small size and deep embedding within the non-conductive adhesive sample. The representative nanorods were highlighted by yellow arrows. As shown in the EDS-F mapping images, the A0 group contained no fluorine, while in the AH1, AH2, and AH3 groups, the red points representing fluorine distribution confirmed the homogeneous dispersion of the FHA nanorods within the adhesive. Evaluation of colloidal stability Figure displays the typical graphs of separation analysis for the adhesives containing 10 wt% different FHA powders. The transmittance at every height of the test tube changed minimally over time, indicating a high colloidal stability of the composite adhesives. Minor differences in average transmission rates among the three groups were observed, which could be attributed to differences in nanorod size. The transmission rate increased as the particle size decreased. The SBS test Figure shows the statistical analysis of SBS values in the four groups. Shapiro-Wilk test revealed the normal distribution of the data ( p > 0.05). Levene’s test confirmed the equality of variances ( p > 0.05), so the one-way ANOVA was used to compare mean SBS values of the four groups. Tukey-HSD method shows that the mean SBS values in groups AH1, AH2, and AH3 (8.79 MPa, 9.08 MPa, and 8.85 MPa) were significantly lower than group AH0 (11.33 MPa). No significant differences were observed among the three groups: AH1, AH2, and AH3. Failure mode analysis Fisher’s exact test indicated that there was no significant association between adhesive groups and failure modes ( P > 0.05). The mixed failure mode was predominant in all the four groups. The A0 group exhibited a greater proportion of the mixed failure mode compared to the other three groups; however, the differences were not statistically significant, as determined by the Kruskal-Wallis test (Table ). Representative microscopic images of the adhesive failure mode and the mixed failure mode were shown in Supplementary Fig. . Evaluation of the light-curing property Figure a shows the transition of the adhesives from a liquid to a solid state after the curing process. In each group of adhesives, a certain degree of fluidity was retained at 5 s, which was nearly lost by 10 s. By 20 s, all groups of adhesives have fully solidified. Figure b illustrates the FTIR spectra for the four groups at different states: uncured, cured for 5 s, 10 s, and 20 s. As the light curing time increased, the intensity of the reactive peaks (1638 cm − 1 ) gradually decreased. Notably, the differences between uncured-5 s and 5 s-10 s were more pronounced, while the reactive peak intensities of 10 s and 20 s were close. Table presents the degrees of monomer conversion for various groups at different curing times. The DC % at 10 s for each group was more than twice that at 5 s, whereas no statistically significant differences were noted between 10 s and 20 s. Furthermore, no significant differences were observed among the four groups at each time point ( P > 0.05). Degradation and release of ions As shown in Fig. a, the weight loss ratio increased during the first 15 days, after which the rate of increase slowed. Figure b shows that the pH values rose with the incorporation of FHAs, with the supernatants of the AH3 samples being more alkaline. ICP and IC tests indicated that the composite adhesives released calcium and fluorine ions over an extended period, although the release rate gradually decreased (Fig. c, d). Notably, the AH1, AH2, and AH3 groups exhibited a similar trend in calcium ion release, while the AH3 group released a greater amount of fluorine ions during the same immersion period. Apatite-forming ability in vitro Figure a shows the surface morphology of each adhesive sample after immersion in the SOF for 7 days. No mineral deposits were observed in the experimental adhesive (A0) group. In contrast, abundant micro-spherical crystals precipitated on the surfaces of the FHA-containing adhesives (AH1, AH2, and AH3). The deposits in groups AH1, AH2, and AH3 were similar in shape but obviously different in number. The AH3 group showed the dense surface precipitation, nearly entirely covered by micro-spherical crystals, while the AH1 and AH2 groups had fewer deposits. EDS measurements, as shown in Fig. b, was used to quantitatively analyze the calcium and phosphorus contents on the specimen surfaces. Tukey’s post hoc test indicated that the most Ca 2+ were absorbed onto the specimen surfaces in the AH3 group, followed by the AH2 and AH1 groups, with almost no absorption observed in the A0 group. The representative atomic percentages and Ca/P ratios in the mineralization sites are shown in the Table . The Ca/P atomic ratio exhibited a similar trend to that of the Ca atomic percentages, measuring 1.40, 1.46, and 1.62 for the AH1, AH2, and AH3 groups, respectively. As shown in Fig. c, the XRD patterns of regenerated crystals on the surfaces of AH1, AH2, and AH3 specimens displayed prominent characteristic diffraction peaks, (002), (211), (202), (222), (213), and (004), corresponding to those of hydroxyapatite (JCPDF #01-074-0566), indicating that the precipitated crystals were mainly composed of hydroxyapatites. Considering that there was almost no calcium deposition in the A0 group, EDS and XRD measurements were not conducted for this group. In vivo enamel remineralization in a rat oral environment model As shown in Fig. c, the surface of intact enamel was smooth, whereas the demineralized surface exhibited a rough texture with a fish-scale enamel-rod structure. Square enamel disk samples were placed in the oral cavities of rats for the experiment (Fig. a, b). After 7 days of incubation, the A0 group showed minimal mineral deposition in the interred spaces, and the profile of the enamel rods remained clearly visible (Fig. c). In contrast, the AH1, AH2, and AH3 groups showed abundant mineral deposits in the spaces between adjacent enamel rods, which obscured the rod profiles. The AH3 group, in particular, displayed the dense deposits, with the interrod spaces fully filled with mineral crystals, showing a homogeneous appearance similar to that of intact enamel. Figure d presents SEM images of cross sections at a 100-nm depth from the margin. In the A0 group, few mineral deposits were observed, whereas the AH3 group exhibited significantly more mineral deposits, rendering the interrod spaces nearly invisible. The EDS analysis of the mineral deposits in the interrod spaces was shown in the Fig. e. Tukey’s post hoc test was used to compare the relative Ca atomic ratios between different specimen groups. A significant decrease in relative Ca content in demineralized enamel was observed compared to intact enamel. The mean value of the relative Ca atonic ratio in the A0 group was slightly higher than that in the demineralized group; however, no significant difference was observed. In contrast, the Ca content was significantly higher in groups AH1, AH2 and AH3 than that in group A0. In particular, the AH3 group absorbed the most Ca ions onto the enamel surfaces, though a significant difference still exists between AH3 and intact enamel. The representative atomic percentages and Ca/P ratios in the mineralization sites are shown in the Table . Notably, the Ca/P ratio of the AH3 group was close to 1.67, which is characteristic of mature hydroxyapatite . The XRD patterns of precipitated crystals on the specimen surfaces are shown in Fig. f. Characteristic diffraction peaks (002), (211), (202), (213), and (004) were observed, corresponding to those of crystalline hydroxyapatite (JCPDF #01-074-0565). Some peaks, including (002) and (004), diminished after demineralization, but after incubation for 7 days, groups AH1, AH2, and AH3 displayed XRD patterns similar to that of intact enamel. Fig. g illustrates the surface micro-hardness of the samples before demineralization (VH1), after demineralization (VH2), and after 7 days of treatment (VH3). The VH1 or VH2 values showed no significant differences across all groups. The VH3 values for groups A0, AH1, AH2, and AH3 were 203.9 ± 10.8, 229.0 ± 11.8, 269.8 ± 12.1, and 293.7 ± 9.1 (mean ± SD). Significant differences in VH3 values were observed among the four groups. The TEM micrographs displayed that uniform nanorods with different sizes were synthesized (Fig. a). ImageJ Software was used to determine the size parameters (Fig. c, Supplementary Fig. ). The results revealed that the lengths of nanorods (H1, H2, and H3) were 505.31 ± 104.43 nm, 111.27 ± 22.89 nm, and 66.21 ± 12.68 nm, respectively, while the diameters were 22.28 ± 4.34 nm, 9.62 ± 1.71 nm, and 10.34 ± 1.56 nm ( n = 50, mean ± SD). The aspect ratios were 24.93 ± 1.25, 11.7 ± 2.78, and 6.6 ± 1.52 ( n = 50, mean ± SD). Therefore, as the fluorine doping level was increased, the length and aspect ratio of the FHA nanorods decreased. As shown in Fig. b, the EDS analysis revealed that the Ca/F atomic ratios of the final products (16.8, 5.9, and 3.7) were close to the theoretical Ca/F atomic ratios of the mixture of Ca(NO 3 ) 2 ·4H 2 O, NaF, and Na 3 PO 4 ·12H 2 O (17.8, 5.9, and 3.6). The P/F and Ca/P atomic ratios also yielded similar results, indicating that nearly all the main elements were converted into the final products. The XRD patterns of the nanorods corresponded to the hexagonal phase of hydroxyapatite [Ca 10 (PO 4 ) 6 (OH) 2 , JCPDF #01-074-0566] (Fig. d). With an increase in the fluorine doping level, all the peaks moved to higher degrees; this suggested that the substitution of F − for -OH caused the lattice volume to shrink. Meanwhile, the FTIR spectrum of the nanorods revealed the absorption bands of -OH, P-O, -CH 3 , and -CH 2 groups (Fig. e). With an increase in the fluorine doping level, the intensity of the absorption peak associated with -OH decreased sharply, while the other peaks remained almost unchanged, indicating the successful substitution of F − for -OH. Figure a shows that the A0 group displayed a smooth surface without any fillers. In the AH1, AH2, and AH3 groups, FHA nanorods were uniformly distributed within the adhesive. The profiles of some nanorods in the AH2 and AH3 groups were unclear due to their small size and deep embedding within the non-conductive adhesive sample. The representative nanorods were highlighted by yellow arrows. As shown in the EDS-F mapping images, the A0 group contained no fluorine, while in the AH1, AH2, and AH3 groups, the red points representing fluorine distribution confirmed the homogeneous dispersion of the FHA nanorods within the adhesive. Figure displays the typical graphs of separation analysis for the adhesives containing 10 wt% different FHA powders. The transmittance at every height of the test tube changed minimally over time, indicating a high colloidal stability of the composite adhesives. Minor differences in average transmission rates among the three groups were observed, which could be attributed to differences in nanorod size. The transmission rate increased as the particle size decreased. Figure shows the statistical analysis of SBS values in the four groups. Shapiro-Wilk test revealed the normal distribution of the data ( p > 0.05). Levene’s test confirmed the equality of variances ( p > 0.05), so the one-way ANOVA was used to compare mean SBS values of the four groups. Tukey-HSD method shows that the mean SBS values in groups AH1, AH2, and AH3 (8.79 MPa, 9.08 MPa, and 8.85 MPa) were significantly lower than group AH0 (11.33 MPa). No significant differences were observed among the three groups: AH1, AH2, and AH3. Fisher’s exact test indicated that there was no significant association between adhesive groups and failure modes ( P > 0.05). The mixed failure mode was predominant in all the four groups. The A0 group exhibited a greater proportion of the mixed failure mode compared to the other three groups; however, the differences were not statistically significant, as determined by the Kruskal-Wallis test (Table ). Representative microscopic images of the adhesive failure mode and the mixed failure mode were shown in Supplementary Fig. . Figure a shows the transition of the adhesives from a liquid to a solid state after the curing process. In each group of adhesives, a certain degree of fluidity was retained at 5 s, which was nearly lost by 10 s. By 20 s, all groups of adhesives have fully solidified. Figure b illustrates the FTIR spectra for the four groups at different states: uncured, cured for 5 s, 10 s, and 20 s. As the light curing time increased, the intensity of the reactive peaks (1638 cm − 1 ) gradually decreased. Notably, the differences between uncured-5 s and 5 s-10 s were more pronounced, while the reactive peak intensities of 10 s and 20 s were close. Table presents the degrees of monomer conversion for various groups at different curing times. The DC % at 10 s for each group was more than twice that at 5 s, whereas no statistically significant differences were noted between 10 s and 20 s. Furthermore, no significant differences were observed among the four groups at each time point ( P > 0.05). As shown in Fig. a, the weight loss ratio increased during the first 15 days, after which the rate of increase slowed. Figure b shows that the pH values rose with the incorporation of FHAs, with the supernatants of the AH3 samples being more alkaline. ICP and IC tests indicated that the composite adhesives released calcium and fluorine ions over an extended period, although the release rate gradually decreased (Fig. c, d). Notably, the AH1, AH2, and AH3 groups exhibited a similar trend in calcium ion release, while the AH3 group released a greater amount of fluorine ions during the same immersion period. Figure a shows the surface morphology of each adhesive sample after immersion in the SOF for 7 days. No mineral deposits were observed in the experimental adhesive (A0) group. In contrast, abundant micro-spherical crystals precipitated on the surfaces of the FHA-containing adhesives (AH1, AH2, and AH3). The deposits in groups AH1, AH2, and AH3 were similar in shape but obviously different in number. The AH3 group showed the dense surface precipitation, nearly entirely covered by micro-spherical crystals, while the AH1 and AH2 groups had fewer deposits. EDS measurements, as shown in Fig. b, was used to quantitatively analyze the calcium and phosphorus contents on the specimen surfaces. Tukey’s post hoc test indicated that the most Ca 2+ were absorbed onto the specimen surfaces in the AH3 group, followed by the AH2 and AH1 groups, with almost no absorption observed in the A0 group. The representative atomic percentages and Ca/P ratios in the mineralization sites are shown in the Table . The Ca/P atomic ratio exhibited a similar trend to that of the Ca atomic percentages, measuring 1.40, 1.46, and 1.62 for the AH1, AH2, and AH3 groups, respectively. As shown in Fig. c, the XRD patterns of regenerated crystals on the surfaces of AH1, AH2, and AH3 specimens displayed prominent characteristic diffraction peaks, (002), (211), (202), (222), (213), and (004), corresponding to those of hydroxyapatite (JCPDF #01-074-0566), indicating that the precipitated crystals were mainly composed of hydroxyapatites. Considering that there was almost no calcium deposition in the A0 group, EDS and XRD measurements were not conducted for this group. As shown in Fig. c, the surface of intact enamel was smooth, whereas the demineralized surface exhibited a rough texture with a fish-scale enamel-rod structure. Square enamel disk samples were placed in the oral cavities of rats for the experiment (Fig. a, b). After 7 days of incubation, the A0 group showed minimal mineral deposition in the interred spaces, and the profile of the enamel rods remained clearly visible (Fig. c). In contrast, the AH1, AH2, and AH3 groups showed abundant mineral deposits in the spaces between adjacent enamel rods, which obscured the rod profiles. The AH3 group, in particular, displayed the dense deposits, with the interrod spaces fully filled with mineral crystals, showing a homogeneous appearance similar to that of intact enamel. Figure d presents SEM images of cross sections at a 100-nm depth from the margin. In the A0 group, few mineral deposits were observed, whereas the AH3 group exhibited significantly more mineral deposits, rendering the interrod spaces nearly invisible. The EDS analysis of the mineral deposits in the interrod spaces was shown in the Fig. e. Tukey’s post hoc test was used to compare the relative Ca atomic ratios between different specimen groups. A significant decrease in relative Ca content in demineralized enamel was observed compared to intact enamel. The mean value of the relative Ca atonic ratio in the A0 group was slightly higher than that in the demineralized group; however, no significant difference was observed. In contrast, the Ca content was significantly higher in groups AH1, AH2 and AH3 than that in group A0. In particular, the AH3 group absorbed the most Ca ions onto the enamel surfaces, though a significant difference still exists between AH3 and intact enamel. The representative atomic percentages and Ca/P ratios in the mineralization sites are shown in the Table . Notably, the Ca/P ratio of the AH3 group was close to 1.67, which is characteristic of mature hydroxyapatite . The XRD patterns of precipitated crystals on the specimen surfaces are shown in Fig. f. Characteristic diffraction peaks (002), (211), (202), (213), and (004) were observed, corresponding to those of crystalline hydroxyapatite (JCPDF #01-074-0565). Some peaks, including (002) and (004), diminished after demineralization, but after incubation for 7 days, groups AH1, AH2, and AH3 displayed XRD patterns similar to that of intact enamel. Fig. g illustrates the surface micro-hardness of the samples before demineralization (VH1), after demineralization (VH2), and after 7 days of treatment (VH3). The VH1 or VH2 values showed no significant differences across all groups. The VH3 values for groups A0, AH1, AH2, and AH3 were 203.9 ± 10.8, 229.0 ± 11.8, 269.8 ± 12.1, and 293.7 ± 9.1 (mean ± SD). Significant differences in VH3 values were observed among the four groups. In this study, uniform FHA nanorods of varying sizes were synthesized by adjusting the fluorine doping level, and the synergistic effects of both the fluoride content and FHA nanorod size on enamel remineralization were investigated for the first time, by incorporating these FHA nanorods into the dental adhesive. Characterization analysis was used to confirm the crystalline phase, chemical composition, size of the prepared FHA nanorods and distribution of FHA powders within adhesives. The SBS test was used to assess whether the bond strength of all three composite adhesives satisfied clinical requirements. Ion release tests were used to examine the release of mineral ions to help investigate the mechanism of remineralization. SEM and XRD were used to qualitatively analyze the morphology and crystalline phase of the minerals on the specimen surfaces. EDS analysis was used to quantitatively evaluate the mineralization ability. For the synthesis of the HA nanorods, the hydrothermal route utilizing an oleic acid system is among the commonly used techniques . As the fluorine doping level increased, the length and aspect ratio of nanorods decreased, resulting in an increased surface-to-volume ratio. The increased specific surface area due to fluorine doping was consistent with a previous study . The changes in the structure were further analyzed through XRD and FTIR. The XRD patterns displayed that the intensity of all peaks rose with the increase of the fluorine doping level, indicating a reduction in lattice volume, as previously described . The FTIR spectrum revealed the decreased intensity of the absorption peak associated with –OH, as fluorine doping was increased. These results indicated that the reduction in length was probably attributed to the substitution of F − at -OH sites along the long c-axis, which slowed the growth rate in this orientation . For the characterization of composite adhesives, SEM and EDS were used to examine the dispersion of FHA nanorods in adhesives. The SEM images (Fig. a-d) revealed a relatively homogeneous distribution of FHA nanorods, with minimal agglomeration. However, the profile of some nanorods in the AH2 and AH3 groups was obscured due to their small size and deep embedding within the non-conductive adhesive. EDS-F mapping confirmed the distribution, as F only existed in the FHA nanorods. The solubility test revealed that they could be dispersed in non-polar solvents such as cyclohexane (Supplementary Fig. ), which explains why they dispersed effectively in organic adhesives. For the evaluation of the light-curing properties, the tilt test was conducted to assess the state of the adhesive after the curing process, inspired by previous studies . Our findings indicated that all groups of adhesives were in a liquid state before curing and completely transitioned to a solid state after 20 s of curing. This suggested that the components of the adhesive dissolved before curing but failed to form a stable structure until after the curing process. To quantitatively evaluate the degrees of monomer conversion for each group at different curing durations, FTIR was employed, as it is recognized as an effective and precise method for this purpose . The results indicated that there were no significant differences among the four groups at different curing times, suggesting that the incorporation of FHA filler did not markedly interfere with the photopolymerization of the adhesive. Additionally, the DC % at 10 s for each group was more than twice that at 5 s, while the DC % at 20 s approached that at 10 s. This finding was consistent with previous studies indicating that the change in DC % with curing time was nonlinear; beyond a certain curing threshold, the mobility of free radicals initiating polymerization became restricted . Consequently, excessively long curing times may not be an effective strategy, as it does not improve the degree of polymerization significantly and may harm the pulp vitality due to overheating at localized sites. For the preparation of the demineralization model in this study, a 37% phosphoric acid solution was employed according to established protocols . This methodology led to the decalcification and dissolution of hydroxyapatite on the enamel surface, which effectively simulated the initial stages of enamel caries induced by bacterial acid . Previous research displayed that it could reflect the pathological state of early enamel caries and facilitated the induction of caries activity . For the SBS test, although there are no universally accepted optimal values of orthodontic bond strength, a range of 5.9–7.8 MPa is typically considered to be the minimum range appropriate for bracket bonding, while achieving satisfactory clinical performance . However, SBS > 12 MPa has been reported to cause enamel damage . For this reason, most in vitro bracket bond strength studies use values in the range of 6–12 MPa . In this study, the SBS values were significantly reduced due to the incorporation of 10 wt% FHA powder into the adhesive, but the values (8.79–9.08 MPa) of the composite adhesives were still within the appropriate range. However, as the incorporation ratio was increased to 20 wt%, the SBS values (6.36–6.53 MPa) decreased more significantly and were too low to satisfy the requirements for clinical use (Supplementary Fig. ). For the failure mode analysis, the mixed failure mode was predominant in all four groups and no significant association was found between adhesive groups and failure modes. Although the proportion of the mixed failure mode in Group A0 was higher than in the other groups, the differences were not statistically significant. Previous studies proved a higher percentage of the adhesive failure mode related to low bond strength and the mixed failure mode related to high bond strength . Our results were consistent with those findings. The SBS of Group A0 was greater than that of the other groups, while also exhibiting a higher proportion of the mixed failure mode. The adhesive remnants characteristic of the mixed failure mode indicated a weak bond between the bracket and the adhesive, which was favorable as it minimized damage to the enamel during the debonding process . Nevertheless, the residual adhesive required additional removal, thereby increasing the workload for orthodontists and exposing enamel to increased abrasion. For the mineralization assessment, morphology may be a feature of the mineral change . In the in vitro experiment, no precipitation was observed on the surface of the adhesive without fillers. In contrast, micro-spherical crystal minerals were observed on the surfaces of adhesives containing FHA. As the fluorine doping level of FHA was increased, the crystals forming on the surfaces became denser. The in vivo experiment demonstrated a similar trend: as the fluorine doping level of FHA was increased, the interrod space was filled by more mineralized crystals, thus getting less distinct. To quantitatively analyze the composition of Ca 2+ and PO 3 4− adsorbed onto the adhesive and enamel surfaces, EDS was employed to examine the precipitated crystals. The results indicated that as the fluorine-doping level increased, more Ca 2+ were absorbed onto these surfaces. Notably, mineral crystals in the group containing 10 wt% fluorine-doped HA had the highest Ca content and a Ca/P ratio nearest to the mature hydroxyapatite ratio of 1.67 , exhibiting superior mineralization performance in vitro and in vivo. Additionally, the XRD results further confirmed that the mineralized crystals on the surfaces were hydroxyapatites. In the in vivo test, the XRD patterns showed that peaks (002) and (004) were weakened after demineralization. The ratio of the diffraction intensities of peaks (002) and (211) was utilized to assess the degree of orientation . After 7 d of incubation, the ratios of groups AH1, AH2, and AH3 were higher than that of group A0, indicating that the incorporation of FHA promoted the newly formed crystals to orient along the crystallographic c-axis . The micro-hardness test demonstrated that the surface hardness of samples before or after demineralization was nearly the same across all groups, indicating a consistent baseline and a standardized demineralization procedure. After demineralization, the hardness decreased significantly, then increased to varying degrees after treatment with different adhesives in vivo. Among the four groups, the AH3 group exhibited the highest hardness value after 7 days of incubation, suggesting more effective remineralization according to previous studies . Early-stage remineralization is crucial for managing hypocalcified areas and improving acid resistance in teeth, which is beneficial for completing orthodontic treatment and increasing the number of caries-free patients . All composite adhesives doped with FHA exhibited improved early mineralization efficacy at the first week, with the group containing 10 wt% fluorine-doped FHA showing superior performance. The mechanism by which composite adhesives containing various FHA nanorods promoted enamel remineralization may be as follows: As shown in Fig. c and d, the long-term release of calcium and fluorine ions might result in the supersaturation of mineral ions in specific areas, thereby facilitating the nucleation and growth of HA crystals in the interrod space . Meanwhile, the FHA-containing adhesives provided a suitable substrate for heterogeneous nucleation. FHA particles, with their negatively charged surfaces, could bind with Ca 2+ in biological fluids. With the accumulation of Ca 2+ , the calcium-rich FHAs developed a positively charged surface, which then bound with PO 4 3− , forming amorphous calcium phosphate layers and, eventually, stable hydroxyapatites . Moreover, fluorine ions had a greater affinity for binding with Ca 2+ to form fluorapatite, which had higher density and chemical stability compared to hydroxyapatites due to stronger electrostatic interactions between fluorine and adjacent ions . The abovementioned mechanism suggests that the fluorine content, surface charge, and size of nanorods play a role in remineralization. More F − can be released from adhesives with a higher fluorine content and remineralization can be promoted. Similarly, particles with a higher net negative charge are more likely to bind with Ca 2+ to form mineral layers. Furthermore, the size of nanorods impacts ion release. Javid et al. reported that the nanoparticles with a larger surface-to-volume ratio solvated ions into the medium more easily due to stronger interactions with the surrounding medium. In this study, the 10 wt% F-doped FHA nanorods exhibited a lower net negative surface charge (Supplementary Table ) but demonstrated a higher surface-to-volume ratio and a higher fluorine content, thus promoting mineralization. Our study presents the first investigation of both the application of size-controllable FHA nanorods in adhesives to enhance enamel remineralization and the synergistic effects of fluorine content and HA nanorod size on this process. Three groups of FHA nanorods with varying fluorine doping levels (2 wt%, 6 wt%, and 10 wt%) and sizes were synthesized. As the fluorine doping level increased, the lengths of the nanorods decreased. Among them, the 10 wt% F-doped FHA nanorods with shorter lengths exhibited superior mineralization performance, based on the most Ca 2+ depositions in both in vitro and in vivo experiments. Notably, previous studies mainly employed in vitro experiments to assess enamel mineralization capability . Considering the differences in factors such as temperature, pH variations, fluid flow, bacteria, and enzymes between in vitro conditions and the oral cavity, an animal experiment was conducted in this study to establish a physiological oral environment, which made the conclusions more reliable. However, some limitations still exist in our study. The EDS analysis used to assess mineralization in vitro and in vivo is a semi-quantitative analysis on localized areas, while newly forming minerals were not distributed evenly. Therefore, more reliable quantitative methods need to be explored to assess surface remineralization. Furthermore, preclinical large animal experiments and clinical trials should be conducted to further assess the clinical applicability of the adhesives. Additionally, this study didn’t involve toxicity tests for these novel composite adhesives, which are crucial for clinical applications. This aspect should be addressed in future research. In this study, three groups of FHA nanorods with varying sizes were synthesized, modulated by the fluorine doping levels (2 wt%, 6 wt%, and 10 wt%). The synergistic effects of the fluorine content and FHA nanorod size on enamel remineralization were first investigated by incorporating these nanorods into dental adhesives. The null hypothesis that the composite adhesives containing FHA nanorods would not maintain adequate shear bond strength or enhance enamel remineralization was rejected. Among all the nanorods, the 10 wt% F-doped HA nanorods with shorter lengths could promote more formation of hydroxyapatites on the adhesive and enamel surfaces. The application of these nanorods in dental adhesives created a novel composite adhesive with remineralization capability, while maintaining adequate bond strength. Consequently, this novel composite adhesive holds the potential to promote enamel remineralization for clinical use, particularly for fixed orthodontic treatment. Further clinical investigations are needed to validate these conclusions and assess the actual clinical application value. Below is the link to the electronic supplementary material. Supplementary Material 1 |
Risk-based stratification triaging system in pediatric urology: what COVID-19 pandemic has taught us | 708f8237-9504-4dd6-962b-e428594f8a7f | 7910196 | Pediatrics[mh] | SARS-COV-2 pandemic has affected the population worldwide requiring social distancing, quarantine and isolation as strategies to limit spread of the virus. Health systems have been forced to redistribute resources toward emergency services and intensive care units. In an effort to preserve resources and protect patients and providers, most elective procedures have been postponed and while others have been performed using regional anesthesia in an effort to shorten hospital stays . In adult surgery, clear guidance on how to decide which procedures should be prioritized were published during the pandemic . However, in pediatric urology these protocols are not applicable worldwide . Care of children with pediatric urologic anomalies varies widely based on availability of pediatric expertise in the local healthcare system. As a result, provision of care to these patients is often delayed, placing them at an increased risk for needing urgent care. These delays are only further exacerbated by the pandemic. It is unknown how much longer it will take once pandemic surge is contained and a plateau or recovery phase begins. As a result, the impact on morbidity and prognosis on our pediatric urological population is unknown . It is imperative to swiftly triage patient care in a safe and efficient manner without compromising quality. The purpose of this article is to demonstrate the impact of a modified triage scoring system for pediatric urology patients with urgent clinical conditions and the impact of ERAS ® principles on the burden of care to the healthcare system during COVID-19 pandemic.
Background and triaging system creation After mandatory social isolation policies were published by local authorities, an immediate moratorium on elective surgery was instituted, creating an urgent need to triage patients. For that reason, we initially created a list of all pending cases and reviewed each case individually looking at patient’s clinical status and indications for surgery. Implementation of a modified version of the Medically Necessary, Time-Sensitive Procedures: Scoring System (MeNTS) was created and applied for the present study. The reasons to modify the score were based on the fact that implementation of the originally proposed MeNTS scoring system did not allow us to individualize patient’s clinical needs nor the urgency of each procedure. The lack of a clear cutoff value for the MeNTS did not allow us to really triage our patients and also, MeNTS score was never designed for the pediatric population. Each case was individually evaluated by the authors and other providers, all members of the COVID-19 surgical response team. Revision and triaging of our patients, was done following the same proposed protocol as described originally by Prachand et al. The modification to scoring was based on increased weight (60% of the total score) applied to patient’s condition and the impact on the long-term prognosis. This percentage was chosen in agreement by all the authors and other members of the department. The rational was supported after careful estimation on how important the individual clinical scenario was. Standardized clinical simulation scenarios were made with same condition and same procedure but with different clinical prognosis and needs. Increasing percentages were used to calculate the total score and we found that to really discriminate for individual clinical needs, a 60% of the total score should be given to patient’s individual condition and prognosis. The potential burden to the health system (30% of total score) was also accounted for in the modification as it was absent in the original scoring scheme . The reason for this was also carefully analyzed by the team members following the same aforementioned rationale. The remaining 10% of the total score was estimated by patient’s own risk factors (age, comorbidities and risk of COVID-19 infection at the time of surgery (Table ). This last percentage was supported by available literature that reports low morbidity to pediatric patients infected with SARS-COV-2 . After scoring system had been proposed and final score weight had been proposed, a validation phase was then carried out. Validation of the adjustments was performed using the available cases booked for surgery as well as our historical cases that had been operated since the beginning of the year (January 2020). Validation was independently performed by two blinded evaluators. A total of 180 cases were used for validation. The score was designed in a way that the lower the score the higher the priority the patient for surgical intervention with higher negative impact if surgery had to be postponed. For final triaging and ranking, the order and priority for surgery was defined by ascending scores with the lowest being the ones to be performed first. For those cases where discrepancies in the final score of more than 2 points were identified, evaluators reviewed and solved discrepancies. Patient care and triaging system implementation Between the 20th of March 2020 and the 24th of April 2020 and during mandatory lockdown, a total of 49 patients were triaged using the modified triage scoring system. Patients included were all those that presented to the emergency department with a surgical urological condition and those that had been booked prior to the beginning of the pandemic and were waiting to be operated. All cases were reviewed individually. We collected demographic and clinical data. Once all cases had been assessed by the adjusted triaging system, we prioritized the patients in ascending order based on the final score. Every 2 weeks, cases were re-triaged as clinical conditions could have changed impacting their prioritization score. To reduce the burden to the system, we implemented ERAS ® protocols. These were protocols that were not being used prior to the pandemic. Since our scoring system had been modified and 10% of the total score depended on how much burden would be put on the facilities and health system if surgery were to be performed, we looked into novel and different ways to reduce this. We found that final scores on the simulation scenarios and validation phase were consistently lower for all of the ambulatory procedures. For that reason, we decided to implement ERAS principles to our cases and explore the potential change from inpatient to ambulatory procedures . Perioperative pain control was of critical importance to achieve this objective by applying regional anesthesia/nerve blocks to minimize opioid utilization and improve recovery times. Prior to the pandemic, procedures such as ureteral reimplants, pyeloplasties and retrograde intrarenal surgery (RIRS) were admitted to the inpatient unit. Application of ERAS ® principles allowed transition of inpatient procedures to the outpatient setting. Another change to our practice included that for all of these cases a postoperative follow-up was monitored by phone or telemedicine at 24 and 48 h. Parents/caregivers were provided with specific instructions after surgery on how to monitor urine output, how to care for drains or catheters and on how to manage pain with oral pain medications. No opioids were used for any of the ambulatory patients. For those who required double-J stent placement, we removed them after 72 h by pulling the strings without the need for an additional surgical intervention. We had no Salle stents available and all pyeloplasties were performed leaving a regular double-J stent in place.
After mandatory social isolation policies were published by local authorities, an immediate moratorium on elective surgery was instituted, creating an urgent need to triage patients. For that reason, we initially created a list of all pending cases and reviewed each case individually looking at patient’s clinical status and indications for surgery. Implementation of a modified version of the Medically Necessary, Time-Sensitive Procedures: Scoring System (MeNTS) was created and applied for the present study. The reasons to modify the score were based on the fact that implementation of the originally proposed MeNTS scoring system did not allow us to individualize patient’s clinical needs nor the urgency of each procedure. The lack of a clear cutoff value for the MeNTS did not allow us to really triage our patients and also, MeNTS score was never designed for the pediatric population. Each case was individually evaluated by the authors and other providers, all members of the COVID-19 surgical response team. Revision and triaging of our patients, was done following the same proposed protocol as described originally by Prachand et al. The modification to scoring was based on increased weight (60% of the total score) applied to patient’s condition and the impact on the long-term prognosis. This percentage was chosen in agreement by all the authors and other members of the department. The rational was supported after careful estimation on how important the individual clinical scenario was. Standardized clinical simulation scenarios were made with same condition and same procedure but with different clinical prognosis and needs. Increasing percentages were used to calculate the total score and we found that to really discriminate for individual clinical needs, a 60% of the total score should be given to patient’s individual condition and prognosis. The potential burden to the health system (30% of total score) was also accounted for in the modification as it was absent in the original scoring scheme . The reason for this was also carefully analyzed by the team members following the same aforementioned rationale. The remaining 10% of the total score was estimated by patient’s own risk factors (age, comorbidities and risk of COVID-19 infection at the time of surgery (Table ). This last percentage was supported by available literature that reports low morbidity to pediatric patients infected with SARS-COV-2 . After scoring system had been proposed and final score weight had been proposed, a validation phase was then carried out. Validation of the adjustments was performed using the available cases booked for surgery as well as our historical cases that had been operated since the beginning of the year (January 2020). Validation was independently performed by two blinded evaluators. A total of 180 cases were used for validation. The score was designed in a way that the lower the score the higher the priority the patient for surgical intervention with higher negative impact if surgery had to be postponed. For final triaging and ranking, the order and priority for surgery was defined by ascending scores with the lowest being the ones to be performed first. For those cases where discrepancies in the final score of more than 2 points were identified, evaluators reviewed and solved discrepancies.
Between the 20th of March 2020 and the 24th of April 2020 and during mandatory lockdown, a total of 49 patients were triaged using the modified triage scoring system. Patients included were all those that presented to the emergency department with a surgical urological condition and those that had been booked prior to the beginning of the pandemic and were waiting to be operated. All cases were reviewed individually. We collected demographic and clinical data. Once all cases had been assessed by the adjusted triaging system, we prioritized the patients in ascending order based on the final score. Every 2 weeks, cases were re-triaged as clinical conditions could have changed impacting their prioritization score. To reduce the burden to the system, we implemented ERAS ® protocols. These were protocols that were not being used prior to the pandemic. Since our scoring system had been modified and 10% of the total score depended on how much burden would be put on the facilities and health system if surgery were to be performed, we looked into novel and different ways to reduce this. We found that final scores on the simulation scenarios and validation phase were consistently lower for all of the ambulatory procedures. For that reason, we decided to implement ERAS principles to our cases and explore the potential change from inpatient to ambulatory procedures . Perioperative pain control was of critical importance to achieve this objective by applying regional anesthesia/nerve blocks to minimize opioid utilization and improve recovery times. Prior to the pandemic, procedures such as ureteral reimplants, pyeloplasties and retrograde intrarenal surgery (RIRS) were admitted to the inpatient unit. Application of ERAS ® principles allowed transition of inpatient procedures to the outpatient setting. Another change to our practice included that for all of these cases a postoperative follow-up was monitored by phone or telemedicine at 24 and 48 h. Parents/caregivers were provided with specific instructions after surgery on how to monitor urine output, how to care for drains or catheters and on how to manage pain with oral pain medications. No opioids were used for any of the ambulatory patients. For those who required double-J stent placement, we removed them after 72 h by pulling the strings without the need for an additional surgical intervention. We had no Salle stents available and all pyeloplasties were performed leaving a regular double-J stent in place.
A total 49 patients, 40 boys and 9 girls with a mean age 6.47 years (13 days old–17 years of age) were included in this series. Distribution of cases and their average score is presented in Table . Based on our modified version of MeNTS, we found that all cases with a score of 12 points or less were treated emergently and this correlated completely with our clinical assessment. Twenty-three cases that required immediate surgical management due to their clinical condition had scores of less than 12. A total of 4 cases had scores between 13 and 15 and a total of 22 patients had a score above 16. Average score obtained for acute scrotum/testicular torsion was 9.3. Median score for cases with active infection of the urinary tract requiring surgical management was 10.6. Kidney and ureteral stone-related procedures had a mean score of 10.0. All other cases that required clinical urgent management had scores below 12. Cases with hypospadias had a median score of 19.5 and circumcisions without acute urinary retention had a median score of 19. Score results were not the same for each procedure and did reflect the individualized patient’s current clinical condition and priority. For example, on Table , the circumcision with the lowest score, was a boy with severe chronic balanitis that was in urinary retention. All other circumcisions were non-emergent and no impact on their clinical prognosis would have been seen if postponed after 6 weeks.
COVID-19 pandemic has created a need to re-invent the way we practice medicine. Given the burden on healthcare systems and the risk to patient and staff, elective surgery was suspended immediately stop elective surgery. The impact of this decision created unanticipated impacts on waiting times that were already excessive, especially in low- to mid-income countries. As a guide, different American and European surgical associations including the American College of Surgeons, published guidelines to select and triage surgical interventions . None of these guidelines was specific to pediatric urologic conditions and focused their design on the triage of cases based on procedure and not individualized patient’s clinical conditions . Recommendations for urological conditions were created for the adult population making them less applicable to the pediatric patient. Low- or mid-income countries have specific and unique limitations that make implementation of guidelines from a higher resourced very difficult. Reduced access to subspecialized trained personnel in the appropriate setting creates longer waiting times, more complicated surgical repair for the patient living in low-income countries . Most recently, a publication from Quaedackers et al. made recommendations during COVID-19 pandemic, specifically on pediatric urological conditions and how to prioritize them based on procedure type and the urgency of the procedure without accounting for other factors such as the unique clinical situation of the patient and comorbidities and the possible burden to the health system if a procedure was to be performed . We present a novel scoring system that individualizes the triage of patients based on the impact of postponing surgery and considers their unique clinical condition assisting the surgeon prioritize their cases for patient with pediatric urologic conditions. The reported complication rates from COVID-19 infection are low in children . Thus, there is benefit in performing pediatric urological surgery during this pandemic to avoid further delays in care and associated complications. The argument to modify the original MeNTS scoring system was the impact of age which impacted the triage score in the adult population given their risk of morbidity and mortality associated with COVID. While age may impact the decision to defer a procedure in the adult population, its contribution to the pediatric risk profile is less important. Prachand’s triage score was not originally designed for pediatric conditions and our modifications to this scoring system gave more weight to the impact of postponing surgical intervention of congenital anomalies on subsequent long-term outcomes . Our scoring system allowed triage of patients based on individual clinical condition as oppose to triage based on procedure only. We also took into consideration the potential impact and burden on the health care system as part of the scoring system and was factored into prioritizing the procedure. After reviewing the triage scores of our cohort, patients with scores below 12 required emergent treatment such as septic patients with need for surgical intervention, testicular torsion or obstructing kidney stones. Patients with triage scores between 13 and 15 could be deferred up to 6 weeks without a negative impact on their prognosis but could not be deferred longer. Based on our results, a good example is the case of an indwelling double-J stent that if left longer it can cause complications. Potential changes that could be considered to reduce the score even further for these specific situations would be to use strings to pull them without the need for another intervention or use Salle stents when possible. Patients with scores of 16 or greater could safely wait to be operated after 6 weeks without having a negative impact on their prognosis. A good example is a patient with hypospadias where deferring his care will not make the surgical procedure more difficult and the prognosis will not change overall if treatment is performed later. As a result, the triage scores of 12 and 16 were used as the cutoff values to define high (≤ 12), medium (13–15) or low (16 and greater) priority to triage cases. The patient’s clinical condition impacted the triage score independent from the type of surgical intervention. For example, in our cohort, the case for bilateral laparoscopic orchiopexy (case number 3 on Table ) had pre-operative imaging suggesting the presence of intraparenchymal gonadal neoplasia. For this reason, the score was significantly lower when compared to other cases of undescended testicle cases. In another example, a patient with acute urinary retention was prioritized for an emergent circumcision, typically a low priority procedure. In addition, patients’ double-J stents in place with prolonged indwelling times and potential for significant calcification tended to have lower scores. Conditions typically not urgent become so when care is delayed which is common in low-resource settings due to poor access to healthcare access. Furthermore, it is unknown how the COVID-19 pandemic will further delay care in these low-resource settings and affect with the long-term outcomes of children congenital anomalies. Surgery is instrumental in a reducing the cost and burden to the health care system associated with illness arising from delayed treatment of congenital anomalies on the health care system . The impact on the health outcomes of an already undertreated population compounded by the COVID pandemic are likely significant. For this reason, performing a triage system that accounts for patients’ clinical condition, the type of surgery and the impact on the health care system allows for safe and efficient surgical management of patients with congenital anomalies in an effort to reduce the burden to the health care system long term. A significant amount of urological congenital anomalies are time sensitive and care should not be delayed. Given the current pandemic, relocation of resources implies changes in how priority is distributed amongst the entire population. This is even more true in low- or mid-income countries. Prior to the pandemic, there was a group of procedures that were performed as inpatients such as ureteral reimplants, pyeloplasties, and retrograde intrarenal lithotripsies. Since one of our goals with this project was to be able to prioritize cases individualizing their clinical status without generating a higher burden to our hospital during the pandemic. Implementation of ERAS ® principles helped in reducing this burden by increasing our outpatients from 68 to 90.4%. Our results also show how changes in managing protocols allow to reduce the burden without affecting patient’s safety. An example is pyeloplasties. Before the pandemic, all cases were managed as inpatients for at least 24 h. Considering the need to avoid exposure of patients and their families to SARS-COV-2 while being at the hospital and also by trying to reduce the demand of hospital resources, we made a decision to perform these interventions whenever clinically possible as outpatient procedures. Available literature has become widely accepted and important to reduce hospital stay, postoperative complications and costs to the healthcare system . We decided to focus on intraoperative elements that could be implemented, in an attempt to perform all cases as outpatient procedures. Regional anesthesia (quadratus lumborum, transverse abdominus and pudendal blocks) allowed our patients to be ready for discharge without any need for opioids. In addition, the possibility of minimally invasive procedures, judicious prevention of hypothermia and management of fluids as well as avoidance of drains or catheters made it a feasible approach . Although these measures were successfully implemented, most of current available data about ERAS ® implementation on pediatric urology is insufficient . Although a lot of debate around whether or not, laparoscopic surgery may increase the risk of COVID-19 transmission to the surgical team, there are no definitive data on how to proceed. We performed all cases with droplet precautions using full face masks with EPA P100 of N95 respirators. Massive implementation of COVID-19 testing prior to surgery might be a possibility, resources for such measures in low-resource setting care systems are not sustainable. All changes to medical care that have taken place to reduce the impact and control the pandemic have shown us how important it is to have a flexible and open-minded approach to innovate and adapt. Our results reflect the need to adapt and these adaptations can result in changes in surgical practice that will continue to improve value in healthcare. This study demonstrates the feasibility of implementing fast track surgery care model with a reproducible triage scoring system for patients with congenital urologic anomalies. While the triage system was used only congenital urological anomalies, we believe that this scoring triage system can be applied more broadly to other congenital anomalies. Ultimately, more studies and larger series are needed to better refine such scoring systems. Our experience and the experience of others have demonstrated that of scoring systems can drive decision making during situations in which access to healthcare resources is limited.
Our results present modified triage tool for patients with congenital urologic anomalies. Cutoff values of 12 and 16 allow to prioritize allocation of resources without deferring surgical repairs of congenital anomalies that otherwise would be affected if surgery had to be postponed based on recommendations during the COVID-19 pandemic. Implementation of ERAS ® principles allowed for these procedures to be done in the outpatient setting, preserving valuable healthcare resources.
|
Clinical Follow-Up and Postmortem Findings in a Cat That Was Cured of Feline Infectious Peritonitis with an Oral Antiviral Drug Containing GS-441524 | 1dc923d9-a0f1-4ff5-9d60-0aa29df03d95 | 9506130 | Forensic Medicine[mh] | Feline infectious peritonitis (FIP) caused by the feline coronavirus (FCoV) is an infectious disease that occurs in felids worldwide. Infection with wildtype FCoV initially only causes a harmless intestinal infection. Mutation of the virus within the host, however, can lead to the disease FIP, which once clinically apparent is always fatal within a short period of time if left untreated . Hitherto identified mutations mainly result in changes in coronaviral spike proteins, which enable the virus to replicate effectively within macrophages and to spread within the cat [ , , ]. Subsequent activation of the immune system leads to extensive cytokine release, and thereby exaggerated multisystemic inflammatory lesions. The average survival time without effective treatment is only 8 days after diagnosis , and most cats have to be euthanized early due to their severe condition. However, recent studies have demonstrated the efficacy of antiviral compounds containing the nucleoside analog GS-441524 in cats with FIP [ , , ]. Although successful clinical recovery from FIP has previously been reported , this case report is the first description of the complete recovery in a cat whose tissues could be examined after a fatal road traffic accident via necropsy, including histopathology as well as FCoV immunohistochemistry (IHC) and quantitative reverse transcription polymerase chain reaction (RT-qPCR). In the first controlled study (performed by the same study group) using an oral multicomponent compound called Xraphconn ® , provided by Mutian Life Sciences Limited, containing GS-441524, 18 cats with naturally occurring and confirmed FIP were treated daily over 84 days . All 18 cats recovered with dramatic resolution of all clinical and laboratory parameters, disappearance of effusion, and complete improvement of neurological signs, if present. Quantitative assessment revealed a large reduction in viral loads (across all measured compartments) within the first few days of treatment. Treatment with Xraphconn ® containing GS-441524 was highly effective against FIP, without causing clinically relevant adverse effects . One cat participating in the abovementioned study died in an unobserved road traffic accident 164 days after the end of treatment. The aim of the present case report was (1) to describe the clinical course during and after treatment, (2) to screen for pathological sequelae of FIP in a cat treated with an oral antiviral drug by examining tissue samples via necropsy and histopathology, and (3) to search for FCoV antigen and viral RNA by IHC and RT-qPCR, respectively.
2.1. Signalment and History A male, neutered, 6-month-old European Shorthair cat was initially presented to a local veterinarian in February 2021. Three of the eight litter mates had died of FIP. One of the siblings that also suffered from confirmed FIP (ocular and neurological manifestation) was another study participant, and was also cured . According to the owner, the cat developed clinical signs of recurrent fever, lethargy, and lack of appetite at the end of January 2021. The cat tested negative for feline immunodeficiency virus (FIV) and feline leukemia virus (FeLV). On admission to the local veterinarian at the beginning of February 2021, the cat showed signs of anisocoria, with a relatively miotic, round, and poorly responsive pupil in the left eye (OS). The OS further revealed rubeosis iridis, with marked thickening and bulging of the iris and fibrin precipitates within the anterior eye segment. The intraocular pressure at this time was within normal limits (OS, 11 mmHg; right eye (OD), 16 mmHg). Blood work at the initial examination revealed regenerative anemia (hematocrit, 21.2%; reference range (RR), 29.7–44.5%) and a reticulocyte count of 50.2 × 10 9 /L; all other hematology parameters were unremarkable. The altered hematologic parameters were consistent with FIP. However, the cat did not show signs of neutrophilia or lymphopenia, which can occur in cats with FIP . In the follow-up examination (10 days later), the left globe appeared larger than the right globe. Fibrin deposition within the anterior segment had increased, and was accompanied by a severe hyphema. Fundoscopy revealed partial retinal detachment. The intraocular pressure had increased to 23 mmHg OS (OD 13 mmHg). The OS was enucleated after a third examination 5 days later, due to continuous deterioration and evidence of high-grade anterior uveitis with an intraocular pressure of 48 mmHg (OD 16 mmHg). The enucleated OS was subjected to histopathological examination. Marked pyogranulomatous uveitis and optic neuritis with retinal detachment were identified. FCoV IHC (as described in 2.5) revealed multiple intralesional immunostaining-positive macrophages, confirming, in combination with the ophthalmologic examination, an ocular manifestation of FIP. Additionally, FIP was confirmed by a positive RT-qPCR result of the ocular tissue (as described in 2.6 and shown in Figure 7). 2.2. Improvement of Clinical Signs, Ultrasonographic Findings, and Laboratory Abnormalities during Treatment Before starting treatment (day 0), a complete physical examination, including determination of the Karnofsky score (see ) , as well as hematology, serum biochemistry, and a detailed abdominal ultrasound, was performed. The Karnofsky score modified for cats was used to evaluate the general condition and well-being of the cats. The score ranges from 0% (dead), to 100%, which corresponds to a cat with healthy normal general condition . In the clinic, the cat presented with reduced general condition, pale mucous membranes, dehydration, a body temperature of 39.1 °C, and a body condition score of 3/9 ( A). The cat had a Karnofsky score of 70% on day 0. At presentation, and the start of the study, the cat had a body weight of 1.8 kg, measured using a baby scale (AE Adam MTB 20 baby scale, Felde, Germany). Physical examination and determination of the Karnofsky score were performed daily during hospitalization in the clinic (day 0 to day 7) and at all rechecks on days 14, 28, 56, 83, and 168. The cat was orally treated by daily administration of the multicomponent drug Xraphconn ® (Mutian Life Sciences Limited, Nantong, China) containing the nucleoside analog GS-441524, for 84 days. Due to the ocular manifestation, a dose of supposedly 10 mg/kg (according to the manufacturer) was chosen, with the drug being administered according to manufacturer’s instructions . During hospitalization, the cat received supportive fluid therapy comprising Ringer’s lactate with potassium supplementation at 20 mval/L to control dehydration at individual dosage, calculated by rehydration and maintenance needs. On the first day of treatment (day 0), the cat developed a fever (40.5 °C), whereupon it received a single injection of metamizole (30 mg/kg) intravenously (IV). From the second day of treatment onward, the cat’s appetite improved, and it started to gain weight. The cat was discharged from the clinic on day 7 with a body weight of 2.1 kg. At home, the weight continued to increase steadily, and the cat doubled the initial weight on day 56. Twelve weeks after the end of treatment (day 168), the cat had reached a weight of 4.0 kg ( B and A and ). The Karnofsky score increased to 80% on day 1, and reached 100% on day 7 ( B). Body temperature decreased to 38.5 °C on day 1, and remained normal for the rest of the study period ( ). Abdominal ultrasound was performed using the Logiq E9 ultrasound machine (GE Healthcare) and an 8-MHz microconvex probe, with the cat in dorsal recumbency after clipping the fur. Upon presentation (day 0), the most notable finding was bilateral renomegaly. Regarding longitudinal measurements, the left and right kidney were 4.7 cm and 4.6 cm in size, respectively ( A,B), with a hypoechoic subcapsular rim on both sides. The surface of the left kidney was irregular, and the cortical parenchyma of both kidneys appeared hyperechoic and mottled. There was poor corticomedullary definition and a small amount of anechoic fluid in the retroperitoneal space. Intestinal lymph nodes were mildly enlarged with a homogeneous texture. On day 7, the lengths of the left and right kidneys had decreased to 3.8 and 4.2 cm, respectively ( C,D). The cortex of both kidneys had a homogenous texture, and both kidneys had a distinct corticomedullary definition. The retroperitoneal fluid was no longer visible. A poorly defined medullary rim sign was observed in the right kidney on day 7 ( D) and in the left kidney on day 14 ( E). Both kidneys were considered normal in size, structure, texture, and echogenicity on day 56. The intestinal lymph nodes remained mildly enlarged throughout the study period. Hematology was performed on days 0, 2, 4, 7, 14, 28, 56, 83, and 168 using an automatic analyzer (Cell-Dyn 3500, Abott Laboratories, Chicago, IL, USA). Differential blood count was additionally performed manually on blood smears exposed to Haema Quick Staining/Diff-Quick staining (LT-SYS ® , Eberhard Lehmann GmbH, Berlin, Germany) if hematology parameters were abnormal. On day 0, the cat presented with severe non-regenerative, hypochromic, and microcytic anemia, and moderate thrombocytopenia. On day 2, hematocrit and reticulocyte counts increased, indicating early regeneration. On day 7, the cat was discharged from the clinic with a hematocrit of 0.252 L/L. On day 56, anemia had resolved ( and C). By day 2, the lymphocytes, which were initially within the lower RR (day 0), had transitioned into mild lymphocytosis. Thrombocyte count was within the RR. Throughout the rest of the treatment period, lymphocyte counts indicated mild lymphocytosis, but 12 weeks after the end of treatment, the lymphocyte count was within the RR ( and D). All other hematology parameters were within the RR throughout the entire study period ( ). Serum biochemistry parameters were measured on days 0, 4, 7, 14, 28, 56, 83, and 168 using an automatic analyzer (Hitachi 911, Roche, Grenzach-Wyhlen, Germany). Symmetric dimethylarginine (SDMA) concentration was analyzed at IDEXX Diavet AG (Bäch, Switzerland) using a high-throughput immunoassay, and serum amyloid A (SAA) concentration was determined using a latex agglutination turbidimetric immunoassay reaction (LZ Test SAA, Eiken Chemical Co., Ltd., Tokyo, Japan) on a cobas ® c 501 clinical chemistry analyzer (Roche Diagnostics AG, Rotkreuz, Switzerland). On day 0, the cat showed signs of mild hyperbilirubinemia, mild hyperproteinemia, and mild hypoalbuminemia ( and E–G). SAA was low ( and J) and SDMA was in the upper RR ( ). On day 4, hyperbilirubinemia was no longer observed (until the end of the observation period) ( E). Furthermore, the alkaline phosphatase activity was mildly elevated on day 28, and urea concentration was mildly decreased on day 0. Total protein concentration continued to increase (both globulin and albumin concentrations) until day 7. From there on, until the end of therapy, globulin concentration decreased until it was finally within the RR on day 28. Albumin concentration continued to increase until the end of treatment ( F–H). SDMA concentration was within the RR at all times during treatment. SAA concentration increased to a maximum value on day 14, but was within the RR on day 168 ( J). All other parameters were within the RR. 2.3. Changes in Viral Loads and Anti-FCoV Antibody Titers The courses of the viral load in blood (on days 0, 4, 7, 14, 28, 56, 83, and 168) and feces (on days 0, 1, 2, 3, 4, 5, 6, 7, 14, 28, 56, 83, and 168) were analyzed by RT-qPCR as previously described . Fecal samples were collected using voided samples (on days 0, 2, 3, 4, 5, 6, 7, 14, 28, 56, and 83) or fecal swabs (on days 1 and 168). The viral RNA load in blood before treatment was 11,473 copies/mL blood. On day 4, only 229 FCoV RNA copies/mL blood were detectable. From day 7 onward, no FCoV RNA was detectable in the blood. In feces, excretion of 3437 viral RNA copies/g feces was detectable on day 0. On day 1, only 53 FCoV RNA copies/fecal swab could be detected by RT-qPCR. From day 2, until the end of treatment, viral RNA was no longer detectable in feces ( A). Anti-FCoV antibody titers in serum (on days 0, 7, 14, 28, 56, 83, and 168) were determined by indirect immunofluorescence assay (IFA) as previously described [ , , , , ]. The cat exhibited very high anti-FCoV antibody titer levels at the beginning (1:6400 from day 0 until day 14) of the treatment period. From day 56 on, the antibody titer levels decreased to 1:400 ( B). 2.4. Necropsy and Histopathology A total of 164 days after completion of antiviral treatment, the cat went missing without preceding clinical signs. When the cat was found dead next to the road by the owners, it was immediately submitted to necropsy. A full postmortem examination was performed within 24 h after death. Upon dissection and gross examination, paired samples were taken from all visceral organs and tissues to be (1) snap-frozen for molecular analysis and (2) transferred into 10% neutral-buffered formalin for histopathology and IHC. Fixed samples were trimmed and underwent automated tissue processing, paraffin embedding, and sectioning at 3 µm slice thickness. Sections were routinely stained with hematoxylin–eosin for histopathological evaluation. Further sections of liver and intestine underwent Giemsa and Gram staining. Upon external inspection and superficial dissection, the carcass had undergone rigor mortis. The body showed prototypic lesions associated with road traffic accidents, including superficial abrasions, subcutaneous and intramuscular hematomas of the head and neck area, and frayed claws. Death occurred due to forced ventral hyperflexion of the head and cervical spine, leading to the luxation of the atlanto-occipital joint and complete spinal cord tear at the medullospinal junction. The other parts of the head, including the OD, were unremarkable. Upon dissection of the body cavities, there were no indications of FIP. In particular, there were no effusions or serosal and subserosal changes. However, generalized lymphadenomegaly of both the internal and peripheral lymph nodes was observed, which was most extensive in the mesenteric lymph nodes, as well as swelling of the tonsils. The spleen only showed mild splenomegaly. Respiratory tract and cardiovascular system, as well as the gastrointestinal tract, liver, and pancreas, were unremarkable. Both kidneys were normal in size; cortex and medulla were clearly delineated and highly unremarkable. No retroperitoneal fluids were evident, and the lower urinary tract was normal. Histopathology confirmed the peracute medullospinal injury and excluded pre-existent central nervous system (CNS) changes. The enlargement of the lymphatic tissues histologically corresponded to a severe generalized follicular lymphoid hyperplasia with clearly delineated germinal centers, mantle and marginal zones, and distinct periarteriolar lymphocyte sheaths ( A,B,D). Histological examination of the other organs also confirmed the absence of FIP-associated changes. The lungs instead presented with a moderate multifocal alveolar emphysema ( A). Liver sections showed mild oligofocal lymphocytic infiltrates within portal areas and limiting plates, consistent with mild chronic portal and periportal hepatitis. Moreover, there were occasional, randomly distributed and variably sized foci of lytic hepatocellular necroses associated with individual degenerate polymorphonuclear neutrophils and lymphohistiocytic infiltrates ( B, dashed line). The distribution was typical for the hematogenous spread of bacteria from the guts via the portal vein after passing the mucosal barrier. Accordingly, there were some foci within better-preserved areas of the intestinal mucosa, with an overall increase in the densities of lymphocytes and plasma cells within the propria, accompanied by degenerate polymorphonuclear neutrophils. Hematoxylin–eosin sections, and those stained via Giemsa and Gram staining techniques, exhibited rather diffuse bacterial overgrowths, with mixed morphology ranging from cocci to elongated rods. During necropsy, the OD was removed, and was found to lack any changes indicative of FIP, in contrast to the OS, which had been enucleated prior to antiviral treatment, and showed a fibrotic scar in the area of the ciliary body ( ). 2.5. IHC for FCoV Antigen Tissue sections from all areas sampled (including tonsils, mandibular lymph nodes, mesenteric lymph nodes, spleen, mesenterium, stomach, duodenum, jejunum, cecum, colon, rectum, kidneys, liver, pancreas, brain, spinal cord, OD) underwent immunohistochemical investigation, and tested negative for FCoV antigen expression ( and ). IHC was performed using FIPV3-70 monoclonal antibodies (Linaris Biologische Produkte GmbH, Dossenheim, Germany), on formalin-fixed, paraffin-embedded tissue sections as described previously . Negative controls were included in every IHC staining process. These were two brain sections from a cat with confirmed FIP affecting the central nervous system, into which the antibody was substituted using phosphate-buffered saline (PBS), and by an irrelevant mouse monoclonal antibody (Bo-18). Additionally, and to ensure adequate performance of the antibody, a positive tissue control (tissue from a cat with confirmed FIP) was included in every IHC run. Samples were considered positive for FIP in IHC if typical histopathological lesions were present (e.g., pyogranulomatous and fibrinonecrotic tissue lesions at predilection sites with the exclusion of other pathogens), and FCoV antigen was detected within macrophages in these lesions. Samples were considered negative for FIP in IHC if no histopathological lesions suggestive of FIP were detected, and if FCoV antigens were absent in all tissue samples, including lymph nodes ( ). Additional sections from the spleen, lymph nodes, and liver were stained for lymphocyte markers CD3, CD20, and CD79a. In lymphatic tissues, cell phenotypes segregated with the distribution of physiological T cell and B cell areas. Liver infiltrates mainly consisted of T cells, with only a few B lymphocytes. 2.6. Tissue FCoV RT-qPCR From each sampled organ (mandibular lymph nodes, mesenteric lymph nodes, jejunum, duodenum, cecum, colon, rectum, spleen, kidneys, liver, brain, and OD), 30 mg of frozen tissue was transferred to a soft tissue homogenizing tube of the Precellys Lysing Kit CK14 (Labgene Scientific SA, Châtel-St-Denis, Switzerland), and 600 µL of buffer RLT of the RNeasy Mini Kit (Qiagen AG, Hombrechtikon, Switzerland) containing 1% beta-mercaptoethanol (GBiosciences, St. Louis, MO, USA) was added. Tissues were homogenized twice for 1 min at 5000 Hertz on the Precellys 24 tissue homogenizer (Labgene Scientific SA, Châtel-Saint-Denis, Switzerland), followed by a centrifugation at 17,601 x g for 3 min. RNA was extracted from the 600 µL supernatant using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. RNA was eluted in 30 µL RNase/DNase-free water and stored at −80 °C until further use. All samples were tested undiluted and diluted (1:5) for FCoV RNA by RT-qPCR as described previously . Additionally, all RNA samples from tissues were tested undiluted and diluted (1:5) for the presence of the 18S rRNA housekeeping gene to test for sufficient TNA and absence of potential PCR inhibition. The master mix consisted of 1X Ag-Path RT-PCR buffer (AgPath-IDTM One-step RT-PCR Kit; Applied Biosystems, Rotkreuz, Switzerland), 1.0 μL Array Script reverse transcriptase and AmpliTaq Gold DNA polymerase (AgPath-IDTM One-step RT-PCR Kit; Applied Biosystems), 1X 18s rRNA Dye Mix (VIC/MGB) EUK (Applied Biosystems), and nuclease-free H 2 O was added to a final volume of 20 μL. All FCoVRT-qPCRs were run with 5 μL of TNA in a final volume of 25 μL. Positive and negative controls were run in parallel using a ABI 7500 Fast instrument (Applied Biosystems). An FCoV RNA standard curve was run in parallel as a positive control and to determine the viral RNA copy number . As negative controls, an extraction control (PBS) and DNase/RNase-free water were used. In the 18S rRNA RT-qPCR all tissue tested positive, showing no inhibition at a 1:5 dilution. The lower the cycle threshold (CT) value, the higher the viral load. The cycling conditions were the same as for the FCoV RT-qPCR. All examined tissue samples, including the OD, were negative for FCoV RNA ( ). In comparison, the OS, enucleated before treatment, tested positive for FCoV, with a cycle threshold (CT) value of 25.38.
A male, neutered, 6-month-old European Shorthair cat was initially presented to a local veterinarian in February 2021. Three of the eight litter mates had died of FIP. One of the siblings that also suffered from confirmed FIP (ocular and neurological manifestation) was another study participant, and was also cured . According to the owner, the cat developed clinical signs of recurrent fever, lethargy, and lack of appetite at the end of January 2021. The cat tested negative for feline immunodeficiency virus (FIV) and feline leukemia virus (FeLV). On admission to the local veterinarian at the beginning of February 2021, the cat showed signs of anisocoria, with a relatively miotic, round, and poorly responsive pupil in the left eye (OS). The OS further revealed rubeosis iridis, with marked thickening and bulging of the iris and fibrin precipitates within the anterior eye segment. The intraocular pressure at this time was within normal limits (OS, 11 mmHg; right eye (OD), 16 mmHg). Blood work at the initial examination revealed regenerative anemia (hematocrit, 21.2%; reference range (RR), 29.7–44.5%) and a reticulocyte count of 50.2 × 10 9 /L; all other hematology parameters were unremarkable. The altered hematologic parameters were consistent with FIP. However, the cat did not show signs of neutrophilia or lymphopenia, which can occur in cats with FIP . In the follow-up examination (10 days later), the left globe appeared larger than the right globe. Fibrin deposition within the anterior segment had increased, and was accompanied by a severe hyphema. Fundoscopy revealed partial retinal detachment. The intraocular pressure had increased to 23 mmHg OS (OD 13 mmHg). The OS was enucleated after a third examination 5 days later, due to continuous deterioration and evidence of high-grade anterior uveitis with an intraocular pressure of 48 mmHg (OD 16 mmHg). The enucleated OS was subjected to histopathological examination. Marked pyogranulomatous uveitis and optic neuritis with retinal detachment were identified. FCoV IHC (as described in 2.5) revealed multiple intralesional immunostaining-positive macrophages, confirming, in combination with the ophthalmologic examination, an ocular manifestation of FIP. Additionally, FIP was confirmed by a positive RT-qPCR result of the ocular tissue (as described in 2.6 and shown in Figure 7).
Before starting treatment (day 0), a complete physical examination, including determination of the Karnofsky score (see ) , as well as hematology, serum biochemistry, and a detailed abdominal ultrasound, was performed. The Karnofsky score modified for cats was used to evaluate the general condition and well-being of the cats. The score ranges from 0% (dead), to 100%, which corresponds to a cat with healthy normal general condition . In the clinic, the cat presented with reduced general condition, pale mucous membranes, dehydration, a body temperature of 39.1 °C, and a body condition score of 3/9 ( A). The cat had a Karnofsky score of 70% on day 0. At presentation, and the start of the study, the cat had a body weight of 1.8 kg, measured using a baby scale (AE Adam MTB 20 baby scale, Felde, Germany). Physical examination and determination of the Karnofsky score were performed daily during hospitalization in the clinic (day 0 to day 7) and at all rechecks on days 14, 28, 56, 83, and 168. The cat was orally treated by daily administration of the multicomponent drug Xraphconn ® (Mutian Life Sciences Limited, Nantong, China) containing the nucleoside analog GS-441524, for 84 days. Due to the ocular manifestation, a dose of supposedly 10 mg/kg (according to the manufacturer) was chosen, with the drug being administered according to manufacturer’s instructions . During hospitalization, the cat received supportive fluid therapy comprising Ringer’s lactate with potassium supplementation at 20 mval/L to control dehydration at individual dosage, calculated by rehydration and maintenance needs. On the first day of treatment (day 0), the cat developed a fever (40.5 °C), whereupon it received a single injection of metamizole (30 mg/kg) intravenously (IV). From the second day of treatment onward, the cat’s appetite improved, and it started to gain weight. The cat was discharged from the clinic on day 7 with a body weight of 2.1 kg. At home, the weight continued to increase steadily, and the cat doubled the initial weight on day 56. Twelve weeks after the end of treatment (day 168), the cat had reached a weight of 4.0 kg ( B and A and ). The Karnofsky score increased to 80% on day 1, and reached 100% on day 7 ( B). Body temperature decreased to 38.5 °C on day 1, and remained normal for the rest of the study period ( ). Abdominal ultrasound was performed using the Logiq E9 ultrasound machine (GE Healthcare) and an 8-MHz microconvex probe, with the cat in dorsal recumbency after clipping the fur. Upon presentation (day 0), the most notable finding was bilateral renomegaly. Regarding longitudinal measurements, the left and right kidney were 4.7 cm and 4.6 cm in size, respectively ( A,B), with a hypoechoic subcapsular rim on both sides. The surface of the left kidney was irregular, and the cortical parenchyma of both kidneys appeared hyperechoic and mottled. There was poor corticomedullary definition and a small amount of anechoic fluid in the retroperitoneal space. Intestinal lymph nodes were mildly enlarged with a homogeneous texture. On day 7, the lengths of the left and right kidneys had decreased to 3.8 and 4.2 cm, respectively ( C,D). The cortex of both kidneys had a homogenous texture, and both kidneys had a distinct corticomedullary definition. The retroperitoneal fluid was no longer visible. A poorly defined medullary rim sign was observed in the right kidney on day 7 ( D) and in the left kidney on day 14 ( E). Both kidneys were considered normal in size, structure, texture, and echogenicity on day 56. The intestinal lymph nodes remained mildly enlarged throughout the study period. Hematology was performed on days 0, 2, 4, 7, 14, 28, 56, 83, and 168 using an automatic analyzer (Cell-Dyn 3500, Abott Laboratories, Chicago, IL, USA). Differential blood count was additionally performed manually on blood smears exposed to Haema Quick Staining/Diff-Quick staining (LT-SYS ® , Eberhard Lehmann GmbH, Berlin, Germany) if hematology parameters were abnormal. On day 0, the cat presented with severe non-regenerative, hypochromic, and microcytic anemia, and moderate thrombocytopenia. On day 2, hematocrit and reticulocyte counts increased, indicating early regeneration. On day 7, the cat was discharged from the clinic with a hematocrit of 0.252 L/L. On day 56, anemia had resolved ( and C). By day 2, the lymphocytes, which were initially within the lower RR (day 0), had transitioned into mild lymphocytosis. Thrombocyte count was within the RR. Throughout the rest of the treatment period, lymphocyte counts indicated mild lymphocytosis, but 12 weeks after the end of treatment, the lymphocyte count was within the RR ( and D). All other hematology parameters were within the RR throughout the entire study period ( ). Serum biochemistry parameters were measured on days 0, 4, 7, 14, 28, 56, 83, and 168 using an automatic analyzer (Hitachi 911, Roche, Grenzach-Wyhlen, Germany). Symmetric dimethylarginine (SDMA) concentration was analyzed at IDEXX Diavet AG (Bäch, Switzerland) using a high-throughput immunoassay, and serum amyloid A (SAA) concentration was determined using a latex agglutination turbidimetric immunoassay reaction (LZ Test SAA, Eiken Chemical Co., Ltd., Tokyo, Japan) on a cobas ® c 501 clinical chemistry analyzer (Roche Diagnostics AG, Rotkreuz, Switzerland). On day 0, the cat showed signs of mild hyperbilirubinemia, mild hyperproteinemia, and mild hypoalbuminemia ( and E–G). SAA was low ( and J) and SDMA was in the upper RR ( ). On day 4, hyperbilirubinemia was no longer observed (until the end of the observation period) ( E). Furthermore, the alkaline phosphatase activity was mildly elevated on day 28, and urea concentration was mildly decreased on day 0. Total protein concentration continued to increase (both globulin and albumin concentrations) until day 7. From there on, until the end of therapy, globulin concentration decreased until it was finally within the RR on day 28. Albumin concentration continued to increase until the end of treatment ( F–H). SDMA concentration was within the RR at all times during treatment. SAA concentration increased to a maximum value on day 14, but was within the RR on day 168 ( J). All other parameters were within the RR.
The courses of the viral load in blood (on days 0, 4, 7, 14, 28, 56, 83, and 168) and feces (on days 0, 1, 2, 3, 4, 5, 6, 7, 14, 28, 56, 83, and 168) were analyzed by RT-qPCR as previously described . Fecal samples were collected using voided samples (on days 0, 2, 3, 4, 5, 6, 7, 14, 28, 56, and 83) or fecal swabs (on days 1 and 168). The viral RNA load in blood before treatment was 11,473 copies/mL blood. On day 4, only 229 FCoV RNA copies/mL blood were detectable. From day 7 onward, no FCoV RNA was detectable in the blood. In feces, excretion of 3437 viral RNA copies/g feces was detectable on day 0. On day 1, only 53 FCoV RNA copies/fecal swab could be detected by RT-qPCR. From day 2, until the end of treatment, viral RNA was no longer detectable in feces ( A). Anti-FCoV antibody titers in serum (on days 0, 7, 14, 28, 56, 83, and 168) were determined by indirect immunofluorescence assay (IFA) as previously described [ , , , , ]. The cat exhibited very high anti-FCoV antibody titer levels at the beginning (1:6400 from day 0 until day 14) of the treatment period. From day 56 on, the antibody titer levels decreased to 1:400 ( B).
A total of 164 days after completion of antiviral treatment, the cat went missing without preceding clinical signs. When the cat was found dead next to the road by the owners, it was immediately submitted to necropsy. A full postmortem examination was performed within 24 h after death. Upon dissection and gross examination, paired samples were taken from all visceral organs and tissues to be (1) snap-frozen for molecular analysis and (2) transferred into 10% neutral-buffered formalin for histopathology and IHC. Fixed samples were trimmed and underwent automated tissue processing, paraffin embedding, and sectioning at 3 µm slice thickness. Sections were routinely stained with hematoxylin–eosin for histopathological evaluation. Further sections of liver and intestine underwent Giemsa and Gram staining. Upon external inspection and superficial dissection, the carcass had undergone rigor mortis. The body showed prototypic lesions associated with road traffic accidents, including superficial abrasions, subcutaneous and intramuscular hematomas of the head and neck area, and frayed claws. Death occurred due to forced ventral hyperflexion of the head and cervical spine, leading to the luxation of the atlanto-occipital joint and complete spinal cord tear at the medullospinal junction. The other parts of the head, including the OD, were unremarkable. Upon dissection of the body cavities, there were no indications of FIP. In particular, there were no effusions or serosal and subserosal changes. However, generalized lymphadenomegaly of both the internal and peripheral lymph nodes was observed, which was most extensive in the mesenteric lymph nodes, as well as swelling of the tonsils. The spleen only showed mild splenomegaly. Respiratory tract and cardiovascular system, as well as the gastrointestinal tract, liver, and pancreas, were unremarkable. Both kidneys were normal in size; cortex and medulla were clearly delineated and highly unremarkable. No retroperitoneal fluids were evident, and the lower urinary tract was normal. Histopathology confirmed the peracute medullospinal injury and excluded pre-existent central nervous system (CNS) changes. The enlargement of the lymphatic tissues histologically corresponded to a severe generalized follicular lymphoid hyperplasia with clearly delineated germinal centers, mantle and marginal zones, and distinct periarteriolar lymphocyte sheaths ( A,B,D). Histological examination of the other organs also confirmed the absence of FIP-associated changes. The lungs instead presented with a moderate multifocal alveolar emphysema ( A). Liver sections showed mild oligofocal lymphocytic infiltrates within portal areas and limiting plates, consistent with mild chronic portal and periportal hepatitis. Moreover, there were occasional, randomly distributed and variably sized foci of lytic hepatocellular necroses associated with individual degenerate polymorphonuclear neutrophils and lymphohistiocytic infiltrates ( B, dashed line). The distribution was typical for the hematogenous spread of bacteria from the guts via the portal vein after passing the mucosal barrier. Accordingly, there were some foci within better-preserved areas of the intestinal mucosa, with an overall increase in the densities of lymphocytes and plasma cells within the propria, accompanied by degenerate polymorphonuclear neutrophils. Hematoxylin–eosin sections, and those stained via Giemsa and Gram staining techniques, exhibited rather diffuse bacterial overgrowths, with mixed morphology ranging from cocci to elongated rods. During necropsy, the OD was removed, and was found to lack any changes indicative of FIP, in contrast to the OS, which had been enucleated prior to antiviral treatment, and showed a fibrotic scar in the area of the ciliary body ( ).
Tissue sections from all areas sampled (including tonsils, mandibular lymph nodes, mesenteric lymph nodes, spleen, mesenterium, stomach, duodenum, jejunum, cecum, colon, rectum, kidneys, liver, pancreas, brain, spinal cord, OD) underwent immunohistochemical investigation, and tested negative for FCoV antigen expression ( and ). IHC was performed using FIPV3-70 monoclonal antibodies (Linaris Biologische Produkte GmbH, Dossenheim, Germany), on formalin-fixed, paraffin-embedded tissue sections as described previously . Negative controls were included in every IHC staining process. These were two brain sections from a cat with confirmed FIP affecting the central nervous system, into which the antibody was substituted using phosphate-buffered saline (PBS), and by an irrelevant mouse monoclonal antibody (Bo-18). Additionally, and to ensure adequate performance of the antibody, a positive tissue control (tissue from a cat with confirmed FIP) was included in every IHC run. Samples were considered positive for FIP in IHC if typical histopathological lesions were present (e.g., pyogranulomatous and fibrinonecrotic tissue lesions at predilection sites with the exclusion of other pathogens), and FCoV antigen was detected within macrophages in these lesions. Samples were considered negative for FIP in IHC if no histopathological lesions suggestive of FIP were detected, and if FCoV antigens were absent in all tissue samples, including lymph nodes ( ). Additional sections from the spleen, lymph nodes, and liver were stained for lymphocyte markers CD3, CD20, and CD79a. In lymphatic tissues, cell phenotypes segregated with the distribution of physiological T cell and B cell areas. Liver infiltrates mainly consisted of T cells, with only a few B lymphocytes.
From each sampled organ (mandibular lymph nodes, mesenteric lymph nodes, jejunum, duodenum, cecum, colon, rectum, spleen, kidneys, liver, brain, and OD), 30 mg of frozen tissue was transferred to a soft tissue homogenizing tube of the Precellys Lysing Kit CK14 (Labgene Scientific SA, Châtel-St-Denis, Switzerland), and 600 µL of buffer RLT of the RNeasy Mini Kit (Qiagen AG, Hombrechtikon, Switzerland) containing 1% beta-mercaptoethanol (GBiosciences, St. Louis, MO, USA) was added. Tissues were homogenized twice for 1 min at 5000 Hertz on the Precellys 24 tissue homogenizer (Labgene Scientific SA, Châtel-Saint-Denis, Switzerland), followed by a centrifugation at 17,601 x g for 3 min. RNA was extracted from the 600 µL supernatant using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. RNA was eluted in 30 µL RNase/DNase-free water and stored at −80 °C until further use. All samples were tested undiluted and diluted (1:5) for FCoV RNA by RT-qPCR as described previously . Additionally, all RNA samples from tissues were tested undiluted and diluted (1:5) for the presence of the 18S rRNA housekeeping gene to test for sufficient TNA and absence of potential PCR inhibition. The master mix consisted of 1X Ag-Path RT-PCR buffer (AgPath-IDTM One-step RT-PCR Kit; Applied Biosystems, Rotkreuz, Switzerland), 1.0 μL Array Script reverse transcriptase and AmpliTaq Gold DNA polymerase (AgPath-IDTM One-step RT-PCR Kit; Applied Biosystems), 1X 18s rRNA Dye Mix (VIC/MGB) EUK (Applied Biosystems), and nuclease-free H 2 O was added to a final volume of 20 μL. All FCoVRT-qPCRs were run with 5 μL of TNA in a final volume of 25 μL. Positive and negative controls were run in parallel using a ABI 7500 Fast instrument (Applied Biosystems). An FCoV RNA standard curve was run in parallel as a positive control and to determine the viral RNA copy number . As negative controls, an extraction control (PBS) and DNase/RNase-free water were used. In the 18S rRNA RT-qPCR all tissue tested positive, showing no inhibition at a 1:5 dilution. The lower the cycle threshold (CT) value, the higher the viral load. The cycling conditions were the same as for the FCoV RT-qPCR. All examined tissue samples, including the OD, were negative for FCoV RNA ( ). In comparison, the OS, enucleated before treatment, tested positive for FCoV, with a cycle threshold (CT) value of 25.38.
This case report describes a cat that participated in a clinical trial investigating the efficacy of an oral antiviral drug to treat FIP. Without treatment, virtually all cats suffering from FIP die, making FIP one of the most lethal diagnoses in feline medicine. The survival rate of the 18 cats in this study, however, was 100% . The cat described here was included in the study as it fulfilled the inclusion criteria of (1) a diagnosis of FIP established by IHC, (2) negative test results for FIV and FeLV infection, and (3) absence of other severe diseases. After initial presentation with anterior uveitis, recurrent fever, apathy, and lack of appetite, the cat showed a very swift response to treatment, with rapid improvement of clinical and laboratory parameters leading to full, relapse-free recovery. The cat was treated with 10 mg/kg GS-441524 according to the manufacturer, but it has to be considered that recent additional analysis of the provided drug suggested that a tablet of the multicomponent drug Xraphconn ® contains more GS-441524 than officially stated by the manufacturer (data not shown, personal communication J. Horak). As the analyses show, it is difficult to rely on statements about an unapproved drug. Different illegal antiviral drugs are manufactured under non-standardized conditions. Cats receiving GS-441524 from their owners are treated at variable doses due to the variety of manufactures that produce the drug under uncontrolled conditions. The cats in the cited study were treated with a significantly higher dose than assumed, and only mild side effects occurred . Whether the cat would have been virus-free with a lower dose requires further research. The main clinical abnormality of FIP in the cat presented in this case report was anterior uveitis, manifesting as anisocoria, aqueous flare, and an optic neuritis with retinal detachment, which is consistent with the typical clinical signs observed in cats with ocular FIP . FIP is the most commonly identified cause of uveitis in young cats. According to a study of the North Carolina Veterinary School, 19 of 120 cats (15.8%) with uveitis had FIP . A study from the UK showed a similar distribution pattern, with FIP being the cause of uveitis in 15/92 cats (16.3%) . Confirmation of FIP as the cause of uveitis is often difficult. Clinical signs are not pathognomonic, and aqueous humor analysis, including cytology and screening for infectious diseases is often unspecific [ , , ]. Detection of FCoV antigens within macrophages via immunostaining of ocular tissue can confirm the diagnosis, although false-positive results are possible using immunocytochemistry on aqueous humor . In the cat of the present case report, the OS was enucleated to confirm FIP. Histopathology of the OS revealed marked pyogranulomatous uveitis, pyogranulomatous neuritis of the optic nerve, and retinal detachment. In IHC, multiple macrophages were positive for FCoV antigen. In contrast, the OD had shown no abnormalities on ophthalmic examinations, although at necropsy, a fibrotic scar could be seen on the OD in the area of the ciliary body. It is conceivable that the virus infiltrated not only the OS, but also the contralateral eye, and that these “microlesions” were too minor to be clinically conspicuous and recognized by ophthalmic examination. In both IHC for FCoV antigen and tissue-based FCoV detection using RT-qPCR, FIP could no longer be detected in the enucleated eye postmortem, suggesting cure by treatment with Xraphconn ® . Thus, the scarring could be an indication of previous lesions caused by FIP after healing. A case report from the US described treatment of four cats with FIP and neurological and/or ocular signs . The cats were treated with GS-441524 via subcutaneous injections at a dose of 5–10 mg/kg, applied once daily for at least 12 weeks. All of the four cats responded to treatment initially, including remission of ocular signs. Serial ophthalmic examinations revealed healing of ocular changes presenting as chorioretinal scars, similar to the changes in the cat described in the present case report. However, one cat of the case series of experienced two relapses, and was ultimately euthanized after two courses of treatment. Postmortem examination revealed lymphocytic, histiocytic uveitis, and choroiditis, and viral antigens could be detected in various tissues, including the eye, by IHC. This could either be caused by viral persistence or recurrent FCoV infection and mutation. This cat received GS-441524 at a lower dose (5 mg/kg) than the cat described in this current case report. It is known that drug levels of GS-441524 in aqueous humor are lower than in serum ; thus, it is likely that the higher dose used in the present study was more effective . This was corroborated by the fact that results from both IHC and RT-qPCR conducted on ocular and other tissues were negative for FCoV antigen and RNA, respectively, in the cat in the present report. Further studies are needed to investigate whether an intermediate dose of GS-441524 might be adequate to permanently stop viral replication in cats with ocular FIP. The medullary rim signs, which was visible in both kidneys, also disappeared towards the end of treatment. Medullary rim signs can have various causes, but has been described in association with FIP. In a retrospective study including 243 cats showing medullary rim signs, 15 of these cats were finally diagnosed with FIP . Therefore, treatment with GS-441524 likely also cured FIP-associated changes in the cat’s kidneys. No residual FIP lesions were present in the cat 164 days after the end of treatment with the multicomponent drug Xraphconn ® , apart from a generalized lymphadenopathy due to massive lymphoid hyperplasia. The involvement of mesenteric lymph nodes could partially resemble a consequence to the presumed mild suppurative bacterial enteritis. Gastrointestinal infection and the subsequent portogenic involvement of the liver, on the other hand, cannot explain lymphoid hyperplasia at distant sites such as the tonsils, and mandibular and superficial cervical lymph nodes. The presence of lymphadenopathy could be an incidental unrelated finding, possibly caused by a recent (re)infection with FCoV, or could be an indication of a “long FIP syndrome”. It is possible that some cats have a genetic predisposition of developing FIP. According to current theory on FIP pathogenesis, FIP occurs within cats that are genetically predisposed to being unable to control viral replication effectively, resulting in uncontrolled virus replication and increased opportunity for mutations. These mutations lead to a switch in pathogenicity of the virus, resulting in a variant that is able to efficiently replicate in macrophages. It is the ability to replicate in macrophages in an uncontrolled fashion that distinguishes FIP-causing variants from low-pathogenic feline coronavirus isolates . To further characterize the predisposition to developing FIP, single-nucleotide polymorphisms (SNP) in the feline IFN-γ gene were previously investigated, and certain genotypes were described as FIP-susceptible factors . It was also demonstrated that cats with FIP have lower IFN-γ production than cats infected with FCoV without FIP [ , , , ]. In the present cat, there were other known cases of FIP in the cat’s family history, which suggests a genetic predisposition of this family. However, FCoV infection neither as relapse nor as newly acquired was found by IHC or RT-PCR that would explain the lymphadenopathy. Lymphadenopathy could be considered as a sign of an exaggerated genetically conditioned reaction of the immune system. Also, persistence of lymphadenopathy after elimination of the FCoV could be discussed as a “long FIP syndrome”. In human medicine, cases have been described in which viral RNA persisted after clinical resolution of an acute infection, such as Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the long COVID syndrome ; however, what is detected by RT-PCR is mostly fragmented RNA strains . In the present case report, RT-PCR did not test positive in any tissue of the cats, and thus, no fragmented persisting RNA was found. Nonetheless, lymphadenopathy might be an aftereffect of FIP, without the presence of virus. Further studies are needed to prove whether these changes are due to a long-term adverse effect of treatment or potentially associated with a “long FIP syndrome”. The initially seen fluid in the retroperitoneal space was not present in the postmortem examination. In FIP, lesions are induced by immune complexes deposited at the wall of blood vessels, subsequently activating the complement cascade, and damaging vascular tissues . In addition to the spleen, the omentum and the mesenteric lymph nodes are tissues with high viral loads . However, no virus was presented in the lymphoid tissue of the cat analyzed here. Moreover, it remains to be determined whether the preceding FIP might have paved the way for intestinal infection e.g., by affecting local or systemic immunocompetence. FCoV was no longer detected in any tissue of the cat. In addition, viral load decreased in the blood within a short period of time after treatment initiation, and fecal shedding stopped by day 2. These results indicate that the cat was 100% cured of FIP. The antibody titer decreased during treatment, although it did not become negative. Possibly, the titer would have continued to decrease after infection was cleared, as described previously in FCoV-infected cats without FIP. Anti-FCoV antibodies can sometimes be measured after the clearance of (harmless) FCoV infection for several months, and this is not a sign of viral persistence . FIP relapses after treatment with GS-441524 have been described in a few cases . In these cases, the question arises whether a reinfection with FCoV and a new mutation took place, or whether the virus could not be completely eliminated in these cats and was still present in individual tissues. The presently described cat was 100% free of the virus, making relapses after appropriate treatment very unlikely. Unfortunately, the drug used for the treatment of FIP in this cat is currently not legally available for veterinary use in most countries, forcing well-meaning owners to self-diagnose and treat their cats based on judgment of non-veterinary lay people and social media groups. Thus, there is an urgent need for respective official bodies and industry to work towards a swift licensing process of the drug, so that it can be legally used by veterinary experts to offer supervised treatment to cats suffering from FIP. The limitation of this case report is that only one cat was examined via necropsy. The other study cats are still alive at the time of publication.
This is the first report describing clinical and laboratory as well as postmortem findings in a cat cured of FIP that subsequently died accidentally. GS-441524 was highly effective in this cat, and neither signs of FIP nor FCoV RNA or antigens could be detected postmortem. GS-441524 is currently the most effective treatment of FIP, and should be licensed for veterinary use as soon as possible.
|
Impact of land-use and fecal contamination on | 2494edab-c657-47e2-bb42-d440f958f1d5 | 11685573 | Microbiology[mh] | Escherichia coli is metabolically flexible, enabling it to colonize the gastrointestinal tract of various mammal and bird species . This makes it a useful indicator of fecal contamination in microbial water quality assessments , signalling the possible presence of other less common fecal pathogens such as Shiga toxin-producing E. coli (STEC), Salmonella , Campylobacter , Cryptosporidium , and Giardia which are more fastidious in their growth requirements . In New Zealand, routine pathogen modelling and risk-based approaches have been used to determine generic E. coli freshwater concentrations that indicate when recreational activity crosses an infection risk threshold for campylobacteriosis . Thus, regular monitoring of E. coli using standard culture-based methods provides a framework for determining potential health risks from contaminated water and the impact of land management practices on microbial water quality trends. Recent evidence suggests that two distinct ‘naturalized’ Escherichia types can persist in freshwater environments long after fecal contamination events – . One type is the fecal E. coli that alert authorities to persistent populations of E. coli derived from fecal contamination events and hence potential pathogen presence. The second type are represented by cryptic clades of Escherichia , several of which have now been characterized as Escherichia whittamii (Clade II) , , Escherichia marmotae (Clade V) and Escherichia ruysiae (Clades III/IV) . High resolution whole genome sequence (WGS) analysis of Escherichia species indicates they are genetically distant from E. coli , but phenotypically indistinct, when using routine culture-based methods for microbial water quality assessments – . The scarcity of these environmental Escherichia species in previous studies , – suggest they represent a lower likelihood of potential health risk, although they have been isolated from immunocompromised patients . Several avian/wildlife hosts , , – have been identified, but the association of these environmental Escherichia species with waterborne pathogens and hence utility in indicating the overall public health risk has not yet been established for microbial water quality assessments . Isolate level approaches for E. coli , are insufficient for understanding population diversity in complex samples such as feces or water where the presence of certain E. col i genotypes/phenotypes in low numbers can make detection and isolation difficult using routine culture media – . Molecular methods targeting the hypervariable gene gnd (encoding 6-phosphogluconate dehydrogenase), allow the detection of a wide variety of distinct E. coli variants , . By targeting a short 284 bp amplicon of the 1407 bp full length gene, over 700 different gnd sequence types (gSTs) from Escherichia species have been recognized and cataloged in a database (gndDb) to provide primary subtyping , . These gSTs include E. coli from pathogenic and commensal/environmental sources and phenotypically indistinguishable Escherichia species ( E. marmotae , E. ruysiae , and E. whittamii ) collected from various geographical regions. In addition, sequence analysis of pooled metabarcoded gnd amplicons from environmental samples allows for in-depth population analysis of Escherichia species providing diversity information relevant to human health risk investigations and revealing the presence of environmental Escherichia populations unrelated to human/livestock fecal inputs . Metabarcoding for E. coli population analysis overcomes the limitations of culture-based analysis, which can be slow and laborious, enabling the study of heterogeneous populations. It also supports a longitudinal sampling approach across multiple sites or the assessment of the impact of conceptual or defined water quality mitigations on E. coli populations . Escherichia populations identified through water quality monitoring assessments from targeted sample sites and matrices (including water, soil, sediment, aquatic biofilm, and feces) provide data that can help understand factors such as potential sources and drivers influencing freshwater quality especially for public health risk assessments. This study addresses a significant knowledge gap by applying a holistic amplicon metabarcoding approach to explore Escherichia population composition and diversity in a range of environmental samples from freshwater sites with contrasting fecal indicator bacteria (FIB) loadings and land-use types, including an intensively managed mammal-free native forest site and pastoral farming sites. By providing baseline data on Escherichia diversity in relation to indicators of fecal microbial contamination and diverse fecal sources, this study offers insights for future public health and ecological investigations and the impact of faecal sources on freshwater systems. Metabarcoding revealed Escherichia diversity across all sites and most samples In total 26 biofilm, 34 soil, 35 sediment, and 47 water samples were collected together with 47 opportunistic fecal samples (Fig. , Supplementary Table ). Boiled lysates of sample enrichments were used as DNA templates to generate individual barcoded amplicons and target detection of viable bacteria, analogous to standard culture-based microbial water quality assessment methods (e.g., IDEXX Colilert-18 ® and Quanti-Tray/2000 ® ). Barcoded gnd amplicons were obtained from samples recovered from all sites and most samples (171 of 189, 90.5%), including all biofilm ( n = 26), most feces (44 of 47, 93.6%), sediment (28 of 35, 80.0%), water (44 of 47, 93.6%) and soil (29 of 34, 85.3%) samples. PCR negative samples indicative of low E. coli concentrations, were from Site 1 (3 sediment, 4 soil, 1 feces), Site 2 (1 feces, 1 soil), Site 3 (2 sediment), Site 4 (1 sediment, 1 feces) and Site 5 (1 sediment, 3 water). Compositional analysis of samples identifies distinct Escherichia community profiles E. coli community analysis was undertaken by targeting the hypervariable gnd allele from environmental samples (Supplementary Table ). In total 2.493 million reads from 168 samples (> 100 reads) averaging 14,839 reads per library (maximum: 38,104 median: 14,101) (Supplementary Table ) were mapped against reference gnd sequences (gndDb) . Overall, 952 distinct ASVs were identified (Supplementary Table ), and on average there were 50 ASVs per environmental sample ( n = 168), with a maximum of 240 separate ASVs identified from one water sample (Site 2) (Supplementary Table ). Rarefaction analysis performed across samples generated flattened curves indicating that the amplicon sequencing depth was sufficient to accurately estimate ASV richness and reflected that the observed diversity was close to the true diversity of the respective target environmental communities (Supplementary Figure ). By mapping ASVs against gndDb, ASVs were able to be separated into three groups (Table ) represented by those which matched E. coli gnd sequences (e.g. ‘ gnd sequence types’ or gSTs), those which matched non- E. coli Escherichia species gnd sequences from gndDb, and the remaining gnd ASVs from other Enterobacteriaceae with no match to gndDb , . Most reads (80.6%) matched E. coli gnd sequences, 10.3% matched non- E. coli Escherichia species, and the remaining 9.1% other Enterobacteriaceae (Table ), but gSTs matching gndDb only formed 44.7% of all ASVs. All gSTs included in the mock community positive control libraries were detected in accordance with anticipated read numbers according to previous studies . The gST with the highest relative abundance at the read level was gST537 (3.4%, 83,967 reads) which was found in 28.6% samples (48 of 168, average 1,749 reads) across all five sites, mainly from water ( n = 22) and biofilm ( n = 15), but also from feces ( n = 6), sediment ( n = 4) and soil ( n = 1) (Supplemental Table ) and matched a gST from gndDb corresponding to E. marmotae (Clade V). Including gST537, 22 gSTs representing 17 E. marmotae , 4 E. ruysiae (Clades III and IV) and a single E. whittamii (Clade II) were identified from gnd metabarcoding sequencing data. Overall non- E. coli Escherichia species-associated gSTs were identified from 53.6% (90 of 168) of samples, especially water (81.8%, 36 of 44) and biofilm (84.0%, 21 of 25), but were less common in sediment (44.4%, 12 of 27), feces (36.4%, 16 of 44), and soil (17.9%, 5 of 28). The top 20 most abundant gSTs (830,000 reads), consisting of 17 E. coli and 3 E. marmotae -associated gSTs (Supplementary Table ), accounted for 33.3% of the total reads. Agglomerated reads of the top 20 gSTs from the five sample types across individual sites suggested some site and sample specificity (Fig. ) and overall prevalence data were examined in more detail using linear mixed effects models (Supplementary Table ). When compared with reads from Site 1 as the control variable, the prevalence of 95.0% (19 of 20) of gSTs was significantly different across all sites ( p < 0.05) (Supplementary Table a). Overall, the prevalence of E. coli B2 phylotype gST036 (Site 5), and E. marmotae gSTs, gST536 (Site 5), gST537 (Site 5) and gST548 (Sites 3 and 5) was significantly less ( p < 0.05) compared to samples from Site 1 (Supplementary Table a). In contrast the prevalence of E. coli B1 phylotypes gST152, gST218, gST251 and gST267 was significantly greater across all individual sites (Supplementary Table S6a). Similarly, when ‘Visit’ as a random variable was nested within ‘Site’, and gST prevalence from fecal samples were included as the control variable, the prevalence of 90% (18 of 20) of the top 20 gSTs was significantly different across at least one sample ( p < 0.05) and 20 (100.0%) differed across all samples ( p < 0.05) (Supplementary Table b). Eight gSTs (gST019, gST070, gST152, gST161, gST231, gST321, gST514 and gST522) were significantly less prevalent ( p < 0.05) in soil compared to feces. Eleven gSTs (55%) including, the three E. marmotae gSTs, had a significantly greater prevalence ( p < 0.05) from biofilm samples compared to feces, and of these eleven, ten also had a significantly greater prevalence from water samples ( p < 0.05) (Supplementary Table b). The amplicon groups from each library corresponding to E. coli and non- E. coli Escherichia species gSTs were compared to generic E. coli MPN per 100 mL concentrations from the corresponding water samples (determined using Colilert-18 ® /Quanti-Tray/2000 ® ) as described in our previous work . A positive relationship (R² = 0.76) was evident where increasing numbers of generic E. coli (culture-based method) broadly corresponded with E. coli and non- E. coli Escherichia species gST diversity using this metabarcoding method (Fig. ). gnd profiles reveal Escherichia population richness and diversity across different sample types and sites Subsequent alpha and beta diversity analysis of all gnd sequence types (gSTs) ( n = 426) matching Escherichia coli , and Escherichia species ( E. marmotae , E. ruysiae and E. whittamii ) was quantified by Chao1, Shannon indices and Principal Component Analysis (PCA) plots (Figs. 4, 5 and 6). Site 1 (low-impact native forest) samples were associated with less diverse populations compared to other livestock impacted sites with higher E. coli loadings ( p < 0.001) (Figs. and A and B). Water and biofilm samples had increased richness and within sample population diversity (Shannon indices) compared to soil, sediment and feces samples ( p < 0.01) (Fig. C and D). The richness of gSTs corresponding to only non- E. coli Escherichia species ( E. marmotae , E. ruysiae , and E. whittamii ) ( n = 22) was increased for Site 4 ( p < 0.01) and reduced for Site 5 ( p < 0.01), compared to Site 1 (Fig. A). Similarly, compared to Site 1, there was an increased average Shannon diversity index for Site 4 ( p < 0.05) but reduced for Site 5 ( p < 0.05) (Fig. B). Compared to water, there was a reduced richness of Escherichia species gSTs in feces ( p < 0.05) (Fig. C), and reduced Shannon diversity in feces ( p < 0.001) and sediment ( p < 0.05) (Fig. D) supporting the hypothesis that feces are not the predominant source of non- E. coli Escherichia species gSTs. Subsequent PCA to examine between community diversity of the sample libraries constituting all Escherichia coli and non- E. coli Escherichia species gSTs ( n = 426) from each respective library ( n = 168) confirmed evidence of reduced sample gnd -amplicon community diversity from the low-impact native forest headwater site (Site 1), and differentiated ASV populations from higher order waterways (Sites 2, 3 and 4) with the constructed wetland (Site 5) (Fig. ). Although only E . coli ( n = 404) and non- E. coli Escherichia species gST ( n = 22) groupings (Table ) were included in the alpha and beta diversity analysis (Figs. and ), the addition of the unassigned ASVs in subsequent analysis to demonstrate Shannon, Chao1 and PCA diversity of populations including all 952 ASVs had minimal impact on the resulting data (Supplementary Figures and ). There was also a significant association between Shannon diversity of all water samples and respective sample site (ANOVA, p < 0.001) with overall diversity of Site 1 less than the other four sites ( p < 0.01), and Site 5 less than Site 2, Site 3 and Site 4 ( p < 0.01), but there was no interaction between Shannon diversity and overall catchment size (ha) (ANOVA, p > 0.05). Using multivariate analysis of variance (PERMANOVA) to assess the proportion of ASV variation between variables (sites, visits, sample type), there was an interaction between site and sample, indicative of a difference between sites, difference between visits, difference between samples, and a variation of the difference between samples across the separate sites ( p < 0.001). Site contributed most variation (9.32%), followed by sample (7.97%), visit (1.89%), and sample within site (6.72%); residuals accounted for 74.1% reflecting the extreme variability in the diversity of gnd ASV populations. Specific gSTs infer E. coli persistence and prevalence from sample sites Identification of gSTs from cultured isolates and subsequent analysis of WGS data indicated the presence and implied long term persistence of clonal E. coli isolates of gST535, gST161, and gST251. In this metabarcoding analysis, the ASV corresponding to gST535 was identified from 24 sample enrichments (6 biofilm, 3 sediment, 1 soil, 9 water and 5 feces) (Supplementary Table ). It was absent from Sites 1 and 5, recovered from six samples at Site 2, but was most abundant from adjoining Site 3 ( n = 9) and Site 4 ( n = 9). All five fecal samples where gST535 reads were recorded, were from Sites 3 and 4 and originated from possum ( n = 2), avian ( n = 2) or rat ( n = 1) fecal material providing further evidence of potential wildlife contamination at these two sites. Similarly, gST251 was recovered from 21 of 28 visits (75.0%) and present in 63 samples consisting of all the different sample types (Supplementary Table ) but was absent from Site 1. gST161 was also identified from all sample types and from all five sites (Supplementary Table ). Escherichia diversity across all sites and most samples In total 26 biofilm, 34 soil, 35 sediment, and 47 water samples were collected together with 47 opportunistic fecal samples (Fig. , Supplementary Table ). Boiled lysates of sample enrichments were used as DNA templates to generate individual barcoded amplicons and target detection of viable bacteria, analogous to standard culture-based microbial water quality assessment methods (e.g., IDEXX Colilert-18 ® and Quanti-Tray/2000 ® ). Barcoded gnd amplicons were obtained from samples recovered from all sites and most samples (171 of 189, 90.5%), including all biofilm ( n = 26), most feces (44 of 47, 93.6%), sediment (28 of 35, 80.0%), water (44 of 47, 93.6%) and soil (29 of 34, 85.3%) samples. PCR negative samples indicative of low E. coli concentrations, were from Site 1 (3 sediment, 4 soil, 1 feces), Site 2 (1 feces, 1 soil), Site 3 (2 sediment), Site 4 (1 sediment, 1 feces) and Site 5 (1 sediment, 3 water). Escherichia community profiles E. coli community analysis was undertaken by targeting the hypervariable gnd allele from environmental samples (Supplementary Table ). In total 2.493 million reads from 168 samples (> 100 reads) averaging 14,839 reads per library (maximum: 38,104 median: 14,101) (Supplementary Table ) were mapped against reference gnd sequences (gndDb) . Overall, 952 distinct ASVs were identified (Supplementary Table ), and on average there were 50 ASVs per environmental sample ( n = 168), with a maximum of 240 separate ASVs identified from one water sample (Site 2) (Supplementary Table ). Rarefaction analysis performed across samples generated flattened curves indicating that the amplicon sequencing depth was sufficient to accurately estimate ASV richness and reflected that the observed diversity was close to the true diversity of the respective target environmental communities (Supplementary Figure ). By mapping ASVs against gndDb, ASVs were able to be separated into three groups (Table ) represented by those which matched E. coli gnd sequences (e.g. ‘ gnd sequence types’ or gSTs), those which matched non- E. coli Escherichia species gnd sequences from gndDb, and the remaining gnd ASVs from other Enterobacteriaceae with no match to gndDb , . Most reads (80.6%) matched E. coli gnd sequences, 10.3% matched non- E. coli Escherichia species, and the remaining 9.1% other Enterobacteriaceae (Table ), but gSTs matching gndDb only formed 44.7% of all ASVs. All gSTs included in the mock community positive control libraries were detected in accordance with anticipated read numbers according to previous studies . The gST with the highest relative abundance at the read level was gST537 (3.4%, 83,967 reads) which was found in 28.6% samples (48 of 168, average 1,749 reads) across all five sites, mainly from water ( n = 22) and biofilm ( n = 15), but also from feces ( n = 6), sediment ( n = 4) and soil ( n = 1) (Supplemental Table ) and matched a gST from gndDb corresponding to E. marmotae (Clade V). Including gST537, 22 gSTs representing 17 E. marmotae , 4 E. ruysiae (Clades III and IV) and a single E. whittamii (Clade II) were identified from gnd metabarcoding sequencing data. Overall non- E. coli Escherichia species-associated gSTs were identified from 53.6% (90 of 168) of samples, especially water (81.8%, 36 of 44) and biofilm (84.0%, 21 of 25), but were less common in sediment (44.4%, 12 of 27), feces (36.4%, 16 of 44), and soil (17.9%, 5 of 28). The top 20 most abundant gSTs (830,000 reads), consisting of 17 E. coli and 3 E. marmotae -associated gSTs (Supplementary Table ), accounted for 33.3% of the total reads. Agglomerated reads of the top 20 gSTs from the five sample types across individual sites suggested some site and sample specificity (Fig. ) and overall prevalence data were examined in more detail using linear mixed effects models (Supplementary Table ). When compared with reads from Site 1 as the control variable, the prevalence of 95.0% (19 of 20) of gSTs was significantly different across all sites ( p < 0.05) (Supplementary Table a). Overall, the prevalence of E. coli B2 phylotype gST036 (Site 5), and E. marmotae gSTs, gST536 (Site 5), gST537 (Site 5) and gST548 (Sites 3 and 5) was significantly less ( p < 0.05) compared to samples from Site 1 (Supplementary Table a). In contrast the prevalence of E. coli B1 phylotypes gST152, gST218, gST251 and gST267 was significantly greater across all individual sites (Supplementary Table S6a). Similarly, when ‘Visit’ as a random variable was nested within ‘Site’, and gST prevalence from fecal samples were included as the control variable, the prevalence of 90% (18 of 20) of the top 20 gSTs was significantly different across at least one sample ( p < 0.05) and 20 (100.0%) differed across all samples ( p < 0.05) (Supplementary Table b). Eight gSTs (gST019, gST070, gST152, gST161, gST231, gST321, gST514 and gST522) were significantly less prevalent ( p < 0.05) in soil compared to feces. Eleven gSTs (55%) including, the three E. marmotae gSTs, had a significantly greater prevalence ( p < 0.05) from biofilm samples compared to feces, and of these eleven, ten also had a significantly greater prevalence from water samples ( p < 0.05) (Supplementary Table b). The amplicon groups from each library corresponding to E. coli and non- E. coli Escherichia species gSTs were compared to generic E. coli MPN per 100 mL concentrations from the corresponding water samples (determined using Colilert-18 ® /Quanti-Tray/2000 ® ) as described in our previous work . A positive relationship (R² = 0.76) was evident where increasing numbers of generic E. coli (culture-based method) broadly corresponded with E. coli and non- E. coli Escherichia species gST diversity using this metabarcoding method (Fig. ). profiles reveal Escherichia population richness and diversity across different sample types and sites Subsequent alpha and beta diversity analysis of all gnd sequence types (gSTs) ( n = 426) matching Escherichia coli , and Escherichia species ( E. marmotae , E. ruysiae and E. whittamii ) was quantified by Chao1, Shannon indices and Principal Component Analysis (PCA) plots (Figs. 4, 5 and 6). Site 1 (low-impact native forest) samples were associated with less diverse populations compared to other livestock impacted sites with higher E. coli loadings ( p < 0.001) (Figs. and A and B). Water and biofilm samples had increased richness and within sample population diversity (Shannon indices) compared to soil, sediment and feces samples ( p < 0.01) (Fig. C and D). The richness of gSTs corresponding to only non- E. coli Escherichia species ( E. marmotae , E. ruysiae , and E. whittamii ) ( n = 22) was increased for Site 4 ( p < 0.01) and reduced for Site 5 ( p < 0.01), compared to Site 1 (Fig. A). Similarly, compared to Site 1, there was an increased average Shannon diversity index for Site 4 ( p < 0.05) but reduced for Site 5 ( p < 0.05) (Fig. B). Compared to water, there was a reduced richness of Escherichia species gSTs in feces ( p < 0.05) (Fig. C), and reduced Shannon diversity in feces ( p < 0.001) and sediment ( p < 0.05) (Fig. D) supporting the hypothesis that feces are not the predominant source of non- E. coli Escherichia species gSTs. Subsequent PCA to examine between community diversity of the sample libraries constituting all Escherichia coli and non- E. coli Escherichia species gSTs ( n = 426) from each respective library ( n = 168) confirmed evidence of reduced sample gnd -amplicon community diversity from the low-impact native forest headwater site (Site 1), and differentiated ASV populations from higher order waterways (Sites 2, 3 and 4) with the constructed wetland (Site 5) (Fig. ). Although only E . coli ( n = 404) and non- E. coli Escherichia species gST ( n = 22) groupings (Table ) were included in the alpha and beta diversity analysis (Figs. and ), the addition of the unassigned ASVs in subsequent analysis to demonstrate Shannon, Chao1 and PCA diversity of populations including all 952 ASVs had minimal impact on the resulting data (Supplementary Figures and ). There was also a significant association between Shannon diversity of all water samples and respective sample site (ANOVA, p < 0.001) with overall diversity of Site 1 less than the other four sites ( p < 0.01), and Site 5 less than Site 2, Site 3 and Site 4 ( p < 0.01), but there was no interaction between Shannon diversity and overall catchment size (ha) (ANOVA, p > 0.05). Using multivariate analysis of variance (PERMANOVA) to assess the proportion of ASV variation between variables (sites, visits, sample type), there was an interaction between site and sample, indicative of a difference between sites, difference between visits, difference between samples, and a variation of the difference between samples across the separate sites ( p < 0.001). Site contributed most variation (9.32%), followed by sample (7.97%), visit (1.89%), and sample within site (6.72%); residuals accounted for 74.1% reflecting the extreme variability in the diversity of gnd ASV populations. E. coli persistence and prevalence from sample sites Identification of gSTs from cultured isolates and subsequent analysis of WGS data indicated the presence and implied long term persistence of clonal E. coli isolates of gST535, gST161, and gST251. In this metabarcoding analysis, the ASV corresponding to gST535 was identified from 24 sample enrichments (6 biofilm, 3 sediment, 1 soil, 9 water and 5 feces) (Supplementary Table ). It was absent from Sites 1 and 5, recovered from six samples at Site 2, but was most abundant from adjoining Site 3 ( n = 9) and Site 4 ( n = 9). All five fecal samples where gST535 reads were recorded, were from Sites 3 and 4 and originated from possum ( n = 2), avian ( n = 2) or rat ( n = 1) fecal material providing further evidence of potential wildlife contamination at these two sites. Similarly, gST251 was recovered from 21 of 28 visits (75.0%) and present in 63 samples consisting of all the different sample types (Supplementary Table ) but was absent from Site 1. gST161 was also identified from all sample types and from all five sites (Supplementary Table ). E. coli is typically used to indicate contamination of freshwater due to mammal and bird feces. However, any individual freshwater grab sample may be contaminated with diverse E. coli and other microbial pathogens from multiple fecal sources, including human (sewage discharges, septic systems and urban run-off), agricultural (livestock fecal run-off), and wildlife sources (fecal contamination from birds, wild mammals etc.). Our previous work examining freshwater Escherichia isolates (20 isolates from each freshwater sample) indicated that the Shannon diversity of E. coli phylotypes is correlated with E. coli concentration (MPN/100 mL) . For the study of E. coli communities, the use of culture media and labour-intensive recovery of individual microbial isolates from freshwater samples impacted by diverse fecal sources may exclude bacterial subtypes present at low relative abundance , . A more holistic metabarcoding approach for FIB community analysis considers the richness and diversity of E. coli , differentiating non- E. coli Escherichia species from sample matrices (water, soil, sediment, aquatic biofilm, and feces), which in previous studies has provided important links between E. coli concentrations (Fig. ), diverse fecal sources and inference of pathogen contamination levels . Moreover, this amplicon sequencing approach is sufficiently robust to assess human health risk with the identification and differentiation of taxonomically distinct Escherichia species, which although phenotypically indistinguishable from E. coli using routine microbial water quality assessment methods, are potentially less pathogenic to humans. This information can provide a more complete understanding of the overall extent of waterborne fecal contamination and assist with targeted mitigation strategies to reduce the sources of contamination and protect public health . Thus, this study extends our previous culture-based investigations, and geographically constrained studies investigating FIB diversity , , . Low impact sites where introduced mammal pests are actively excluded (Site 1), with fewer anticipated fecal sources and lower overall E. coli loads, provide a unique opportunity to study environmental contamination from native forest catchments. These managed ‘natural’ sites provide information on baseline E. coli concentrations and microbe diversity to compare with samples obtained from ‘impacted’ pastoral or urban sites exposed to a greater variety of fecal sources . The use of metabarcoding methods to assess Escherichia diversity can also provide insights into impacts of contrasting land-uses. Previous sampling of freshwater from genuine low impact sites (native or exotic forest sites) has identified distinct phylotype variability, such as increased prevalence of phylotype B2, and fewer fecal contaminants from pastoral/urban sources . In the current study, examination of Escherichia communities from the native forest site (Site 1) was associated with reduced alpha diversity measures (Fig. A and B) in line with lower E. coli contamination levels. These findings support evidence from other studies demonstrating lower E. coli phylotype diversity, and lower prevalence of bacterial pathogens from low impact sites , . By comparing E. coli phylogenies described previously from these same samples analysed in this study, further links can be made at the isolate level from WGS data, and at the population level, with gnd amplicon sequencing studies. For example, there was a higher prevalence of gST036 in Site 1 compared to Site 5 ( p < 0.05) (Supplementary Table a). The corresponding gST036 E. coli isolates have been characterized as phylotype B2 , an E. coli phylogeny previously associated with low impact sites , , but also associated with extra-intestinal disease and a multidrug resistance phenotype . Similarly, gST548-positive isolates are indicative of E. marmotae , and compared to Site 1, also had a decreased prevalence in Site 3 and Site 5, which were associated with livestock inputs ( p < 0.05) (Supplementary Table a). In contrast, other gSTs, such as gST152, gST161, gST218 and gST251, had a significantly increased prevalence in Sites 2–5 compared to Site 1 ( p < 0.05) (Supplementary Table b). These gSTs correspond to E. coli phylotype B1 , a generalist phylotype more commonly associated with human and animal fecal samples , , aged fecal sources and freshwaters with diverse fecal sources . Overall, there was good correlation between the number of gSTs (matching those from the gndDb database) identified from water samples, and E. coli MPN per 100 mL concentration data, indicating that as the E. coli concentration and potentially the diversity of fecal sources increase, so too does the diversity of Escherichia populations (Fig. ). Datapoints with increased E. coli concentration and lower diversity may represent samples with high E. coli levels from fewer fecal sources, such as localized point source discharge or aged fecal sources with a subset of persistent and multiplying E. coli strains. For example, Site 5 had high E. coli loads originating from cattle feces from a single farm deposited near the constructed wetland site. Conversely, those datapoints from Sites 2, 3 and 4, with increased E. coli population diversity and lower E. coli concentrations may correspond to contamination from multiple farms with more diverse fecal sources in larger waterways with more mixing and uniform distribution of diverse E. coli populations. The average level of E. coli diversity from fecal samples was lower than from water. Waterbodies are often impacted by run-off from multiple fecal sources with consequent mixing and transport of contaminants from diverse sources. In contrast, the mammal/avian gastrointestinal tract is a niche impacted by competing metabolic and colonisation factors that potentially limits the numbers and diversity of different E. coli able to colonize such a dynamic environment . For example, only one and seven gSTs (six E. coli and one E. marmotae ) respectively were identified from two possum fecal samples (S073 and S095) examined in this study where E. coli gST535 was the most abundant gST identified. These findings are corroborated by another study in the vicinity of Sites 3 and 4, which investigated the role of possums as a source of FIB and showed alpha diversity values of E. coli from possum feces were less than water, biofilm and sediment . The endemism of gST535 to wildlife in the vicinity of Sites 3 and 4 is apparent but further studies are required to investigate the role of this gST as a significant FIB at other locales and from other wildlife fecal sources. Ultimately, water is the vehicle in which the mobile fraction of the bacterial population are transported from one site to another with mobilized bacteria settling on new substrata when flow rates decrease. Initial observations indicate an association of E. coli diversity and different fecal sources which can be extrapolated to also imply an increased number of infectious pathogens , . The greatest Escherichia diversity from metabarcoding sequencing data was from water and biofilm samples, compared to soil, sediment and feces (Figs. and ). Substrate surfaces provide attachment sites for bacterial protection from predation in multi-species and multi-strain biofilms established through continuous input of transient bacteria from the wider freshwater catchment and entrainment from the water column , . However, loosely or non-adherent sedimentary bacteria are quickly re-mobilized when flow velocity increases , highlighting the dynamic exchange of bacteria, including E. coli and non- E. coli Escherichia species, between submerged surfaces and the water column . Although the ASVs not assigned to the gndDb database consisted of over half the total ASVs (Table ), their overall proportion to the total number of reads was low (< 10% total reads) and their bearing on alpha and beta diversity was relatively minor (Supplementary Figs. and ). Almost 90% of the uncharacterized ASVs matched non- Escherichia coliforms (e.g. Citrobacter , Enterobacter , Klebsiella , Raoultella ) with the remaining sequences a best match (< 100% nucleotide sequence identity) to E. coli . These unassigned E. coli ASVs may represent further environmental E. coli diversity from hitherto uncultured and/or isolates for which no WGS/ gnd sequence data has yet been deposited in reference databases. This presents a valuable research opportunity to further explore Escherichia diversity as sequence data repositories are updated, potentially revealing new strains with ecological or public health significance. With regards to beta diversity, between sample variation demonstrated that low impact Site 1 (native forest) was the least diverse suggesting reduced fecal inputs and/or fewer types of fecal source. The E . coli and non- E. coli Escherichia species populations (Table ) from Sites 2–4 (three freshwater catchments dominated by sheep, beef and dairy cattle from multiple farm premises), clustered separately from those recovered from Site 5 (dairy farm wetland) about 340 km distance away (Fig. ).The constructed wetland at Site 5 is impacted by a small catchment area (2.6 ha), with fewer fecal sources (a single dairy herd with some wild bird fecal deposition) and during the summer months, experiences intermittent senescence with reduced flows and ponding. Nevertheless, multivariate analysis from this work provides some initial evidence that the interaction between site and sample is important for Escherichia population diversity indicating that multiple factors (catchment area, land-use, and diversity of fecal sources) significantly impact E. coli (and non- E. coli Escherichia species) freshwater diversity. Hence more frequent freshwater testing of high-risk sites associated with recreational contact using a suite of methods including standard enumeration (e.g. Colilert) and holistic metabarcoding methods incorporating gnd analysis would provide information on both human health risk as indicated by E. coli per se , and within and between-sample/site Escherichia diversity. Using the metabarcoding method implemented in this study, non- E. coli Escherichia species were identified from all sites and sample types (Fig. ), but were only rarely isolated (14 of 199, 7.04%) from freshwater samples during a larger longitudinal study across 41 sites with historically high E. coli concentrations, where twenty colonies from each individual Colilert TM enrichment sample were phylotyped . Culture-based analysis from our previous work indicated that the recovery of Escherichia species ( E. marmotae and E. ruysiae ) isolates at the individual sample level was uncommon (21 of 189, 11.1%) from the same environmental samples as in the current study . The data from the current amplicon-metabarcoding-based study, however, indicate that the non- E. coli Escherichia species-associated gSTs are more commonly associated with environmental samples than previously thought (53.6%, 90 of 168) with more diverse representation within water and biofilm samples (Fig. C and D), but were less frequently observed from feces, soil and sediment (Supplementary Table b). Unlike the culture-based study where only four colonies from each enrichment were analysed and E. marmotae was only isolated from a single avian feces sample , here 34.0% of fecal samples (16 of 47) contained reads corresponding to non- E. coli Escherichia species. Previous studies have demonstrated that E. marmotae , E. whittamii and E. ruysiae are associated with fecal samples from avian species including the feces/cloacal contents of poultry and wildfowl , , , however it is unknown, whether these Escherichia species are stable colonizers of the avian gastrointestinal tract. Similarly, E. marmotae have also been isolated from mammals, but no long-term colonisation or clinical disease has been observed , , , , suggesting that the mammalian gastrointestinal tract may not be the preferred colonisation site and thus they likely represent a low public health risk. Our previous WGS studies have identified certain Escherichia types as stably maintained in environments outside of the gastrointestinal tract of animals and birds , . Indeed, some E. coli , including gST535, similar at the core genome single nucleotide polymorphism level have been isolated from the environment and wild animal/bird feces over periods of several years , . Metabarcoding analysis of the corresponding enrichments using gnd as a target for amplification, provided unequivocal evidence in this study of the same gSTs being present at the same sites, whilst completely absent from others (Fig. ). Whether wildlife, including avian species, contribute to the high prevalence and maintenance of some E. coli within distinct environmental sites, signifying a degree of endemism, or if wildlife are a spill over host , , requires further study. Evidence using culture-based methods has also demonstrated that poikilothermic invertebrates, such as gastropods, and poikilothermic vertebrates, such as fish and frogs – are additional sources of microbes, including E. coli that could confound health-based water quality-monitoring. To overcome sensitivity issues with amplification of purified DNA from recovered samples, barcoded gnd amplicons were generated from boiled lysates of environmental sample enrichments. These enrichments improve experimental sensitivity providing increased richness where Escherichia are present in environmental matrices at very low levels , . The use of enrichments, however, may obscure the original relative abundance in samples prior to enrichment due to the differential growth rates and competitive exclusion between different Escherichia . Nevertheless, the use of culture enrichments and detection of ASVs strongly implies that the respective Escherichia species types identified are viable bacteria rather than the amplification of free DNA or DNA from non-viable Escherichia in unenriched samples . The increased diversity of gSTs from sites impacted by pastoral farming and with more diverse fecal sources, compared to Site 1 (native forest), suggests that changes in land-use types from forest to open farmland are indicative of fecal pollution represented by increased anthropogenic influences . Confirmation of this observation would require additional community analysis of low-impact sites compared with modified sites. Escherichia populations of environmental samples from the dairy farm wetland (Site 5) were more diverse than the larger catchment of Site 1 (Figs. and ), despite the wetland receiving tile drainage from only a small catchment area. The higher diversity observed in complex E. coli populations, likely indicative of recent fecal inputs, from bovine fecal contamination at Site 5 differentiates these populations, using gnd metabarcoding, from the less diverse naturalized E. coli populations from non-recent fecal contamination events , . These naturalized E. coli subpopulations present at low concentrations may represent those subtypes more adept at environmental survival and persistence in organic-rich extra-intestinal environments. Widely considered a ‘model organism’, E. coli culture collections of clinical isolates from human and animal samples have clearly failed to provide a holistic perspective on the presence and persistence of non- E. coli Escherichia species in wildlife or extraintestinal environments . The use of gnd metabarcoding is a valuable tool for differentiating environmental sources of both naturalized fecal E. coli and other Escherichia species. However, current water quality monitoring methods are inadequate for detection and characterization of non- E. coli Escherichia species whose relevance to water quality and health risk assessments remain uncertain and debated. Addressing this gap is essential as it could otherwise hinder informed decision-making regarding appropriate land-use mitigation strategies for improved water quality management and public health protection. Sample site details The sites and sampling regimen have previously been described . Briefly, samples were obtained from five field sites (Fig. , Supplementary Table ) in the North Island of New Zealand on five or six occasions between August 2017 and June 2018. Four sites were within the Manawatū River catchment: Site 1 (a second order stream site) is within Pūkaha Mount Bruce (942 hectares), an intensively managed native forest reserve with high endemic biodiversity values where trapping of introduced predatory species (mustelids, possums, hedgehogs and rats) is undertaken. Although, this site is free of native and introduced waterfowl species, opportunistic fecal sampling of cattle feces occurred in an adjacent but fenced paddock. Site 2 (Hamua Bridge, Mākākahi River, Tararua), 21.5 km (direct distance) downstream from Site 1, is a fourth order freshwater site where sheep, beef and dairy farming operations form much of the adjacent land-use. Site 3 and the close-by Site 4 (Mākirikiri and Mangatera Streams, Tararua) are south of Dannevirke (population 5200). Livestock farming (sheep, beef and dairy) also dominate the catchment land use for both these streams. Site 3 (Mākirikiri Stream) is immediately downstream from a small stand of native forest and 30 m upstream of the confluence with the Mangatera Stream. Site 4 (Mangatera Stream) is about 50 m upstream of the confluence with the Mākirikiri Stream. Site 5 is a constructed wetland in the Toenepi River catchment in the Waikato, a major New Zealand dairying region, containing 1.06 million cows; 22.6% of the national dairy cattle numbers . Here environmental samples were obtained from the inflow, middle and outlet of the wetland which intercepts and treats subsurface tile drainage waters from a small catchment (2.6 ha) with intensively grazed pasture and which has consistently shown a net export of E. coli , . Additional water samples were obtained from Site 2 (four occasions) and Site 4 (four occasions) on the same day by local environmental agency (Regional Council) staff during their routine monthly monitoring for additional water quality parameters. At Site 5, multiple water ( n = 3), sediment ( n = 3), soil ( n = 2) together with opportunistic fecal samples were collected on each of six visits. Land cover data was retrieved from the Land Cover Database v5.0 , and riverine data from the River Environment Classification 2 (REC2) , using ipyleaflet within Jupyter, and Python v3.9 scripts. Total area of sites ranged from 2.6 to approximately 16,400 ha with catchment lengths of between 0.1 and over 86 km (Table ). As a measure of total area, grassland ranged from approximately 78.0 to 96.0% for each catchment with catchments for Site 2, Site 3 and Site 4 dominated by sheep (69.0 to 74.1%). Site 1 was native forest with no ruminants present. Environmental sampling and bacterial recovery During each visit, water, sediment (not Site 2), soil, periphyton (biofilm material attached to submerged surfaces), and opportunistic fecal samples were collected. E. coli were enumerated (MPN per 100 mL water) using Colilert-18 ® /Quanti-Tray/2000 ® (IDEXX Laboratories, Inc., Maine, US). Further water samples (100 mL) were also filtered through 0.45 μm nitrocellulose filters (Millipore, Thermo Fisher) using positive pressure and the filter enriched in 10 mL EC broth (Oxoid, Hampshire, UK) incubated at 35 °C (18 to 21 h). Freshwater sediment samples (1 g) sieved through a sieve (3 mm mesh), and 1 g soil samples (from within 5 to 10 m of freshwater sample site), were enriched in 9 mL EC broth at 35 °C (18 to 21 h). Biofilm samples (approximately 100 cm 2 ) were obtained by removing a fully submerged rock from the waterway and swabbing using a sterile sponge swab (EZ-Reach Sponge Sampler, World Bioproducts, Washington, USA). At Site 5 (wetland), rocks were absent, therefore biofilm samples were obtained from submerged vegetative material. The sterile sponge swab was stomached for 1 min with 25 mL EC broth and incubated at 35 °C (18 to 21 h). Opportunistic fecal material was obtained using a sterile Amies swab (Copan Diagnostics Inc., Brescia, Italy) or sterile specimen container with scoop cap, diluted 1:100 in EC broth and incubated at 35 °C (18 to 21 h). Crude boiled DNA preparations for use as template DNA in subsequent PCRs were performed from 1 mL overnight cultures. Briefly 1 mL of culture was centrifuged at 13,000 x g and the supernatant removed. The pellet was resuspended in 0.01 M phosphate buffered saline (PBS, pH 7.4) and centrifuged again. The supernatant was removed once more, and the pellet resuspended in 1 mL molecular biology grade water. Bacteria were lysed by heating the resuspended bacteria at 100 °C for 10 min and then stored at −20 °C. Escherichia community analysis For Escherichia community analysis, the 284 bp region of the gnd gene from the crude boiled lysates was amplified using indexed 2gndF and 2gndR PCR primers , . The individually indexed PCR products were pooled to create a mixture of libraries which were run on a single Illumina MiSeq (version 2 chemistry) lane with 2 × 250 bp paired-end sequencing reads (Massey Genome Service, Massey University, Palmerston North, New Zealand). The gnd metabarcoding sequence reads were analysed using the DADA2 package v.1.16.0 and the reads remaining post-processing were used to construct an amplicon sequence variant (ASV) table. Individual ASVs were mapped against gndDb, a database of over 700 separate partial (284 bp) gnd sequence types (gSTs, synonymous with ASVs) , to provide information on the Escherichia community profile associated with individual sample enrichments. Two mock community positive control libraries were prepared with pooled equimolar amounts of separate purified PCR products with a unique combination of barcoded primers from six and seven different E. coli , respectively. The ASV table, sample metadata and the matching gST table were combined to construct a phyloseq object using the R package Phyloseq v.1.32.0 . Chao1 and Shannon alpha diversity metrics, and rarefaction analysis using ‘rarecurve’, were calculated using the vegan R package . Data visualisation was conducted in R version 3.6.2 using a range of packages including ggpubr v.0.4.0 , and ggplot2 . Principal Component Analysis (PCA) of log transformed counts (log(reads + 1)) was used to simplify the population data of each sample to identify trends, and perform permutational multivariate analysis of variance (PERMANOVA) . Linear mixed effects models were applied using the lmeTest R package where the effects of ‘Site’ were included as the explanatory variable in a standard linear regression, including ‘Visit’ as a random effect, with the prevalence of the top 20 most abundant gSTs. In a separate analysis, any association between the prevalence of each of the top 20 gSTs and ‘Sample’ was fitted after controlling for variation in ‘Visit’ and ‘Visit’ within ‘Site’. The sites and sampling regimen have previously been described . Briefly, samples were obtained from five field sites (Fig. , Supplementary Table ) in the North Island of New Zealand on five or six occasions between August 2017 and June 2018. Four sites were within the Manawatū River catchment: Site 1 (a second order stream site) is within Pūkaha Mount Bruce (942 hectares), an intensively managed native forest reserve with high endemic biodiversity values where trapping of introduced predatory species (mustelids, possums, hedgehogs and rats) is undertaken. Although, this site is free of native and introduced waterfowl species, opportunistic fecal sampling of cattle feces occurred in an adjacent but fenced paddock. Site 2 (Hamua Bridge, Mākākahi River, Tararua), 21.5 km (direct distance) downstream from Site 1, is a fourth order freshwater site where sheep, beef and dairy farming operations form much of the adjacent land-use. Site 3 and the close-by Site 4 (Mākirikiri and Mangatera Streams, Tararua) are south of Dannevirke (population 5200). Livestock farming (sheep, beef and dairy) also dominate the catchment land use for both these streams. Site 3 (Mākirikiri Stream) is immediately downstream from a small stand of native forest and 30 m upstream of the confluence with the Mangatera Stream. Site 4 (Mangatera Stream) is about 50 m upstream of the confluence with the Mākirikiri Stream. Site 5 is a constructed wetland in the Toenepi River catchment in the Waikato, a major New Zealand dairying region, containing 1.06 million cows; 22.6% of the national dairy cattle numbers . Here environmental samples were obtained from the inflow, middle and outlet of the wetland which intercepts and treats subsurface tile drainage waters from a small catchment (2.6 ha) with intensively grazed pasture and which has consistently shown a net export of E. coli , . Additional water samples were obtained from Site 2 (four occasions) and Site 4 (four occasions) on the same day by local environmental agency (Regional Council) staff during their routine monthly monitoring for additional water quality parameters. At Site 5, multiple water ( n = 3), sediment ( n = 3), soil ( n = 2) together with opportunistic fecal samples were collected on each of six visits. Land cover data was retrieved from the Land Cover Database v5.0 , and riverine data from the River Environment Classification 2 (REC2) , using ipyleaflet within Jupyter, and Python v3.9 scripts. Total area of sites ranged from 2.6 to approximately 16,400 ha with catchment lengths of between 0.1 and over 86 km (Table ). As a measure of total area, grassland ranged from approximately 78.0 to 96.0% for each catchment with catchments for Site 2, Site 3 and Site 4 dominated by sheep (69.0 to 74.1%). Site 1 was native forest with no ruminants present. During each visit, water, sediment (not Site 2), soil, periphyton (biofilm material attached to submerged surfaces), and opportunistic fecal samples were collected. E. coli were enumerated (MPN per 100 mL water) using Colilert-18 ® /Quanti-Tray/2000 ® (IDEXX Laboratories, Inc., Maine, US). Further water samples (100 mL) were also filtered through 0.45 μm nitrocellulose filters (Millipore, Thermo Fisher) using positive pressure and the filter enriched in 10 mL EC broth (Oxoid, Hampshire, UK) incubated at 35 °C (18 to 21 h). Freshwater sediment samples (1 g) sieved through a sieve (3 mm mesh), and 1 g soil samples (from within 5 to 10 m of freshwater sample site), were enriched in 9 mL EC broth at 35 °C (18 to 21 h). Biofilm samples (approximately 100 cm 2 ) were obtained by removing a fully submerged rock from the waterway and swabbing using a sterile sponge swab (EZ-Reach Sponge Sampler, World Bioproducts, Washington, USA). At Site 5 (wetland), rocks were absent, therefore biofilm samples were obtained from submerged vegetative material. The sterile sponge swab was stomached for 1 min with 25 mL EC broth and incubated at 35 °C (18 to 21 h). Opportunistic fecal material was obtained using a sterile Amies swab (Copan Diagnostics Inc., Brescia, Italy) or sterile specimen container with scoop cap, diluted 1:100 in EC broth and incubated at 35 °C (18 to 21 h). Crude boiled DNA preparations for use as template DNA in subsequent PCRs were performed from 1 mL overnight cultures. Briefly 1 mL of culture was centrifuged at 13,000 x g and the supernatant removed. The pellet was resuspended in 0.01 M phosphate buffered saline (PBS, pH 7.4) and centrifuged again. The supernatant was removed once more, and the pellet resuspended in 1 mL molecular biology grade water. Bacteria were lysed by heating the resuspended bacteria at 100 °C for 10 min and then stored at −20 °C. community analysis For Escherichia community analysis, the 284 bp region of the gnd gene from the crude boiled lysates was amplified using indexed 2gndF and 2gndR PCR primers , . The individually indexed PCR products were pooled to create a mixture of libraries which were run on a single Illumina MiSeq (version 2 chemistry) lane with 2 × 250 bp paired-end sequencing reads (Massey Genome Service, Massey University, Palmerston North, New Zealand). The gnd metabarcoding sequence reads were analysed using the DADA2 package v.1.16.0 and the reads remaining post-processing were used to construct an amplicon sequence variant (ASV) table. Individual ASVs were mapped against gndDb, a database of over 700 separate partial (284 bp) gnd sequence types (gSTs, synonymous with ASVs) , to provide information on the Escherichia community profile associated with individual sample enrichments. Two mock community positive control libraries were prepared with pooled equimolar amounts of separate purified PCR products with a unique combination of barcoded primers from six and seven different E. coli , respectively. The ASV table, sample metadata and the matching gST table were combined to construct a phyloseq object using the R package Phyloseq v.1.32.0 . Chao1 and Shannon alpha diversity metrics, and rarefaction analysis using ‘rarecurve’, were calculated using the vegan R package . Data visualisation was conducted in R version 3.6.2 using a range of packages including ggpubr v.0.4.0 , and ggplot2 . Principal Component Analysis (PCA) of log transformed counts (log(reads + 1)) was used to simplify the population data of each sample to identify trends, and perform permutational multivariate analysis of variance (PERMANOVA) . Linear mixed effects models were applied using the lmeTest R package where the effects of ‘Site’ were included as the explanatory variable in a standard linear regression, including ‘Visit’ as a random effect, with the prevalence of the top 20 most abundant gSTs. In a separate analysis, any association between the prevalence of each of the top 20 gSTs and ‘Sample’ was fitted after controlling for variation in ‘Visit’ and ‘Visit’ within ‘Site’. Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 |
Primary robot-assisted laparoscopic partial nephrectomy for hemorrhage secondary to angiomyolipoma: a retrospective study from a large tertiary hospital in China | 4bc93fb6-d755-415a-9a17-f719fd1e28ee | 11439052 | Robotic Surgical Procedures[mh] | Renal angiomyolipoma (RAML)is a benign tri-phasic soft tissue tumor composed of mature adipose tissue, branched capillary-like vessels, and smooth muscles . The incidence of RAML is approximately 0.13%, slightly higher than renal cell carcinomas (0.11%) . Hemorrhage secondary to angiomyolipoma(HSA)is one of the most severe complications of RAML. According to clinical Statistics, up to 50% of patients whose tumors are larger than 40 mm can develop HSA, leading to 33% of patients with HSA developing hypovolemic shock , , seriously threatening the life and safety of patients. Since Gettman et al. reported it in 2004, Robot-assisted laparoscopic partial nephrectomy(RALPN)has been widely used in treating kidney masses and has become a standard treatment – . Primary RALPN excises RAML and clears perirenal hematoma completely, reduces the incidence of infection, dramatically reduces the probability of secondary bleeding, and remains renal function normal both immediately, postoperatively, and one year after the surgery, which is a potential treatment method for HSA. However, due to the lack of similar studies, the safety and efficacy of primary RALPN for HSA cannot be fully proved. Herein, we conducted this retrospective study and investigated the safety and hemostatic efficacy of primary RALPN for HSA by assessing fourteen patients’ preoperative conditions and postoperative outcomes.
Patients From July 2017 to October 2023, fourteen patients were referred for primary RALPN due to HSA in the Department of Urology of Zhejiang Provincial People’s Hospital. A general examination cleared all patients of a history of coagulation disorders, infection, or cardiac problems. Serum creatinine, hemoglobin, and glomerular filtration rate were collected through blood tests. The blood routine examination revealed decreased hemoglobin in all patients, and CT indicated retroperitoneal hematoma. In patients with shock, immediate intravenous rehydration with a balanced salt solution was used to correct shock. CT was used to delineate the size and situation of the renal lesion regarding HSA. The patients later underwent definitive surgical extirpation using a primary RALPN within a week following Hemorrhage. The perioperative status of primary RALPN was analyzed to evaluate the efficacy and safety of Primary RALPN in treating Hemorrhage secondary to angiomyolipoma. The study was approved by the ethics committees in Zhejiang Provincial People’s Hospital (Approval No. 2021QT082 ) and all experiments were performed in accordance with relevant guidelines and regulations. Surgery procedures RALPNs were performed by three highly experienced robotic surgeons, and more than 500 robotic surgeries have been executed so far. The specific surgical procedure is as follows: Positioning and Trocar placement: After successful anesthesia, the patients were secured in a flank position with a lateral tilt. A 5-mm trocar was used for abdominal paracentesis to establish CO2 pneumoperitoneum of 14 mm Hg lateral to the rectus muscle, at the level of the umbilicus, and then a robotic camera port (Da Vinci robot SI:12 mm; Da Vinci robot XI:8 mm) was placed at the same position; a 30° lens was introduced to place the other ports. Three 8-mm ports were inserted 10 cm below the xiphoid at the ventral midline, below the 12th rib costochondral margin at the mid-clavicle, and 8 cm away from the camera port at the anterior axillary line, respectively. A 12-mm assisted port was placed 3 cm above the umbilicus. If the tumor was located on the right side, a 5 mm Trocar was inserted under the xiphoid process to lift the liver for exposure. Dissection of the angiomyolipoma and renal artery: The camera port was placed to provide excellent vision of the extent of the retroperitoneal hematoma. For right-sided neoplasms, the liver was retracted anteriorly and cephalad; for left-sided neoplasms, the spleen and pancreas were reflected medially. The intestinal loop was pushed open, and the paracolic sulci posterior wall peritoneum was opened to the colon. Push the colon away to the ventromedial side to expose the Gerota fascia. The renal hilum was thoroughly dissected anteriorly and posteriorly, and the renal artery was isolated at the level of the renal hilum. Then, the Gerota fascia was opened, and the perirenal hematoma was revealed and evacuated by scraping and suctioning with a suction device. The angiomyolipoma of the kidney was fully exposed. Resection of the angiomyolipoma and suture of the wound: The laparoscopic bulldog clamp clamped the renal artery, and the warm ischemia time (WIT) was noted. The angiomyolipoma was scraped away with Maryland bipolar and monopolar scissors. The exposed blood vessels on the wound were cauterized for hemostasis using Maryland bipolar. The repair was performed with 4 − 0 Vicryl sutures if the collection system was exposed. After that, 2 − 0 and 0 absorbable barbed sutures were performed for inner and outer running stitches, respectively. Subsequently, the vascular bulldog clamp was removed, and the timing of WIT was completed. Finally, the Da Vinci device was undocked after the Endocatch bag removed the resected angiomyolipoma. The perioperative CT and surgical procedures are depicted in Fig. . Data collection and follow-up As of October 2023, all medical records were reviewed. Information such as operation time, warm ischemia time, mean blood loss, intraoperative blood transfusion volume, postoperative pathology, intraoperative complications, postoperative complication score, and length of hospital stay were collected. The postoperative complication score was based on the Clavien-Dindo scale of surgical complications. Assessment of renal function included creatinine (Cr, umol/L) and glomerular filtration rate (ECT-GFR, ml/min) measurements before and after surgery. Failure of treatment was defined as significant complications associated with surgery or Secondary bleeding. Patients generally received outpatient urological follow-up on the first, third, and sixth months and at 6-month intervals after that or sooner when clinically indicated.
From July 2017 to October 2023, fourteen patients were referred for primary RALPN due to HSA in the Department of Urology of Zhejiang Provincial People’s Hospital. A general examination cleared all patients of a history of coagulation disorders, infection, or cardiac problems. Serum creatinine, hemoglobin, and glomerular filtration rate were collected through blood tests. The blood routine examination revealed decreased hemoglobin in all patients, and CT indicated retroperitoneal hematoma. In patients with shock, immediate intravenous rehydration with a balanced salt solution was used to correct shock. CT was used to delineate the size and situation of the renal lesion regarding HSA. The patients later underwent definitive surgical extirpation using a primary RALPN within a week following Hemorrhage. The perioperative status of primary RALPN was analyzed to evaluate the efficacy and safety of Primary RALPN in treating Hemorrhage secondary to angiomyolipoma. The study was approved by the ethics committees in Zhejiang Provincial People’s Hospital (Approval No. 2021QT082 ) and all experiments were performed in accordance with relevant guidelines and regulations.
RALPNs were performed by three highly experienced robotic surgeons, and more than 500 robotic surgeries have been executed so far. The specific surgical procedure is as follows: Positioning and Trocar placement: After successful anesthesia, the patients were secured in a flank position with a lateral tilt. A 5-mm trocar was used for abdominal paracentesis to establish CO2 pneumoperitoneum of 14 mm Hg lateral to the rectus muscle, at the level of the umbilicus, and then a robotic camera port (Da Vinci robot SI:12 mm; Da Vinci robot XI:8 mm) was placed at the same position; a 30° lens was introduced to place the other ports. Three 8-mm ports were inserted 10 cm below the xiphoid at the ventral midline, below the 12th rib costochondral margin at the mid-clavicle, and 8 cm away from the camera port at the anterior axillary line, respectively. A 12-mm assisted port was placed 3 cm above the umbilicus. If the tumor was located on the right side, a 5 mm Trocar was inserted under the xiphoid process to lift the liver for exposure. Dissection of the angiomyolipoma and renal artery: The camera port was placed to provide excellent vision of the extent of the retroperitoneal hematoma. For right-sided neoplasms, the liver was retracted anteriorly and cephalad; for left-sided neoplasms, the spleen and pancreas were reflected medially. The intestinal loop was pushed open, and the paracolic sulci posterior wall peritoneum was opened to the colon. Push the colon away to the ventromedial side to expose the Gerota fascia. The renal hilum was thoroughly dissected anteriorly and posteriorly, and the renal artery was isolated at the level of the renal hilum. Then, the Gerota fascia was opened, and the perirenal hematoma was revealed and evacuated by scraping and suctioning with a suction device. The angiomyolipoma of the kidney was fully exposed. Resection of the angiomyolipoma and suture of the wound: The laparoscopic bulldog clamp clamped the renal artery, and the warm ischemia time (WIT) was noted. The angiomyolipoma was scraped away with Maryland bipolar and monopolar scissors. The exposed blood vessels on the wound were cauterized for hemostasis using Maryland bipolar. The repair was performed with 4 − 0 Vicryl sutures if the collection system was exposed. After that, 2 − 0 and 0 absorbable barbed sutures were performed for inner and outer running stitches, respectively. Subsequently, the vascular bulldog clamp was removed, and the timing of WIT was completed. Finally, the Da Vinci device was undocked after the Endocatch bag removed the resected angiomyolipoma. The perioperative CT and surgical procedures are depicted in Fig. .
As of October 2023, all medical records were reviewed. Information such as operation time, warm ischemia time, mean blood loss, intraoperative blood transfusion volume, postoperative pathology, intraoperative complications, postoperative complication score, and length of hospital stay were collected. The postoperative complication score was based on the Clavien-Dindo scale of surgical complications. Assessment of renal function included creatinine (Cr, umol/L) and glomerular filtration rate (ECT-GFR, ml/min) measurements before and after surgery. Failure of treatment was defined as significant complications associated with surgery or Secondary bleeding. Patients generally received outpatient urological follow-up on the first, third, and sixth months and at 6-month intervals after that or sooner when clinically indicated.
During the study period, 132 patients diagnosed with hamartoma in our hospital were analyzed retrospectively, 118 patients who did not meet the inclusion criteria were excluded, and 14 were included. Patient demographics are described in Table . The median age of patients was 43.5 years, and the median initial tumor size was 80 mm (range: 57–145 mm). The median R.E.N.A.L. was 9(range:7–11). All patients were treated with Primary RALPN, and all operations were completed. None of the patients in this study had Tuberous sclerosis(TS). The median operative time was 150 min, and the median intraoperative blood loss was 300 ml. Two patients received blood transfusions: one received 880 ml plasma, and the second received 220 ml plasma and 2U red blood cells(Table ). The median warm ischemia time was 24.5 min. The median length of postoperative hospital stay was 6.5 days. Pathological results showed thirteen cases of typical AML and one epithelioid subtype. After surgery, two patients (14.3%) observed minor adverse events within 30 days of RALPN and were assessed as Clavien-Dindo (I) One patient (7.1%) developed chylous drainage after surgery, improved with fasting and parenteral nutrition, and was evaluated as Clavien-Dindo (II) One patient (7.1%) developed Pleural effusion, improved with thoracocentesis, and was assessed as Clavien-Dindo III.Creatinine and eGFR before and after surgery were available in seven patients. The other seven data sets were missing due to other reasons. Compared with the preoperative data, the median postoperative creatinine(ΔCr), glomerular filtration rate(ΔeGFR), and postoperative drainage time rate were decreased by -4.5(-19.50-10.10) umol/L, 6.81 (-37.74 ~ 29.38) ml/ (min*1.73 m^2), and 5 (3–14) days, respectively. No local recurrence, related death, urinoma, or urinary fistula was observed during a median follow-up period of 3.2(1.0-6.2) years.
Hemorrhage secondary to angiomyolipoma(HSA) is one of the most severe complications of RAML, characterized by acute episodes of spontaneous, non-traumatic renal Hemorrhage into the subcapsular and perirenal spaces. Clinical symptoms are summarized as Lenk’s triad, including acute flank pain, abdominal tenderness, and signs of internal bleeding, which can be fatal if not treated promptly and aggressively . The literature showed that 83% of cases presented with acute onset of flank pain, and 19% had hematuria . However, in our study, ten patients (71.4%) were admitted to the hospital for initial symptoms of flank pain and four (28.6%) for abdominal pain, but none had hematuria. We hypothesized that Hemorrhage due to Hamartoma rupture may enter the retroperitoneal space without penetrating the collecting system, and hematuria was avoided. For patients with HSA, the first step is to treat HSA as safely and effectively as possible without affecting the patient’s life, and the second is to preserve the renal function as much as possible. Selective arterial embolization (SAE) is the clinic’s most common treatment option because it can effectively prevent and treat HSA and induce tumor shrinkage simultaneously . However, SAE does not address the root cause of renal hamartoma and has severe hidden dangers. First, 35.9% of embolizations develop Self-limiting post-embolization syndrome, with further morbidity developed in 6.9% . Second, up to 31% of post-SAE patients may require secondary treatment due to revascularization of abnormal blood vessels or regeneration of new pathological blood vessels after SAE . Furthermore, SAE often expands embolization to entirely stop bleeding, especially for relatively complex renal AML, resulting in excessive nephron loss or even more renal dysfunction. Considering the complications of SAE, primary SAE treatment followed by second-stage surgery is recommended. However, it also raises new questions: First, although preoperative embolization controls bleeding, it also inevitably damages renal function, especially for the relatively complex HSA that requires extensive embolization. Second, when SAE is performed, if the patient is not actively bleeding, the procedure may not be able to find significant blood vessels for embolization. Furthermore, deciding how long the patient will be suitable for the surgery after embolization is difficult. If the time between embolization and surgery is too short, the patient’s renal function may be further damaged, affecting the patient’s prognosis. If the time between embolization and surgery is long, as time passes after the embolization, dense tissue adhesion after bleeding can make separation between the kidney and the tumor difficult and may result in partial kidney removal in intraoperative. Hernandez et al. described 4 cases of transperitoneal laparoscopic nephrectomy performed 3 or 6 months after a spontaneous hemorrhage. The authors highlight the difficulty of dissection due to the fibrosis secondary to the resorption of the hematoma. In addition, the risk of infection increases over time. As a result, primary SAE treatment and second-stage surgery are challenging to implement, and preoperative embolization does not appear to be the best choice for some patients. The timeframe between hemorrhage and intervention is vitally essential for primary surgery patients.In our study, surgery is usually arranged on the next day for patients who see a doctor on a working day, but for some well-prepared patients, the operation can be completed within 24 h. For patients who visit on a rest day, the surgery will be scheduled for the next working day. The scheduling of these surgical plans depends mainly on the following points: First, the urology department of our hospital has two robotic devices(Da Vinci robot SI and Da Vinci robot XI) and enough specialist nurses, which provides a sufficient practical basis for surgery. Second, Antishock therapy also requires sufficient time to keep the patient’s vital signs stable. Futhermore, for some patients whose vital signs are stable, it is desirable to observe them for a period of time before performing surgery. Some patients choose to seek treatment outside the hospital or watch and wait after bleeding, which also leads to a longer time between bleeding and surgery. But no matter what, the time interval between bleeding and surgery should be limited to less than a week to prevent the effects of hematoma adhesion. Surgical emergency treatment of RAML rupture and hemorrhage has been used in clinical practice for many years, and the efficacy is safe and reliable. However, the laparoscopic technique has not been widely accepted because laparoscopic treatment for HSA is still challenging and time-consuming under the constraint of warm ischemia time (WIT). First, for patients with significant tumor rupture accompanied by perirenal or retroperitoneal hematoma, the retroperitoneal space was relatively narrow, and the hematoma itself occupied part of it, thus affecting the surgical operation. Second, hematoma due to tumor rupture hindered the correct surgical field vision. Furthermore, Laparoscopy is challenging for emergency surgery because of its high technical level and experience requirements. Compared with the laparoscopic technique, the flexible “wristed” instruments and 3-D optical imaging of the Da Vinci surgical platform provide precise suturing, spot small or hidden bleeding spots, and stop the bleeding by switching the electrocoagulation system in time. In this study, we adopted a 4-arm configuration, which has the following benefits: Firstly, with four arms, one arm can be dedicated to the camera(Da Vinci robot SI:12 mm; Da Vinci robot XI:8 mm), providing continuous and high-definition visualization, while the other three can handle different surgical instruments(The left hand: Maryland bipolar; The right hand: Monopolar scissors; Fourth arm: Non-destructive gripper). This separation of functions allows for more precise, coordinated movements and complex maneuvers during the procedure, such as tumor resection and renal reconstruction. Secondly, surgeons do not need to switch instruments as frequently, which reduces the time spent managing tools, minimizes disruptions to the surgical workflow, and lessens surgeon fatigue. This contributes to better performance and decision-making during lengthy procedures. Finally, the 4-arm configuration allows for better placement and angling of instruments within the confined space of the abdominal cavity, leading to improved access to the tumor and surrounding tissues. Many studies have shown that robot-assisted surgery has advantages for tumor mass treatments, such as a shorter learning curve, less intraoperative bleeding, shorter WIT, and faster postoperative recovery than traditional Laparoscopic management , . Accordingly, robot-assisted laparoscopic surgery may overcome challenges such as narrow operating space and offer an alternative surgical approach for treating HSA. Then, in order to further the pursuit of surgical results, some researchers have explored innovative imaging technologies, such as highly accurate three-dimensional virtual models (3DVMs) , . Following their studies, patients who used RAPN with 3DVM assistance had a lower incidence of global ischemia and enhanced 12-month functional preservation. Our views are similar. For suitable patients, preoperative 3DVMs may help reduce intraoperative bleeding and reduce surgical difficulty, thereby improving patient prognosis and enhancing postoperative functional preservation, but further clinical studies are needed. WIT is another crucial limiting factor for surgery. During the operation, the renal artery should be clipped by a laparoscopic bulldog clamp in time to prevent continuing bleeding, and WIT was noted. Studies have reported that every additional minute of WIT is associated with a 6% increase in the risk of new-onset stage IV chronic kidney disease during the follow-up . In our studies, angiomyolipoma was resected, and median warm ischemia time was 24.5 min, which was controllable. Compared with the report of Ploumidis A used RALPN for HSA, our WIT is more prolonged, probably due to the following reasons. First, the continuous bleeding of angiomyolipoma hindered the correct surgical field vision. Second, the nearness of the angiomyolipoma’s deepest portion to the collecting system increases the risk of urinary leakage and urinary exosmosis, prolonging operation time. In addition, inflammatory lymph nodes surrounding the vascular structures made the dissection more difficult. Nevertheless, the WIT of primary RALPN is controllable, effectively protecting kidney function, as shown by the postoperative renal function being within the normal range. We systematically searched the Medline electronic database and summarized the case series on treatments for HSA. According to our search results, only a few reports used Primary RALPN to treat HSA , , while most other studies used laparoscopic management or open surgery , , , as listed in Table . Most studies have a sample size of less than 4, and we are the only known study that used RALPN for HSA with a sample size of more than 10. In the studies by Pal, A.K., not all patients used robot-assisted laparoscopic management but multiple surgical procedures. Compared with Pal A.K.’s study, our operating time and WIT are shorter than theirs, probably due to the smaller size of our tumors. A larger tumor volume might mean more intraoperative blood loss and longer operative time to resect the tumor and suture. In the complication of our study, two patients were assessed as Clavien-Dindo I for low calcium and Low-grade fever; two patients were assessed as Clavien-Dindo II and III because of chylous drainage and Pleural effusion, respectively. Similar to our results, In the studies by Pal A.K. and Ploumidis A., the patients used RA and did not develop significant postoperative complications requiring acute surgical reintervention. Overall, Primary RALPN excises RAML, clears perirenal hematoma completely, reduces the absorption of necrotic material, and allows the effective reconstruction of the renal collecting duct and cortex in a complex abdominal environment in time. Compared with laparoscopic management, primary RALPN has a relatively short operating time, less medium blood loss amount, and less incidence of blood transfusion, and is a safe technique associated with a low complication rate. However, Several limitations should be taken into account when interpreting our findings. Firstly, this was a retrospective study with a small sample size, even though this is the first study with the largest sample size in the literature. Further study with sufficient sample size is needed to confirm our findings. Secondly, the surgery regimen was at the discretion of the urologist without a prospective protocol. In addition, the single study center could not fully prove the effectiveness of the primary RALPN, for some factors may affect the outcomes, such as urologists’ competencies and supportive contexts. Future studies require large samples of prospective randomized trials in multiple centers with long-term follow-up schemes.
Primary RALPN was safe and effective in treating Hemorrhage secondary to angiomyolipoma, which may be considered an alternative to selective renal artery embolization. Taking into account more effective evacuation of hematoma, and more complete angiomyolipoma resection, primary RALPN may be a better treatment for patients with HSA.
|
210e7faf-4ddb-49ee-931c-eaaf9092e398 | 11936199 | Cytology[mh] | Cancer is characterized by uncontrolled growth and proliferation, leading to the formation of malignant tumors that can invade adjacent tissues. Conventional chemotherapy lacks specificity, causing severe damage to both tumor and normal cells. This necessitates the development of more effective delivery systems to reduce the side effects and increase the efficiency of the antitumor drugs . The use of nanotechnology in medicine, particularly for drug delivery, has rapidly expanded in recent years . Nanoparticles have been used to improve the delivery of poorly water-soluble drugs by enabling faster dissolution in the bloodstream . Cell culture systems have significantly impacted biomedical research by reducing reliance on animal models, leading to important discoveries in drug development and exploring different disease mechanisms. Historically, two-dimensional (2D) cell cultures have been widely used for several decades and are considered a cornerstone of in vitro research . However, growing cells as monolayers on plastic surfaces fail to accurately replicate the in vivo conditions, especially the biological activity of therapeutic agents and cellular heterogeneity in oncological studies. This obstacle is related to the inability of the 2D cultures to replicate the three-dimensional (3D) microenvironments of tumor tissues, including biomedical signaling, structural organization, and cell-cell and cell-matrix interactions . Researchers overcome these limitations by developing three-dimensional culture (3D) cell cultures, commonly referred to as 3D models, which exhibit greater physiological relevance and have led to advancements in our understanding of various biological mechanisms in healthy and diseased tissues . These 3D cell culture systems can recapitulate key features of solid tumors that mimic the in vivo environment, including tumor morphology, gradients of chemical and biological factors, and dynamic and reciprocal interactions between tumors and their stroma. Consequently, offering insights into understanding cell behavior, drug penetration, and therapeutic responses. Nanotechnology and its applications, like liposomal-based nanoparticles, have provided more effective systems for dealing with the native structure of the tumor. Nanoparticles can interact with the tumor mass and its surrounding vasculature via the enhanced permeability and retention effect (EPR). Their biocompatible and biodegradable structures can reduce toxicity and improve selectivity in therapeutic applications . Doxorubicin (Dox), an anthracycline antibiotic, was first isolated from the pigment-producing Streptomyces bacteria early in the 1960s. Currently, it is clinically used to treat solid tumors of the breast, bile ducts, prostate, uterus, ovary, stomach, and liver, as well as hematological malignancies such as lymphoblastic leukemia . Integrating 3D cell culture systems with nanotechnology-based drug delivery systems can contribute to bridging the gap between in vitro models and in vivo conditions that ultimately advance the translational potential of cancer therapies. Therefore, this study aims to investigate the performance of liposomal drug delivery systems in both 2D and 3D cell culture models. Additionally, the study seeks to validate the 3D model as a physiologically relevant platform that better replicates the tumor microenvironment compared to traditional 2D assays, offering a more accurate tool for preclinical drug testing.
2.1. Cell lines and reagents Human glioblastoma cell line (U-87MG) (ATCC HTB-14) and Human dermal fibroblast cell line (HDF) (ATCC PCS-201-012) were obtained from ATCC. Collagen type I and Doxorubicin were purchased from Sigma-Aldrich (Poole, UK). Doxil ® 20mg\10ml was purchased from Johnson & Johnson (USA). 2.2. Cell culture condition U-87-MG and HDF were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 100 mg/mL streptomycin, 100 U/mL penicillin, 2 mM L-glutamine, and 10% fetal bovine serum. The cell lines were incubated at standard culture conditions at 37°C in a humidified atmosphere containing 5% CO₂. Cells were subcultured at a 70% confluency. 2.3. Spheroid formation Tumor spheroids were generated using the hanging drops technique by seeding U87-MG and HDF cells at a ratio of 70:30 (1 × 10 4 cells/drop) [ – ]. The drops were loaded into the lid of Petri dishes and incubated at 37 °C, 5% CO 2 for 3 days. To maintain humidity, 5 mL of Phosphate Buffered Saline (PBS) was added to the bottom of the dish. After incubation, the formation of spheroids was monitored and prepared for loading into agarose for further cellular experiments. 2.4. 3D cellular uptake studies 2.4.1. Cellular uptake of 3D spheroids by flow cytometry. Agarose gel (0.125%) was prepared by weighing an adequate amount of agarose powder and dissolving it in an appropriate volume of cell culture medium. Wells of a 24-well culture plate was loaded with 250 µL of agarose solution as a bottom layer. After agarose solidification, spheroids from hanging drops were transferred into each well. Another 250 µL of agarose solution was added to each well and set to solidify. Finally, 500 µL of medium was added to each well. Spheroids were treated with free Doxorubicin (FD) (1 µM and 2 µM) and Doxil ® (1 µM and 2 µM Doxorubicin equivalent) and incubated for 4 h and 24 h. After incubation, 13 Spheroids were extracted from each well. Spheroids were disaggregated using 0.5% trypsin. After trypsinization, cells were centrifuged at 300xg for 5 min, and the pellet was resuspended in 200 µL PBS. Then, 1 × 10 4 events were counted using FACS Canto II and analyzed with BD FACSDiva™ software version 8.0 (BD, USA). The excitation\emission wavelengths (λ 470/λ 595 nm) for Doxorubicin were used to measure fluorescence intensity (FI). The previous experiment was performed in triplicates. 2.4.2. Cellular uptake of 3D spheroids by confocal microscopy. Spheroids were incubated and treated in an agarose gel medium, as previously mentioned in Section . The spheroid was then transferred into a shape-untreated well plate. 4% formalin was added to each spheroid for fixation. (1:1500) DAPI\PBS of DAPI stain was added for each well and incubated in the dark. After 10 min, spheroids were ready to be replaced on a clean glass slide. A small amount of Dako mounting medium as a drop was added and incubated for 15 min before the Z-stacking examination by Zeiss LSM780 confocal microscope system (Carl Zeiss AG, Germany). 2.5. 2D cellular uptake studies 2.5.1. 2D cellular uptake by flow cytometry. In each well of 12-well plate, 1.25 × 10 5 U87-MG cells were seeded and incubated for 24h. 1mL of (1 µM and 1 µM) FD and Doxil ® was added and incubated for 4 and 24h. All the previous experiments were performed in a triplicate. 0.5% trypsin was used to harvest and collect the cells. The excitation\emission wavelength of Dox was detected using flow cytometry to measure FI as described in Section . 2.5.2. 2D cellular uptake by confocal microscopy. In this experiment, U87-MG (1.2 × 10 5 ) cells were seeded into individual wells of 6-well plates containing sterile coverslips and incubated for 24 h. Next, 1 mL of FD and Doxil ® , at concentrations of 1 µM and 2 µM, was added to each well and incubated for 4 and 24 h. This procedure was performed in triplicate. To fix U87-MG cells, 4% formalin was added, and then each well was treated with DAPI stain (diluted 1:1500 in PBS) and incubated for 15 min at room temperature in the dark. After fixation, the coverslips were placed face down on clean glass slides with a drop of Dako mounting medium. Finally, images were captured using a Zeiss Plan Apo 63x/1.40 oil lens with 405 nm and 595 nm lasers to excite the DAPI-stained nuclei and Dox. Fluorescent emission was detected at 495 nm for DAPI and 470 nm for Dox, respectively. 2.6. MTT assay A 96-well culture plate was used to seed (8 × 10^3) U87-MG cells, which were incubated for 24 h. The cells were then subjected to the following treatments: untreated, FD at concentrations ranging from 62.5 to 2000 nM, and Doxil ® at concentrations ranging from 62.5 to 2000 nM. After 72h incubation, the treatment was removed, followed by adding 100μL of culture medium and 15 μL of 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyltetrazolium bromide (MTT) solution. After 3h of incubation, the medium was removed, and the cells were mixed with 50μL of dimethyl sulphoxide (DMSO) to dissolve the formazan. The absorbance was measured at a wavelength of 570 nm using a microplate reader (Synergy™ HTX by BioTek Instruments Inc, USA), and the IC 50 values were subsequently determined. All experiments were performed in triplicate. 2.7. 3D invasion assay Collagen medium was prepared using a type I collagen solution derived from bovine skin [ , , ]. The chambers of an 8-chamber cover glass were loaded with collagen medium to generate the first layer (200 µL) and allowed to polymerize. After polymerization, spheroids were placed into each chamber. An additional layer of collagen medium (200 µL) was added to each well as the second layer and allowed to polymerize. Finally, media and drugs were added as follows: untreated spheroids (control), spheroids with free Dox (FD) at concentrations of 62.5 nM and 125 nM, and spheroids with Doxil ® at concentrations of 250 nM, 400 nM, and 2 µM. All experiments were performed in triplicate, incubated for 4 days, and images were captured daily. Data were analyzed using ImageJ software. Spheroids were observed using an inverted light microscopy at a 10 × objective lens. Images were captured daily using Nikon Eclipse camera (TE2000-4, Panasonic Lumix DMC-GF6). 2.8. Wound healing assay In a 6-well culture plate, U87-MG (250 × 10 3 ) cells were seeded into each well and incubated for 24 h. After incubation, a wound (scratch) was inflicted using a 200 μL pipette tip on the cell monolayer, followed by a washing step with PBS to remove cell debris. Next, the media and drugs were replaced to treat the cells as untreated cells (control), cells with FD at concentrations of 62.5nM and 125 nM, and cells with Doxil ® at concentrations of 250 nM and 400nM. The cells were then incubated for 3 days, with all treatments performed in triplicate. Images of the wounds were taken at 0, 3, 6, 9, and 24h. Wounds were observed using inverted light microscopy at a 10 × objective lens. Images were captured daily using a Nikon Eclipse camera (TE2000-4, Panasonic Lumix DMC-GF6), and the wound areas were analyzed using ImageJ software. 2.9. Statistical analysis A two-way ANOVA test was used to analyze the data results; the experiment was performed in triplicates (n = 3). The P values significance were determined as follows: * P ≤ 0.05, ** P ≤ 0. 01, *** P ≤ 0.001 and **** P ≤ 0.0001.
Human glioblastoma cell line (U-87MG) (ATCC HTB-14) and Human dermal fibroblast cell line (HDF) (ATCC PCS-201-012) were obtained from ATCC. Collagen type I and Doxorubicin were purchased from Sigma-Aldrich (Poole, UK). Doxil ® 20mg\10ml was purchased from Johnson & Johnson (USA).
U-87-MG and HDF were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 100 mg/mL streptomycin, 100 U/mL penicillin, 2 mM L-glutamine, and 10% fetal bovine serum. The cell lines were incubated at standard culture conditions at 37°C in a humidified atmosphere containing 5% CO₂. Cells were subcultured at a 70% confluency.
Tumor spheroids were generated using the hanging drops technique by seeding U87-MG and HDF cells at a ratio of 70:30 (1 × 10 4 cells/drop) [ – ]. The drops were loaded into the lid of Petri dishes and incubated at 37 °C, 5% CO 2 for 3 days. To maintain humidity, 5 mL of Phosphate Buffered Saline (PBS) was added to the bottom of the dish. After incubation, the formation of spheroids was monitored and prepared for loading into agarose for further cellular experiments.
2.4.1. Cellular uptake of 3D spheroids by flow cytometry. Agarose gel (0.125%) was prepared by weighing an adequate amount of agarose powder and dissolving it in an appropriate volume of cell culture medium. Wells of a 24-well culture plate was loaded with 250 µL of agarose solution as a bottom layer. After agarose solidification, spheroids from hanging drops were transferred into each well. Another 250 µL of agarose solution was added to each well and set to solidify. Finally, 500 µL of medium was added to each well. Spheroids were treated with free Doxorubicin (FD) (1 µM and 2 µM) and Doxil ® (1 µM and 2 µM Doxorubicin equivalent) and incubated for 4 h and 24 h. After incubation, 13 Spheroids were extracted from each well. Spheroids were disaggregated using 0.5% trypsin. After trypsinization, cells were centrifuged at 300xg for 5 min, and the pellet was resuspended in 200 µL PBS. Then, 1 × 10 4 events were counted using FACS Canto II and analyzed with BD FACSDiva™ software version 8.0 (BD, USA). The excitation\emission wavelengths (λ 470/λ 595 nm) for Doxorubicin were used to measure fluorescence intensity (FI). The previous experiment was performed in triplicates. 2.4.2. Cellular uptake of 3D spheroids by confocal microscopy. Spheroids were incubated and treated in an agarose gel medium, as previously mentioned in Section . The spheroid was then transferred into a shape-untreated well plate. 4% formalin was added to each spheroid for fixation. (1:1500) DAPI\PBS of DAPI stain was added for each well and incubated in the dark. After 10 min, spheroids were ready to be replaced on a clean glass slide. A small amount of Dako mounting medium as a drop was added and incubated for 15 min before the Z-stacking examination by Zeiss LSM780 confocal microscope system (Carl Zeiss AG, Germany).
Agarose gel (0.125%) was prepared by weighing an adequate amount of agarose powder and dissolving it in an appropriate volume of cell culture medium. Wells of a 24-well culture plate was loaded with 250 µL of agarose solution as a bottom layer. After agarose solidification, spheroids from hanging drops were transferred into each well. Another 250 µL of agarose solution was added to each well and set to solidify. Finally, 500 µL of medium was added to each well. Spheroids were treated with free Doxorubicin (FD) (1 µM and 2 µM) and Doxil ® (1 µM and 2 µM Doxorubicin equivalent) and incubated for 4 h and 24 h. After incubation, 13 Spheroids were extracted from each well. Spheroids were disaggregated using 0.5% trypsin. After trypsinization, cells were centrifuged at 300xg for 5 min, and the pellet was resuspended in 200 µL PBS. Then, 1 × 10 4 events were counted using FACS Canto II and analyzed with BD FACSDiva™ software version 8.0 (BD, USA). The excitation\emission wavelengths (λ 470/λ 595 nm) for Doxorubicin were used to measure fluorescence intensity (FI). The previous experiment was performed in triplicates.
Spheroids were incubated and treated in an agarose gel medium, as previously mentioned in Section . The spheroid was then transferred into a shape-untreated well plate. 4% formalin was added to each spheroid for fixation. (1:1500) DAPI\PBS of DAPI stain was added for each well and incubated in the dark. After 10 min, spheroids were ready to be replaced on a clean glass slide. A small amount of Dako mounting medium as a drop was added and incubated for 15 min before the Z-stacking examination by Zeiss LSM780 confocal microscope system (Carl Zeiss AG, Germany).
2.5.1. 2D cellular uptake by flow cytometry. In each well of 12-well plate, 1.25 × 10 5 U87-MG cells were seeded and incubated for 24h. 1mL of (1 µM and 1 µM) FD and Doxil ® was added and incubated for 4 and 24h. All the previous experiments were performed in a triplicate. 0.5% trypsin was used to harvest and collect the cells. The excitation\emission wavelength of Dox was detected using flow cytometry to measure FI as described in Section . 2.5.2. 2D cellular uptake by confocal microscopy. In this experiment, U87-MG (1.2 × 10 5 ) cells were seeded into individual wells of 6-well plates containing sterile coverslips and incubated for 24 h. Next, 1 mL of FD and Doxil ® , at concentrations of 1 µM and 2 µM, was added to each well and incubated for 4 and 24 h. This procedure was performed in triplicate. To fix U87-MG cells, 4% formalin was added, and then each well was treated with DAPI stain (diluted 1:1500 in PBS) and incubated for 15 min at room temperature in the dark. After fixation, the coverslips were placed face down on clean glass slides with a drop of Dako mounting medium. Finally, images were captured using a Zeiss Plan Apo 63x/1.40 oil lens with 405 nm and 595 nm lasers to excite the DAPI-stained nuclei and Dox. Fluorescent emission was detected at 495 nm for DAPI and 470 nm for Dox, respectively.
In each well of 12-well plate, 1.25 × 10 5 U87-MG cells were seeded and incubated for 24h. 1mL of (1 µM and 1 µM) FD and Doxil ® was added and incubated for 4 and 24h. All the previous experiments were performed in a triplicate. 0.5% trypsin was used to harvest and collect the cells. The excitation\emission wavelength of Dox was detected using flow cytometry to measure FI as described in Section .
In this experiment, U87-MG (1.2 × 10 5 ) cells were seeded into individual wells of 6-well plates containing sterile coverslips and incubated for 24 h. Next, 1 mL of FD and Doxil ® , at concentrations of 1 µM and 2 µM, was added to each well and incubated for 4 and 24 h. This procedure was performed in triplicate. To fix U87-MG cells, 4% formalin was added, and then each well was treated with DAPI stain (diluted 1:1500 in PBS) and incubated for 15 min at room temperature in the dark. After fixation, the coverslips were placed face down on clean glass slides with a drop of Dako mounting medium. Finally, images were captured using a Zeiss Plan Apo 63x/1.40 oil lens with 405 nm and 595 nm lasers to excite the DAPI-stained nuclei and Dox. Fluorescent emission was detected at 495 nm for DAPI and 470 nm for Dox, respectively.
A 96-well culture plate was used to seed (8 × 10^3) U87-MG cells, which were incubated for 24 h. The cells were then subjected to the following treatments: untreated, FD at concentrations ranging from 62.5 to 2000 nM, and Doxil ® at concentrations ranging from 62.5 to 2000 nM. After 72h incubation, the treatment was removed, followed by adding 100μL of culture medium and 15 μL of 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyltetrazolium bromide (MTT) solution. After 3h of incubation, the medium was removed, and the cells were mixed with 50μL of dimethyl sulphoxide (DMSO) to dissolve the formazan. The absorbance was measured at a wavelength of 570 nm using a microplate reader (Synergy™ HTX by BioTek Instruments Inc, USA), and the IC 50 values were subsequently determined. All experiments were performed in triplicate.
Collagen medium was prepared using a type I collagen solution derived from bovine skin [ , , ]. The chambers of an 8-chamber cover glass were loaded with collagen medium to generate the first layer (200 µL) and allowed to polymerize. After polymerization, spheroids were placed into each chamber. An additional layer of collagen medium (200 µL) was added to each well as the second layer and allowed to polymerize. Finally, media and drugs were added as follows: untreated spheroids (control), spheroids with free Dox (FD) at concentrations of 62.5 nM and 125 nM, and spheroids with Doxil ® at concentrations of 250 nM, 400 nM, and 2 µM. All experiments were performed in triplicate, incubated for 4 days, and images were captured daily. Data were analyzed using ImageJ software. Spheroids were observed using an inverted light microscopy at a 10 × objective lens. Images were captured daily using Nikon Eclipse camera (TE2000-4, Panasonic Lumix DMC-GF6).
In a 6-well culture plate, U87-MG (250 × 10 3 ) cells were seeded into each well and incubated for 24 h. After incubation, a wound (scratch) was inflicted using a 200 μL pipette tip on the cell monolayer, followed by a washing step with PBS to remove cell debris. Next, the media and drugs were replaced to treat the cells as untreated cells (control), cells with FD at concentrations of 62.5nM and 125 nM, and cells with Doxil ® at concentrations of 250 nM and 400nM. The cells were then incubated for 3 days, with all treatments performed in triplicate. Images of the wounds were taken at 0, 3, 6, 9, and 24h. Wounds were observed using inverted light microscopy at a 10 × objective lens. Images were captured daily using a Nikon Eclipse camera (TE2000-4, Panasonic Lumix DMC-GF6), and the wound areas were analyzed using ImageJ software.
A two-way ANOVA test was used to analyze the data results; the experiment was performed in triplicates (n = 3). The P values significance were determined as follows: * P ≤ 0.05, ** P ≤ 0. 01, *** P ≤ 0.001 and **** P ≤ 0.0001.
3.1. Spheroids formation Tumor spheroids were formed by hanging drop technique after three days of incubation; each drop in the petri dish contained the same uniform size of spheroids. Cells of 2D culture models lack cell-cell and cell-matrix interactions, which are presented in native tumors. In contrast, 3D culture models have an opportunity to culture cancer cells alone or with different cell types. Thus, it can closely mimic the native environment of tumors . U87-MG cells were co-cultured with HDF and successfully formed a compacted spheroid structure to investigate tumor behavior and interaction in vitro . Several studies focused on the role of tumor microenvironment, which consists of cellular and non-cellular components, and how it affects cells’ behaviors such as proliferation, migration, and therapeutic resistance . Some of these studies illustrated the effect of co-culturing cells; breast cancer cells exhibit features reflective of ductal carcinoma when co-culturing with luminal cells, myoepithelial cells, and stromal fibroblasts as 3D model . Ewing tumor closely resemble patient tumors in cell-cell junctions and proliferative index . These studies’ findings show that 3D culture models allow the evaluation of tumor microenvironment effects on in vitro oncology studies. 3.2. Cellular uptake 3.2.1. 2D and 3D cellular uptake by flow cytometry. Flow cytometry analysis method was employed to quantify the cellular uptake and penetration of Dox in both two-dimensional (2D) and three-dimensional (3D) cell culture systems. Dox, a potent anticancer therapy, exhibits intrinsic fluorescence with an emission signal at 595nm when excited with a 470nm laser. This valuable fluorescence characteristic is an important tool for understanding the cellular dynamics and mechanisms of action in biomedical research and imaging [ – ]. In the present study, the cellular uptake of Dox and the percentage of positive cells with Dox following the treatment with free Dox (FD) and Doxil ® were evaluated and compared to untreated cells control. Cells in both 2D and 3D culture systems were treated for 4h and 24h with Dox concentrations of 1 µM and 2 µM, which were selected based on clinical relevance and optimal signal detection . Notably, the highest cellular uptake of Dox was observed at a concentration of 2 µM in both 2D and 3D models and . Moreover, the results showed that 2D cultures demonstrated a significantly higher Dox uptake and percent of positive cells compared to 3D cells . illustrates 2D and 3D model cell uptake for 2 µM concentrations at 24 h incubation. The fold increase in fluorescent intensity was calculated as: (FIs of FD or Doxil ® – FI of untreated cells)/FI of untreated cells). 3.2.2. 2D cellular uptake by confocal microscopy. Fluorescent microscopy visualized the localization of DAPI (blue-stained nuclei) and Dox PE-A (red signal) after cellular uptake. FD-treated cells showed more signal localization than those for Doxil ® -treated cells, even at low concentrations, as shown in and . 3.2.3. 3D cellular uptake by confocal microscopy. Spheroid’s structure could affect FD and Doxil ® penetration and uptake; Z-stacking by SLCM was performed to examine these interactions. Spheroids were treated with (1 µM and 2 µM) of drug concentration for 24h. FD highly diffused through the spheroid’s layer; meanwhile, Doxil ® localization clearly appeared within the spheroid’s surface layers. Doxil ® and FD interaction and diffusion are demonstrated in . Doxorubicin-loaded PEGylated liposome (Doxil ® ) has been broadly used to treat many cancer types . Flow cytometry was applied to investigate the Doxorubicin uptake by measuring FI by cells of 2D and 3D models treated with FD and Doxil ® . 2D and 3D cells showed a higher uptake for FD compared to Doxil ® in an effective manner. Several studies on the efficacy of Doxil ® have suggested that the mechanism behind the slow release of Doxorubicin from the liposome is not yet fully understood. The insufficient release of Dox from Doxil ® can be attributed to ammonium sulfate gradient loading of Doxil ® , which results in Dox precipitation and stacking in liposomal core as bundles of fibers, and to the high cholesterol level that makes such a phase transition missed. These factors contribute to the slow Dox release, which can hinder drug passage. Thus, a small amount of drug molecules can pass the lipid membrane . Efficacious therapy is not achieved only by targeted delivery of nanoparticles. Sufficient release of the encapsulated drug into the target tissue is equally important . The flow cytometry results also demonstrated that 2D cells had a higher Doxil ® uptake than 3D cells; one explanation could be due to the inability of 2D monolayer cancer cells to capture the intertumoral transport (a critical mechanism of drug-tumor interaction at the tumor site) that can adequately determine the drug efficacy compared to 3D cultures . To track the intracellular localization of Doxorubicin after it was taken up by the cells, its natural fluorescent property was utilized, and FM and CLSM were conducted for this purpose. For 2D monolayer cells, more Dox was detected in cells exposed to FD than those treated with Doxil ® . This was cleared by visualizing that the red fluorescent material significantly accumulated in the nuclei of FD-treated cells, while Doxil ® -treated cells showed less Dox accumulation. The intracellular Dox distribution seemingly depends on how the Dox was delivered to the cells. One of the possible explanations is that, after cells were exposed to FD and Doxil ® , FD diffused more easily into cells, and because of its high affinity toward DNA, it can accumulate in the nuclei. On the other hand, many vesicles could also be transported from inside the cell to the outside by exocytosis, reducing the amount of Doxil ® that was taken up by endocytosis . Z-stacking of 3D spheroids by CSLM was performed to evaluate the treatment’s ability to penetrate and diffuse through the spheroid structure; FD showed a broad and deep diffusion within the spheroid’s layers from the surface, whereas Doxil ® could not diffuse and spread deep enough as the Dox red fluorescence was visualized only on the outer surface layers ( ). This can be due to the spheroid’s structure, which mimics in vivo tumor mass architecture, as well as NPs penetration properties since the cell’s dense packing makes a transport barrier, preventing the nanoparticle from deep diffusion into the spheroidal core; thus, NPs penetration is limited to the outer layers. This proves that spheroid represents a suitable model to test nanoparticle efficacy compared to 2D cultures . 3.2.4. MTT assay. The percentage of viable cells was evaluated after the performance of the MTT assay to estimate the treatment’s effect on cellular metabolism. The results were then calculated with respect to untreated cells (untreated, 100%) as follows: Percentage of viable cells = (OD of treated cells/ OD of untreated cells × 100%). It is important to mention that the reduction in the MTT signal following DOX treatment may reveal DOX-mediated mitochondrial disruption, including reduced respiration and energy production, in addition to the potential effect on cell viability . FD tended to exhibit more metabolic disruption with an IC 50 value of 0.25 µM, while Doxil ® had IC 50 of approximately 0.7 µM. demonstrates the cell metabolic activity for each treatment. Doxorubicin, an antibiotic and anticancer chemotherapy agent, gains its cytotoxicity effect through the intercalation ability within DNA, inhibiting the synthesis of DNA and RNA strands. Furthermore, it inhibits topoisomerase II enzyme activity and induces apoptosis . After flow cytometry and fluorescent microscopy assessment for the cellular uptake, an MTT assay was performed to evaluate the metabolic activity as an indicator of the treatment effect. Cells were treated with FD and Doxil ® at concentrations ranging from 2 µM con to 0.62 µM. After 72h of incubation, significant differences were observed between the treatments at different concentrations. FD showed higher metabolic disruption with an IC 50 of approximately 250nM, while the IC 50 of Doxil ® was 700nM, which could be accounted for the optimal binding and diffusivity of FD among the partial drug release and diffusion of Doxil ® through tissue . 3.2.5. 3D invasion model. A 3D collagen invasion assay was performed to understand in vivo conditions adequately. Cells were treated with chosen concentrations lower than the IC 50 values (125nM, 62.5nM) and (400nM and 250nM) for each FD and Doxil ® , respectively. Also, 2 µM concentration was used for more comparable data. After that, the total invasion area (TIA) was calculated. shows the invasion area for FD and Doxil ® and compares treatments versus time. Cellular interaction with Doxil ® was found to be less tight than with FD. After conducting TIA measurements and analyzing the data, it was observed that the interaction between FD and cells was significantly higher than the interaction between Doxil ® and cells. Additionally, a minor difference was noted in the interaction between Doxil ® -treated cells and untreated cells. 3.2.6. Wound healing assay. The wound healing assay was conducted as a 2D model in which cells were treated with specific concentrations of FD and Doxil ® , which were below the IC 50 values: (125nM, 62.5nM) and (400nM and 250nM), respectively. The scratch area was then calculated. shows that FD effectively inhibited cell migration, while Doxil ® was less efficient. The Figure also illustrates the correlation between the migration area following treatment with FD and Doxil ® and the elapsed time. Developing a cellular model that can recapitulate tumors in vivo conditions would facilitate monitoring cell-cell communication and cell-drug interaction for a selected treatment. 3D cell culture models could offer this instead of traditional 2D ones . The matrix-embedded model was chosen to represent the 3D model. U87-MG cells were cultured into 3D spheroids and embedded within the collagen matrix to mimic the ECM surrounding the tumor, followed by FD and Doxil ® treatment. According to the TIA observed after 4 days of incubation, there is a higher significant difference in FD-cell interaction compared to Doxil ® -cell interaction, while less was observed between Doxil ® -treated cells and untreated cells. Meanwhile, in wound healing assay, a chosen 2D model was performed to represent cell migration by creating a cell-free area in the monolayer cells . Doxil ® showed the opposite of its interaction with the 3D model with the observation of a slightly higher significance in the scratch area when Doxil ® -treated cells were compared to untreated cells. These findings could be explained according to a major aspect: the ECM consists of interconnected networks of collagen fibers that somehow could determine the diffusion complexity and block the larger nanoparticles penetration into tumors . Many previous studies have proved the role of ECM on drug efficacy. One of these studies found that collagen digestion increased the drug diffusion deeper into the tumor, thus facilitating molecule delivery . This finding can justify the highly significant differences between 2D Doxil® treated cells, which lack the connectivity of ECM found in the 3D model and work as a barrier for Doxil ® particles to diffuse and reach the spheroids .
Tumor spheroids were formed by hanging drop technique after three days of incubation; each drop in the petri dish contained the same uniform size of spheroids. Cells of 2D culture models lack cell-cell and cell-matrix interactions, which are presented in native tumors. In contrast, 3D culture models have an opportunity to culture cancer cells alone or with different cell types. Thus, it can closely mimic the native environment of tumors . U87-MG cells were co-cultured with HDF and successfully formed a compacted spheroid structure to investigate tumor behavior and interaction in vitro . Several studies focused on the role of tumor microenvironment, which consists of cellular and non-cellular components, and how it affects cells’ behaviors such as proliferation, migration, and therapeutic resistance . Some of these studies illustrated the effect of co-culturing cells; breast cancer cells exhibit features reflective of ductal carcinoma when co-culturing with luminal cells, myoepithelial cells, and stromal fibroblasts as 3D model . Ewing tumor closely resemble patient tumors in cell-cell junctions and proliferative index . These studies’ findings show that 3D culture models allow the evaluation of tumor microenvironment effects on in vitro oncology studies.
3.2.1. 2D and 3D cellular uptake by flow cytometry. Flow cytometry analysis method was employed to quantify the cellular uptake and penetration of Dox in both two-dimensional (2D) and three-dimensional (3D) cell culture systems. Dox, a potent anticancer therapy, exhibits intrinsic fluorescence with an emission signal at 595nm when excited with a 470nm laser. This valuable fluorescence characteristic is an important tool for understanding the cellular dynamics and mechanisms of action in biomedical research and imaging [ – ]. In the present study, the cellular uptake of Dox and the percentage of positive cells with Dox following the treatment with free Dox (FD) and Doxil ® were evaluated and compared to untreated cells control. Cells in both 2D and 3D culture systems were treated for 4h and 24h with Dox concentrations of 1 µM and 2 µM, which were selected based on clinical relevance and optimal signal detection . Notably, the highest cellular uptake of Dox was observed at a concentration of 2 µM in both 2D and 3D models and . Moreover, the results showed that 2D cultures demonstrated a significantly higher Dox uptake and percent of positive cells compared to 3D cells . illustrates 2D and 3D model cell uptake for 2 µM concentrations at 24 h incubation. The fold increase in fluorescent intensity was calculated as: (FIs of FD or Doxil ® – FI of untreated cells)/FI of untreated cells). 3.2.2. 2D cellular uptake by confocal microscopy. Fluorescent microscopy visualized the localization of DAPI (blue-stained nuclei) and Dox PE-A (red signal) after cellular uptake. FD-treated cells showed more signal localization than those for Doxil ® -treated cells, even at low concentrations, as shown in and . 3.2.3. 3D cellular uptake by confocal microscopy. Spheroid’s structure could affect FD and Doxil ® penetration and uptake; Z-stacking by SLCM was performed to examine these interactions. Spheroids were treated with (1 µM and 2 µM) of drug concentration for 24h. FD highly diffused through the spheroid’s layer; meanwhile, Doxil ® localization clearly appeared within the spheroid’s surface layers. Doxil ® and FD interaction and diffusion are demonstrated in . Doxorubicin-loaded PEGylated liposome (Doxil ® ) has been broadly used to treat many cancer types . Flow cytometry was applied to investigate the Doxorubicin uptake by measuring FI by cells of 2D and 3D models treated with FD and Doxil ® . 2D and 3D cells showed a higher uptake for FD compared to Doxil ® in an effective manner. Several studies on the efficacy of Doxil ® have suggested that the mechanism behind the slow release of Doxorubicin from the liposome is not yet fully understood. The insufficient release of Dox from Doxil ® can be attributed to ammonium sulfate gradient loading of Doxil ® , which results in Dox precipitation and stacking in liposomal core as bundles of fibers, and to the high cholesterol level that makes such a phase transition missed. These factors contribute to the slow Dox release, which can hinder drug passage. Thus, a small amount of drug molecules can pass the lipid membrane . Efficacious therapy is not achieved only by targeted delivery of nanoparticles. Sufficient release of the encapsulated drug into the target tissue is equally important . The flow cytometry results also demonstrated that 2D cells had a higher Doxil ® uptake than 3D cells; one explanation could be due to the inability of 2D monolayer cancer cells to capture the intertumoral transport (a critical mechanism of drug-tumor interaction at the tumor site) that can adequately determine the drug efficacy compared to 3D cultures . To track the intracellular localization of Doxorubicin after it was taken up by the cells, its natural fluorescent property was utilized, and FM and CLSM were conducted for this purpose. For 2D monolayer cells, more Dox was detected in cells exposed to FD than those treated with Doxil ® . This was cleared by visualizing that the red fluorescent material significantly accumulated in the nuclei of FD-treated cells, while Doxil ® -treated cells showed less Dox accumulation. The intracellular Dox distribution seemingly depends on how the Dox was delivered to the cells. One of the possible explanations is that, after cells were exposed to FD and Doxil ® , FD diffused more easily into cells, and because of its high affinity toward DNA, it can accumulate in the nuclei. On the other hand, many vesicles could also be transported from inside the cell to the outside by exocytosis, reducing the amount of Doxil ® that was taken up by endocytosis . Z-stacking of 3D spheroids by CSLM was performed to evaluate the treatment’s ability to penetrate and diffuse through the spheroid structure; FD showed a broad and deep diffusion within the spheroid’s layers from the surface, whereas Doxil ® could not diffuse and spread deep enough as the Dox red fluorescence was visualized only on the outer surface layers ( ). This can be due to the spheroid’s structure, which mimics in vivo tumor mass architecture, as well as NPs penetration properties since the cell’s dense packing makes a transport barrier, preventing the nanoparticle from deep diffusion into the spheroidal core; thus, NPs penetration is limited to the outer layers. This proves that spheroid represents a suitable model to test nanoparticle efficacy compared to 2D cultures . 3.2.4. MTT assay. The percentage of viable cells was evaluated after the performance of the MTT assay to estimate the treatment’s effect on cellular metabolism. The results were then calculated with respect to untreated cells (untreated, 100%) as follows: Percentage of viable cells = (OD of treated cells/ OD of untreated cells × 100%). It is important to mention that the reduction in the MTT signal following DOX treatment may reveal DOX-mediated mitochondrial disruption, including reduced respiration and energy production, in addition to the potential effect on cell viability . FD tended to exhibit more metabolic disruption with an IC 50 value of 0.25 µM, while Doxil ® had IC 50 of approximately 0.7 µM. demonstrates the cell metabolic activity for each treatment. Doxorubicin, an antibiotic and anticancer chemotherapy agent, gains its cytotoxicity effect through the intercalation ability within DNA, inhibiting the synthesis of DNA and RNA strands. Furthermore, it inhibits topoisomerase II enzyme activity and induces apoptosis . After flow cytometry and fluorescent microscopy assessment for the cellular uptake, an MTT assay was performed to evaluate the metabolic activity as an indicator of the treatment effect. Cells were treated with FD and Doxil ® at concentrations ranging from 2 µM con to 0.62 µM. After 72h of incubation, significant differences were observed between the treatments at different concentrations. FD showed higher metabolic disruption with an IC 50 of approximately 250nM, while the IC 50 of Doxil ® was 700nM, which could be accounted for the optimal binding and diffusivity of FD among the partial drug release and diffusion of Doxil ® through tissue . 3.2.5. 3D invasion model. A 3D collagen invasion assay was performed to understand in vivo conditions adequately. Cells were treated with chosen concentrations lower than the IC 50 values (125nM, 62.5nM) and (400nM and 250nM) for each FD and Doxil ® , respectively. Also, 2 µM concentration was used for more comparable data. After that, the total invasion area (TIA) was calculated. shows the invasion area for FD and Doxil ® and compares treatments versus time. Cellular interaction with Doxil ® was found to be less tight than with FD. After conducting TIA measurements and analyzing the data, it was observed that the interaction between FD and cells was significantly higher than the interaction between Doxil ® and cells. Additionally, a minor difference was noted in the interaction between Doxil ® -treated cells and untreated cells. 3.2.6. Wound healing assay. The wound healing assay was conducted as a 2D model in which cells were treated with specific concentrations of FD and Doxil ® , which were below the IC 50 values: (125nM, 62.5nM) and (400nM and 250nM), respectively. The scratch area was then calculated. shows that FD effectively inhibited cell migration, while Doxil ® was less efficient. The Figure also illustrates the correlation between the migration area following treatment with FD and Doxil ® and the elapsed time. Developing a cellular model that can recapitulate tumors in vivo conditions would facilitate monitoring cell-cell communication and cell-drug interaction for a selected treatment. 3D cell culture models could offer this instead of traditional 2D ones . The matrix-embedded model was chosen to represent the 3D model. U87-MG cells were cultured into 3D spheroids and embedded within the collagen matrix to mimic the ECM surrounding the tumor, followed by FD and Doxil ® treatment. According to the TIA observed after 4 days of incubation, there is a higher significant difference in FD-cell interaction compared to Doxil ® -cell interaction, while less was observed between Doxil ® -treated cells and untreated cells. Meanwhile, in wound healing assay, a chosen 2D model was performed to represent cell migration by creating a cell-free area in the monolayer cells . Doxil ® showed the opposite of its interaction with the 3D model with the observation of a slightly higher significance in the scratch area when Doxil ® -treated cells were compared to untreated cells. These findings could be explained according to a major aspect: the ECM consists of interconnected networks of collagen fibers that somehow could determine the diffusion complexity and block the larger nanoparticles penetration into tumors . Many previous studies have proved the role of ECM on drug efficacy. One of these studies found that collagen digestion increased the drug diffusion deeper into the tumor, thus facilitating molecule delivery . This finding can justify the highly significant differences between 2D Doxil® treated cells, which lack the connectivity of ECM found in the 3D model and work as a barrier for Doxil ® particles to diffuse and reach the spheroids .
Flow cytometry analysis method was employed to quantify the cellular uptake and penetration of Dox in both two-dimensional (2D) and three-dimensional (3D) cell culture systems. Dox, a potent anticancer therapy, exhibits intrinsic fluorescence with an emission signal at 595nm when excited with a 470nm laser. This valuable fluorescence characteristic is an important tool for understanding the cellular dynamics and mechanisms of action in biomedical research and imaging [ – ]. In the present study, the cellular uptake of Dox and the percentage of positive cells with Dox following the treatment with free Dox (FD) and Doxil ® were evaluated and compared to untreated cells control. Cells in both 2D and 3D culture systems were treated for 4h and 24h with Dox concentrations of 1 µM and 2 µM, which were selected based on clinical relevance and optimal signal detection . Notably, the highest cellular uptake of Dox was observed at a concentration of 2 µM in both 2D and 3D models and . Moreover, the results showed that 2D cultures demonstrated a significantly higher Dox uptake and percent of positive cells compared to 3D cells . illustrates 2D and 3D model cell uptake for 2 µM concentrations at 24 h incubation. The fold increase in fluorescent intensity was calculated as: (FIs of FD or Doxil ® – FI of untreated cells)/FI of untreated cells).
Fluorescent microscopy visualized the localization of DAPI (blue-stained nuclei) and Dox PE-A (red signal) after cellular uptake. FD-treated cells showed more signal localization than those for Doxil ® -treated cells, even at low concentrations, as shown in and .
Spheroid’s structure could affect FD and Doxil ® penetration and uptake; Z-stacking by SLCM was performed to examine these interactions. Spheroids were treated with (1 µM and 2 µM) of drug concentration for 24h. FD highly diffused through the spheroid’s layer; meanwhile, Doxil ® localization clearly appeared within the spheroid’s surface layers. Doxil ® and FD interaction and diffusion are demonstrated in . Doxorubicin-loaded PEGylated liposome (Doxil ® ) has been broadly used to treat many cancer types . Flow cytometry was applied to investigate the Doxorubicin uptake by measuring FI by cells of 2D and 3D models treated with FD and Doxil ® . 2D and 3D cells showed a higher uptake for FD compared to Doxil ® in an effective manner. Several studies on the efficacy of Doxil ® have suggested that the mechanism behind the slow release of Doxorubicin from the liposome is not yet fully understood. The insufficient release of Dox from Doxil ® can be attributed to ammonium sulfate gradient loading of Doxil ® , which results in Dox precipitation and stacking in liposomal core as bundles of fibers, and to the high cholesterol level that makes such a phase transition missed. These factors contribute to the slow Dox release, which can hinder drug passage. Thus, a small amount of drug molecules can pass the lipid membrane . Efficacious therapy is not achieved only by targeted delivery of nanoparticles. Sufficient release of the encapsulated drug into the target tissue is equally important . The flow cytometry results also demonstrated that 2D cells had a higher Doxil ® uptake than 3D cells; one explanation could be due to the inability of 2D monolayer cancer cells to capture the intertumoral transport (a critical mechanism of drug-tumor interaction at the tumor site) that can adequately determine the drug efficacy compared to 3D cultures . To track the intracellular localization of Doxorubicin after it was taken up by the cells, its natural fluorescent property was utilized, and FM and CLSM were conducted for this purpose. For 2D monolayer cells, more Dox was detected in cells exposed to FD than those treated with Doxil ® . This was cleared by visualizing that the red fluorescent material significantly accumulated in the nuclei of FD-treated cells, while Doxil ® -treated cells showed less Dox accumulation. The intracellular Dox distribution seemingly depends on how the Dox was delivered to the cells. One of the possible explanations is that, after cells were exposed to FD and Doxil ® , FD diffused more easily into cells, and because of its high affinity toward DNA, it can accumulate in the nuclei. On the other hand, many vesicles could also be transported from inside the cell to the outside by exocytosis, reducing the amount of Doxil ® that was taken up by endocytosis . Z-stacking of 3D spheroids by CSLM was performed to evaluate the treatment’s ability to penetrate and diffuse through the spheroid structure; FD showed a broad and deep diffusion within the spheroid’s layers from the surface, whereas Doxil ® could not diffuse and spread deep enough as the Dox red fluorescence was visualized only on the outer surface layers ( ). This can be due to the spheroid’s structure, which mimics in vivo tumor mass architecture, as well as NPs penetration properties since the cell’s dense packing makes a transport barrier, preventing the nanoparticle from deep diffusion into the spheroidal core; thus, NPs penetration is limited to the outer layers. This proves that spheroid represents a suitable model to test nanoparticle efficacy compared to 2D cultures .
The percentage of viable cells was evaluated after the performance of the MTT assay to estimate the treatment’s effect on cellular metabolism. The results were then calculated with respect to untreated cells (untreated, 100%) as follows: Percentage of viable cells = (OD of treated cells/ OD of untreated cells × 100%). It is important to mention that the reduction in the MTT signal following DOX treatment may reveal DOX-mediated mitochondrial disruption, including reduced respiration and energy production, in addition to the potential effect on cell viability . FD tended to exhibit more metabolic disruption with an IC 50 value of 0.25 µM, while Doxil ® had IC 50 of approximately 0.7 µM. demonstrates the cell metabolic activity for each treatment. Doxorubicin, an antibiotic and anticancer chemotherapy agent, gains its cytotoxicity effect through the intercalation ability within DNA, inhibiting the synthesis of DNA and RNA strands. Furthermore, it inhibits topoisomerase II enzyme activity and induces apoptosis . After flow cytometry and fluorescent microscopy assessment for the cellular uptake, an MTT assay was performed to evaluate the metabolic activity as an indicator of the treatment effect. Cells were treated with FD and Doxil ® at concentrations ranging from 2 µM con to 0.62 µM. After 72h of incubation, significant differences were observed between the treatments at different concentrations. FD showed higher metabolic disruption with an IC 50 of approximately 250nM, while the IC 50 of Doxil ® was 700nM, which could be accounted for the optimal binding and diffusivity of FD among the partial drug release and diffusion of Doxil ® through tissue .
A 3D collagen invasion assay was performed to understand in vivo conditions adequately. Cells were treated with chosen concentrations lower than the IC 50 values (125nM, 62.5nM) and (400nM and 250nM) for each FD and Doxil ® , respectively. Also, 2 µM concentration was used for more comparable data. After that, the total invasion area (TIA) was calculated. shows the invasion area for FD and Doxil ® and compares treatments versus time. Cellular interaction with Doxil ® was found to be less tight than with FD. After conducting TIA measurements and analyzing the data, it was observed that the interaction between FD and cells was significantly higher than the interaction between Doxil ® and cells. Additionally, a minor difference was noted in the interaction between Doxil ® -treated cells and untreated cells.
The wound healing assay was conducted as a 2D model in which cells were treated with specific concentrations of FD and Doxil ® , which were below the IC 50 values: (125nM, 62.5nM) and (400nM and 250nM), respectively. The scratch area was then calculated. shows that FD effectively inhibited cell migration, while Doxil ® was less efficient. The Figure also illustrates the correlation between the migration area following treatment with FD and Doxil ® and the elapsed time. Developing a cellular model that can recapitulate tumors in vivo conditions would facilitate monitoring cell-cell communication and cell-drug interaction for a selected treatment. 3D cell culture models could offer this instead of traditional 2D ones . The matrix-embedded model was chosen to represent the 3D model. U87-MG cells were cultured into 3D spheroids and embedded within the collagen matrix to mimic the ECM surrounding the tumor, followed by FD and Doxil ® treatment. According to the TIA observed after 4 days of incubation, there is a higher significant difference in FD-cell interaction compared to Doxil ® -cell interaction, while less was observed between Doxil ® -treated cells and untreated cells. Meanwhile, in wound healing assay, a chosen 2D model was performed to represent cell migration by creating a cell-free area in the monolayer cells . Doxil ® showed the opposite of its interaction with the 3D model with the observation of a slightly higher significance in the scratch area when Doxil ® -treated cells were compared to untreated cells. These findings could be explained according to a major aspect: the ECM consists of interconnected networks of collagen fibers that somehow could determine the diffusion complexity and block the larger nanoparticles penetration into tumors . Many previous studies have proved the role of ECM on drug efficacy. One of these studies found that collagen digestion increased the drug diffusion deeper into the tumor, thus facilitating molecule delivery . This finding can justify the highly significant differences between 2D Doxil® treated cells, which lack the connectivity of ECM found in the 3D model and work as a barrier for Doxil ® particles to diffuse and reach the spheroids .
Our study involved the use of 2D and 3D cell culture models to evaluate the efficiency of nanoparticle-based drug delivery systems. We developed a 3D model replicating the in vivo conditions of tumor structure and extracellular matrix to assess the delivery of liposomal nanoparticles to spheroids through a collagen matrix. This 3D model proved to be more informative than traditional 2D models as it provided a better understanding of nanoparticle interactions. We compared the interactions of liposomal-Doxil ® with cellular targets in both 2D and 3D models and found that the interaction of nanoparticles with 2D cells was more straightforward than that with 3D models. However, the 3D model could better recapitulate the in vivo microenvironment than the 2D model. Our study also revealed the need for modifications to the liposomal-Doxil ® to achieve sufficient drug release. We recommend additional modifications, such as surface modification for active targeting and modification in loading and encapsulation of Dox, to achieve efficient release by the delivery system. Finally, we emphasize the importance of using 3D models to understand better the interplay between nano-technological, physiochemical, and biological principles of the tumor microenvironment, which may improve the in vivo efficacy of nanoparticles.
|
|
Studying the added effect of sum-of-segments biometry to modern intraocular lens power calculation formulas for short eyes | f8af1e19-05bd-4d98-b1e3-5c1841c4a018 | 11816771 | Surgical Procedures, Operative[mh] | Accurate assessment of the axial length (AXL) is crucial for calculating the power of intraocular lenses (IOLs). Mistakes in measuring AXL can lead to notable refractive errors after surgery . Optical biometry has established itself as the preferred method for AXL measurements, offering exceptional reproducibility and precision . Most of the optical biometers currently on the market employ a composite refractive index for the entire axial length (AXL) to transform optical path length into geometrical distance. The ARGOS (Alcon Laboratories, Inc., Fort Worth, TX) is an optical biometer featuring a swept light source that operates at an infrared wavelength of 1060 nm, allowing for higher acquisition rates in cases of denser cataracts. Unlike the majority of other devices, ARGOS utilizes a sum-of-segments approach for AXL measurement, applying a segment-specific refractive index instead of a composite one. It assigns distinct refractive indices for each part of the eye: cornea at 1.376, aqueous humor at 1.336, lens at 1.410, and vitreous at 1.336. This sum-of-segments technique has led to longer AXL measurements in shorter eyes, while resulting in shorter AXL measurements in longer eyes. This could be attributed to the significantly larger percentage of contribution of lens thickness to the AXL in shorter eyes and the comparatively larger percentage of the vitreous cavity in longer AXL. In shorter eyes, a longer AXL results in a lower power of the intraocular lens (IOL), this can lead to hyperopic errors . It is worth mentioning that the AXL measurement error can contribute to inaccuracy in IOL power in 36% of cases . Currently, several intraocular lens (IOL) power calculation formulas provide the option to utilize a sum-of-segments axial length (AXLsos) as input. The online IOL calculator from the European Society of Cataract and Refractive Surgery (ESCRS) features some of these formulas, including the Cooke K6 , EVO 2.0, and PEARL-DGS formulas (accessible at: https://iolcalculator.escrs.org/ ). Additionally, the Barrett formula has been updated and is now incorporated into the ARGOS biometer, introducing the Barrett true axial length (BTAL) formula. This new formula has resolved the concerns regarding the accuracy of conventional Barrett Universal II formula calculated using the ARGOS anterior chamber depth, therefore aims to minimize refractive prediction errors in eyes measured with the ARGOS biometer . However, there is a scarcity of literature discussing the outcomes of these novel formulas that permit the use of AXLsos in IOL power calculations. The aim of this study was to study the added effect of sum-of-segments biometry to modern intraocular lens power calculation formulas for eyes with short axial length. This was a retrospective case series that included 99 eyes from 99 patients. The included patients were older than 18 years of age, phakic patients with cataract and AXL less than or equal 22.0 mm. The patients included in this study underwent a standard and uneventful phacoemulsification procedure, followed by the implantation of a hydrophobic single-piece acrylic intraocular lens (Alcon AcrySof model SA60AT), within the capsular bag. Following the procedure, the patients were scheduled for a final follow-up appointment and provided their written consent to participate in the research. Exclusion criteria for the study included patients who experienced intraoperative complications that could compromise postoperative biometric measurements, those with inadequate visual acuity that would impede proper postoperative refraction, and individuals with other ocular conditions affecting biometric assessments, such as corneal scarring or lens dislocation. A review of the medical records of the patients, spanning from January 2021 to October 2024, was conducted. Demographic information, including age and sex, was documented, along with biometric parameters such as axial length (AXL), keratometric readings (K), anterior chamber depth (ACD), lens thickness, central corneal thickness, and white-to-white diameter. Preoperative AXL measurements were conducted utilizing the ARGOS biometer (Alcon, Inc., Fort Worth, TX), an advanced swept-source optical coherence tomography (SS-OCT) biometer operating at a wavelength of 1060 nm. This device employs sum-of-segments biometry along with a segmental refractive index. The mean of three high-quality scans was documented. All patients underwent standard phacoemulsification without complications, followed by the implantation of a foldable hydrophobic acrylic IOL, and were monitored in the postoperative period. At 4 to 6 weeks after surgery, manifest refraction was assessed. The refractive error was subsequently converted to spherical equivalent (SE) and recorded, calculated as SE = spherical power + (cylinder power/2). The results of various IOL power calculation formulas were analyzed in this study. Two integrated formulas from ARGOS were employed: Barrett Universal II (BUII) and BTAL. Additionally, the following formulas were utilized: Cooke K6 , EVO 2.0, and PEARL DGS, all of which can be accessed through the ESCRS IOL calculator website ( https://iolcalculator.escrs.org/ ). The last three formulas available on the ESCRS online IOL calculator were applied both with (Cooke K6sos, EVO 2.0sos, and PEARL-DGSsos) and without the selection of the ARGOS AXLsos option. The Cooke K6 formula employs the general vergence formula, thereby mitigating some limitations associated with thick-lens, ray-tracing, and artificial intelligence formulas; it is also accessible online at ( https://cookeformula.com/ ) . The PEARL-DGS formula, which stands for Postoperative Spherical Equivalent prediction using Artificial intelligence and Linear algorithms, was developed by Debellemanière, Gatinel, and Saad. This thick lens formula utilizes artificial intelligence to estimate the distance between the posterior corneal surface and the anterior IOL surface (theoretical internal lens position) and is available online at www.iolsolver.com . EVO (Emmetropia Verifying Optical) 2.0 is a contemporary thick-lens formula based on the theory of emmetropization, which can also be found online at https://www.evoiolcalculator.com/calculator.aspx . The initial A-constant for most formulas was set at 118.8, while for the BUII and BTAL formulas, the initial Lens Factor (LF) was 1.74. These lens constants were updated from the ARGOS biometer and the online site of “User group for Laser Interference Biometry” (ULIB), available at http://ocusoft.de/ulib/c1.htm . Refractive prediction error (PE) was determined by calculating the difference in spherical equivalent between the value predicted by the formula and the actual value measured 4 to 6 weeks after surgery. The absolute prediction error (APE) was derived by converting the PE into an absolute figure. The main outcomes assessed included the median absolute prediction error, the mean absolute prediction error, and the percentage of cases falling within 0.25, 0.5 D, 1 D, and 2 D of the desired refraction. Data analysis was performed using the Social Sciences SPSS Statistics for Windows (version 26.0; SPSS Inc., Chicago, IL, USA). The quantitative data were described in terms of their range, median, mean, and standard deviation. The normality of the dataset was evaluated using the Kolmogorov–Smirnov test. Friedman's ANOVA test was utilized to compare different means. The Wilcoxon signed-rank test for paired samples was used to assess the medians within the same group. Additionally, the Cochran’s Q test was applied to examine the distribution of cases within the specified refraction range. Statistical significance was established when the p value was less than 0.05, 95% confidence interval. Eyetemis web-based analysis software) was used to double check the results of spherical equivalent prediction errors . This study included 99 eyes from 99 patients. The mean age was 53.7 ± 7.1 years (range from 43 to 69 years). The study included 50 males and 49 females. Table shows the demographic and biometric data of the included patients ( n = 99). Table lists the arithmetic mean prediction errors of the included formulas. The ANOVA showed that the difference was statistically significant ( p = 0.017). The PEARL-DGS sos was the only formula to show a myopic mean prediction error (−0.25 ± 0.36 D). The Cooke K6 formula showed the highest hyperopic mean prediction error (0.55 ± 0.35 D), followed by EVO 2.0, Cooke K6 sos , and BUII formulas. The mean and median absolute errors (MAE and MedAE) for the various formulas are shown in Table . Table displays the number of cases within ± 0.25 D, ± 0.5 D, and ± 1.0 D of the target refraction. The Friedman's ANOVA test showed statistically significant differences ( p = 0.014). Post-hoc analysis showed statistically significant differences between PEARL-DGS and PEARL-DGSsos, Cooke K6 and Cooke K6sos, BUII and BTAL, and EVO 2.0 and EVO 2.0sos. The Cochran's Q test was used to analyze the number of cases falling within the targeted refraction range; the results indicated a statistically significant difference ( p < 0.05). EVO 2.0 sos and PEARL-DGS sos formulas showed the least mean and median absolute errors. The median absolute error (MedAE) for EVO 2.0 sos and PEARL-DGS sos was 0.29 and 0.31 D respectively. EVO 2.0 sos and PEARL-DGS sos formulas had the highest cases within ± 0.25 D of the intended refraction (45.45% and 42.42%, respectively). Cooke K6 formula had the least cases within ± 0.25 D of the intended refraction (15.50%). EVO 2.0 sos , Cooke K6 sos, and PEARL-DGS sos formulas had 100% of cases within ± 1.0 D of the intended refraction. Cooke K6 formula had the least cases within ± 1.0 D of the intended refraction (87.88%). All of the included formulas had 100% of cases within ± 2.0 D of the target refraction. Subgroup analysis of eyes with AXL 21 mm or shorter ( n = 57) was done. The mean AXL was 20.41 ± 0.40 mm (range 19.80 to 21.00 mm). Table shows the mean and median absolute errors for the included formulas for eyes ≤ 21 mm AXL. Table also displays the number of cases within ± 0.25 D, ± 0.5 D, and ± 1.0 D of the target refraction for eyes ≤ 21 mm AXL. The PEARL-DGS sos formula showed the only myopic mean prediction error (−0.23 ± 0.37 D). The Cooke K6 and EVO 2.0 formulas showed the highest hyperopic mean prediction error (0.58 ± 0.36 D and 0.52 ± 0.42 D respectively). The MAE and MedAE for the various formulas are shown in Table for eyes ≤ 21 mm. Table displays the number of cases within ± 0.25 D, ± 0.5 D, and ± 1.0 D of the target refraction for eyes shorter than or equal 21 mm. The Friedman's ANOVA test showed statistically significant differences ( p = 0.023). The Cochran’s Q test was used to analyze the number of cases falling within the targeted refraction range; the results indicated a statistically significant difference ( p < 0.05). PEARL-DGS sos and EVO 2.0 sos formulas showed the lowest median absolute errors (0.28 and 0.29 D, respectively). Cooke K6, EVO 2.0, and BUII had the highest MedAE (0.66, 0.60, and 0.55 D, respectively). PEARL-DGS sos and EVO 2.0 sos formulas had the highest cases within ± 0.25 D of the intended refraction (47.37%), followed by BTAL (42.11%). Cooke K6 formula had the least cases within ± 0.25 D of the intended refraction (5.26%) followed by EVO 2.0 and Cooke K6 sos (15.79%). For eyes ≤ 21 mm, Cooke K6 sos, EVO 2.0 sos, and PEARL-DGS sos formulas had 100% of cases within ± 1.0 D of the target refraction. All of the included formulas had 100% of cases within ± 2.0 D of the target refraction for eyes ≤ 21 mm. The ARGOS optical biometer provides a unique method that measures the AXL that uses a sum-of-segments concept or a segmental refractive index instead of the composite one used by most of the other optical biometers. This resulted in a longer AXL in the eyes with short axial length which lead to a lower IOL power . The authors in the current study used the cutoffs for eyes with short axial length of 22.00 mm as reported by Shammas and Jabre . Not all contemporary IOL power calculation formulas provide the option to utilize sum-of-segments axial length (AXLsos). The Cooke K6, EVO 2.0, and PEARL-DGS formulas do offer this option. These formulas can be accessed on their respective websites as well as on the online ESCRS IOL calculator platform. The Barrett Universal II formula features a modified version, known as the BTAL formula, which is integrated into the ARGOS biometer and employs AXLsos. This study aims to evaluate the accuracy of incorporating AXLsos in the aforementioned formulas, comparing results with and without the sum-of-segments option. The authors conducted a back-calculation of the predicted outcomes from various formulas and juxtaposed these with the actual postoperative refraction. The IOL constants utilized were those that are routinely applied in clinical practice to evaluate outcomes in real-world scenarios. In the current study, it was noted that the PEARL-DGS sos was the only formula to show a myopic mean prediction error. All the formulas without SOS correction option (Cooke K6, followed by EVO 2.0, Cooke K6 sos , and BUII formulas had hyperopic mean prediction error. By choosing SOS option or using BTAL, the outcome turned into less hyperopic which is more desirable. This suggests that when less hyperopia or myopia is desired, it is better to choose SOS option or BTAL when using an axial length measured by ARGOS. It is worth mentioned here, that hyperopic shift is an undesirable outcome. Kato et al. evaluated the accuracy of BTAL and EVO formulas using segmental refractive index in comparison to the conventional BUII. They reported that the mean arithmetic error differed significantly among the 3 formulas in eyes with short axial length, BU II giving 0.32 ± 0.40 D, BTAL 0.22 ± 0.37 D, and EVO 0.10 ± 0.37 D ( P < 0.0001). This is similar to the results of the current study, where BUII formula showed hyperopic outcome that decreased with BTAL and EVO 2.0 sos . In the current study, all formulas with SOS option and BTAL performed well, with almost all the cases within ± 1 D of intended refraction. The EVO 2.0 sos , PEARL-DGS sos , and Cooke K6 sos formulas yielded the lowest MedAE in both the whole pool of cases and the subgroup with 21 mm or shorter AXL. Using BTAL or the SOS option in the other included formulas decreased the MedAE in a more evident way in the subgroup of short AXL ≤ 21 mm. Shammas et al. analyzed the accuracy of many newer IOL power formulas using SOS biometry including BUII, BTAL, K6, EVO, and PEARL-DGS. The authors classified eyes with short axial length into 2 groups: short with an AXL ≤ 22.5 mm and very short with an AXL ≤ 22.0 mm. They reported a MedAE for short and very short eyes of 0.31 D and 0.35 D with BU II, 0.30 D and 0.32 D with BTAL, 0.26 D and 0.26 D with Cooke K6, 0.28 D and 0.33 D with EVO, and 0.27 D and 0.27 D with PEARL-DGS, respectively. The current study reported slightly higher values for MedAE for eyes with short axial length ≤ 22.0 mm. The difference could be attributed to the lower mean AXL in this study with more included eyes shorter than 21 mm. Their mean AXL for eyes with short axial length was (22.00 ± 0.38 mm, range from 20.75 to 22.49 mm) versus mean AXL of 20.89 ± 0.67 mm (range from 19.80 to 21.98 mm) in the current study. Miyamoto et al. aimed at verifying the accuracy of BTAL. They included 356 Japanese eyes with mean AXL 23.84 ± 1.16 mm. The MAEs for BTAL and BUII were 0.225 ± 0.179 D and 0.219 ± 0.168 D, respectively. This was less than the reported MAEs for the current study for BTAL and BUII which were 0.42 ± 0.27 D and 0.40 ± 0.29 D, respectively. This is due to the difference between the mean AXL between the 2 studies. Blehm et al. reported that the predictability of ARGOS measurements and the BUII formula in eyes with short axial length implanted with an extended depth of focus IOL was moderate with a prediction error of 0.33 ± 0.33 D. The percentage of eyes in their study with ≤ 0.5 D of MRSE was 74% for eyes with short axial length ≤ 22.5 mm (mean AXL was 22.21 ± 0.24 mm). The current study showed a lower percentage of cases with ± 0.50 D of intended refraction with the BUII (60.61% in the whole cases and 42.11% in the subgroup with shorter eyes ≤ 21.0 mm). Blehm et al. reported a lower MedAE for the BUII (0.27 D) than the current study (0.34 D for the whole cases and 0.55 D for subgroup ≤ 21.0 mm). This is due to the shorter mean AXL in the current study and the more included cases less than 21.0 mm. Blehm et al. in another study compared the refractive predictability of ARGOS measurements with BUII and BTAL formulas in a large sample (445 eyes) of long (≥ 24.5 mm), medium, and short AXL eyes. They included 75 eyes with short axial length (≤ 22.5 mm) with mean AXL of 22.08 ± 0.45 mm. They reported a hyperopic mean arithmetic error for both BUII and BTAL of 0.34 ± 0.48 D and 0.15 ± 0.46 D, respectively. They reported MAE of 0.45 ± 0.37 D and 0.37 ± 0.31 D for BUII and BTAL, respectively. Their results were comparable to our results especially in the less hyperopic BTAL mean arithmetic error in comparison to the BUII formula despite their higher mean AXL. It was noted that both eyes from certain patients were included in the analysis, potentially introducing bias. Additionally, it was indicated that the data examined originated from a single site, which may limit its relevance to other surgeons. It is noteworthy to mention different approaches to evaluate the accuracy of IOL power calculation formulas and to optimize the lens constants. Gatinel et al. concluded that prioritizing standard deviation minimization before adjusting the mean prediction error significantly improved the precision of the selected IOL power calculation formulas, which enhanced postoperative refractive outcomes. Lagenbucher et al. investigated the performance of a simple strategy for formula constant optimization. Stopyra et al. used root mean square absolute error (RMSAE) as a primary outcome. Stopyra W in another study used the agreement interval in Bland–Altman analysis. The current study showed some points of strength, including a large proportion of eyes with short axial length of 21 mm or less. This study presented the results of IOL power calculation formulas, both with and without the option of the SOS even when utilizing the AXLsos measurement. A notable aspect of the study is the simultaneous strength and limitation associated with the use of lens constants without additional optimization. The authors emphasized the necessity of reporting actual clinical practice outcomes based on the constants already employed in the ARGOS machine and available on the ULIB website. Only one IOL model was investigated, this IOL is a spherical model, which is likely to generate positive spherical aberrations (the results may be slightly different with aspheric IOLS). Another possible limitation identified was the absence of comparisons with other contemporary formulas, such as Kane and Hill RBF 3.0, as well as the retrospective design of the study. The authors opted to focus solely on formulas that included the SOS option to evaluate its impact on outcomes with and without this feature. To further improve the accuracy of refractive cataract surgery, advancements in IOL manufacturing technology, such as the introduction of 0.25-D increments, would be beneficial in enhancing postoperative patient satisfaction. In conclusion, PEARL-DGS sos was the only formula to show a myopic mean prediction error. Using BTAL and SOS option in Cooke K6, EVO 2.0, and PEARL-DGS formulas decreased the undesirable hyperopic shift in the mean prediction error. This effect was more evident in shorter eyes ≤ 21.0 mm. |
Internal Limiting Membrane Peeling in Primary Rhegmatogenous Retinal Detachment: Functional and Morphologic Results | cef20726-fc16-413d-8204-d80b6b840854 | 11825206 | Surgical Procedures, Operative[mh] | Pars plana vitrectomy (ppV) currently represents the most popular and successful surgical approach to rhegmatogenous retinal detachment (RRD) . While its use and technical aspects are standardized internationally, a multitude of procedural subtleties based on the individual surgeonʼs discretion have been described. These include, for example, the use of 23-, 25-, or 27-gauge trocars , the use of air or gas tamponades , or the addition of macular surgery. Concerning the latter, peeling of the internal limiting membrane (ILM) represents the most frequently performed addendum in retinal detachment (RD) repair. Nevertheless, its potential risks and benefits are debated controversially. A large benefit of ILM peeling in RRD repair includes the prevention of postoperative fibrotic proliferation on the macular surface. The formation of epiretinal membranes (ERMs) is one of the most common postoperative complications after vitrectomy for RD. In 7 – 15% of vitrectomies for RD, an ERM forms postoperatively , , . The resulting membrane can exert high tensile forces on the underlying retinal layer and alter its anatomy . As a result, visual acuity (VA) gain after surgery is limited, metamorphopsia can occur, and in some cases, reoperation with peeling of the ERMs is required . In the development of ERMs, vitreous cortex remnants (VCRs) play a major role by serving as a scaffold on the ILM for fibrocellular proliferation , , , . Peeling of the ILM results in removal of the VCRs and thus a reduced proliferative stimulus , , , , , . On the other hand, the ILM is a physiologic component of the retinal architecture. Thus, ILM removal also carries some risks. Among these are modified postoperative retinal anatomy with foveal shifting, focal retinal hemorrhage, and dissociated optic nerve fiber layer (DONFL) , , , , . Functionally, paracentral scotomas and reduced central retinal sensitivity may occur , . In addition, removal of the ILM in detached retina is surgically challenging and much more complex compared to an attached retina . The aim of the present study was to compare the outcomes of ILM peeling in proliferative vitreoretinopathy (PVR) B RRD with uncomplicated RRD without ILM peeling concerning VA and retinal reattachment rates. Patient selection For this retrospective cohort study, the Smart Eye Database (SmEyeDat) of the University Hospital was screened for patients with primary RRD repair between January 2020 and May 2023 by a single surgeon. The study adhered to the declaration of Helsinki and local ethics committee approval was obtained. Provided written informed consent for treatment was obtained and analysis was performed after anonymization. Screening included all patients with at least a follow-up of 6 months. We included macular on and off patients. Eyes with a history of trauma, uveitis, proliferative vitreoretinopathy C, or high myopia > − 6 D were excluded. ILM peeling was performed based on the surgeonʼs discretion using Membrane Blue as dye (DORC, Zuidland, Netherlands). Patients were retrospectively divided into two groups. All patients with ILM peeling during RD surgery and PVR B reaction were included in group 1, and patients who did not undergo ILM peeling with no PVR were included in group 2. PVR B reaction was categorized intraoperative by the surgeon. Intraoperative optical coherence tomography (OCT) was performed in every patient and used for grading. Intraoperative OCT was used to rule out a concomitant macular hole. Before and after surgery, each patient underwent a standardized examination including refraction-based VA testing, air-puff noncontact tonometry, slit lamp examination, and fundoscopy. Postoperative OCT was performed using the Heidelberg Spectralis (Heidelberg Engineeringʼs Spectralis HRA + OCT, Heidelberg, Germany) OCT. Clinical data was collected, including age, sex, date of RD surgery, best-corrected visual acuity (BCVA) at each follow-up visit, postoperative ERM formation, retinal redetachment, macular edema, and OCT parameters such as central subfield thickness (CST). Decimal VA was converted to logarithm of the minimum angle resolution (logMAR) units for analysis. Hand motion and finger counting was converted to 2.3 and 1.9 as previously described . Optical coherence tomography Imaging was performed as previously described elsewhere . Spectral-domain OCT was performed using the Spectralis HRA + OCT (Heidelberg Engineering) system, including a volume scan (49 B-scans). Auto segmentation was proofed by a physician. Heidelberg Spectralis SD-OCT (Heidelberg Engineering) was used to evaluate the postoperative presence of ERMs, foveal profile, CST, presence of macular edema, and the integrity of the ellipsoid zone (EZ). Data analysis and statistics Data management was performed with Microsoft Excel Version 16.72 (Microsoft Corp., Redmond, WA, USA) for Mac. For statistical analyses, IBM SPSS Statistics 28 (IBM Germany GmbH, Ehningen, Deutschland) was used. As the data was not normally distributed, nonparametric tests were applied. The differences between the groups were assessed using the Mann-Whitney U test for nominal scaled data and non-normally distributed variables. Continuous variables are expressed as the mean ± standard deviation and categorial variables as frequencies. The significance level was set at p < 0.05. For this retrospective cohort study, the Smart Eye Database (SmEyeDat) of the University Hospital was screened for patients with primary RRD repair between January 2020 and May 2023 by a single surgeon. The study adhered to the declaration of Helsinki and local ethics committee approval was obtained. Provided written informed consent for treatment was obtained and analysis was performed after anonymization. Screening included all patients with at least a follow-up of 6 months. We included macular on and off patients. Eyes with a history of trauma, uveitis, proliferative vitreoretinopathy C, or high myopia > − 6 D were excluded. ILM peeling was performed based on the surgeonʼs discretion using Membrane Blue as dye (DORC, Zuidland, Netherlands). Patients were retrospectively divided into two groups. All patients with ILM peeling during RD surgery and PVR B reaction were included in group 1, and patients who did not undergo ILM peeling with no PVR were included in group 2. PVR B reaction was categorized intraoperative by the surgeon. Intraoperative optical coherence tomography (OCT) was performed in every patient and used for grading. Intraoperative OCT was used to rule out a concomitant macular hole. Before and after surgery, each patient underwent a standardized examination including refraction-based VA testing, air-puff noncontact tonometry, slit lamp examination, and fundoscopy. Postoperative OCT was performed using the Heidelberg Spectralis (Heidelberg Engineeringʼs Spectralis HRA + OCT, Heidelberg, Germany) OCT. Clinical data was collected, including age, sex, date of RD surgery, best-corrected visual acuity (BCVA) at each follow-up visit, postoperative ERM formation, retinal redetachment, macular edema, and OCT parameters such as central subfield thickness (CST). Decimal VA was converted to logarithm of the minimum angle resolution (logMAR) units for analysis. Hand motion and finger counting was converted to 2.3 and 1.9 as previously described . Imaging was performed as previously described elsewhere . Spectral-domain OCT was performed using the Spectralis HRA + OCT (Heidelberg Engineering) system, including a volume scan (49 B-scans). Auto segmentation was proofed by a physician. Heidelberg Spectralis SD-OCT (Heidelberg Engineering) was used to evaluate the postoperative presence of ERMs, foveal profile, CST, presence of macular edema, and the integrity of the ellipsoid zone (EZ). Data management was performed with Microsoft Excel Version 16.72 (Microsoft Corp., Redmond, WA, USA) for Mac. For statistical analyses, IBM SPSS Statistics 28 (IBM Germany GmbH, Ehningen, Deutschland) was used. As the data was not normally distributed, nonparametric tests were applied. The differences between the groups were assessed using the Mann-Whitney U test for nominal scaled data and non-normally distributed variables. Continuous variables are expressed as the mean ± standard deviation and categorial variables as frequencies. The significance level was set at p < 0.05. Baseline demographics In total, 26 patients with 26 eyes were included in our study. All patients with ILM peeling during RD surgery and PVR B reaction were included in group 1 (13 patients), and patients who did not undergo ILM peeling with no PVR were included in group 2 (13 patients). The mean age was similar in group 1 (62.38 ± 8.61 years) and group 2 (63.77 ± 7.98) (p = 0.65). Characteristics of groups 1 and 2, including sex, age, baseline VA, preoperative lens status, and foveal involvement, did not differ significantly within the two groups and are summarized in . Preoperative PVR B features are also listed in . Surgical outcomes All eyes underwent RD surgery with vitrectomy including intraocular tamponade. Primary success rates were 92.3% in groups 1 and 2 (12/13) (p = 1.0). The distribution of intraocular tamponades did not differ significantly between groups and is listed in (p = 0.96). In group 1, five patients had a combined phakovitrectomy surgery, and in group 2, three patients (p = 0.84). All phakic patients received combined phakovitrectomy, except for one 51-year old patient in group 2 who only underwent vitrectomy. Postoperatively, this patient had a slight posterior subcapsular cataract, which, however, was visually insignificant, with a VA of 0.1 logMAR postoperatively. There were no severe intraoperative complications. Mean follow-up period was 555.08 ± 376.62 days in group 1 and 808.31 ± 402.12 days in group 2 in total and did not significantly vary in the two groups (p = 0.13). OCT outcomes Postoperative ERM formation based on OCT was significantly reduced by ILM peeling (p = 0.04). Postoperative ERM formation was found in one patient in group 1 and six patients in group 2. Two patients in group 2 had the need for a second ERM peeling surgery. shows exemplary postoperative OCT scans. The mean CST in group 1 was 339.08 ± 50.14 µm and in group 2 315.62 ± 33.53 µm; there was no significant difference (p = 0.06). A postoperative cystoid macular edema was found in five patients in group 1 and in two patients in group 2 (p = 0.34). Treatment was needed in most patients and included Nevanac eye drops (Novartis AG, Basel, Switzerland), one patient additionally needed peribulbar triamcinolone, and one patient received a dexamethasone implant (Ozurdex, Abbvie, North Chicago, IL, USA). To evaluate the outer retina, the integrity of the EZ was analyzed. In group 1, two patients and in group 2, three patients showed an EZ defect in the OCT scan; there was no significant difference (p = 0.76). OCT outcome parameters are shown in . Visual acuity VA was tested at every visit. The baseline VA was 1.03 ± 0.96 in group 1 and 0.84 ± 0.97 logMAR in group 2; there was no significant difference (p = 0.39). We analyzed VA postoperatively, as well as the VA gain between baseline and postoperative visit ( , , and ). Postoperative VA was 0.26 ± 0.29 in group 1 and 0.15 ± 0.17 in group 2 (p = 0.125). The VA gain between the baseline visit and postoperative follow-up visit in group 1 was − 0.77 ± 0.97 compared to − 0.7 ± 0.89 in group 2 (p = 0.920). In total, 26 patients with 26 eyes were included in our study. All patients with ILM peeling during RD surgery and PVR B reaction were included in group 1 (13 patients), and patients who did not undergo ILM peeling with no PVR were included in group 2 (13 patients). The mean age was similar in group 1 (62.38 ± 8.61 years) and group 2 (63.77 ± 7.98) (p = 0.65). Characteristics of groups 1 and 2, including sex, age, baseline VA, preoperative lens status, and foveal involvement, did not differ significantly within the two groups and are summarized in . Preoperative PVR B features are also listed in . All eyes underwent RD surgery with vitrectomy including intraocular tamponade. Primary success rates were 92.3% in groups 1 and 2 (12/13) (p = 1.0). The distribution of intraocular tamponades did not differ significantly between groups and is listed in (p = 0.96). In group 1, five patients had a combined phakovitrectomy surgery, and in group 2, three patients (p = 0.84). All phakic patients received combined phakovitrectomy, except for one 51-year old patient in group 2 who only underwent vitrectomy. Postoperatively, this patient had a slight posterior subcapsular cataract, which, however, was visually insignificant, with a VA of 0.1 logMAR postoperatively. There were no severe intraoperative complications. Mean follow-up period was 555.08 ± 376.62 days in group 1 and 808.31 ± 402.12 days in group 2 in total and did not significantly vary in the two groups (p = 0.13). Postoperative ERM formation based on OCT was significantly reduced by ILM peeling (p = 0.04). Postoperative ERM formation was found in one patient in group 1 and six patients in group 2. Two patients in group 2 had the need for a second ERM peeling surgery. shows exemplary postoperative OCT scans. The mean CST in group 1 was 339.08 ± 50.14 µm and in group 2 315.62 ± 33.53 µm; there was no significant difference (p = 0.06). A postoperative cystoid macular edema was found in five patients in group 1 and in two patients in group 2 (p = 0.34). Treatment was needed in most patients and included Nevanac eye drops (Novartis AG, Basel, Switzerland), one patient additionally needed peribulbar triamcinolone, and one patient received a dexamethasone implant (Ozurdex, Abbvie, North Chicago, IL, USA). To evaluate the outer retina, the integrity of the EZ was analyzed. In group 1, two patients and in group 2, three patients showed an EZ defect in the OCT scan; there was no significant difference (p = 0.76). OCT outcome parameters are shown in . VA was tested at every visit. The baseline VA was 1.03 ± 0.96 in group 1 and 0.84 ± 0.97 logMAR in group 2; there was no significant difference (p = 0.39). We analyzed VA postoperatively, as well as the VA gain between baseline and postoperative visit ( , , and ). Postoperative VA was 0.26 ± 0.29 in group 1 and 0.15 ± 0.17 in group 2 (p = 0.125). The VA gain between the baseline visit and postoperative follow-up visit in group 1 was − 0.77 ± 0.97 compared to − 0.7 ± 0.89 in group 2 (p = 0.920). Our study shows that, in skilled hands, additional ILM peeling in RRD with PVR B does not negatively influence VA outcomes or anatomical success, while it dramatically reduces the postoperative formation of ERMs, which, in the long term, reduces the burden of reoperation and VA decline in patients with successful RD repair. Moreover, no negative impact on ILM peeling on OCT biomarkers demonstrating a healthy retinal architecture was found, except DONFL, which was seen in 38.5% of patients in group 1; however, VA did not significantly differ between our groups. Concerning clinical decision-making, our study adds evidence supporting ILM peeling in cases with primary RD with PVR B reaction. It seems to dramatically reduce the redetachment rate. While redetachment in PVR B cases has been described to be up to 34% , we only found a redetachment in 7.7% of our ILM peeled PVR B cases. We thereby could achieve a reduction of the redetachment rate to the levels of uncomplicated RDs. As our data suggest, this benefit is not outweighed by potential anatomical damage, for example, foveal shifting, focal retinal hemorrhage, and DONFL , , , , . Concerning foveal shifting, the retina itself is already displaced due to the RD, which probably renders this effect secondary to ILM peeling less important. Many retinal surgeons even advocate performing ILM peeling in patients with foveal splitting, a special anatomical configuration where the demarcation line of the RD goes directly through the fovea and can cause severe postoperative metamorphopsia . Besides, potential negative effects of ILM peeling, such as paracentral scotomas and reduced central retinal sensitivity, may also be caused by the RD itself , . The efficacy of simultaneous ILM peeling with removal of the ERM to prevent recurrence has been demonstrated in a number of studies in recent years. However, most of these studies refer to idiopathic and not secondary ERMs . Currently, there is controversy regarding the potential beneficial effects of ILM peeling during RD surgery other than ERM recurrence reduction . Based on recent literature, many authors recommend ILM peeling only for patients with complicated RD , . As a limitation, our study provides primary outcomes at follow-ups of 6 months. Although our follow-up is relatively short, most ERM formations are detected within the first 3 months postoperatively , , , . So with a mean follow-up of 681.69 ± 389.94 days, we should have detected most of the ERM formations. Metamorphopsia was, unfortunately, not recorded in all patients, so it was omitted from the analysis. However, a large percentage of ERMs remain asymptomatic and OCT imaging provides a sensitive diagnosis , . As a caveat, longer-term retinal changes secondary to any intervention involving the ILM might only be obvious after many months or years. For example, Klaas et al. recently suggested that structural changes, e.g., foveal asymmetry secondary to posterior vitreous detachment, can cause a domino effect that can cause lamellar or full-thickness macular holes after a long period of 6 – 27 months . This can also apply to ILM removal, which may cause substantial histologic damage to the Mueller cells within the ILM. The mechanical stress of the basal lamina caused by peeling can be transmitted to deeper retinal layers and lead to modifications of retinal structures . Lamas-Francis et al. performed a meta-analysis in 2023 examining studies with ILM peeling in RD surgery, finding no significant difference in terms of VA outcome depending on the peeling condition. However, in terms of redetachment rate, a significant reduction was found in patients with ILM peeling . Several other studies also demonstrated a reduction in ERM formation with ILM peeling but found no significant difference in postoperative VA , , . Obata et al. even observed poorer VA in patients with ILM peeling and macula-OFF RD . Eissa et al. found a worse postoperative VA and macular sensitivity in patients with ILM removal. Additionally, they described retinal pits in all patients with ILM peeling, which they explained by more mechanical trauma during surgery. Nevertheless, there was no correlation of dimples with the VA or macular sensitivity . However, there are further studies that demonstrated a better visual outcome after ILM peeling , . Nam and Kim described a better postoperative VA in macula-ON patients when ILM peeling was performed. They attribute this to the prevention of ERM formation, which could reduce VA postoperatively . Although ILM peeling might not lead to a better postoperative VA, it can lead to better anatomical success and macular compliance by decreasing the contraction forces in PVR retinal redetachment . Consistent with the studies mentioned above, postoperative VA was not significantly worse in patients who underwent ILM peeling than in patients without peeling in our study. Regarding the redetachment rate, we could not find any difference between the two groups. Additionally, our study is limited by the retrospective design, the relatively small sample size, and the surgeonʼs bias of selecting patients for ILM Peeling. In summary, we found that ILM peeling in primary rhegmatogenous ablatio with PVR B can significantly reduce the risk of ERMs in high-risk patients and improve their prognosis comparable to uncomplicated RRD. No negative effects on functional outcome were detected. Further studies with larger cohorts, longer follow-up, and additional functional testing (e.g., microperimetry) are needed to investigate the long-term impact of ILM peeling on visual function and retinal redetachment. Conclusion Already known: The formation of ERMs is one of the most common postoperative complications after vitrectomy for RD. The benefits of ILM peeling for maculopathies such as a macular hole or advanced macular pucker. Newly described: This study suggests that intraoperative removal of the ILM in PVR B RRD can improve functional and morphological outcomes to levels obtained in uncomplicated RRD without PVR. ILM peeling does not appear to negatively affect postoperative VA. |
Does alveolar ridge preservation reduce the need for sinus floor elevation: A comparative study to spontaneous healing | ca6533ec-9f92-48dc-8bcd-e492604616e7 | 11843591 | Surgical Procedures, Operative[mh] | INTRODUCTION Numerous studies and reviews have shown that tooth extractions cause changes in alveolar bone morphology with structural and dimensional adjustments. , , , Bone resorptions were observed 6 months after tooth removal, with horizontal dimensional reduction of 29%–63% and vertical dimensional reduction of 11%–22%, while two‐thirds of the reductions occurred within the first 3 months. After tooth extraction, the bundle bone that lines the periodontal ligament around the tooth and anchors the sharpey fibers in the bone is resorbed and replaced with woven bone, causing a vertical reduction of the alveolar crest. Subsequently reabsorption occurs from the outer surfaces of both bone walls. Additionally Araujo and colleagues reported a larger ridge reduction in the molar than in the front region. Furthermore, a process called pneumatization takes place in the maxillary sinus. It describes a physiological process of the maxillary sinus characterized by the expansion of the sinus over time. Several factors have been linked to its occurrence, including bone density, respiratory air pressure within the sinus in combination with extraction of posterior teeth, heredity, and previous sinus surgery. It is likely that the sinus membrane will shift more coronally and occupy the space of the extraction socket if there is no or only a thin bundle bone between the sinus and the root apex as the extraction socket cannot resist the pressure from the sinus. , Depending upon individual local and systematic factors, the degree of bone remodeling may vary but generally results in both horizontal and vertical irreversible alveolar ridge reduction. The loss of bone and the susceptibility of the maxillary sinus to pneumatization after a tooth extraction may limit the subsequent placement of an implant and the preservation of the bony structures is a prerequisite for a functional and aesthetic result. Implants are advised to be provided with bone walls about 1‐ to 2‐mm wide on buccal and lingual aspects in order to maintain a stable bone height. Thus, many authors suggest preserving the alveolar ridge by using various bone grafts, such as bovine‐derived xenografts, autologous bone, allografts, and alloplasts which promote hard tissue formation or proposed the use of different types of implant placement. , A variety of successful procedures have been reported for maintaining the anatomical dimensions of the alveolar ridge and previous data indicated that alveolar ridge preservation (ARP) could prevent the pneumatization process and therefore reduce the need for subsequent sinus augmentation. , , Some authors demonstrated that compared to non‐grafted sites, the profile of ridges treated by means of ridge preservation is better preserved than those without grafts and previous systematic reviews have shown solid evidence supporting their success. , , However, the treatment could neither completely prevent volume nor the contour alterations. The aim of this study was to investigate the influence of the alveolar ridge preservation after tooth extraction in the posterior maxilla with a bovine bone substitute and a collagen membrane on the need for sinus lift augmentation compared to spontaneous healing. The primary outcome criterion is the avoidance of an internal or external sinus lift. Secondarily, the vertical bone loss in the area of the extraction socket was measured. Furthermore, the success rate of the inserter implant in this study is investigated. MATERIALS AND METHODS 2.1 Study design and population This study was designed as a prospective non‐randomized controlled clinical trial with two parallel groups. The study was approved by the ethical commission relevant state medical chambers (No. 202‐15160). The CONSORT flowchart for non‐randomized controlled clinical trial is presented in Figure . This study was performed from July 8, 2021 to January 22, 2024. All the patients were older than 18 years and provided informed consent for the surgery. Patients were recruited from the patient pool of the Department of Oral and Maxillofacial surgery—plastic surgery of the University. The study protocol was approved by the Ethics Committee of the University Medical Centre under the number: 2020‐15160. Patients with posterior teeth or roots in the posterior maxilla candidate to extraction were allocated into two groups. Allocation was performed by assigning the first 22 patients, with a total of 31 teeth to be extracted, to the test group, while the subsequent 18 patients were assigned to the control group. Allocation was performed by assigning the first 22 patients, with a total of 31 teeth to be extracted, to the test group, while the subsequent 18 patients, with a total of 22 teeth to be extracted, were assigned to the control group. In some patients, more than one tooth was extracted, but no more than three teeth were extracted from a patient. An implant was placed in every region where a tooth was removed. The study was designed as a controlled clinical trial. Patients of test group were treated by alveolar ridge preservation using bovine bone grafting material (Straumann® XenoFlex) and porcine collagen membrane (Straumann® Jason® membrane). In patients of the control group, no grafting was performed. After 6 months of healing Cone‐beam computed tomography's (CBCT) were taken, for implant planning. The necessity of sinus floor elevation between the test and control group was determined depending on the extent of available vertical bone before implant placement (treatment modalities for implant placement) and changes in vertical bone dimensions were measured radiologically and compared at baseline and after implant surgery. Implant survival, success rates, and radiographic changes at the crestal bone level by peri‐apical x‐ray and panoramic tomography were evaluated. 2.2 Inclusion criteria The following inclusion criteria were applied: Patients aged ≥18 years. Patients requiring extraction of maxillary posterior teeth (molars, first premolars and second premolars). Patients willing to participate in this study and no systematic or local conditions that would preclude them from implant therapy. Patients who were healthy. No systematic or local conditions presenting a contraindication to extraction and implant placement. 2.3 Exclusion criteria The following exclusion criteria were applied: General contraindication to implant surgery. Extraction of more than three adjacent teeth. Pregnant or nursing women. Uncontrolled diabetes. History of oromaxillofacial radiation therapy. Smoking. Osteogenesis‐related diseases or medications affecting bone formation. Pathologic conditions of the maxillary sinus. Untreated severe periodontitis with poor oral hygiene. Teeth with symptoms of an acute infection such as tapping sensitivity or in conjunction with an abscess. 2.4 Handling of correlated data Among 40 patients, 53 posterior teeth were extracted. The test group included 22 patients with a total of 31 teeth to be extracted, while the control group consisted of 18 patients with 22 teeth to be extracted. 2.5 Treatment modalities for implant placement Implants with a diameter of 3.5, 3.75, 4, and 4.5 mm and length of 6, 8, 10, and 12 mm were used depending on the bone level. The treatment strategies for implant surgery and the need for sinus lifts were prepared according to the residual bone height (RBH), assessed from CBCT images taken 6 months after extraction. The following protocol was used for further patient management: ≥6 mm residual vertical bone height: placement of Straumann® BLX Implant 3–5 mm residual vertical bone height: internal sinus lift with simultaneous implant insertion 0–3 mm residual vertical bone height: external sinus lift with two‐staged implant insertion after 4 months 2.6 Treatment procedure 2.6.1 Tooth extraction Extractions of the tooth which had to be removed were performed as atraumatically as possible using periotomes, elevators, and extraction force. After tooth extraction, the socket was carefully curetted, and all granulation tissue was removed (Figures , , , and ). If the patient belonged to the control group, the wound edges of the alveolus were adapted after sufficient coagulation with the help of adaptation sutures using non‐resorbable suture material Ethilon™ (5‐0, Ethicon®, Germany). In the test group, the bovine bone graft substitute was placed directly into the alveolus using an applicator and covered with a resorbable membrane (Figures and ). Perioperatively, the patients in the test group received an oral administration of prophylactic antibiotic therapy of 2.0 g Amoxicillin/Clavulanic Acid (875 mg/125 mg) 1 h before surgery. No attempt was made to augment the ridge vertically above the height of the crest. The wound margins were covered over the membrane with non‐resorbable suture material Ethilon™ (5‐0, Ethicon®, Germany). Primary wound closure was not attempted. For both groups, Valsalva maneuver was performed to check whether Schneiderian membrane was damaged. A recall appointment was scheduled 7–14 days after the surgery for suture removal. The membrane barrier was left in place in all cases. 2.6.2 Evaluations after the first surgery (tooth extraction) All patients were recalled evaluating the wound healing. A CBCT was conducted 6 month‐post‐operatively for the purpose of planning implant placement. 2.6.3 Implant insertion All patients received an oral administration of prophylactic antibiotic therapy of 2.0 g Amoxicillin/Clavulanic Acid (875 mg/125 mg) 1 h before surgery. In case of allergy, Clindamycin 600 mg 1 h before surgery was given. Individualized oral hygiene instructions were provided prior to any surgical intervention and patients had to rinse with 0.2% chlorhexidine solution for 1 min. Conventional crestal incisions were applied under local anesthesia (Ultracain® adrenaline 1:200,000, Septodont, Niederkassel, Germany), followed by the reflect of full thickness soft tissue flap in healed ridge. Sequential osteotomy and implant insertion is performed according to the manufacture's guidelines (Straumann® BLX, Institut Straumann AG, Switzerland) (Figures , , , and ). In case of insufficient vertical bone height as shown in the treatment modalities for implant placement, a sinus lift procedure was performed. Mid crestal incision was made with mesial vertical releasing incision. The internal sinus floor elevation began with the crestal incision and the preparation of the soft tissue. The crestal approach was performed through the pilot hole until just before the Schneiderian membrane. By using condensers and osteotomes, which contribute both to the compaction of the bone and to the targeted perforation and elevation of the sinus floor, the sinus floor could be elevated by a few millimeters so that the implant could be inserted at the desired length. For the external sinus floor elevation, a small window was created on the side of the maxillary sinus using a round diamond bur under copious saline irrigation. The Schneiderian membrane was carefully elevated using different sinus elevation curettes. Following complete elevation of the membrane, the sinus cavity was augmented by bovine bone grafting material which was mixed with saline taking care not to perforate the sinus membrane during packing of the graft. Wound closure was achieved using resorbable suture Vicryl™ (4‐0, Ethicon®, Norderstedt, Germany). All implants were prosthetically restored (Figures and ). Follow‐up of the restored implant took place 1 year postoperatively. Patient from the test group Patient from the control group Study design and population This study was designed as a prospective non‐randomized controlled clinical trial with two parallel groups. The study was approved by the ethical commission relevant state medical chambers (No. 202‐15160). The CONSORT flowchart for non‐randomized controlled clinical trial is presented in Figure . This study was performed from July 8, 2021 to January 22, 2024. All the patients were older than 18 years and provided informed consent for the surgery. Patients were recruited from the patient pool of the Department of Oral and Maxillofacial surgery—plastic surgery of the University. The study protocol was approved by the Ethics Committee of the University Medical Centre under the number: 2020‐15160. Patients with posterior teeth or roots in the posterior maxilla candidate to extraction were allocated into two groups. Allocation was performed by assigning the first 22 patients, with a total of 31 teeth to be extracted, to the test group, while the subsequent 18 patients were assigned to the control group. Allocation was performed by assigning the first 22 patients, with a total of 31 teeth to be extracted, to the test group, while the subsequent 18 patients, with a total of 22 teeth to be extracted, were assigned to the control group. In some patients, more than one tooth was extracted, but no more than three teeth were extracted from a patient. An implant was placed in every region where a tooth was removed. The study was designed as a controlled clinical trial. Patients of test group were treated by alveolar ridge preservation using bovine bone grafting material (Straumann® XenoFlex) and porcine collagen membrane (Straumann® Jason® membrane). In patients of the control group, no grafting was performed. After 6 months of healing Cone‐beam computed tomography's (CBCT) were taken, for implant planning. The necessity of sinus floor elevation between the test and control group was determined depending on the extent of available vertical bone before implant placement (treatment modalities for implant placement) and changes in vertical bone dimensions were measured radiologically and compared at baseline and after implant surgery. Implant survival, success rates, and radiographic changes at the crestal bone level by peri‐apical x‐ray and panoramic tomography were evaluated. Inclusion criteria The following inclusion criteria were applied: Patients aged ≥18 years. Patients requiring extraction of maxillary posterior teeth (molars, first premolars and second premolars). Patients willing to participate in this study and no systematic or local conditions that would preclude them from implant therapy. Patients who were healthy. No systematic or local conditions presenting a contraindication to extraction and implant placement. Exclusion criteria The following exclusion criteria were applied: General contraindication to implant surgery. Extraction of more than three adjacent teeth. Pregnant or nursing women. Uncontrolled diabetes. History of oromaxillofacial radiation therapy. Smoking. Osteogenesis‐related diseases or medications affecting bone formation. Pathologic conditions of the maxillary sinus. Untreated severe periodontitis with poor oral hygiene. Teeth with symptoms of an acute infection such as tapping sensitivity or in conjunction with an abscess. Handling of correlated data Among 40 patients, 53 posterior teeth were extracted. The test group included 22 patients with a total of 31 teeth to be extracted, while the control group consisted of 18 patients with 22 teeth to be extracted. Treatment modalities for implant placement Implants with a diameter of 3.5, 3.75, 4, and 4.5 mm and length of 6, 8, 10, and 12 mm were used depending on the bone level. The treatment strategies for implant surgery and the need for sinus lifts were prepared according to the residual bone height (RBH), assessed from CBCT images taken 6 months after extraction. The following protocol was used for further patient management: ≥6 mm residual vertical bone height: placement of Straumann® BLX Implant 3–5 mm residual vertical bone height: internal sinus lift with simultaneous implant insertion 0–3 mm residual vertical bone height: external sinus lift with two‐staged implant insertion after 4 months Treatment procedure 2.6.1 Tooth extraction Extractions of the tooth which had to be removed were performed as atraumatically as possible using periotomes, elevators, and extraction force. After tooth extraction, the socket was carefully curetted, and all granulation tissue was removed (Figures , , , and ). If the patient belonged to the control group, the wound edges of the alveolus were adapted after sufficient coagulation with the help of adaptation sutures using non‐resorbable suture material Ethilon™ (5‐0, Ethicon®, Germany). In the test group, the bovine bone graft substitute was placed directly into the alveolus using an applicator and covered with a resorbable membrane (Figures and ). Perioperatively, the patients in the test group received an oral administration of prophylactic antibiotic therapy of 2.0 g Amoxicillin/Clavulanic Acid (875 mg/125 mg) 1 h before surgery. No attempt was made to augment the ridge vertically above the height of the crest. The wound margins were covered over the membrane with non‐resorbable suture material Ethilon™ (5‐0, Ethicon®, Germany). Primary wound closure was not attempted. For both groups, Valsalva maneuver was performed to check whether Schneiderian membrane was damaged. A recall appointment was scheduled 7–14 days after the surgery for suture removal. The membrane barrier was left in place in all cases. 2.6.2 Evaluations after the first surgery (tooth extraction) All patients were recalled evaluating the wound healing. A CBCT was conducted 6 month‐post‐operatively for the purpose of planning implant placement. 2.6.3 Implant insertion All patients received an oral administration of prophylactic antibiotic therapy of 2.0 g Amoxicillin/Clavulanic Acid (875 mg/125 mg) 1 h before surgery. In case of allergy, Clindamycin 600 mg 1 h before surgery was given. Individualized oral hygiene instructions were provided prior to any surgical intervention and patients had to rinse with 0.2% chlorhexidine solution for 1 min. Conventional crestal incisions were applied under local anesthesia (Ultracain® adrenaline 1:200,000, Septodont, Niederkassel, Germany), followed by the reflect of full thickness soft tissue flap in healed ridge. Sequential osteotomy and implant insertion is performed according to the manufacture's guidelines (Straumann® BLX, Institut Straumann AG, Switzerland) (Figures , , , and ). In case of insufficient vertical bone height as shown in the treatment modalities for implant placement, a sinus lift procedure was performed. Mid crestal incision was made with mesial vertical releasing incision. The internal sinus floor elevation began with the crestal incision and the preparation of the soft tissue. The crestal approach was performed through the pilot hole until just before the Schneiderian membrane. By using condensers and osteotomes, which contribute both to the compaction of the bone and to the targeted perforation and elevation of the sinus floor, the sinus floor could be elevated by a few millimeters so that the implant could be inserted at the desired length. For the external sinus floor elevation, a small window was created on the side of the maxillary sinus using a round diamond bur under copious saline irrigation. The Schneiderian membrane was carefully elevated using different sinus elevation curettes. Following complete elevation of the membrane, the sinus cavity was augmented by bovine bone grafting material which was mixed with saline taking care not to perforate the sinus membrane during packing of the graft. Wound closure was achieved using resorbable suture Vicryl™ (4‐0, Ethicon®, Norderstedt, Germany). All implants were prosthetically restored (Figures and ). Follow‐up of the restored implant took place 1 year postoperatively. Patient from the test group Patient from the control group Tooth extraction Extractions of the tooth which had to be removed were performed as atraumatically as possible using periotomes, elevators, and extraction force. After tooth extraction, the socket was carefully curetted, and all granulation tissue was removed (Figures , , , and ). If the patient belonged to the control group, the wound edges of the alveolus were adapted after sufficient coagulation with the help of adaptation sutures using non‐resorbable suture material Ethilon™ (5‐0, Ethicon®, Germany). In the test group, the bovine bone graft substitute was placed directly into the alveolus using an applicator and covered with a resorbable membrane (Figures and ). Perioperatively, the patients in the test group received an oral administration of prophylactic antibiotic therapy of 2.0 g Amoxicillin/Clavulanic Acid (875 mg/125 mg) 1 h before surgery. No attempt was made to augment the ridge vertically above the height of the crest. The wound margins were covered over the membrane with non‐resorbable suture material Ethilon™ (5‐0, Ethicon®, Germany). Primary wound closure was not attempted. For both groups, Valsalva maneuver was performed to check whether Schneiderian membrane was damaged. A recall appointment was scheduled 7–14 days after the surgery for suture removal. The membrane barrier was left in place in all cases. Evaluations after the first surgery (tooth extraction) All patients were recalled evaluating the wound healing. A CBCT was conducted 6 month‐post‐operatively for the purpose of planning implant placement. Implant insertion All patients received an oral administration of prophylactic antibiotic therapy of 2.0 g Amoxicillin/Clavulanic Acid (875 mg/125 mg) 1 h before surgery. In case of allergy, Clindamycin 600 mg 1 h before surgery was given. Individualized oral hygiene instructions were provided prior to any surgical intervention and patients had to rinse with 0.2% chlorhexidine solution for 1 min. Conventional crestal incisions were applied under local anesthesia (Ultracain® adrenaline 1:200,000, Septodont, Niederkassel, Germany), followed by the reflect of full thickness soft tissue flap in healed ridge. Sequential osteotomy and implant insertion is performed according to the manufacture's guidelines (Straumann® BLX, Institut Straumann AG, Switzerland) (Figures , , , and ). In case of insufficient vertical bone height as shown in the treatment modalities for implant placement, a sinus lift procedure was performed. Mid crestal incision was made with mesial vertical releasing incision. The internal sinus floor elevation began with the crestal incision and the preparation of the soft tissue. The crestal approach was performed through the pilot hole until just before the Schneiderian membrane. By using condensers and osteotomes, which contribute both to the compaction of the bone and to the targeted perforation and elevation of the sinus floor, the sinus floor could be elevated by a few millimeters so that the implant could be inserted at the desired length. For the external sinus floor elevation, a small window was created on the side of the maxillary sinus using a round diamond bur under copious saline irrigation. The Schneiderian membrane was carefully elevated using different sinus elevation curettes. Following complete elevation of the membrane, the sinus cavity was augmented by bovine bone grafting material which was mixed with saline taking care not to perforate the sinus membrane during packing of the graft. Wound closure was achieved using resorbable suture Vicryl™ (4‐0, Ethicon®, Norderstedt, Germany). All implants were prosthetically restored (Figures and ). Follow‐up of the restored implant took place 1 year postoperatively. Patient from the test group Patient from the control group OUTCOME MEASUREMENTS 3.1 Radiographic analysis To evaluate the vertical bone height, x‐ray images were examined. However, this examination is limited due to the low resolution and the fact that the 3‐dimensional event is reduced to a 2‐dimensional image. For the presented study, both panoramic slice images and intraoral dental radiographs were taken and evaluated. Compared to the intraoral dental radiographs, the panoramic radiograph has a lower level of detail reproduction. Due to the known implant dimension, it was possible to correct the measured values. For this purpose, the implant length shown from the abutment to the apex was measured on the computer and compared with the length specified by the manufacturer and adjusted. The vertical bone height was then measured according to this standard, which made it possible to correct the typical image distortion errors. The radiograph taken before the tooth was extracted was used as the basis for the comparison and the bone level was determined mesially and distally of the tooth. The radiograph directly after insertion of the implant was then used as the comparison image and the bone level was determined mesially and distally of the implant. The mean values of the mesial and distal vertical bone height were then determined preoperatively and postoperatively and compared with each other to assess whether the bone height had increased, decreased, or remained the same. It is to be noted that for the present study, only vertical height changes could be analyzed. Because some cases ( n = 11) underwent sinus floor elevations prior or in combination to implant placement and thus gained new vertical bone height, these cases were excluded from the analysis of the radiographic examination to determine bone resorption. Thus, the analysis focused exclusively on those cases who did not require further augmentation procedures. All measurements were performed by the same single, experienced investigator. 3.2 Success criteria of the inserted implant In 1990, the success criteria, which adhered to the definition outlined by Buser and colleagues, were described as follows: The implant is in situ. No mobility of the implant. No peri‐implant infection with putrid secretion. No persistent peri‐implant radiotranslucency. No persistent discomfort such as pain, foreign body sensation, or dyesthesia. 3.3 Statistical analysis The statistical analyses were performed using the SPSS program, version 27 (IBM®, USA) for the statistical analysis of the data. The study is descriptive and exploratory in nature with the primary initial hypothesis that ridge preservation prevents the need for sinus floor elevation. Means, standard deviations and confidence intervals were calculated for each factor. Descriptive statistical analysis involved determining mean values and standard deviations for continuous variables and evaluating frequency distributions for categorical variables. Normality of continuous variables was checked using Shapiro–Wilk tests. p ‐Values were derived using t ‐tests or Mann–Whitney U test for continuous variables that followed an apparently normal distribution, Mann–Whitney U tests for continuous variables not normally distributed, and Chi‐square tests with Fischer's exact test for categorical variables. All p ‐values were two‐sided. Box plots and bar charts were created to graphically illustrate differences. The alpha error was set at 0.05. Radiographic analysis To evaluate the vertical bone height, x‐ray images were examined. However, this examination is limited due to the low resolution and the fact that the 3‐dimensional event is reduced to a 2‐dimensional image. For the presented study, both panoramic slice images and intraoral dental radiographs were taken and evaluated. Compared to the intraoral dental radiographs, the panoramic radiograph has a lower level of detail reproduction. Due to the known implant dimension, it was possible to correct the measured values. For this purpose, the implant length shown from the abutment to the apex was measured on the computer and compared with the length specified by the manufacturer and adjusted. The vertical bone height was then measured according to this standard, which made it possible to correct the typical image distortion errors. The radiograph taken before the tooth was extracted was used as the basis for the comparison and the bone level was determined mesially and distally of the tooth. The radiograph directly after insertion of the implant was then used as the comparison image and the bone level was determined mesially and distally of the implant. The mean values of the mesial and distal vertical bone height were then determined preoperatively and postoperatively and compared with each other to assess whether the bone height had increased, decreased, or remained the same. It is to be noted that for the present study, only vertical height changes could be analyzed. Because some cases ( n = 11) underwent sinus floor elevations prior or in combination to implant placement and thus gained new vertical bone height, these cases were excluded from the analysis of the radiographic examination to determine bone resorption. Thus, the analysis focused exclusively on those cases who did not require further augmentation procedures. All measurements were performed by the same single, experienced investigator. Success criteria of the inserted implant In 1990, the success criteria, which adhered to the definition outlined by Buser and colleagues, were described as follows: The implant is in situ. No mobility of the implant. No peri‐implant infection with putrid secretion. No persistent peri‐implant radiotranslucency. No persistent discomfort such as pain, foreign body sensation, or dyesthesia. Statistical analysis The statistical analyses were performed using the SPSS program, version 27 (IBM®, USA) for the statistical analysis of the data. The study is descriptive and exploratory in nature with the primary initial hypothesis that ridge preservation prevents the need for sinus floor elevation. Means, standard deviations and confidence intervals were calculated for each factor. Descriptive statistical analysis involved determining mean values and standard deviations for continuous variables and evaluating frequency distributions for categorical variables. Normality of continuous variables was checked using Shapiro–Wilk tests. p ‐Values were derived using t ‐tests or Mann–Whitney U test for continuous variables that followed an apparently normal distribution, Mann–Whitney U tests for continuous variables not normally distributed, and Chi‐square tests with Fischer's exact test for categorical variables. All p ‐values were two‐sided. Box plots and bar charts were created to graphically illustrate differences. The alpha error was set at 0.05. RESULTS 4.1 Participants and clinical findings The outcome measurements and characteristics of examined cases are presented in Tables and . Out of 40 patients, 53 posterior teeth were extracted. The test group included 22 patients with a total of 31 extracted teeth, while the control group included 18 patients, with a total of 22 extracted teeth. In some cases ( n = 11), more than one tooth was extracted. In none of the cases a perforation of the Schneiderian membrane was observed during the extraction, the ARP, and the sinus lift. No infection in both groups, neither in the group with the exposed collagen membrane was observed. 4.2 Sinus lift augmentation Figure visualizes the number of cases where sinus lifts were needed. A total of 11 sinus floor elevations were performed: 10 internal and one external. In the control group, seven sinus floor elevations were performed, while only four were performed in the study group. The only external sinus floor elevation was performed in the control group. Less sinus lifts were performed in the test group (odds ratio 0.32; 95% CI: 0.08, 1.26). However, the Alveolar Ridge Preservation using bovine bone substitute material was thus unable to prevent the need for sinus floor elevation, as no statistically significant differences in the number of sinus floor elevations performed could be determined between the control group and the test group ( p = 0.168, Table ). 4.3 Implant survival All the surgeries related to implant insertion had an uneventful healing without any complication. All inserted implants in the test group as well as in the control group were as expected osseointegrated and received, after an average healing period of 5 months the definitive crowns, with one exception in the test group where one implant was early lost, because of failed osseointegration. The survival rate of the implants in the control group was 100%, while it was 96.77% in the test group (Table ). According to the criteria of Buser and colleagues, a success rate of 100% was determined for the implants in the control group. In the test group, however, the success rate was 96.77% (Table ). 4.4 Radiographic analysis In the control group, the mean value of the radiographically measured bone height in (mesial and distal) was 11.13 ± 2.12 mm preoperatively before tooth extraction, while it was 11.3 ± 2.17 mm postoperatively after implant placement. In contrast, the mean value in the test group was 11.78 ± 3.09 mm preoperatively and 11.92 ± 2.79 mm postoperatively (Table and Figure ). The analysis does not include the regions which required a sinus floor elevation prior to implant placement. Thus, the analysis refers to 15 teeth from the control group and 27 teeth from the test group. A t ‐test demonstrated that there was no significant difference in the changes in pre‐ and postoperative bone height between the control group and the test group (95% CI: −1.01, 1.08; p = 0.951, Table ). Participants and clinical findings The outcome measurements and characteristics of examined cases are presented in Tables and . Out of 40 patients, 53 posterior teeth were extracted. The test group included 22 patients with a total of 31 extracted teeth, while the control group included 18 patients, with a total of 22 extracted teeth. In some cases ( n = 11), more than one tooth was extracted. In none of the cases a perforation of the Schneiderian membrane was observed during the extraction, the ARP, and the sinus lift. No infection in both groups, neither in the group with the exposed collagen membrane was observed. Sinus lift augmentation Figure visualizes the number of cases where sinus lifts were needed. A total of 11 sinus floor elevations were performed: 10 internal and one external. In the control group, seven sinus floor elevations were performed, while only four were performed in the study group. The only external sinus floor elevation was performed in the control group. Less sinus lifts were performed in the test group (odds ratio 0.32; 95% CI: 0.08, 1.26). However, the Alveolar Ridge Preservation using bovine bone substitute material was thus unable to prevent the need for sinus floor elevation, as no statistically significant differences in the number of sinus floor elevations performed could be determined between the control group and the test group ( p = 0.168, Table ). Implant survival All the surgeries related to implant insertion had an uneventful healing without any complication. All inserted implants in the test group as well as in the control group were as expected osseointegrated and received, after an average healing period of 5 months the definitive crowns, with one exception in the test group where one implant was early lost, because of failed osseointegration. The survival rate of the implants in the control group was 100%, while it was 96.77% in the test group (Table ). According to the criteria of Buser and colleagues, a success rate of 100% was determined for the implants in the control group. In the test group, however, the success rate was 96.77% (Table ). Radiographic analysis In the control group, the mean value of the radiographically measured bone height in (mesial and distal) was 11.13 ± 2.12 mm preoperatively before tooth extraction, while it was 11.3 ± 2.17 mm postoperatively after implant placement. In contrast, the mean value in the test group was 11.78 ± 3.09 mm preoperatively and 11.92 ± 2.79 mm postoperatively (Table and Figure ). The analysis does not include the regions which required a sinus floor elevation prior to implant placement. Thus, the analysis refers to 15 teeth from the control group and 27 teeth from the test group. A t ‐test demonstrated that there was no significant difference in the changes in pre‐ and postoperative bone height between the control group and the test group (95% CI: −1.01, 1.08; p = 0.951, Table ). DISCUSSION After tooth extraction, the bundle bone that lines the periodontal ligament around the tooth and anchors the sharpey fibers in the bone is usually resorbed and replaced with woven bone, causing a vertical and horizontal reduction of the alveolar crest. Furthermore, loss of the vertical dimension especially in the posterior maxilla is the result of a combination of the resorption of the bundle bone and sinus pneumatization. In order to achieve the best possible aesthetic outcome through functional prosthetic rehabilitation supported by osseointegrated implants, it may be crucial to preserve the alveolar ridge dimensions and various studies have already confirmed the effectiveness of ARP in reducing the bone loss after tooth removal. , , Even though the ARP cannot not fully prevent the resorption of the alveolar bone, studies have shown, that the volume of the alveolus is significantly less reduced compared to spontaneous healing. The present clinical trial investigated the clinical benefit of Alveolar Ridge Preservation by reducing the resorption of the bundle bone and for acting against the pneumatization in comparison to spontaneous healing without any alveolar filling material. The idea was if with the ARP the need of a sinus floor augmentation can be prevented. The results of this controlled prospective study indicated that Alveolar Ridge Preservation in the posterior maxilla, reduced need for sinus augmentation procedures, however not statistically relevant ( p = 0.168, Table ). After tooth extraction, the socket in the test group was filled with bovine bone graft substitute and covered with a resorbable porcine membrane Although no primary wound closure was attempted and the membranes were exposed in oral cavity after suturing, new soft tissue formation was observed without any signs of inflammation in all cases. These findings were confirmed by recent studies which showed that exposed biodegradable membranes do not negatively affect bone regeneration of the fresh extraction socket and soft tissue formation. Furthermore, in the present study, new soft tissue formation without signs of inflammation or immunological reaction was observed up to the follow‐up period of 6 months. These data are confirmed by a systematic review by Wang and Lang, which showed that primary wound closure had no positive effect on the preservation of the alveolar ridge. Although it is not a standard of precaution, all patients received antibiotics. It is commonly reported in the literature and seems to be indicated to prescribe antibiotics for extractions and dental implant placement when biomaterials are used to prevent complications. With the intention of limiting the experimental bias antibiotics were also given to the control group. During the follow‐up period, no postsurgical wound healing complications were observed. At surgical re‐entry after 6 months, all alveolar sockets were filled with hard tissue in both groups. Titanium implants were placed without any additional augmentations except for four patients in the test group (12.9%) and seven (31.82%) in the control group who received further sinus lift augmentation. Of these 11 patients who required further augmentation procedures in the form of a sinus lift, 10 were internal sinus lifts and 1 was an external sinus lift. The external sinus lift had to be performed in the control group (Figure ). Even though sinus augmentation procedures are an effective and successful treatment option to increase the alveolar bone height and thus enable to place implants in the posterior maxilla, , they have a noticeable risk of complications such as perforation of the sinus membrane or infection. The present study demonstrated that ARP indicated to lower the need for sinus lift augmentations, however not statistically different on a p value of 0.05 ( p = 0.168, Table ). This occurrence is in line with findings reported by Lee and colleagues. The authors showed that 15.8% of patients who did not receive ARP required an additional sinus lift. In comparison, only 6.5% of patients in the group with ARP required an additional sinus lift, however not statistically relevant. These findings are also presented with the previous study by Barone and colleagues who reported that only 7% of sites in the test group who received ARP with corticocancellous porcine bone, and a collagen membrane required an additional bone augmentation compared to 42% in the control group who did not receive any grafting material. However, it should be noted that in some of the extracted teeth the alveolar walls were not intact and patients who had suffered a traumatic extraction were included in the study population. In addition, smoking was not an exclusion criterion in this study. Patients who smoked less than 10 cigarettes per day were only asked to stop smoking before and after surgery. This could not be verified. In a prospective study conducted by Saldanha and colleagues, it was shown that smoking can negatively influence dimensional changes and the healing process of the extraction sockets. Moreover implant survival can also be significantly negatively influenced by smoking. In a study conducted by Daftari and colleagues, it was also shown that smoking inhibits revascularization of bone grafts, largely due to its vasoconstrictor effect on the arteries. Therefore, non‐smokers were selected in the present study to avoid a major factor that could interfere with or disrupt normal bone graft healing. In the present study, no vertical bone resorption was recorded in test and in control sites after 6‐months follow‐up (Figure ). Furthermore, there was no statistically significant distinction observed in the vertical alterations of the ridge between the two groups (95% CI: −1.01, 1.08; p = 0.951). The test group and the control group almost maintained the vertical bone level without significant change (test group: 11.78 ± 3.09 mm preoperatively before tooth extraction and 11.92 ± 2.79 mm postoperatively after implant placement, control group: 11.13 ± 2.12 mm preoperatively, while it was 11.3 ± 2.17 mm postoperatively, Figure ). These findings are not in agreement with the results from Cha and colleagues who reported a significantly larger bone height in the test group who received ARP using collagenated bovine bone mineral and a resorbable collagen membrane in comparison to the control group. Furthermore, a meta‐analysis by Willenbacher and colleagues found that ARP preserved an average of approximately 1.31–1.54 mm more bone width and 0.91–1.12 mm more bone height compared to spontaneous healing. Although the study underlined that ARP cannot completely stop alveolar ridge resorption, it can limit it more effectively though the use of bone substitute materials. The total survival rate of the implants placed was 100% in the control group and 96.77% in the test group. One implant was lost in the test group (length: 8 mm, diameter: 3.5 mm) during the study. Implants inserted in augmented areas with ARP showed a high success rate, despite direct contact with bovine bone substitute. Similar results were obtained in the study conducted by Barone and colleagues. This study compared the success rates for Implants placed in augmented versus non‐augmented extraction sockets. At the 3‐year follow‐up, the combined success rate of the implants for both groups was recorded at 95%. The analysis of the implant success rates for both groups, considering the length and diameter of the implants, can be summarized as follows: The overall survival rate of all implants in the test and control group was 98.12%. The survival rate of the implants with a length of 6 and 10 mm was 100% after 12 months. Due to the loss of one implant in the group of 8 mm implants, the survival rate at this site was 96.3%. Other studies also confirm the high implant survival rates of different implant lengths placed after ARP , , and spontaneous healing. The survival rate of the implants with diameters of 3.75, 4, and 4.5 mm in both groups was 100%. In contrast, the survival rate for the group of implants with a diameter of 3.5 mm was 0%. However, this is because only one implant with such a diameter was implanted. Here too, the overall survival rate was 98.12%. These survival rates are consistent with those found in the literature. In 2012, Al‐Nawas and colleagues showed that the implant diameter of 3.5 or 4 mm had no significant influence on the survival rate of the implants after an average duration of 108 months. This finding was also confirmed by other studies. , When asked in which cases alveolar ridge preservation has a benefit for adequate preservation of hard and soft tissue, the biotypes of soft tissue described by Olsson and colleagues , could be of great importance for prognosis. A thin soft tissue, especially in the maxilla can be associated with thin vestibular bone, which is rapidly resorbed as a consequence of tooth extraction. In a thick biotype, on the other hand, resorption after tooth extraction may play a rather subordinate role due to the thick vestibular bone. Thus, the significance of ARP could be significantly higher in a thin gingival biotype than in a thick biotype. However, a systematic review and meta‐analysis by Avila‐Ortiz and colleagues found that ridge regions with a buccal bone thickness of more than 1.0 mm exhibited more favorable ridge preservation outcomes than regions with a thinner buccal wall. This is in line with the results of Nevins and colleagues where it was shown that although ARP cannot preserve the buccal bone wall in all cases, the volume of the alveolus is significantly less reduced compared to spontaneous healing. This research acknowledges several limitations that must be taken into consideration. It is important to remark that one of the potential limitations of this study is the small sample size in both groups. These results reveal the need for further research involving a larger patient population to support and substantiate our findings, as the limited number of patients in our study prevents drawing definitive, meaningful conclusions. The necessity for further investigation is especially relevant considering the trends observed in our results, which, although suggestive, require more extensive data for final confirmation. Furthermore, the discrepancy in outcomes between our study and others may be due to this difference in patient sample size. Therefore, studies involving a larger number of patients may not only confirm the trends we observed, but also provide insight into the comparative results seen in different studies. Another limitation of the current study is the absence of an analysis of horizontal bone loss. To comprehensively evaluate changes in horizontal bone width, 3‐dimensional imaging, such as beam‐computed tomography (CBCT), is essential. Due to ethical considerations and the need to minimize exposure to radiation, CBCT scans were only performed prior to implant placement for the purpose of implant planning. Therefore, additional 3‐dimensional imaging would be necessary to allow an in‐depth analysis of horizontal bone loss. The lack of these subsequent imaging sessions hinders our ability to evaluate bone changes in the horizontal dimension over time. CONCLUSION Within the limits of this study, the use of bovine bone substitute and a porcine resorbable membrane after tooth extraction in the posterior maxilla seems to reduce the need for sinus augmentation compared to spontaneous healing, although the difference was not statistically significant. Nevertheless, the ARP in the test group made external sinus floor elevation unnecessary compared to the control group. The change in radiographically measured bone height pre‐ and postoperatively showed no significant difference between the two groups. Further investigation with a larger sample size is recommended to confirm these findings. Khoury, Elias Jean‐Jacques—Concept/Design, Data collection, Drafting article. Sagheb, Keyvan—Concept/Design, Drafting article, Critical revision of article, Approval of article, Data collection, Surgical procedures. Al‐Nawas, Bilal—Concept/Design, Drafting article, Critical revision of article, Approval of article, Data collection, Surgical procedures. König, Jochem—Concept/Design, Data analysis/interpretation. Schiegnitz, Eik—Concept/Design, Drafting article, Critical revision of article, Approval of article, Data collection, Surgical procedures. Funding secured by Straumann, Switzerland. Elias Jean‐Jacques Khoury declares that he has no conflict of interest. Keyvan Sagheb reports lectures, personal fees and/or grants from Dentsply, Geistlich, and Straumann outside the submitted work. Bilal Al‐Nawas reports lectures, personal fees and/or grants from Camlog, Dentsply, Geistlich, Medartis, Straumann and Zimmer outside the submitted work. Jochem König declares that he has no conflict of interest. Eik Schiegnitz reports lectures, personal fees and/or grants from Dentsply, Geistlich, Medartis, Septodont and Straumann outside the submitted work. |
Endothelial GSDMD underlies LPS-induced systemic vascular injury and lethality | ab208835-1af8-437d-994e-653c84389df7 | 11948583 | Pathologic Processes[mh] | Sepsis is a life-threatening organ dysfunction caused by an aberrant host immune response to infection by pathogenic microorganisms ( ). Approximately 48.9 million people worldwide are estimated to suffer from sepsis annually, approximately 22.5% of whom die, and the incidence and mortality of sepsis are still increasing ( , ). Endothelial cells, as vascular barriers, are responsible for maintaining vascular homeostasis and the normal physiological function of multiple organs ( ). During sepsis, vascular endothelial cell injury may lead to impaired microcirculation, tissue hypoperfusion, and organ failure ( , ). To date, considerable attention has been focused on improving endothelial damage to treat sepsis ( , ). However, few treatments targeting sepsis-induced endothelial injury have improved survival in large randomized clinical trials ( , ). Bacterial endotoxin (lipopolysaccharide [LPS]) is a main component of the outer membrane of Gram-negative bacteria and a potent immunostimulant ( ). Gram-negative bacteria lyse and release large amounts of LPS into the circulation after infection, initiating a septicemic cascade ( , ). Caspase-11 initiates the innate immune response once it senses LPS ( – ). Activated caspase-11 cleaves gasdermin D (GSDMD) into the N-terminal GSDMD fragment (GSDMD-N), which triggers the formation of pores in the plasma membrane, causing pyroptosis and the secretion of proinflammatory interleukin-1β (IL-1β) into the circulation ( – ). High-mobility group box 1 (HMGB1), a DNA-binding protein, is abundantly expressed in the nucleus and regulates the immune response intracellularly and extracellularly upon infection ( – ). Both neutralizing HMGB1 and hepatocyte-specific Hmgb1 deletion obviously reduce LPS-induced lethality ( – ). In vitro, LPS-induced HMGB1 release from hepatocytes requires the activation of the caspase-11/GSDMD signaling pathway, and HMGB1 subsequently delivers extracellular LPS into the cytosol of lung endothelial cells to induce endothelial pyroptosis ( ). The results of a recent study suggested that endothelial conditional Caspase-11 deletion evidently reduced endotoxemia-induced lung microvascular injury and improved mouse survival from 0% to 50%–60% ( ). Global knockout of Gsdmd protects against lethal endotoxemia caused by LPS challenge ( , ). However, the role of endothelial GSDMD in LPS-induced endothelial injury and lethality and its regulatory mechanisms in vivo need to be further clarified. Here, we demonstrated that hepatocytic GSDMD-mediated release of HMGB1 bound with the receptor for advanced glycation end products (RAGE) and subsequently promoted vascular endothelial GSDMD activation and endothelial damage in mice induced by LPS or live bacteria, resulting in systemic vascular injury, acute lung injury (ALI), and death. Furthermore, conditional deletion or inhibition of endothelial Gsdmd protected mice from lethal endotoxic shock and sepsis.
Endothelial Gsdmd deletion prevents endothelial damage-mediated vascular injury and death in endotoxemia. Endothelial cells and monocytes/macrophages primarily and actively participate in infection-initiated immune responses ( ). GSDMD is required for LPS-induced pyroptosis and death ( , ). To identify which specific GSDMD-expressing cells determine lethal endotoxemia, we constructed global Gsdmd -knockout ( Gsdmd –/– ) mice and crossed endothelial cell–specific Cre transgenic ( Tie2 Cre/+ ) mice or myeloid cell–specific Cre transgenic ( Lyz2 Cre/+ ) mice with Gsdmd fl/fl mice harboring loxP -flanked (floxed) alleles of Gsdmd to obtain endothelial Gsdmd -deficient ( Gsdmd fl/fl Tie2 Cre/+ ) mice, myeloid cell Gsdmd -deficient ( Gsdmd fl/fl Lyz2 Cre/+ ) mice, and their Cre -negative littermates ( Gsdmd fl/fl mice) ( , A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.182398DS1 ). We then intraperitoneally injected wild-type (WT) mice, Gsdmd –/– mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, and Gsdmd fl/fl mice with LPS and observed its impact on survival for at least 1 week. Compared with that of Gsdmd fl/fl mice, the percentage survival of Gsdmd fl/fl Tie2 Cre/+ mice improved from 10% to 100%, which was comparable to that of Gsdmd –/– mice ( ). However, no difference in survival was observed between Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice ( ). Considering the gradual decrease in the survival rate of the mice 16 hours after the intraperitoneal injection of LPS, we used this key time point to observe the pathophysiological changes in endotoxemic mice. Circulating levels of IL-1β, which is released during pyroptotic cell death ( ), were determined. No difference in the circulating IL-1β level was observed between Gsdmd –/– mice before and after LPS injection ( ). Compared with that in Gsdmd fl/fl mice, the plasma IL-1β concentration was obviously lower in Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd fl/fl Lyz2 Cre/+ mice following exposure to LPS but was significantly greater than the respective baseline level ( ). Compared with those in the PBS-injected mice, lung edema and lung microvascular permeability were significantly aggravated in the WT mice, Gsdmd fl/fl mice, and Gsdmd fl/fl Lyz2 Cre/+ mice after treatment with LPS but were prevented in the Gsdmd –/– mice and Gsdmd fl/fl Tie2 Cre/+ mice ( ). As a whole organ and the largest artery, the aorta is a major vessel that delivers oxygenated blood from the left ventricle to multiple systemic organs ( ). We investigated aortic changes in endotoxemic mice. Compared with PBS, LPS clearly increased aortic permeability in WT mice, which was consistent with the findings in Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice ( ). Both global and endothelial Gsdmd deficiency significantly inhibited aortic permeability, indicating the integrity of the endothelial barrier ( ). A significant increase in the expression and activation of GSDMD, which is expressed mainly in vascular endothelial cells, was observed in WT mice after 16 hours of treatment with LPS ( ). However, LPS failed to change the endothelial GSDMD level in Gsdmd –/– mice ( ). The level of endothelial GSDMD was elevated in Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice 16 hours after LPS injection but not in Gsdmd fl/fl Tie2 Cre/+ mice ( ). These results suggest that endothelial Gsdmd deletion protects LPS-treated mice against systemic vascular injury and lethality. Hepatocyte Hmgb1 deficiency prevents vascular injury and death in endotoxemia. HMGB1, which binds to LPS, regulates endothelial pyroptosis, causing endothelial injury in vitro ( ). Circulating HMGB1 is derived mainly from hepatocytes during endotoxemia ( ). To identify the effects of hepatocytic HMGB1 on the vascular system in vivo, we crossed hepatocyte-specific Cre transgenic ( Alb Cre/+ ) mice with Hmgb1 fl/fl mice harboring floxed alleles of Hmgb1 to generate hepatocellular Hmgb1 -deficient ( Hmgb1 fl/fl Alb Cre/+ ) mice and their Cre -negative littermates ( Hmgb1 fl/fl mice) ( ). The survival of Hmgb1 fl/fl Alb Cre/+ mice was significantly improved compared with that of Hmgb1 fl/fl mice with endotoxemia ( ). No significant differences in the circulating HMGB1 or IL-1β concentrations were detected among Hmgb1 fl/fl Alb Cre/+ mice that were intraperitoneally administered LPS or PBS ( , C and D). Compared with those in Hmgb1 fl/fl mice, the LPS-induced plasma HMGB1 and IL-1β levels were significantly lower in Hmgb1 fl/fl Alb Cre/+ mice ( , C and D). The conditional deletion of Hmgb1 in hepatocytes clearly alleviated pulmonary edema, pulmonary microvascular permeability, and aortic permeability in endotoxemic mice ( , E–H). Thus, hepatocyte-derived HMGB1 causes systemic vascular injury and death in endotoxemia. HMGB1 interacts with RAGE and subsequently participates in endothelial GSDMD-mediated vascular injury in endotoxemia. RAGE and Toll-like receptor 4 (TLR4) act as pivotal HMGB1 receptors in the pathogenesis of inflammatory diseases ( , ). We investigated the roles of these 2 transmembrane receptors in the endothelium during endotoxemia. Five-week-old WT mice were injected with a null adenoassociated virus serotype 9 (AAV9) vector or an endothelial conditional Tlr4 or Rage shRNA-knockdown AAV9 vector via the tail vein, and these mice were treated with LPS 6 weeks later. Endothelial Rage knockdown significantly improved survival from 10% to 80% in LPS-treated mice ( ). No significant difference was detected between the mice injected with the null AAV9 vector and those injected with the endothelial cell–specific Tlr4 shRNA-knockdown AAV9 vector ( ). Therefore, we further assessed the effects of the HMGB1/RAGE axis on endothelial GSDMD-induced vascular injury via in vivo experiments. Recombinant HMGB1 (rHMGB1) protein was administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after LPS injection. Compared with the vehicle, the rHMGB1 protein decreased the survival rate of the LPS-treated mice from 10% to 0%, which was significantly improved from 0% to 70% by the endothelial Rage shRNA-knockdown AAV9 vector ( ). Compared with the vehicle, the rHMGB1 protein significantly aggravated lung edema and increased lung microvascular permeability and aortic permeability, and these effects were reversed by inhibiting endothelial RAGE expression ( ). Consistently, administration of the endothelial Rage -knockdown AAV9 vector significantly reduced the endothelial GSDMD level and the subsequent release of plasma IL-1β in LPS-treated mice stimulated with the rHMGB1 protein ( ). To verify the necessity of the interaction of HMGB1 with RAGE in vascular injury, we also induced endotoxemia by intratracheally instilling LPS, which allowed the LPS to be evenly distributed in the lungs of the mice. Compared with the null AAV9 vector, the rHMGB1 protein–induced survival was significantly improved by the endothelial Rage shRNA-mediated knockdown of the AAV9 vector in LPS-treated mice ( ). The effects of rHMGB1 on pulmonary edema, pulmonary microvascular permeability, aortic permeability, and the IL-1β concentration in LPS-induced mice were reversed by endothelial RAGE expression inhibition ( , B–F). These data indicate that the interaction between HMGB1 and RAGE contributes to endothelial GSDMD-mediated systemic vascular injury and death in endotoxemia. RAGE binds to various damage-associated molecular patterns, such as HMGB1 ( ). FPS-ZM1, a RAGE inhibitor ( – ), was used to further elucidate whether the binding of HMGB1 to RAGE determines the vascular injury function of the HMGB1/RAGE axis in endotoxemia. WT mice were intraperitoneally injected with FPS-ZM1 (3 mg/kg) or dimethyl sulfoxide (DMSO) at 72, 48, 24, and 1 hour before LPS injection and at 24, 48, 72, 96, 120, 144, and 168 hours after LPS injection. rHMGB1 protein was administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after LPS injection. rHMGB1 protein–induced survival was significantly improved by FPS-ZM1 in LPS-treated mice ( ). The effects of the rHMGB1 protein on lung edema, lung microvascular permeability, and aortic permeability in LPS-induced mice were reversed by FPS-ZM1 ( , B–E). Consistently, the administration of FPS-ZM1 significantly reduced the release of plasma IL-1β in the LPS-treated mice stimulated with the rHMGB1 protein ( ). Therefore, the binding of HMGB1 to RAGE is indispensable for systemic vascular injury in lethal endotoxic shock. Alveolar epithelial cells provide protection against environmental insults, regulate water and ion transport, and produce pulmonary surfactants to maintain alveolar homeostasis ( ). Injury to alveolar epithelial cells is an important factor in the occurrence and development of ALI ( ). The absence of Rage protects mice from lethal endotoxemia and sepsis ( , ). Rage mRNA is expressed in type II alveolar epithelial cells, and the protein expression level of RAGE in these cells steadily increases in response to LPS treatment ( , ). Further exploration of the role of type II alveolar epithelial RAGE in endotoxemia is interesting. Five-week-old WT mice were injected with a null AAV9 vector, an endothelial conditional Rage shRNA-knockdown AAV9 vector, or a type II alveolar epithelial conditional Rage shRNA-knockdown AAV9 vector via the tail vein, and these mice were treated with LPS 6 weeks later. rHMGB1 protein was administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after LPS injection. Compared with the null AAV9 vector, the type II alveolar epithelial conditional Rage shRNA-mediated AAV9 vector increased survival from 0% to 30% in rHMGB1 protein–induced mice treated with LPS intraperitoneally or increased survival from 0% to 20% in rHMGB1 protein–induced mice treated intratracheally with LPS; however, no significant difference was observed between these groups ( ). Compared with the type II alveolar epithelial conditional Rage shRNA-knockdown AAV9 vector, the endothelial conditional Rage shRNA-knockdown AAV9 vector significantly improved survival in rHMGB1 protein–induced mice treated with LPS intraperitoneally or intratracheally ( ). Thus, endothelial RAGE rather than alveolar epithelial RAGE is essential for lethal endotoxemia. Hepatocyte Gsdmd deletion regulates HMGB1 release and inhibits vascular damage in endotoxemia. Hepatocyte-specific deletion of Caspase-11 reduces the release of circulating HMGB1 and promotes survival from 0% to approximately 30% in LPS-treated mice ( ). However, the effects of hepatocyte Gsdmd deletion on circulating HMGB1 levels and survival in endotoxin-treated mice are unknown. The GSDMD-N in the liver was significantly increased at 1 hour after treatment with LPS, which was earlier than the increase in plasma HMGB1 levels at 2 hours ( ). We crossed Alb Cre/+ mice with Gsdmd fl/fl mice to generate hepatocellular Gsdmd -deficient ( Gsdmd fl/fl Alb Cre/+ ) mice and their Cre -negative littermates ( Gsdmd fl/fl mice) ( ). Compared with those in Gsdmd fl/fl mice, the LPS-induced plasma HMGB1 and IL-1β levels were significantly lower in Gsdmd fl/fl Alb Cre/+ mice, and no significant differences in the circulating HMGB1 and IL-1β concentrations were detected between Gsdmd fl/fl Alb Cre/+ mice administered LPS or PBS ( ). LPS-induced death in Gsdmd fl/fl Alb Cre/+ mice was prevented, which was consistent with the findings in Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd –/– mice ( ). The conditional deletion of Gsdmd in hepatocytes clearly alleviated pulmonary edema and pulmonary microvascular permeability in endotoxemic mice ( ). Compared with those in Gsdmd fl/fl mice, LPS-induced aortic permeability and endothelial GSDMD levels were significantly lower in Gsdmd fl/fl Alb Cre/+ mice ( ). These results indicate that hepatocyte GSDMD activation triggered HMGB1 release and caused endothelial GSDMD-mediated vascular injury and death in endotoxemia. Hepatocyte GSDMD regulates endothelial GSDMD-mediated vascular injury in an HMGB1-dependent manner in endotoxemia. The above data demonstrate that the plasma HMGB1 level is predominantly regulated by hepatocyte GSDMD activation in LPS-induced mice. In addition, HMGB1 delivers extracellular LPS into the cytosol to promote pulmonary endothelial pyroptosis in vitro ( ). However, in vivo studies are needed to elucidate whether hepatocytic GSDMD regulates endothelial GSDMD-mediated vascular injury through the release of HMGB1 in endotoxemia. rHMGB1 protein can effectively increase plasma HMGB1 level in Gsdmd fl/fl Alb Cre/+ mice treated with LPS ( ). Compared with vehicle administration, treatment with the rHMGB1 protein resulted in a decrease in survival and an evident increase in the IL-1β concentration in Gsdmd fl/fl Alb Cre/+ mice treated with LPS ( ). Compared with those in Gsdmd fl/fl mice, pulmonary edema, pulmonary microvascular permeability, and aortic permeability were obviously lower in Gsdmd fl/fl Alb Cre/+ mice with endotoxemia, and these effects were reversed by the injection of the rHMGB1 protein ( ). Compared with vehicle administration, rHMGB1 protein administration increased the endothelial GSDMD level in Gsdmd fl/fl Alb Cre/+ mice treated with LPS ( ). Collectively, the results of the in vivo experiments suggest that endothelial GSDMD-mediated systemic vascular injury and lethality are dependent on hepatocytic GSDMD-mediated HMGB1 release in endotoxemia. Endothelial GSDMD contributes to vascular injury and death in sepsis. We also assessed the role of endothelial GSDMD in sepsis. The cecal slurry (CS) model involves the intraperitoneal administration of the cecal contents of a euthanized animal into another animal and is used to induce polymicrobial sepsis ( , ). We intraperitoneally injected CS into WT mice, Gsdmd –/– mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, and Gsdmd fl/fl mice. Compared with that of Gsdmd fl/fl mice, the survival rate of Gsdmd fl/fl Tie2 Cre/+ mice improved from 0% to 90% in sepsis, which was comparable to that of Gsdmd –/– mice ( ). Because decreased survival of the mice was observed 16 hours after the intraperitoneal injection of CS, we performed biochemical and pathological tests at this time. No significant change in the circulating IL-1β level in Gsdmd –/– mice was observed after treatment with CS ( ). Compared with that in Gsdmd fl/fl mice, the plasma IL-1β concentration was obviously lower in Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd fl/fl Lyz2 Cre/+ mice following exposure to CS but significantly greater than the corresponding baseline level ( ). Compared with those of Gsdmd fl/fl mice, the lung edema, lung microvascular permeability, and aortic permeability of Gsdmd fl/fl Tie2 Cre/+ mice were significantly alleviated in sepsis ( ). Survival, ALI, and systemic vascular injury were comparable between Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice with sepsis ( ). These results suggest that endothelial Gsdmd deletion improved the integrity of the endothelial barrier, which protected mice against ALI and death in sepsis. The HMGB1/RAGE signaling pathway regulates vascular injury through endothelial GSDMD in sepsis. To explore the regulatory mechanism of vascular injury in sepsis, WT mice were intraperitoneally injected with FPS-ZM1 (3 mg/kg) or DMSO at 72, 48, 24, and 1 hour before CS injection and at 24, 48, 72, 96, 120, 144, and 168 hours after CS injection. rHMGB1 protein was subsequently administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after CS injection. The survival of rHMGB1 protein–treated WT mice with sepsis significantly improved from 0% to 50% in response to FPS-ZM1 injection compared with that in response to DMSO injection ( ). Compared with those of the DMSO-treated mice, the rHMGB1 protein–induced lung edema, lung microvascular permeability, aortic permeability, and plasma IL-1β levels of the WT mice with sepsis were obviously reversed by FPS-ZM1 ( , B–F). These data indicate that the binding of HMGB1 to RAGE causes systemic vascular injury and death in sepsis. We further identified the underlying mechanism of the HMGB1/RAGE signaling pathway in vascular damage in sepsis. Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd fl/fl mice were administered CS and then treated intravenously with 5 μg of the rHMGB1 protein at 2, 16, 28, and 40 hours after CS injection. Compared with those in Gsdmd fl/fl mice, rHMGB1 protein–induced lethality, pulmonary edema, pulmonary microvascular permeability, aortic permeability, and IL-1β concentration were significantly lower in Gsdmd fl/fl Tie2 Cre/+ mice ( ). Therefore, endothelial GSDMD mediates the regulatory effects of the HMGB1/RAGE axis on vascular injury and death in sepsis. Hepatocyte GSDMD regulates vascular injury through the release of HMGB1 in sepsis. Compared with that in Gsdmd fl/fl mice, the CS-induced plasma HMGB1 level was significantly lower in Gsdmd fl/fl Alb Cre/+ mice, and no significant difference in the circulating HMGB1 concentration was detected between Gsdmd fl/fl Alb Cre/+ mice administered CS or 5% dextrose ( ). Compared with those in Gsdmd fl/fl mice, improved survival and significantly decreased plasma IL-1β concentrations were observed in Gsdmd fl/fl Alb Cre/+ mice with sepsis, and these effects were reversed by rHMGB1 protein injection ( ). Compared with those in Gsdmd fl/fl mice, lung edema, lung microvascular permeability, and aortic permeability were obviously inhibited in Gsdmd fl/fl Alb Cre/+ mice with sepsis, and these effects were reversed by the injection of the rHMGB1 protein ( ). Collectively, these results suggest that hepatocytic GSDMD is responsible for the release of HMGB1, ultimately resulting in systemic vascular injury and lethality in sepsis. Targeting endothelial GSDMD protected against systemic vascular injury and lethality in endotoxemia and sepsis. GSDMD has emerged as a promising therapeutic target for the treatment of LPS-triggered endotoxemia ( , ). Five-week-old mice were injected with an endothelium-specific Gsdmd shRNA-knockdown AAV9 vector via the tail vein and then treated with LPS after 6 weeks. Compared with the null AAV9 vector, the endothelium-specific Gsdmd shRNA-knockdown AAV9 vector significantly improved the survival of the mice from 10% to 80% and decreased the release of IL-1β during endotoxemia ( ). Compared with those in the null AAV9 vector–treated mice, LPS-induced pulmonary edema and pulmonary microvascular permeability were evidently alleviated in the endothelium-conditioned Gsdmd shRNA-knockdown AAV9 vector–treated mice ( ). The endothelium-conditioned Gsdmd shRNA–mediated knockdown of the AAV9 vector also reduced aortic permeability and endothelial GSDMD levels in LPS-treated mice ( ). The mouse GSDMD recognition motif for inflammatory caspases has been reported, which indicates that the GSDMD cleavage site peptide LLSD directly binds to caspase-11 ( , ). We designed and synthesized a GSDMD activation inhibitor, benzyloxycarbonyl-Leu-Leu-Ser-Asp-fluoromethyl ketone, which targets LLSD from GL Biochem Co., Ltd. and was previously demonstrated to successfully suppress pyroptosis ( ). This GSDMD activation inhibitor was administered intraperitoneally at a dose of 200 μg at 2, 12, 24, and 36 hours after LPS injection. Compared with vehicle injection, the use of a GSDMD activation inhibitor significantly reduced the plasma HMGB1 concentration in endotoxemic mice, which was comparable to the effect of Gsdmd siRNA injection ( ). LPS-induced death and circulating IL-1β levels were significantly lower in the GSDMD activation inhibitor group than in the vehicle group ( ). Compared with the vehicle, the GSDMD activation inhibitor significantly reduced LPS-stimulated lung edema, lung microvascular permeability, and aortic permeability ( ). We also identified the protective role of a GSDMD activation inhibitor during sepsis. Compared with the vehicle, the GSDMD activation inhibitor obviously reduced the mortality rate and plasma HMGB1 and IL-1β concentrations in the mice with sepsis ( and , A and B). Compared with the vehicle, the GSDMD activation inhibitor significantly inhibited pulmonary edema, pulmonary microvascular permeability, and aortic permeability in mice with sepsis ( , C–F). Similar improvements were observed in sepsis model mice after the use of an endothelially conditioned Gsdmd shRNA-knockdown AAV9 vector ( , G–L). Therefore, inhibiting endothelial GSDMD expression and activation decreased vascular injury and improved survival in mice with endotoxemia or sepsis.
Endothelial cells and monocytes/macrophages primarily and actively participate in infection-initiated immune responses ( ). GSDMD is required for LPS-induced pyroptosis and death ( , ). To identify which specific GSDMD-expressing cells determine lethal endotoxemia, we constructed global Gsdmd -knockout ( Gsdmd –/– ) mice and crossed endothelial cell–specific Cre transgenic ( Tie2 Cre/+ ) mice or myeloid cell–specific Cre transgenic ( Lyz2 Cre/+ ) mice with Gsdmd fl/fl mice harboring loxP -flanked (floxed) alleles of Gsdmd to obtain endothelial Gsdmd -deficient ( Gsdmd fl/fl Tie2 Cre/+ ) mice, myeloid cell Gsdmd -deficient ( Gsdmd fl/fl Lyz2 Cre/+ ) mice, and their Cre -negative littermates ( Gsdmd fl/fl mice) ( , A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.182398DS1 ). We then intraperitoneally injected wild-type (WT) mice, Gsdmd –/– mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, and Gsdmd fl/fl mice with LPS and observed its impact on survival for at least 1 week. Compared with that of Gsdmd fl/fl mice, the percentage survival of Gsdmd fl/fl Tie2 Cre/+ mice improved from 10% to 100%, which was comparable to that of Gsdmd –/– mice ( ). However, no difference in survival was observed between Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice ( ). Considering the gradual decrease in the survival rate of the mice 16 hours after the intraperitoneal injection of LPS, we used this key time point to observe the pathophysiological changes in endotoxemic mice. Circulating levels of IL-1β, which is released during pyroptotic cell death ( ), were determined. No difference in the circulating IL-1β level was observed between Gsdmd –/– mice before and after LPS injection ( ). Compared with that in Gsdmd fl/fl mice, the plasma IL-1β concentration was obviously lower in Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd fl/fl Lyz2 Cre/+ mice following exposure to LPS but was significantly greater than the respective baseline level ( ). Compared with those in the PBS-injected mice, lung edema and lung microvascular permeability were significantly aggravated in the WT mice, Gsdmd fl/fl mice, and Gsdmd fl/fl Lyz2 Cre/+ mice after treatment with LPS but were prevented in the Gsdmd –/– mice and Gsdmd fl/fl Tie2 Cre/+ mice ( ). As a whole organ and the largest artery, the aorta is a major vessel that delivers oxygenated blood from the left ventricle to multiple systemic organs ( ). We investigated aortic changes in endotoxemic mice. Compared with PBS, LPS clearly increased aortic permeability in WT mice, which was consistent with the findings in Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice ( ). Both global and endothelial Gsdmd deficiency significantly inhibited aortic permeability, indicating the integrity of the endothelial barrier ( ). A significant increase in the expression and activation of GSDMD, which is expressed mainly in vascular endothelial cells, was observed in WT mice after 16 hours of treatment with LPS ( ). However, LPS failed to change the endothelial GSDMD level in Gsdmd –/– mice ( ). The level of endothelial GSDMD was elevated in Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice 16 hours after LPS injection but not in Gsdmd fl/fl Tie2 Cre/+ mice ( ). These results suggest that endothelial Gsdmd deletion protects LPS-treated mice against systemic vascular injury and lethality.
HMGB1, which binds to LPS, regulates endothelial pyroptosis, causing endothelial injury in vitro ( ). Circulating HMGB1 is derived mainly from hepatocytes during endotoxemia ( ). To identify the effects of hepatocytic HMGB1 on the vascular system in vivo, we crossed hepatocyte-specific Cre transgenic ( Alb Cre/+ ) mice with Hmgb1 fl/fl mice harboring floxed alleles of Hmgb1 to generate hepatocellular Hmgb1 -deficient ( Hmgb1 fl/fl Alb Cre/+ ) mice and their Cre -negative littermates ( Hmgb1 fl/fl mice) ( ). The survival of Hmgb1 fl/fl Alb Cre/+ mice was significantly improved compared with that of Hmgb1 fl/fl mice with endotoxemia ( ). No significant differences in the circulating HMGB1 or IL-1β concentrations were detected among Hmgb1 fl/fl Alb Cre/+ mice that were intraperitoneally administered LPS or PBS ( , C and D). Compared with those in Hmgb1 fl/fl mice, the LPS-induced plasma HMGB1 and IL-1β levels were significantly lower in Hmgb1 fl/fl Alb Cre/+ mice ( , C and D). The conditional deletion of Hmgb1 in hepatocytes clearly alleviated pulmonary edema, pulmonary microvascular permeability, and aortic permeability in endotoxemic mice ( , E–H). Thus, hepatocyte-derived HMGB1 causes systemic vascular injury and death in endotoxemia.
RAGE and Toll-like receptor 4 (TLR4) act as pivotal HMGB1 receptors in the pathogenesis of inflammatory diseases ( , ). We investigated the roles of these 2 transmembrane receptors in the endothelium during endotoxemia. Five-week-old WT mice were injected with a null adenoassociated virus serotype 9 (AAV9) vector or an endothelial conditional Tlr4 or Rage shRNA-knockdown AAV9 vector via the tail vein, and these mice were treated with LPS 6 weeks later. Endothelial Rage knockdown significantly improved survival from 10% to 80% in LPS-treated mice ( ). No significant difference was detected between the mice injected with the null AAV9 vector and those injected with the endothelial cell–specific Tlr4 shRNA-knockdown AAV9 vector ( ). Therefore, we further assessed the effects of the HMGB1/RAGE axis on endothelial GSDMD-induced vascular injury via in vivo experiments. Recombinant HMGB1 (rHMGB1) protein was administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after LPS injection. Compared with the vehicle, the rHMGB1 protein decreased the survival rate of the LPS-treated mice from 10% to 0%, which was significantly improved from 0% to 70% by the endothelial Rage shRNA-knockdown AAV9 vector ( ). Compared with the vehicle, the rHMGB1 protein significantly aggravated lung edema and increased lung microvascular permeability and aortic permeability, and these effects were reversed by inhibiting endothelial RAGE expression ( ). Consistently, administration of the endothelial Rage -knockdown AAV9 vector significantly reduced the endothelial GSDMD level and the subsequent release of plasma IL-1β in LPS-treated mice stimulated with the rHMGB1 protein ( ). To verify the necessity of the interaction of HMGB1 with RAGE in vascular injury, we also induced endotoxemia by intratracheally instilling LPS, which allowed the LPS to be evenly distributed in the lungs of the mice. Compared with the null AAV9 vector, the rHMGB1 protein–induced survival was significantly improved by the endothelial Rage shRNA-mediated knockdown of the AAV9 vector in LPS-treated mice ( ). The effects of rHMGB1 on pulmonary edema, pulmonary microvascular permeability, aortic permeability, and the IL-1β concentration in LPS-induced mice were reversed by endothelial RAGE expression inhibition ( , B–F). These data indicate that the interaction between HMGB1 and RAGE contributes to endothelial GSDMD-mediated systemic vascular injury and death in endotoxemia. RAGE binds to various damage-associated molecular patterns, such as HMGB1 ( ). FPS-ZM1, a RAGE inhibitor ( – ), was used to further elucidate whether the binding of HMGB1 to RAGE determines the vascular injury function of the HMGB1/RAGE axis in endotoxemia. WT mice were intraperitoneally injected with FPS-ZM1 (3 mg/kg) or dimethyl sulfoxide (DMSO) at 72, 48, 24, and 1 hour before LPS injection and at 24, 48, 72, 96, 120, 144, and 168 hours after LPS injection. rHMGB1 protein was administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after LPS injection. rHMGB1 protein–induced survival was significantly improved by FPS-ZM1 in LPS-treated mice ( ). The effects of the rHMGB1 protein on lung edema, lung microvascular permeability, and aortic permeability in LPS-induced mice were reversed by FPS-ZM1 ( , B–E). Consistently, the administration of FPS-ZM1 significantly reduced the release of plasma IL-1β in the LPS-treated mice stimulated with the rHMGB1 protein ( ). Therefore, the binding of HMGB1 to RAGE is indispensable for systemic vascular injury in lethal endotoxic shock. Alveolar epithelial cells provide protection against environmental insults, regulate water and ion transport, and produce pulmonary surfactants to maintain alveolar homeostasis ( ). Injury to alveolar epithelial cells is an important factor in the occurrence and development of ALI ( ). The absence of Rage protects mice from lethal endotoxemia and sepsis ( , ). Rage mRNA is expressed in type II alveolar epithelial cells, and the protein expression level of RAGE in these cells steadily increases in response to LPS treatment ( , ). Further exploration of the role of type II alveolar epithelial RAGE in endotoxemia is interesting. Five-week-old WT mice were injected with a null AAV9 vector, an endothelial conditional Rage shRNA-knockdown AAV9 vector, or a type II alveolar epithelial conditional Rage shRNA-knockdown AAV9 vector via the tail vein, and these mice were treated with LPS 6 weeks later. rHMGB1 protein was administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after LPS injection. Compared with the null AAV9 vector, the type II alveolar epithelial conditional Rage shRNA-mediated AAV9 vector increased survival from 0% to 30% in rHMGB1 protein–induced mice treated with LPS intraperitoneally or increased survival from 0% to 20% in rHMGB1 protein–induced mice treated intratracheally with LPS; however, no significant difference was observed between these groups ( ). Compared with the type II alveolar epithelial conditional Rage shRNA-knockdown AAV9 vector, the endothelial conditional Rage shRNA-knockdown AAV9 vector significantly improved survival in rHMGB1 protein–induced mice treated with LPS intraperitoneally or intratracheally ( ). Thus, endothelial RAGE rather than alveolar epithelial RAGE is essential for lethal endotoxemia.
Hepatocyte-specific deletion of Caspase-11 reduces the release of circulating HMGB1 and promotes survival from 0% to approximately 30% in LPS-treated mice ( ). However, the effects of hepatocyte Gsdmd deletion on circulating HMGB1 levels and survival in endotoxin-treated mice are unknown. The GSDMD-N in the liver was significantly increased at 1 hour after treatment with LPS, which was earlier than the increase in plasma HMGB1 levels at 2 hours ( ). We crossed Alb Cre/+ mice with Gsdmd fl/fl mice to generate hepatocellular Gsdmd -deficient ( Gsdmd fl/fl Alb Cre/+ ) mice and their Cre -negative littermates ( Gsdmd fl/fl mice) ( ). Compared with those in Gsdmd fl/fl mice, the LPS-induced plasma HMGB1 and IL-1β levels were significantly lower in Gsdmd fl/fl Alb Cre/+ mice, and no significant differences in the circulating HMGB1 and IL-1β concentrations were detected between Gsdmd fl/fl Alb Cre/+ mice administered LPS or PBS ( ). LPS-induced death in Gsdmd fl/fl Alb Cre/+ mice was prevented, which was consistent with the findings in Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd –/– mice ( ). The conditional deletion of Gsdmd in hepatocytes clearly alleviated pulmonary edema and pulmonary microvascular permeability in endotoxemic mice ( ). Compared with those in Gsdmd fl/fl mice, LPS-induced aortic permeability and endothelial GSDMD levels were significantly lower in Gsdmd fl/fl Alb Cre/+ mice ( ). These results indicate that hepatocyte GSDMD activation triggered HMGB1 release and caused endothelial GSDMD-mediated vascular injury and death in endotoxemia.
The above data demonstrate that the plasma HMGB1 level is predominantly regulated by hepatocyte GSDMD activation in LPS-induced mice. In addition, HMGB1 delivers extracellular LPS into the cytosol to promote pulmonary endothelial pyroptosis in vitro ( ). However, in vivo studies are needed to elucidate whether hepatocytic GSDMD regulates endothelial GSDMD-mediated vascular injury through the release of HMGB1 in endotoxemia. rHMGB1 protein can effectively increase plasma HMGB1 level in Gsdmd fl/fl Alb Cre/+ mice treated with LPS ( ). Compared with vehicle administration, treatment with the rHMGB1 protein resulted in a decrease in survival and an evident increase in the IL-1β concentration in Gsdmd fl/fl Alb Cre/+ mice treated with LPS ( ). Compared with those in Gsdmd fl/fl mice, pulmonary edema, pulmonary microvascular permeability, and aortic permeability were obviously lower in Gsdmd fl/fl Alb Cre/+ mice with endotoxemia, and these effects were reversed by the injection of the rHMGB1 protein ( ). Compared with vehicle administration, rHMGB1 protein administration increased the endothelial GSDMD level in Gsdmd fl/fl Alb Cre/+ mice treated with LPS ( ). Collectively, the results of the in vivo experiments suggest that endothelial GSDMD-mediated systemic vascular injury and lethality are dependent on hepatocytic GSDMD-mediated HMGB1 release in endotoxemia.
We also assessed the role of endothelial GSDMD in sepsis. The cecal slurry (CS) model involves the intraperitoneal administration of the cecal contents of a euthanized animal into another animal and is used to induce polymicrobial sepsis ( , ). We intraperitoneally injected CS into WT mice, Gsdmd –/– mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, and Gsdmd fl/fl mice. Compared with that of Gsdmd fl/fl mice, the survival rate of Gsdmd fl/fl Tie2 Cre/+ mice improved from 0% to 90% in sepsis, which was comparable to that of Gsdmd –/– mice ( ). Because decreased survival of the mice was observed 16 hours after the intraperitoneal injection of CS, we performed biochemical and pathological tests at this time. No significant change in the circulating IL-1β level in Gsdmd –/– mice was observed after treatment with CS ( ). Compared with that in Gsdmd fl/fl mice, the plasma IL-1β concentration was obviously lower in Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd fl/fl Lyz2 Cre/+ mice following exposure to CS but significantly greater than the corresponding baseline level ( ). Compared with those of Gsdmd fl/fl mice, the lung edema, lung microvascular permeability, and aortic permeability of Gsdmd fl/fl Tie2 Cre/+ mice were significantly alleviated in sepsis ( ). Survival, ALI, and systemic vascular injury were comparable between Gsdmd fl/fl mice and Gsdmd fl/fl Lyz2 Cre/+ mice with sepsis ( ). These results suggest that endothelial Gsdmd deletion improved the integrity of the endothelial barrier, which protected mice against ALI and death in sepsis.
To explore the regulatory mechanism of vascular injury in sepsis, WT mice were intraperitoneally injected with FPS-ZM1 (3 mg/kg) or DMSO at 72, 48, 24, and 1 hour before CS injection and at 24, 48, 72, 96, 120, 144, and 168 hours after CS injection. rHMGB1 protein was subsequently administered intravenously at a dose of 5 μg at 2, 16, 28, and 40 hours after CS injection. The survival of rHMGB1 protein–treated WT mice with sepsis significantly improved from 0% to 50% in response to FPS-ZM1 injection compared with that in response to DMSO injection ( ). Compared with those of the DMSO-treated mice, the rHMGB1 protein–induced lung edema, lung microvascular permeability, aortic permeability, and plasma IL-1β levels of the WT mice with sepsis were obviously reversed by FPS-ZM1 ( , B–F). These data indicate that the binding of HMGB1 to RAGE causes systemic vascular injury and death in sepsis. We further identified the underlying mechanism of the HMGB1/RAGE signaling pathway in vascular damage in sepsis. Gsdmd fl/fl Tie2 Cre/+ mice and Gsdmd fl/fl mice were administered CS and then treated intravenously with 5 μg of the rHMGB1 protein at 2, 16, 28, and 40 hours after CS injection. Compared with those in Gsdmd fl/fl mice, rHMGB1 protein–induced lethality, pulmonary edema, pulmonary microvascular permeability, aortic permeability, and IL-1β concentration were significantly lower in Gsdmd fl/fl Tie2 Cre/+ mice ( ). Therefore, endothelial GSDMD mediates the regulatory effects of the HMGB1/RAGE axis on vascular injury and death in sepsis.
Compared with that in Gsdmd fl/fl mice, the CS-induced plasma HMGB1 level was significantly lower in Gsdmd fl/fl Alb Cre/+ mice, and no significant difference in the circulating HMGB1 concentration was detected between Gsdmd fl/fl Alb Cre/+ mice administered CS or 5% dextrose ( ). Compared with those in Gsdmd fl/fl mice, improved survival and significantly decreased plasma IL-1β concentrations were observed in Gsdmd fl/fl Alb Cre/+ mice with sepsis, and these effects were reversed by rHMGB1 protein injection ( ). Compared with those in Gsdmd fl/fl mice, lung edema, lung microvascular permeability, and aortic permeability were obviously inhibited in Gsdmd fl/fl Alb Cre/+ mice with sepsis, and these effects were reversed by the injection of the rHMGB1 protein ( ). Collectively, these results suggest that hepatocytic GSDMD is responsible for the release of HMGB1, ultimately resulting in systemic vascular injury and lethality in sepsis.
GSDMD has emerged as a promising therapeutic target for the treatment of LPS-triggered endotoxemia ( , ). Five-week-old mice were injected with an endothelium-specific Gsdmd shRNA-knockdown AAV9 vector via the tail vein and then treated with LPS after 6 weeks. Compared with the null AAV9 vector, the endothelium-specific Gsdmd shRNA-knockdown AAV9 vector significantly improved the survival of the mice from 10% to 80% and decreased the release of IL-1β during endotoxemia ( ). Compared with those in the null AAV9 vector–treated mice, LPS-induced pulmonary edema and pulmonary microvascular permeability were evidently alleviated in the endothelium-conditioned Gsdmd shRNA-knockdown AAV9 vector–treated mice ( ). The endothelium-conditioned Gsdmd shRNA–mediated knockdown of the AAV9 vector also reduced aortic permeability and endothelial GSDMD levels in LPS-treated mice ( ). The mouse GSDMD recognition motif for inflammatory caspases has been reported, which indicates that the GSDMD cleavage site peptide LLSD directly binds to caspase-11 ( , ). We designed and synthesized a GSDMD activation inhibitor, benzyloxycarbonyl-Leu-Leu-Ser-Asp-fluoromethyl ketone, which targets LLSD from GL Biochem Co., Ltd. and was previously demonstrated to successfully suppress pyroptosis ( ). This GSDMD activation inhibitor was administered intraperitoneally at a dose of 200 μg at 2, 12, 24, and 36 hours after LPS injection. Compared with vehicle injection, the use of a GSDMD activation inhibitor significantly reduced the plasma HMGB1 concentration in endotoxemic mice, which was comparable to the effect of Gsdmd siRNA injection ( ). LPS-induced death and circulating IL-1β levels were significantly lower in the GSDMD activation inhibitor group than in the vehicle group ( ). Compared with the vehicle, the GSDMD activation inhibitor significantly reduced LPS-stimulated lung edema, lung microvascular permeability, and aortic permeability ( ). We also identified the protective role of a GSDMD activation inhibitor during sepsis. Compared with the vehicle, the GSDMD activation inhibitor obviously reduced the mortality rate and plasma HMGB1 and IL-1β concentrations in the mice with sepsis ( and , A and B). Compared with the vehicle, the GSDMD activation inhibitor significantly inhibited pulmonary edema, pulmonary microvascular permeability, and aortic permeability in mice with sepsis ( , C–F). Similar improvements were observed in sepsis model mice after the use of an endothelially conditioned Gsdmd shRNA-knockdown AAV9 vector ( , G–L). Therefore, inhibiting endothelial GSDMD expression and activation decreased vascular injury and improved survival in mice with endotoxemia or sepsis.
The endothelium is recognized as a fully fledged organ that covers an area of nearly 1,000 m 2 ( , ). The amount of evidence supporting the role of endothelial cells in both physiological and pathological responses to sepsis is continuously growing ( ). Endothelial cells make up 50% of lung cells and are initially exposed to bacteria in the blood, making the lungs the most vulnerable organs to septic damage ( , ). ALI, including its most severe manifestation, acute respiratory distress syndrome, is a major complication and a leading cause of sepsis-related death in clinical practice ( , ). Pathologically, ALI is characterized by damage to the microvascular endothelium and alveolar epithelium and alveolar capillary leakage and fluid accumulation in the alveolar and interstitial space, leading to inflammatory cell infiltration and edema formation ( , ). Endotoxemia has ongoing utility in preclinical research, specifically in examining the acute response associated with the initial stages of sepsis ( ). Hence, a better understanding of the regulatory mechanisms of endothelial damage in endotoxemia and endotoxemia-induced septic lethality is urgently needed. In this study, we demonstrated that endothelial GSDMD, not myeloid cell–derived GSDMD, was responsible for endothelial injury–mediated ALI and lethality in endotoxemia and sepsis. In contrast with previous studies that focused on lung microvascular changes ( , ), our results revealed increased aortic permeability in endotoxemia and sepsis, which was prevented by endothelial Gsdmd deletion. The results of this study suggested that endothelial GSDMD-mediated endothelial pyroptosis causes systemic vascular injury, which may trigger systemic hypoperfusion and organ dysfunction, ultimately leading to death in endotoxemia or sepsis. In this study, endothelial Gsdmd deficiency protected mice against LPS-induced ALI and death but did not reduce the IL-1β concentration to the baseline level. Inconsistent with previous research ( ), our results suggest that the IL-1β level is not a determining factor in endotoxemia-induced ALI or lethality. A cytokine storm is a life-threatening systemic inflammatory syndrome involving elevated levels of circulating cytokines and immune cell hyperactivation that can be triggered by LPS and sepsis ( , ). Cytokine storms can lead to multiorgan dysfunction and even multiorgan failure and lethality if inadequately treated ( ). Myeloid cells, including monocytes, mature macrophages, and granulocytes, are the primary sources of IL-1 family members and tumor necrosis factor–α (TNF-α), which are closely associated with cytokine storm disorders and GSDMD-mediated pyroptosis ( , , – ). In response to acute infections, the production and mobilization of monocyte and neutrophil populations from the bone marrow increase, and these cells are recruited to sites of inflammation to produce and release IL-1 and TNF-α ( ). Although some studies have reported that neutralizing antibodies against both TNF-α and IL-1β improved survival to approximately 60% after intraperitoneal injection of LPS in mice ( , ), neutralization of IL-1β via an IL-1 receptor antagonist did not protect against LPS-induced organ dysfunction ( ). Randomized controlled clinical trials have also indicated the failure of TNF-α or IL-1 blockade to improve outcomes in sepsis patients ( – ). We found that conditional endothelial deletion of Gsdmd completely protected against mortality and obviously alleviated ALI in mice with endotoxemia, while myeloid cell Gsdmd deficiency did not improve ALI or survival. Therefore, endothelial GSDMD activation–mediated endothelial pyroptosis is most likely the decisive cause of endotoxemia-induced death, whereas the cytokine storm driven mainly by the myeloid cell line may aggravate the disease. Huebener et al. ( ) reported that hepatocyte Hmgb1 deficiency reduced circulating HMGB1 levels in LPS-treated mice but had no influence on LPS-induced lethal shock. These results were different from those published by Wang et al. ( ) and Denget al. ( ), in which 70% and 90% of the LPS-treated mice survived following treatment with an HMGB1-neutralizing antibody and hepatocyte-specific Hmgb1 deletion, respectively. Our data revealed that survival improved from 10% to 90% in hepatocyte-conditioned Hmgb1 -deficient mice with endotoxemia, which was also inconsistent with the findings of Huebener et al. ( ). The release of HMGB1 is regulated by GSDMD activation ( ). The direct effects of hepatocyte Gsdmd deletion on circulating HMGB1 levels and survival in LPS-treated mice have not been explored. In our study, the increase in liver GSDMD activation preceded the increase in circulating HMGB1 levels during endotoxemia, and hepatocyte-specific Gsdmd deficiency decreased the plasma HMGB1 concentration and improved the survival of mice from 10% to 100%. Therefore, the inhibition of hepatocyte GSDMD had a greater protective effect on endotoxemia than did hepatocyte-specific Hmgb1 knockout, and other factors that are released may depend on hepatocyte GSDMD activation and have a lethal effect on endotoxemia. Our results indicated that both hepatocyte-specific Gsdmd deletion and endothelial Gsdmd deletion prevented LPS-induced death. Although in vitro experiments have shown that HMGB1 derived from hepatocytes delivers LPS into the cytosol of lung endothelial cells to trigger caspase-11–dependent pyroptosis ( ), the regulatory mechanism between hepatocyte GSDMD and vascular endothelial GSDMD in vivo needs further clarification. We demonstrated that hepatocyte GSDMD activation occurred earlier than vascular endothelial GSDMD activation in LPS-treated mice. In endotoxemia, we demonstrated that hepatocyte GSDMD was responsible for regulating the release of HMGB1. Additionally, we found that hepatocyte Gsdmd deletion inhibited LPS-induced vascular endothelial GSDMD levels and systemic vascular injury and that these effects were reversed by rHMGB1 protein intervention. Therefore, the results of these in vivo experiments verified that hepatocyte GSDMD mediated HMGB1 release and subsequently regulated endothelial GSDMD-mediated vascular injury in endotoxemia. Currently, researchers in the field agree that LPS injection may serve as a model for endotoxic shock but not for sepsis ( ). LPS is a single component of complex pathogen-associated molecular patterns released by Gram-negative organisms ( ). LPS injection neglects the host-pathogen interactions of Gram-positive organisms and polymicrobial sepsis ( ). Numerous clinical trials of antiinflammatory strategies for the treatment of sepsis might be referred to as “graveyards” for pharmaceutical companies, since almost none of these strategies has resulted in obviously improved survival of patients ( ). Cecal contents contain not only live microbes but also particulate matter that assists with bacterial colonization of the peritoneum ( ). In an untreated animal, bacterial colonies can be recovered transiently from the blood and persist in the peritoneum and visceral organs ( ). We used CS to establish a peritoneal sepsis model and further investigated the role of endothelial GSDMD in sepsis. Consistent with our findings in LPS shock, hepatocyte GSDMD-mediated HMGB1 contributed to endothelial GSDMD-mediated systemic vascular damage in sepsis, and endothelial Gsdmd deficiency prevented sepsis-induced lethality. These data suggest that endothelial GSDMD may be an attractive target for treating endotoxemia and sepsis. We used an endothelial cell–specific Gsdmd shRNA-knockdown AAV9 vector to inhibit endothelial GSDMD levels, thereby protecting mice with endotoxemia and sepsis from ALI and mortality. We designed and synthesized a GSDMD activation inhibitor based on the possible conserved inflammatory caspase cleavage site in mouse GSDMD ( ). We previously demonstrated the inhibitory effect of this inhibitor on GSDMD activation by determining LPS-induced lactate dehydrogenase release and propidium iodide staining in vitro ( ). Here, we used a GSDMD inhibitor to prevent endothelial injury, systemic vascular injury, and lethality in mice with endotoxemia and sepsis successfully. Therefore, this research validates endothelial GSDMD as a viable pharmaceutical target and provides a basis for the development of future therapeutics for endotoxemia and endotoxemia-induced septic lethality. There are several limitations to this study. We found that endothelial Gsdmd deletion, rather than myeloid cell Gsdmd deletion, prevented vascular injury and death in sepsis. Consistently, the IL-1β concentration in endothelial Gsdmd -knockout mice was greater than that in myeloid cell Gsdmd -knockout mice. Therefore, the effects of IL-1β on sepsis remain to be further determined. In addition, it remains unknown whether the protective role of endothelial Gsdmd deletion is associated with endothelial pyroptosis-related cytokine storms in sepsis. Although excessive or uncontrolled pyroptosis has a deleterious effect on the host, it has proven to have a game-changing therapeutic effect on pathogenic invasion when controlled ( ). As a critical mechanism of host defense, GSDMD activation–mediated IL-18 release contributes to the killing and clearance of gastrointestinal pathogens in intestinal cells and immune cells, which drives anti-rotavirus immunity and protects mice against rotavirus infection ( ). GSDMD-mediated pyroptosis promotes the Th1 immune response by controlling the release of IL-18, which plays a vital role in clearing the parasite ( ). Upon GSDMD activation, GSDMD-N and other cytosolic contents are released from pyroptotic cells and reduce the number of intracellular and extracellular bacteria by causing host cell death or a direct antibacterial effect ( , ). Gsdmd deficiency led to severe abscess formation, extensive skin damage, bacterial spread, and cellular inflammation in a mouse model of S . aureus skin infection ( ). The activation of GSDMD-dependent pyroptosis and IL-18 secretion has been shown to improve antitumor immunity by maintaining healthy gut microbiota ( , , ). However, further experiments are needed to determine the role of GSDMD in host defense in the future ( ).
Sex as a biological variable. Our study examined male and female animals, and similar findings are reported for both sexes. Animals. C57BL/6J (WT) mice (No. 219) were purchased from Charles River Laboratories Co., Ltd. (Beijing, China). Gsdmd –/– mice and Gsdmd fl/fl mice generated via the CRISPR/Cas9 system were purchased from GemPharmatech Co., Ltd. (Nanjing, China) ( ). Hmgb1 fl/fl mice were gifted by Tadatsugu Taniguchi from the University of Tokyo, Tokyo, Japan ( ). Alb Cre/+ mice (Stock No. 003574), Tie2 Cre/+ mice (Stock No. 008863), and Lyz2 Cre/+ mice (Stock No. 004781) were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA) ( – ). Alb Cre/+ , Tie2 Cre/+ , and Lyz2 Cre/+ mice were crossed with Gsdmd fl/fl mice to generate Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, and Gsdmd fl/fl Lyz2 Cre/+ mice, respectively, and their Cre -negative littermates ( Gsdmd fl/fl mice). Alb Cre/+ mice were crossed with Hmgb1 fl/fl mice to generate Hmgb1 fl/fl Alb Cre/+ mice and their Cre -negative littermates ( Hmgb1 fl/fl mice). WT mice, Gsdmd –/– mice, Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, Gsdmd fl/fl mice, Hmgb1 fl/fl Alb Cre/+ mice, and Hmgb1 fl/fl mice were used for in vivo experiments. The mice were housed under a 12-hour light/12-hour dark cycle in a temperature-controlled specific pathogen–free environment with ad libitum access to a regular chow diet and water. The genotypes were confirmed via PCR as previously described ( ). For identification of the Gsdmd -floxed allele, the primers used were as follows: forward primer, TCTGTTCCCTCCAGCCCTACTTG; reverse primer, CAGCAACCACAGCACTACGTTC. The WT allele corresponded to a band of 223 bp, and the floxed allele yielded a product of 321 bp. The forward primer CGATGGAACGTAGTGCTGTG and reverse primer TCCTTCCCAACCTGCTGTTG were used for genotyping the Gsdmd –/– mice. The WT allele yielded a band of 550 bp, and the conventional knockout allele yielded a band of 423 bp. The forward primer AGCGATGGATTTCCGTCTCTGG and the reverse primer AGCTTGCATGATCTCCGGTATTGAA were used to examine Alb-Cre transgenic mice, Tie2-Cre transgenic mice, and Lyz2-Cre transgenic mice, which resulted in a band of 272 bp, whereas Cre -negative mice presented no band. For identification of the Hmgb1 -floxed allele, the primers used were as follows: forward primer, TGTCATGCCACCCTGAGCAGTT; reverse primer, TGTGCTCCTCCCGGCAAGTT. The WT allele corresponded to a 172 bp band, and the floxed allele yielded a 280 bp product. Endotoxic shock model. For LPS-induced endotoxemia, 10- to 12-week-old WT mice, Gsdmd –/– mice, Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, Gsdmd fl/fl mice, Hmgb1 fl/fl Alb Cre/+ mice, and Hmgb1 fl/fl mice were injected intraperitoneally with a lethal dose of LPS (17.5 mg/kg). LPS-induced endotoxemia was also established via intratracheal delivery. Briefly, WT mice between 10 and 12 weeks of age were anesthetized and subjected to intratracheal administration of LPS (17.5 mg/kg) in 50 μL of sterile PBS. The survival of the mice was observed and recorded every 8 hours. CS model of sepsis. Sepsis was induced in the mice via the CS method described by Rincon et al. ( ). Briefly, 10- to 12-week-old WT mice were euthanized, the skin was opened, and the cecum was excised. The cecal contents were suspended in 5% dextrose to produce a final concentration of 60 mg/mL CS. For sepsis, 10- to 12-week-old WT mice, Gsdmd –/– mice, Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, and Gsdmd fl/fl mice were injected intraperitoneally with a dose of CS (2.5 mg/g). The survival of the mice was observed and recorded every 8 hours. Vascular permeability measurements. An Evans blue–albumin extravasation assay was performed to assess endothelial permeability ( , , ). Briefly, anesthetized mice were injected with 50 μL of 7% Evans blue dye (No. E2129, Merck) via retro-orbital injection. After 10 minutes, the intravascular Evans blue dye was removed via PBS perfusion (20 mL) through the left ventricle. Mouse aortas (from the aortic arch to the iliac arteries) and lungs were harvested, air-dried, weighed, homogenized, and extracted in 0.5 mL and 1 mL formamide (No. 47671, Merck), respectively, for 24 hours at 60°C. The quantity of Evans blue dye in the aorta and lung homogenate supernatants was determined spectrophotometrically at an absorbance of 620 nm. The Evans blue dye content is expressed as micrograms per gram (μg/g) of aorta or lung. Determination of the lung wet/dry weight ratio. The lungs were excised from the mice, and wet weights were immediately measured. After the samples were dried at 60°C for 48 hours, the dry weights were measured. The lung wet/dry weight ratio was then calculated ( – ). AAV9 vector injection. A Rage shRNA-knockdown AAV9 vector, a Tlr4 shRNA-knockdown AAV9 vector, a Gsdmd shRNA-knockdown AAV9 vector, and a null AAV9 vector as a negative control harbored a Tie2 promoter, which contributed to the specific knockdown of target genes in endothelial cells. A Rage shRNA-mediated knockdown AAV9 vector and a null AAV9 vector harbored an Sp-c promoter, which induced specific knockdown of target genes in type II alveolar epithelial cells. All the recombinant AAV9 vectors were constructed by Genomeditech Co., Ltd. (Shanghai, China). Five-week-old WT mice were injected with 100 μL of virus containing 1 × 10 12 vector genomes via the tail vein. siRNA injection. The Gsdmd -siRNA and negative control used for tail vein injection were provided by BiOligo Biotechnology Co., Ltd. (Shanghai, China). Each mouse received a freshly prepared mixture (10 nmol of siRNA dissolved in 200 μL of saline) on days 2, 4, and 6 before endotoxemia or sepsis model construction. Histopathology. The mice were euthanized under deep anesthesia 16 hours after treatment with LPS or CS. Aortas (from the aortic arch to the iliac arteries), and lungs were collected and fixed in 4% paraformaldehyde (No. G1101, Servicebio) for 24 hours. The tissues were then dehydrated through a graded ethanol series, embedded in paraffin wax, and finally sectioned into 3 μm sections. The sections were dewaxed and stained with hematoxylin and eosin (No. G1076, Servicebio), dehydrated in ethanol and n -butanol, and cleared in xylene before being mounted with neutral balsam. Images were captured using a Leica microscope. Dehydrated paraffin sections were pretreated via heat-mediated antigen retrieval with antigen repair buffer (No. B0035, POWERFUL). The slides were incubated with 5% BSA (No. A8010, Solarbio) at room temperature for 30 minutes and stained with a rabbit anti-GSDMD antibody (1:200, No. NBP2-33422, Novus) and a goat anti-platelet endothelial cell adhesion molecule (CD31) antibody (1:200, No. AF3628, R&D Systems, Bio-Techne) overnight at 4°C. The sections were washed 3 times with PBS (No. G0002-2 L, Servicebio), followed by a 1-hour incubation at room temperature with an Alexa Fluor 488–conjugated donkey anti-rabbit IgG (H+L) cross-adsorbed secondary antibody (1:400, No. A-21206, Invitrogen) and an Alexa Fluor 555–conjugated donkey anti-goat IgG (H+L) cross-adsorbed secondary antibody (1:400, No. A-21432, Invitrogen). Nuclei were stained with antifade mounting medium containing DAPI (No. B0025, POWERFUL) at room temperature for 10 minutes. Images were acquired via a Nikon microscope. All histopathological assessments were performed by researchers who were masked to the experimental groups. Western blotting. Mouse aortas (from the aortic arch to the iliac arteries) and livers were surgically removed and lysed as previously described ( ). Proteins (20–40 μg) were separated on 10% or 12.5% sodium dodecyl sulfate-polyacrylamide gels (No. PG113, EpiZyme) and transferred onto polyvinylidene difluoride membranes (No. ISEQ00010, MilliporeSigma). The membranes were blocked with protein-free rapid blocking buffer (No. PS108P, EpiZyme) for 10 minutes at room temperature and incubated overnight at 4°C with primary antibodies, including mouse anti-GSDMD (1:500, No. sc-393656, Santa Cruz Biotechnology) and β-actin (1:20,000, No. BS6007MH, Bioworld) antibodies. The membranes were incubated with Peroxidase AffiniPure goat anti-mouse IgG (H+L) (1:5,000, No. 115-035-003, Jackson ImmunoResearch) secondary antibody for 2 hours at room temperature. Bands were visualized via enhanced chemiluminescence (No. RPN2235, Cytiva). Band intensity was quantified via Quantity One software (Bio-Rad), and the data were normalized against those of β-actin. ELISA. Blood was collected following cardiac puncture and centrifuged at 1,006 g at 4°C for 15 minutes. The levels of IL-1β and HMGB1 in mouse plasma were determined via a commercially available IL-1β ELISA Kit (No. EK201B, MULTISCIENCES) and an HMGB1 ELISA Kit (No. ARG81310 and No. ARG81351, Arigo) according to the manufacturer’s instructions. Statistics. Statistical analysis was performed via GraphPad Prism software, version 8.0. The quantitative data are shown as the means ± SEMs. A log-rank (Mantel-Cox) test was used to compare survival curves. When the data were normally distributed and the variances between groups were equal, data were further analyzed with an unpaired 2-tailed Student’s t test for comparisons of 2 groups or with 1-way ANOVA followed by Bonferroni’s post hoc correction for the comparison of multiple groups. Analyses of the effects of different treatments on mice with different phenotypes were performed via 2-way ANOVA with Bonferroni’s post hoc correction. A P value less than 0.05 was considered significant. Study approval. All the animal studies were approved by the Institutional Animal Care and Use Committee of the Shanghai Research Center for Model Organisms, Shanghai, China. All experimental procedures involving animals were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, revised 1996, National Academies Press). Data availability. The values for all the data points in the graphs are reported in the file.
Our study examined male and female animals, and similar findings are reported for both sexes.
C57BL/6J (WT) mice (No. 219) were purchased from Charles River Laboratories Co., Ltd. (Beijing, China). Gsdmd –/– mice and Gsdmd fl/fl mice generated via the CRISPR/Cas9 system were purchased from GemPharmatech Co., Ltd. (Nanjing, China) ( ). Hmgb1 fl/fl mice were gifted by Tadatsugu Taniguchi from the University of Tokyo, Tokyo, Japan ( ). Alb Cre/+ mice (Stock No. 003574), Tie2 Cre/+ mice (Stock No. 008863), and Lyz2 Cre/+ mice (Stock No. 004781) were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA) ( – ). Alb Cre/+ , Tie2 Cre/+ , and Lyz2 Cre/+ mice were crossed with Gsdmd fl/fl mice to generate Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, and Gsdmd fl/fl Lyz2 Cre/+ mice, respectively, and their Cre -negative littermates ( Gsdmd fl/fl mice). Alb Cre/+ mice were crossed with Hmgb1 fl/fl mice to generate Hmgb1 fl/fl Alb Cre/+ mice and their Cre -negative littermates ( Hmgb1 fl/fl mice). WT mice, Gsdmd –/– mice, Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, Gsdmd fl/fl mice, Hmgb1 fl/fl Alb Cre/+ mice, and Hmgb1 fl/fl mice were used for in vivo experiments. The mice were housed under a 12-hour light/12-hour dark cycle in a temperature-controlled specific pathogen–free environment with ad libitum access to a regular chow diet and water. The genotypes were confirmed via PCR as previously described ( ). For identification of the Gsdmd -floxed allele, the primers used were as follows: forward primer, TCTGTTCCCTCCAGCCCTACTTG; reverse primer, CAGCAACCACAGCACTACGTTC. The WT allele corresponded to a band of 223 bp, and the floxed allele yielded a product of 321 bp. The forward primer CGATGGAACGTAGTGCTGTG and reverse primer TCCTTCCCAACCTGCTGTTG were used for genotyping the Gsdmd –/– mice. The WT allele yielded a band of 550 bp, and the conventional knockout allele yielded a band of 423 bp. The forward primer AGCGATGGATTTCCGTCTCTGG and the reverse primer AGCTTGCATGATCTCCGGTATTGAA were used to examine Alb-Cre transgenic mice, Tie2-Cre transgenic mice, and Lyz2-Cre transgenic mice, which resulted in a band of 272 bp, whereas Cre -negative mice presented no band. For identification of the Hmgb1 -floxed allele, the primers used were as follows: forward primer, TGTCATGCCACCCTGAGCAGTT; reverse primer, TGTGCTCCTCCCGGCAAGTT. The WT allele corresponded to a 172 bp band, and the floxed allele yielded a 280 bp product.
For LPS-induced endotoxemia, 10- to 12-week-old WT mice, Gsdmd –/– mice, Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, Gsdmd fl/fl mice, Hmgb1 fl/fl Alb Cre/+ mice, and Hmgb1 fl/fl mice were injected intraperitoneally with a lethal dose of LPS (17.5 mg/kg). LPS-induced endotoxemia was also established via intratracheal delivery. Briefly, WT mice between 10 and 12 weeks of age were anesthetized and subjected to intratracheal administration of LPS (17.5 mg/kg) in 50 μL of sterile PBS. The survival of the mice was observed and recorded every 8 hours.
Sepsis was induced in the mice via the CS method described by Rincon et al. ( ). Briefly, 10- to 12-week-old WT mice were euthanized, the skin was opened, and the cecum was excised. The cecal contents were suspended in 5% dextrose to produce a final concentration of 60 mg/mL CS. For sepsis, 10- to 12-week-old WT mice, Gsdmd –/– mice, Gsdmd fl/fl Alb Cre/+ mice, Gsdmd fl/fl Tie2 Cre/+ mice, Gsdmd fl/fl Lyz2 Cre/+ mice, and Gsdmd fl/fl mice were injected intraperitoneally with a dose of CS (2.5 mg/g). The survival of the mice was observed and recorded every 8 hours.
An Evans blue–albumin extravasation assay was performed to assess endothelial permeability ( , , ). Briefly, anesthetized mice were injected with 50 μL of 7% Evans blue dye (No. E2129, Merck) via retro-orbital injection. After 10 minutes, the intravascular Evans blue dye was removed via PBS perfusion (20 mL) through the left ventricle. Mouse aortas (from the aortic arch to the iliac arteries) and lungs were harvested, air-dried, weighed, homogenized, and extracted in 0.5 mL and 1 mL formamide (No. 47671, Merck), respectively, for 24 hours at 60°C. The quantity of Evans blue dye in the aorta and lung homogenate supernatants was determined spectrophotometrically at an absorbance of 620 nm. The Evans blue dye content is expressed as micrograms per gram (μg/g) of aorta or lung.
The lungs were excised from the mice, and wet weights were immediately measured. After the samples were dried at 60°C for 48 hours, the dry weights were measured. The lung wet/dry weight ratio was then calculated ( – ).
A Rage shRNA-knockdown AAV9 vector, a Tlr4 shRNA-knockdown AAV9 vector, a Gsdmd shRNA-knockdown AAV9 vector, and a null AAV9 vector as a negative control harbored a Tie2 promoter, which contributed to the specific knockdown of target genes in endothelial cells. A Rage shRNA-mediated knockdown AAV9 vector and a null AAV9 vector harbored an Sp-c promoter, which induced specific knockdown of target genes in type II alveolar epithelial cells. All the recombinant AAV9 vectors were constructed by Genomeditech Co., Ltd. (Shanghai, China). Five-week-old WT mice were injected with 100 μL of virus containing 1 × 10 12 vector genomes via the tail vein.
The Gsdmd -siRNA and negative control used for tail vein injection were provided by BiOligo Biotechnology Co., Ltd. (Shanghai, China). Each mouse received a freshly prepared mixture (10 nmol of siRNA dissolved in 200 μL of saline) on days 2, 4, and 6 before endotoxemia or sepsis model construction.
The mice were euthanized under deep anesthesia 16 hours after treatment with LPS or CS. Aortas (from the aortic arch to the iliac arteries), and lungs were collected and fixed in 4% paraformaldehyde (No. G1101, Servicebio) for 24 hours. The tissues were then dehydrated through a graded ethanol series, embedded in paraffin wax, and finally sectioned into 3 μm sections. The sections were dewaxed and stained with hematoxylin and eosin (No. G1076, Servicebio), dehydrated in ethanol and n -butanol, and cleared in xylene before being mounted with neutral balsam. Images were captured using a Leica microscope. Dehydrated paraffin sections were pretreated via heat-mediated antigen retrieval with antigen repair buffer (No. B0035, POWERFUL). The slides were incubated with 5% BSA (No. A8010, Solarbio) at room temperature for 30 minutes and stained with a rabbit anti-GSDMD antibody (1:200, No. NBP2-33422, Novus) and a goat anti-platelet endothelial cell adhesion molecule (CD31) antibody (1:200, No. AF3628, R&D Systems, Bio-Techne) overnight at 4°C. The sections were washed 3 times with PBS (No. G0002-2 L, Servicebio), followed by a 1-hour incubation at room temperature with an Alexa Fluor 488–conjugated donkey anti-rabbit IgG (H+L) cross-adsorbed secondary antibody (1:400, No. A-21206, Invitrogen) and an Alexa Fluor 555–conjugated donkey anti-goat IgG (H+L) cross-adsorbed secondary antibody (1:400, No. A-21432, Invitrogen). Nuclei were stained with antifade mounting medium containing DAPI (No. B0025, POWERFUL) at room temperature for 10 minutes. Images were acquired via a Nikon microscope. All histopathological assessments were performed by researchers who were masked to the experimental groups.
Mouse aortas (from the aortic arch to the iliac arteries) and livers were surgically removed and lysed as previously described ( ). Proteins (20–40 μg) were separated on 10% or 12.5% sodium dodecyl sulfate-polyacrylamide gels (No. PG113, EpiZyme) and transferred onto polyvinylidene difluoride membranes (No. ISEQ00010, MilliporeSigma). The membranes were blocked with protein-free rapid blocking buffer (No. PS108P, EpiZyme) for 10 minutes at room temperature and incubated overnight at 4°C with primary antibodies, including mouse anti-GSDMD (1:500, No. sc-393656, Santa Cruz Biotechnology) and β-actin (1:20,000, No. BS6007MH, Bioworld) antibodies. The membranes were incubated with Peroxidase AffiniPure goat anti-mouse IgG (H+L) (1:5,000, No. 115-035-003, Jackson ImmunoResearch) secondary antibody for 2 hours at room temperature. Bands were visualized via enhanced chemiluminescence (No. RPN2235, Cytiva). Band intensity was quantified via Quantity One software (Bio-Rad), and the data were normalized against those of β-actin.
Blood was collected following cardiac puncture and centrifuged at 1,006 g at 4°C for 15 minutes. The levels of IL-1β and HMGB1 in mouse plasma were determined via a commercially available IL-1β ELISA Kit (No. EK201B, MULTISCIENCES) and an HMGB1 ELISA Kit (No. ARG81310 and No. ARG81351, Arigo) according to the manufacturer’s instructions.
Statistical analysis was performed via GraphPad Prism software, version 8.0. The quantitative data are shown as the means ± SEMs. A log-rank (Mantel-Cox) test was used to compare survival curves. When the data were normally distributed and the variances between groups were equal, data were further analyzed with an unpaired 2-tailed Student’s t test for comparisons of 2 groups or with 1-way ANOVA followed by Bonferroni’s post hoc correction for the comparison of multiple groups. Analyses of the effects of different treatments on mice with different phenotypes were performed via 2-way ANOVA with Bonferroni’s post hoc correction. A P value less than 0.05 was considered significant.
All the animal studies were approved by the Institutional Animal Care and Use Committee of the Shanghai Research Center for Model Organisms, Shanghai, China. All experimental procedures involving animals were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, revised 1996, National Academies Press).
The values for all the data points in the graphs are reported in the file.
ES, HJ, and ML conceptualized the study. ES, XS, and LW analyzed the data. ES, XS, LW, JX, XC, SX, and ML performed experiments. ES and XS wrote the original draft of the manuscript. ES, XS, LW, and JX reviewed and edited of the manuscript. HJ and ML supervised the study. We assigned the authorship according to the number of tasks undertaken by the author.
Supplemental data Unedited blot and gel images Supporting data values
|
Flourishing and the scope of medicine and public health | fd1b331c-8b15-4050-94a4-9878b8e78d66 | 11187398 | Psychiatry[mh] | I have elsewhere defined ‘flourishing’ as ‘the relative attainment of a state in which all aspects of a person’s life are good, including the contexts in which that person lives’. Defined as such, flourishing is an ideal. It is not something we ever fully attain in this life. We all are flourishing only in a relative sense with respect to that ideal. Flourishing is multi-dimensional. Certain aspects of a person’s life may be good, and others not. Flourishing arguably includes, among other things, one’s happiness, health, meaning, character, relationships and financial resources. Flourishing moreover arguably further extends to the contexts in which a person lives also being good. One might distinguish between flourishing and well-being, insofar as well-being concerns all aspects of a person’s life being good, as they pertain to that individual , whereas flourishing pertains to all aspects of a person’s life being good, including the contexts in which that person lives . We are social and communal beings, and part of our own flourishing is constituted by the well-being of our communities. Flourishing thus encompasses well-being. With this understanding, let us now turn to potential proposals for the aspects of flourishing that arguably fall within the proper scope of each of medicine, psychiatry and clinical counselling, and public health and public policy.
It seems clear that the practice of medicine ought to attend to the maintenance and restoration of the health of the body. However, the role of medicine with regard to aspects of a person’s life that extend beyond the body is more complex. On the one hand, many decisions concerning the health of the body affect an individual’s mental, social and spiritual life as well. In some instances, the promotion of physical health will positively contribute to a patient’s mental and social well-being. However, in other cases, various goods and ends may come into conflict. Various surgeries may extend years of disease-free survival but also seriously compromise a person’s quality of life, or capacity to work, or ability to function sexually. In such cases, the end of bodily health or longevity may come into conflict with other aspects of well-being. Decisions concerning promoting various aspects of the health of the body may potentially adversely affect other aspects of the ‘health of the person’ or that person’s flourishing. While it is reasonable that a clinician should take such implications into account in deciding, along with the patient, on the best course of action, it also seems clear that the role of clinician is not appropriately construed as the maximisation of all aspects of a patient’s flourishing. The physician is not interchangeable with a marital counsellor, priest or career coach. Different institutions and different caring offices have different roles with regard to addressing different aspects of well-being. However, given the implications of medical decisions concerning the health of the body for the health of the person, one way to construe the proper purview of medicine might be as follows : the proper purview of medicine may be taken to be the health of the body, along with those aspects of flourishing that are affected by decisions concerning the health of the body . This in no way makes clinicians responsible for the full flourishing of the person, but nevertheless acknowledges that their actions have a role in the promotion, restoration, and maintenance of such flourishing. Often, the promotion of bodily health will be consonant with the well-being of a person in a broader sense. Putting a cast on a broken arm will in most cases not only foster restored physical well-being but will also eventually facilitate mental and social well-being also. However, when surgeries or medications have serious side effects, various ends and goals can come into conflict, and in such cases, it is important to consider the well-being of patients, and their preferences and goals, and the priorities they give to various ends, more holistically. This, arguably, is part of the purview of medicine. Taking the purview of medicine to be the health of the body along with those aspects of flourishing that are affected by decisions concerning the health of the body need not be seen in conflict with the position taken by others that the end of medicine is health, with health understood as bodily health and in a narrower sense than the WHO definition. Even in maintaining that the end of medicine is bodily health, the purview or scope of considerations for medicine extends beyond this proper end because of the relation of medicine, and the decisions made within medicine, to other ends. When decisions affect a plurality of ends, these other ends must also be taken into account. The consideration of well-being in medical decision-making pertains to the role of clinicians as caregivers. Patients desire care for the whole person and most healthcare practitioners desire to care, not just for the body, but for the person. Patients value their capacity to carry out different roles in life, often within workplaces or family contexts, and their capacity to do so pertains to the well-being of the person. Such care for the person requires that the clinician have some sense of the whole of a person’s life. While specialisation and division of labour within medicine has its advantages with regard to technical capacity, it also has the potential to threaten the understanding of the whole. Attention must be given to a patient’s emotional, relational and potentially spiritual well-being and such matters often cannot be easily addressed or documented in medical records readily transferable across numerous medical practitioners. Addressing the well-being of the patient, even in a limited scope as this pertains to decisions concerning the health of the body, will require time and compassionate care. It will require training beyond narrow technical confines. It will require an understanding of what is constituted by a person’s flourishing; it will require a genuine concern for the well-being of others; and it will require healthcare systems and practices that allow clinicians to be attentive to such matters.
The proposal above concerning the purview of medicine as the health of the body and also those aspects of flourishing that are affected by decisions concerning the health of the body arguably pertains to internal medicine, but the proper purview of psychiatry may be yet broader still. Psychiatry is not infrequently envisioned as addressing mental health, understood broadly, and not only in terms of the health of the brain. Full consideration of mind-body relations extends beyond the scope of this essay. However, mental health, understood in a broader conception of the wholeness of the mind as it pertains to the entire human person, is not coextensive with, but does include a substantial portion of what might be understood by flourishing. While some may embrace a potentially expansive scope for psychiatry, including enhancing numerous aspects of a person’s flourishing, others may consider this beyond the proper purview of psychiatric practice. If, in contrast, mental health is understood in a much narrower sense as wholeness of the mind as it pertains to the proper functioning of the brain, this arguably tightens the conception considerably. Mental health in this narrower sense may be viewed by most as clearly being within the purview of psychiatry. There are complexities as to how wholeness of the mind is to be understood, or what types of mental functioning are to be considered normal, and there will be disputed territory, but the same is true with other aspects of health and medicine. However, restricting the purview of psychiatry to addressing mental health in this narrower sense may be viewed by some as in fact being too restrictive and especially so in light of the connections between a person’s physical, social and spiritual life and their mental well-being. The loss of a particular relationship, or some other good, may lead to a substantial decline in mental well-being. Because of such losses, depressive symptoms may ensue, even if the brain is functioning properly and the mental experience, although negative, may yet still be reasonable and normal in light of that loss, as is arguably often the case with bereavement. Helping a patient through a loss or through a relationship difficulty or through a relatively normal process of grief would be considered by many within the bounds of psychiatric practice. An intermediate position might be to take the purview of psychiatry to be the promotion, maintenance and restoration of the wholeness of the mind as it pertains to the proper functioning of the brain, along with those aspects of flourishing concerning which the patient and clinician together agree through dialogue to address . Such a position would take, at a minimum, as the object of psychiatric care, the wholeness of the mind that arises from the proper function or malfunction of the brain, and would allow the psychiatrist to restrict his or her attention to this narrower conception of mental health. However, it would also allow the joint pursuit, for those psychiatrists and those patients who desired it, of other aspects of the patient’s flourishing. Such a position admittedly introduces potential heterogeneity in the scope of psychiatry across practices. However, it allows a minimum set of goals which psychiatric practice should pursue while also allowing flexibility for the aims of specific psychiatrists or practices, to be considerably more expansive. Such a perspective would open the possibility for, though not require, within the practice of psychiatry, various activities and interventions intended to promote well-being, or to foster virtues that may be related to mental health, rather than more narrowly restricting focus to addressing mental disorders. For such approaches to be effective it would be important that the psychiatrist and the patient be relatively aligned as to these broader aims being a part of the clinical relationship. It would not necessarily be essential that the psychiatrist and patient fully agree on their understanding of flourishing, so long as there were sufficient agreement to pursue particular aspects of well-being valued by the patient. Moreover, there may, in many cases, be considerable overlap in the patient’s and the clinician’s understanding of well-being, even if this agreement is not perfect. Agreement as to joint pursuit of broader aims for the patient’s flourishing will often only be possible through dialogue concerning the life and goals of the patient, and concerning both the patient’s and the clinician’s understanding of well-being. In principle, this could be carried out also within medicine more broadly, though doing so may require the development of a broader set of competencies. The extent to which that might be done may vary by specialty and practice within medicine. While clarification as to how well-being is to be understood and what aspects of well-being are to be pursued may be viewed as a lofty aspiration, psychiatric care will often in practice itself contain an implicit understanding of what goods and ends are being sought. It may be helpful then to more explicitly clarify this, so as to facilitate better care and to come to an awareness of the areas of agreement, along with the differences in perspective, in what the patient and clinician see as the most important goals to pursue. Similar considerations pertain to clinical counselling as well. Within clinical psychology or counselling, the focus will generally be less on the proper functioning of the brain, and more on the person’s life as a whole, on the person’s flourishing. It may be the case that a clinical psychologist or a counsellor does not necessarily feel comfortable addressing all aspects of a person’s flourishing, or all of their various pursuits of well-being, but may feel comfortable addressing some subset of these. Through dialogue, it will often once again be possible to come to an agreement between the counsellor and the patient as to which aspects of a patient’s flourishing are going to be discussed and pursued. An understanding of the patient’s and the counsellor’s various views and values will help clarify the scope of what might be addressed. Consonant values and understandings may help broaden the scope of care, but once again a number of goals and ends, such as improving relationships and growth in character, may be shared even if the counsellor and patient come from somewhat different perspectives. The proper purview of counselling with regard to flourishing might thus be simply taken as those aspects of flourishing concerning which the patient and clinician together agree through dialogue to address .
While it might in principle be possible to achieve some level of agreement on the understanding of flourishing and on the most important values, goods, and ends between a patient and a clinician or potentially even between a patient and an entire practice, these considerations of agreement and consensus become more challenging within the context of public health and local, national and international public policy. In such contexts, decision-making within a pluralistic society must often take place amidst differing and competing visions as to what constitutes the good. As such, public health and public policy priorities are often reduced to matters of physical health and economic considerations. This is arguably often done because these are goods around which it seems to be comparatively easier to attain consensus. Physical health is nearly universally valued and is considered important both in its own right and also in that it often facilitates attaining other ends. Economic considerations likewise constitute important means in the pursuit of numerous, and potentially divergent, goals and ends. However, to reduce public health and public policy considerations to bodily health and economic resources is to effectively embrace a highly impoverished view of human well-being. Public health and public policy efforts ought to aspire to something greater. The difficulty, however, is that in a pluralistic context it can be challenging to navigate competing conceptions of the good. Even in pluralistic contexts, however, there is arguably more potential for achieving consensus concerning well-being than is often acknowledged. While the vast majority of people do indeed value bodily health, and having sufficient financial resources, they also care about more than these. Almost everyone desires to be happy; almost everyone wants to have a sense of meaning and purpose; almost everyone wants to strive to be a good person; almost everyone wants good relationships, and also good communities, contexts and environments. These are other aspects of well-being around which it might be possible to achieve a relatively broad consensus. If this is so, the potential implications for public health and public policy efforts are arguably far-reaching. Rather than a near exclusive focus on physical health and economic considerations, it might instead be proposed that the proper purview of public health and public policy ought to be those aspects of flourishing around which broad societal consensus can be attained . The degree of consensus that can be attained as to which aspects of what is understood as flourishing truly are good will of course vary by context, but in many cases, this would include not only physical health and income, but also happiness and life satisfaction, meaning and purpose, character and virtue, close social relationships, and good communities. The specification and expansion of the ends considered in policy does not necessarily lead to straightforward policy decision-making as certain policies may potentially enhance some aspects of flourishing and hinder others, or may enhance the flourishing of some persons but hinder that of others. What we should be seeking is ultimately the flourishing of the whole of society, and ideally policy efforts would encompass a twofold principle that advancing flourishing should pertain to all people, and also to the whole of the person. Whenever possible, policies should be sought that enhance various aspects of flourishing without hindering other aspects, and that enhance the flourishing of many while not impeding that of others. This is undoubtedly challenging, and conflicts will arise, but a focus on what values are clearly shared and on empowering more local communities may, in many settings, provide helpful ways forward. In any case, an expansion of the ends in view within policy would result in a rather broader set of considerations taken into account in decision-making than is the case at present. Given that as individuals we are ultimately aiming at the health of the person , at flourishing, and that we arguably ought to be doing so collectively as a society, a shift of public health and public policy efforts to promote flourishing, at least insofar as we can obtain general consensus, would seem desirable. There is, moreover, now also ample evidence that various aspects of psychological and social well-being themselves affect both physical health and economic outcomes. The ends that are being sought are consonant with one another. A widening, and a change in emphasis, with regard to public health and public policy would more powerfully advance a broader range of ends.
This paper has offered an outline of a framework concerning the scope of flourishing within medicine, psychiatry and counselling, and public health and public policy. These practices are of course inter-related, and further work on coordinating these practices with one another, and with other practices and institutions that promote flourishing such as schools, workplaces, religious communities and arts organisations, would be valuable. A clearer understanding of what aspects of flourishing do, and do not, lie within the bounds of each institution and practice has the potential to better enable the pursuit of societal well-being.
|
Practical Solutions for Problems in Blood Grouping and Crossmatching | fbb62e4a-a037-4e63-b179-cb9590f94817 | 8886265 | Internal Medicine[mh] | Red blood cell (RBC) compatibility testing is a crucial step for erythrocyte concentrate (EC) transfusion. Blood grouping, antibody screening, antibody identification, direct antiglobulin test (DAT), and crossmatching are different aspects of RBC compatibility testing. This review aims to provide information for practicing hematologists on how to use these tests to solve problems in blood grouping and crossmatching.
Most blood bank laboratories use column agglutination technology, commonly referred to as gel testing or card testing. shows the reaction strengths in gel testing.
There are more than 40 blood group systems and over 300 RBC antigens . The ability of a substance to induce antibody production is called immunogenicity. ABO system antigens and the D antigen from the Rh system are the most immunogenic antigens. Anti-A and anti-B are naturally occurring antibodies and are usually of the immunoglobulin (Ig) M type. Blood grouping involves two steps, forward grouping (reacting anti-A, anti-B, and anti-D antibodies with the person’s RBCs) and reverse grouping (reacting commercially available A- and B-type RBCs with the person’s plasma). Forward grouping reactions in particular must be very strong (+4) and the results of forward and reverse grouping should be complementary, as seen in . Subtypes of A and B antigens can cause weaker or mixed field reactions in forward grouping and sometimes anti-A1 can be identified in reverse grouping. The D antigen has numerous variations that affect serological reactions. If the D antigen is normal in structure but has fewer antigenic sites, it is called weak D (formerly Du). If it has a qualitative structural defect, it is called partial D. The most common partial D variant in white people is DVI. There are some special panels for the differentiation of weak D from partial D, but it is impossible to make this differentiation by routine serological testing. Therefore, we refer to these different types as D variants. People with partial D may produce anti-D when transfused with D-positive EC, while people with weak D will not . The other immunogenic RBC antigens can be identified serologically using specific antibodies. shows the most important antigens and antibodies in transfusion medicine.
A positive DAT shows that the RBCs are coated with antibodies. In DATs, the antigen-antibody reaction occurs in vivo. Adding anti-human globulin (AHG) to RBCs enables the reaction to be visualized in vitro. AHG is an antibody against human antibodies and can be polyspecific or monoclonal against IgG or complement C3.
Antibody screening detects antibodies against antigens other than A and B and is performed by indirect antiglobulin test (IAT) technique. In this test, the person’s plasma is mixed with at least two (preferably four) commercially available O-type RBCs, which should be selected to screen for antibodies against most immunogenic antigens. shows an antibody screening result. Positivity indicates that the patient’s plasma contains antibodies against RBC antigens that are reactive at body temperature. A positive screening test should be followed by antibody identification.
This test is technically the same as antibody screening but is performed with more types of RBCs and aims to identify the antibody(ies) detected in the screening test. The O-type RBCs used in antibody identification tests are collectively referred to as a “panel.” A panel should consist of at least 11 types of RBCs. An example of an antibody identification panel is shown in and . Interpreting the identification panel requires some experience and is rather time-consuming.
CM looks for unexpected antibodies in the recipient’s plasma against the RBCs in the EC. It is done by IAT technique and is thus actually an antibody screening test. A negative result is called CM-compatible.
When there is a problem in EC CM, the physician should ask the following questions: a. Is there a problem with the recipient’s ABO and D (Rh) blood grouping? b. Has the patient received any transfusions or been pregnant before (including miscarriages and abortions)? If so, was that within the last 3 months? c. Does the patient currently have hemolysis? d. What are the patient’s results from DAT and antibody screening (and antibody identification, if available)? e. What is the “pattern” of positive reactions in antibody screening (and antibody identification, if available)?
These problems are usually solved in blood bank laboratories. If the patient has subtype A, it will be safe to transfuse him or her with a CM-compatible O group EC, because A subtype recipients can produce anti-A1. D variant recipients should be considered D-negative when planning transfusions because if they carry a partial D variant, they may produce anti-D if transfused with D-positive EC. This method may result in the unnecessary transfusion of Rh-negative EC to weak D patients but it is a safer approach unless genetic testing can be done. Double populations or mixed field reactions can be seen in patients who have undergone hematopoietic stem cell transplantation (HSCT). shows one example. The blood bank should be informed about these patients and there should be algorithms for EC transfusions during the engraftment period of ABO-incompatible HSCT cases. shows the algorithm that the Turkish Society of Hematology published in 2020 . Ultimately, if the ABO type cannot be determined, CM-compatible O blood group transfusions almost always solve the problem.
Incompatible CM means that the recipient has an antibody against the RBCs in the EC. The first step is to determine whether it is an autoantibody (against the patient’s own RBC antigens), alloantibody (against non-self RBC antigens, i.e., foreign RBC antigens from transfusion or pregnancy), or both. Here are some clinical situations in which the questions above should be asked to plan the safest transfusion for the patient. If the patient has no history of transfusion or pregnancy and has a positive DAT result, the antibodies coating the RBCs should be autoantibodies because there was no exposure to foreign antigens. If the patient has hemolysis, the diagnosis is autoimmune hemolytic anemia (AIHA). In this case, if the patient needs to be transfused, there is no need to waste time trying to find a CM-compatible EC because that will be impossible. Autoantibodies will react with every RBC from every EC. Physicians should take responsibility and transfuse an EC with close follow-up. Such patients will hemolyze these transfused RBCs at the same speed they hemolyze their own RBCs. Autoantibodies are almost always against public antigens. Public antigens are high-frequency antigens, which means they can be found in nearly all RBCs. Therefore, autoantibodies will react and give the same reaction strength with all RBCs. A typical antibody screening result for a patient with autoantibodies can be seen in . Ideally, antibody identification should be performed. All identification panel cells would give the same reaction, including the auto-control ( ). If there is a reaction with all RBCs in the panel, it is called a pan-reactive panel. If the patient has been transfused or pregnant but not within the last 3 months and the DAT is positive, then once again, the antibodies coating the RBCs should be autoantibodies. However, since the patient was exposed to foreign antigens previously, there is a possibility that the patient’s plasma contains alloantibodies. If there are both allo- and autoantibodies, antibody screening and antibody identification will be pan-reactive but we will see a wavy pattern ( ). If the autoantibody is strong and causes a 4+ reaction, it may mask the alloantibodies. To eliminate the autoantibodies, auto-adsorption can be performed. Adsorption is a procedure in which antibodies are adsorbed onto the RBCs. The patient’s own RBCs can be used for this, but since self-RBCs are already coated with antibodies, there will be few antigenic sites left and the procedure must be repeated several times. If the positive reaction seen in the DAT is +3 or +4, it will be very difficult to adsorb the autoantibodies. In such a case, one alternative method is to phenotype the patient and find RBCs that match the patient’s negative antigens to use for adsorption, and another method is to subject the patient’s RBCs to gentle heat elution (see below) and use these “naked” RBCs for adsorption . Antibody identification is done after every adsorption procedure. If there is an autoantibody, positivity will decrease with every adsorption and eventually the panel will be negative. CM can be done with this auto-adsorbed plasma and should be compatible. If there is an alloantibody accompanying the autoantibody, alloantibody(ies) will be left alone after adsorbing and the appearance will be like that in . After that, antibody identification can be done. This process is clearly time-consuming; if a patient with AIHA needs to be transfused, delaying the transfusion can be life-threatening. If the antibody screening panel (and identification panel, if done) is equally pan-reactive and if there is no time to do the tests above, I suggest transfusion of CM-incompatible EC with close monitoring of vital signs. Besides ABO and D antigen matching, Ee, Cc, and K antigen matching is suggested. There is no need to select O RhD-negative blood if there is no doubt in the patient’s blood grouping. If there is a wavy pan-reactive pattern in the antibody screening (and identification panel) ( ) of an AIHA patient and there is no time or opportunity to perform adsorption, then I recommend transfusing the patient with an EC that gives the same reaction strength on DAT or auto-control. If the DAT is positive and the patient was transfused or pregnant within the last 3 months, foreign RBCs may be present in the patient’s circulation. In this case, we cannot easily attribute DAT positivity to autoantibodies; we have to perform elution. Elution is a process in which the antibodies coating the RBCs are dissociated into a liquid called the eluate. After elution, antibody identification is performed on the eluate to identify the antibodies on the RBCs. In clinical practice, if you do not have the time or opportunity to perform elution, and if the patient has received a transfusion due to AIHA within the last 3 months, the DAT was positive before transfusion, and antibody screening suggests an autoantibody, you can proceed with transfusing the patient. In the case of delayed hemolytic reaction (DHR), the DAT is positive, antibody screening results are wavy and positive like in , and the patient has a history of transfusion, usually within the last 15 days. The classical clinical picture will be a patient transfused for some reason other than immune hemolysis and the first transfusion is CM compatible, and then after 7 to 15 days, hemolysis starts, the DAT becomes positive, and antibody screening is positive. The DAT positivity is caused by RBCs from the previously transfused EC. In this case, elution will reveal an alloantibody. Antibodies like Fy a , which may be cleared from the plasma in as little as 3 months, are usually responsible for these reactions. To avoid DHR, if a patient is known to have had an alloantibody in the past, always transfuse with the corresponding antigen-negative EC. If the DAT is negative and antibody screening is positive, it must be due to an alloantibody. The patient will report a history of transfusion and/or pregnancy. Antibody identification should be performed in order to define the antibody(ies) and an EC that is antigen-negative for the corresponding antibody(ies) should be selected for CM. If there is no time or opportunity to identify the antibody, then the only option is to perform CM tests with different units and try to find one that is compatible. If there are negative cell(s) in the antibody screen panel ( ), then one can guess it will not take long to find a CM-compatible EC. If the screening panel is pan-reactive, it would be wise to contact an experienced blood bank laboratory without wasting time trying to find compatible units. Not all alloantibodies have the same hemolytic risk ( ). Rh, Kell, Kidd, and Duffy system antibodies and anti-S or anti-s antibodies in the MNS system can cause clinically significant hemolytic transfusion reactions . In the event of an alloantibody, if the antibody is one of the risky ones or cannot be identified, do not agree to transfuse a CM-incompatible EC unless the patient’s life is in danger. Unlike autoantibodies, hemolysis with alloantibodies is often unpredictable. One or two positive CM-incompatible units may cause a serious hemolytic transfusion reaction. On the other hand, if transfusion is absolutely necessary, than it should be done. If a CM-compatible EC cannot be found for a patient with alloantibody(ies), one option is to perform extensive phenotyping of the recipient. This phenotyping should include the most immunogenic antigens: Ee, Cc, K, Fy a , Fy b , Jk a , Jk b , and Ss. The recipient and donor RBCs should match. For example, if the patient is E-negative, the EC must be E-negative. If the patient has been transfused recently, there may be double populations, indicating antigen mismatch in previous transfusions, which makes phenotyping complicated. Another option is to transfuse the patient with the least incompatible EC and hope for the best. This is definitely not recommended but can be done if transfusion is urgently needed. In this case, the EC should be transfused slowly with close monitoring of the patient. After transfusion of 10-15 mL of EC, testing for intravascular hemolysis with an interim blood sample is strongly advised. However, an uneventful transfusion in this case should not be completely reassuring because it does not guarantee normal RBC survival in vivo. A simple scheme that summarizes how to differentiate auto- and alloantibodies can be seen in .
This review covers the most frequently encountered compatibility testing problems. However, the reader should keep in mind that there are important exceptions in all clinical scenarios. Open communication and collaboration between clinicians and blood bank personnel can solve many problems.
|
Assessment and comparative study of diosgenin doses in alleviating experimental periodontitis | 4af80535-c9a7-4f0a-b6b6-3c184173d966 | 11283694 | Anatomy[mh] | Various factors cause the bone loss including systemic diseases, trauma, osteoporosis and periodontal disease . Bone loss in jaws is critical because especially severe losses leads the loss of function and complicates dental treatment. Therefore, studies have been focused protective and therapeutic treatment of the bone loss . Periodontitis is an inflammatory disease in which the interactions between periodontal bacteria and the host tissue response lead to tissue destruction . Specific groups of oral bacteria populate in dental plaque play a precursor role in the development of periodontal disease, however that once the disease has been stimulated, other factors effect the progression of periodontitis and aggravate the treatment of disease . Increases in oxidative stress, proinflammatory cytokines, and osteoclast cells have major roles in periodontal destruction . Oxidative stress stimulates the transformation of precursor osteoclast cells into mature osteoclasts, leading to pathological changes, followed by the destruction of affected tissue . Reactive oxygen species (ROS) are highly reactive by-products of oxygen metabolism and they have crucial role in various cellular processes as signalling molecules . ROS causes apoptosis by reducing B-cell lymphoma 2 proteins (Bcl-2) and elevating the expression of Bcl-2-associated X protein (BAX) . In addition, the elevation of ROS levels can damage tissue cells by stimulating proinflammatory cytokine cells and modulating the several pathways such as activation of NFκB ligand (RANKL) pathway, decreasing the protective effect of Nuclear factor red line 2 related factor 2 pathway, c-Jun N-terminal kinase signaling pathway, NOD-like receptor protein 3 . RANKL is known a member of the tumor necrosis factor superfamily. RANKL is an apoptosis regulator gene and it is a binding partner of osteoprotegerin. RANKL is expressed by several types of cells, including osteoblasts, osteocytes, fibroblasts, and lymphocytes . RANKL induces the activation of osteoclast cells and osteoclastogenesis because it stimulates the formation of osteoclast precursor cells. RANKL-mediated osteoclastogenesis has a critical role in periodontal destruction (Fig. ) . Growth factors organize cellular activities and improve tissue healing by binding to specific cell receptors. Several studies have used growth factors to enhance periodontal tissue and bone regeneration . BMP-2 belongs to the TGF-β superfamily of proteins and it is a growth factor with roles in tissue regeneration, including the transformation of undifferentiated mesenchymal cells and enhancement of osteoblast differentiation . Furthermore, it stimulates the secretion of several osteoblastic-specific molecules, such as alkaline phosphatase (ALP), osteocalcin (OCN), and type I collagen (Col-1) . Various agents have been used to reduce the effects of ROS on periodontitis, and diosgenin (DG) is one of them. DG is a naturally occurring bioactive steroid saponin. It has been used in several steroidal drugs in the pharmaceutical industry because its chemical structure is similar to the structures of sex hormones . DG exhibits various therapeutic effects, including antioxidative, antidiabetic, anti-inflammatory, and antihyperlipidemic activities . Moreover, DG modulates RANKL and OCN levels, regulates oxidative stress, stimulates signaling in the BMP pathways, and prevents apoptosis . To our knowledge, no study has evaluated the effects of different doses DG treatment on periodontal destruction in rats with systemically healthy. Therefore, it is unclear the influence and mechanisms of different doses DG treatment in systemically healthy rats with periodontitis. Here, we hypothesized that DG has antioxidative, anti-inflammatory and anti-resorptive properties and it could prevent periodontal tissue destruction by decreasing RANKL levels, inhibiting periodontal inflammation and cell apoptosis, and inducing bone formation. This study was performed to investigate the therapeutic effects of DG on ALP, OCN, Col-1, BAX, Bcl-2, BMP-2, and RANKL levels, as well as alveolar bone loss (ABL), in rats with experimental periodontitis to ensure basic information for potential DG application and further researches studies. Animals All experimental procedures in the present study were approved by the University Ethics Committee for Animal Experiments, Denizli (PAUHADYEK-2018/33). The Animal Research: Reporting of In Vivo Experiments guidelines were followed in this study. Thirty-two male Wistar albino rats (4 months old, 350–400 g), which were obtained from Pamukkale University Experimental Surgery Application and Research Center, were used in this study. Before initiation of the experimental procedures, the rats were adapted to the experimental environment for ten days; they were housed separately in cages in a room at 21 ± 2 °C and with a 12-h light:12-h dark cycle. All animals had free access to water and food. G* Power 3.1 software was used to calculating of sample size, considering the global significance level of α = 0.05, a sampling power of 95%, and f = 0.86 . Rats were divided into groups by simple randomization using the coin flip method into four groups ( n = 8/group): control (non-ligated), periodontitis (P; ligature only), DG-48 (ligature + DG 48 mg/kg/day), and DG-96 (ligature + DG 96 mg/kg/day). The DG (Sigma-Aldrich, Saint Louis, MO, USA) was dissolved in distilled water and administered by oral gavage for 29 days, as in previous studies . Rats in the control and P groups were given 1 ml distilled water by oral gavage during the experiment. All rats were sacrificed at day 30 . Before sacrification, 50 mg/kg body weight of ketamine (Eczacibasi Ilac Sanayi, Istanbul, Turkey) and 5 mg/kg xylazine chloride (Virbaxil ® , São Paulo, Brazil) were used for general anesthesia. Hence, all animals were unconscious. The animals were stabilized and their head were placed in the small animal guillotine opening by a specialist animal technician for sacrification. Subsequently, the rats were decapitated rapidly. Induction of periodontitis model The experimental procedure was performed under general anesthesia. 50 and 5 mg/kg body weight of ketamine and xylazine chloride respectively were administered intraperitoneally to provide general anesthesia. The cervical areas of the first lower right and left mandibular molars were submarginally ligatured using a 4 − 0 sterile silk suture (Dogsan Ilac Sanayi, Istanbul, Turkey) to stimulate plaque accumulation and periodontal inflammation. The ligatures were checked daily by two operators to prevent the observer bias (AK and ALA). Three-dimensional imaging A supine-position cone-beam computed tomography (CBCT) unit (Newtom 5G-XL; QR, Verona, Italy) was used for three-dimensional imaging. The smallest field-of-view of this device (6 cm × 6 cm) was chosen; the exposure settings were 100 mm voxel, 110 kV, 11.4 mA, 9.0 s exposure time, 26.0 s scanning time, enhanced scan, boosted dose, and high-resolution (HiRes) mode. The unit’s proprietary software (NNT, version 12.1; QR) was used for image analysis. All specimens were exposed in the same position with the same exposure parameters. A dentomaxillofacial radiologist with 9 years of experience was blinded to the specimens (MO); this radiologist performed all tomographic procedures and analyzed the images. Figure shows three-dimensional reconstructed and cross-sectional slice images. The distance was measured at cementoenamel junction to the alveolar bone crest and averaged across six areas (the mesial, medial, and distal parts of the buccal–lingual surfaces) of the mandibular first molar teeth for evaluating the linear bone loss (in mm). Histopathological method After rats had been sacrificed, mandibular samples were obtained and fixed in 10% neutral-buffered formalin for histopathological evaluation. The samples were decalcified in a solution (Osteofast 1; Biognost, Zagreb, Croatia) for 2 weeks, then routinely processed using automatic tissue processor equipment (Leica ASP300S; Leica Microsystems, Wetzlar, Germany) and immersed in paraffin. Subsequently, a rotary microtome (Leica RM 2155; Leica Microsystems) was used to obtained 5 μm sections from each sample. Each sample was cut along the long axis of the tooth in the mesiodistal direction and stained with hematoxylin and eosin. Histopathological examinations were performed by a single specialist who was blinded to the samples (ÖÖ). Observations were conducted using a light microscope at ×40 magnification, according to a modified version of histopathological scoring criteria established by Leitao et al. . To standardize the data, five areas from each rat were evaluated and their averages were taken. Neutrophil leukocyte infiltrations were specifically assessed. Inflammatory cell infiltrations, alveolar bone resorption, and degeneration and destruction of the cementum were scored according to previous study . Immunohistochemical method The streptoavidin-biotin peroxidase technique was performed to the sections selected for immunohistochemical processing. Sections were immunohistochemically stained at 1/100 dilution for all primary antibodies using anti-BMP 2 (ab59348; all from Abcam plc, Cambridge, UK), anti-RANKL (ab216484), anti-ALP (ab224335), Bax (ab53154), anti-Bcl-2 (ab59348), anti-Col-1 (ab34710) and anti-OCN (ab93876) antibody kits according to the manufacturer’s recommendations. The sections were then embedded with hydrogen peroxide in 3% methanol for 20 min to eliminate activity of endogenous peroxidase. Sections were boiled twice for 5 min with citrate buffer solution and washed in phosphate buffered saline (PBS). The UltraVision Detection System Anti-Polyvalenti HRP Kit (Mouse and Rabbit Specific HRP/DAB Detection Kit-Micro-polymer, ab236466; Abcam plc) was used as the secondary antibody was used as the secondary antibody and 3,3’-diaminobenzidine (DAB) as the chromogen. Sections were incubated with primary antibodies for 60 min at room tempareture. Immunohistochemistry was then performed using biotinylated secondary antibody and streptavidin-alkaline phosphatase conjugate. Sections were incubated with DAB for 3–5 min. For negative controls, an antibody dilution solution was used instead of primary antibodies. Harris haematoxylin was used for contrast staining and slides were examined under a light microscope. Immunohistochemical findings were scored on a scale of 0 to 3, where 0 = no staining, 1 = mild staining, 2 = moderate staining, and 3 = heavy staining . All immunohistochemical evaluations were performed by a specialized pathologist who was blinded to the samples (ÖÖ). Immunohistochemical analyses were performed using ImageJ 1,48 version (National Institutes of Health, Bethesda MD). After the classic microscopic analyses, we obtained histomorphometric and immunohistochemical evaluations using an automated image analysis system (Olympus CX41; Olympus Corporation, Tokyo, Japan). The lesioned area was evaluated using proprietary software (cellSens Life Science Imaging Software System; Olympus Corporation). Statistical analysis The Shapiro–Wilk test was used to assess whether data exhibited normal distributions. The post hoc Duncan multiple comparison test and one-way analysis of variance were used to analyze the ABL. Independent variables (ALP, BAX, Bcl-2, BMP-2, Col-1, OCN, RANKL, and histopathological scores) were evaluated using the Kruskal–Wallis test. All data are reported as means ± standard deviations for each group ( p < 0.05). All analyses were conducted using SPSS software (version 23; IBM Corporation, Armonk, NY, USA). All experimental procedures in the present study were approved by the University Ethics Committee for Animal Experiments, Denizli (PAUHADYEK-2018/33). The Animal Research: Reporting of In Vivo Experiments guidelines were followed in this study. Thirty-two male Wistar albino rats (4 months old, 350–400 g), which were obtained from Pamukkale University Experimental Surgery Application and Research Center, were used in this study. Before initiation of the experimental procedures, the rats were adapted to the experimental environment for ten days; they were housed separately in cages in a room at 21 ± 2 °C and with a 12-h light:12-h dark cycle. All animals had free access to water and food. G* Power 3.1 software was used to calculating of sample size, considering the global significance level of α = 0.05, a sampling power of 95%, and f = 0.86 . Rats were divided into groups by simple randomization using the coin flip method into four groups ( n = 8/group): control (non-ligated), periodontitis (P; ligature only), DG-48 (ligature + DG 48 mg/kg/day), and DG-96 (ligature + DG 96 mg/kg/day). The DG (Sigma-Aldrich, Saint Louis, MO, USA) was dissolved in distilled water and administered by oral gavage for 29 days, as in previous studies . Rats in the control and P groups were given 1 ml distilled water by oral gavage during the experiment. All rats were sacrificed at day 30 . Before sacrification, 50 mg/kg body weight of ketamine (Eczacibasi Ilac Sanayi, Istanbul, Turkey) and 5 mg/kg xylazine chloride (Virbaxil ® , São Paulo, Brazil) were used for general anesthesia. Hence, all animals were unconscious. The animals were stabilized and their head were placed in the small animal guillotine opening by a specialist animal technician for sacrification. Subsequently, the rats were decapitated rapidly. The experimental procedure was performed under general anesthesia. 50 and 5 mg/kg body weight of ketamine and xylazine chloride respectively were administered intraperitoneally to provide general anesthesia. The cervical areas of the first lower right and left mandibular molars were submarginally ligatured using a 4 − 0 sterile silk suture (Dogsan Ilac Sanayi, Istanbul, Turkey) to stimulate plaque accumulation and periodontal inflammation. The ligatures were checked daily by two operators to prevent the observer bias (AK and ALA). A supine-position cone-beam computed tomography (CBCT) unit (Newtom 5G-XL; QR, Verona, Italy) was used for three-dimensional imaging. The smallest field-of-view of this device (6 cm × 6 cm) was chosen; the exposure settings were 100 mm voxel, 110 kV, 11.4 mA, 9.0 s exposure time, 26.0 s scanning time, enhanced scan, boosted dose, and high-resolution (HiRes) mode. The unit’s proprietary software (NNT, version 12.1; QR) was used for image analysis. All specimens were exposed in the same position with the same exposure parameters. A dentomaxillofacial radiologist with 9 years of experience was blinded to the specimens (MO); this radiologist performed all tomographic procedures and analyzed the images. Figure shows three-dimensional reconstructed and cross-sectional slice images. The distance was measured at cementoenamel junction to the alveolar bone crest and averaged across six areas (the mesial, medial, and distal parts of the buccal–lingual surfaces) of the mandibular first molar teeth for evaluating the linear bone loss (in mm). After rats had been sacrificed, mandibular samples were obtained and fixed in 10% neutral-buffered formalin for histopathological evaluation. The samples were decalcified in a solution (Osteofast 1; Biognost, Zagreb, Croatia) for 2 weeks, then routinely processed using automatic tissue processor equipment (Leica ASP300S; Leica Microsystems, Wetzlar, Germany) and immersed in paraffin. Subsequently, a rotary microtome (Leica RM 2155; Leica Microsystems) was used to obtained 5 μm sections from each sample. Each sample was cut along the long axis of the tooth in the mesiodistal direction and stained with hematoxylin and eosin. Histopathological examinations were performed by a single specialist who was blinded to the samples (ÖÖ). Observations were conducted using a light microscope at ×40 magnification, according to a modified version of histopathological scoring criteria established by Leitao et al. . To standardize the data, five areas from each rat were evaluated and their averages were taken. Neutrophil leukocyte infiltrations were specifically assessed. Inflammatory cell infiltrations, alveolar bone resorption, and degeneration and destruction of the cementum were scored according to previous study . The streptoavidin-biotin peroxidase technique was performed to the sections selected for immunohistochemical processing. Sections were immunohistochemically stained at 1/100 dilution for all primary antibodies using anti-BMP 2 (ab59348; all from Abcam plc, Cambridge, UK), anti-RANKL (ab216484), anti-ALP (ab224335), Bax (ab53154), anti-Bcl-2 (ab59348), anti-Col-1 (ab34710) and anti-OCN (ab93876) antibody kits according to the manufacturer’s recommendations. The sections were then embedded with hydrogen peroxide in 3% methanol for 20 min to eliminate activity of endogenous peroxidase. Sections were boiled twice for 5 min with citrate buffer solution and washed in phosphate buffered saline (PBS). The UltraVision Detection System Anti-Polyvalenti HRP Kit (Mouse and Rabbit Specific HRP/DAB Detection Kit-Micro-polymer, ab236466; Abcam plc) was used as the secondary antibody was used as the secondary antibody and 3,3’-diaminobenzidine (DAB) as the chromogen. Sections were incubated with primary antibodies for 60 min at room tempareture. Immunohistochemistry was then performed using biotinylated secondary antibody and streptavidin-alkaline phosphatase conjugate. Sections were incubated with DAB for 3–5 min. For negative controls, an antibody dilution solution was used instead of primary antibodies. Harris haematoxylin was used for contrast staining and slides were examined under a light microscope. Immunohistochemical findings were scored on a scale of 0 to 3, where 0 = no staining, 1 = mild staining, 2 = moderate staining, and 3 = heavy staining . All immunohistochemical evaluations were performed by a specialized pathologist who was blinded to the samples (ÖÖ). Immunohistochemical analyses were performed using ImageJ 1,48 version (National Institutes of Health, Bethesda MD). After the classic microscopic analyses, we obtained histomorphometric and immunohistochemical evaluations using an automated image analysis system (Olympus CX41; Olympus Corporation, Tokyo, Japan). The lesioned area was evaluated using proprietary software (cellSens Life Science Imaging Software System; Olympus Corporation). The Shapiro–Wilk test was used to assess whether data exhibited normal distributions. The post hoc Duncan multiple comparison test and one-way analysis of variance were used to analyze the ABL. Independent variables (ALP, BAX, Bcl-2, BMP-2, Col-1, OCN, RANKL, and histopathological scores) were evaluated using the Kruskal–Wallis test. All data are reported as means ± standard deviations for each group ( p < 0.05). All analyses were conducted using SPSS software (version 23; IBM Corporation, Armonk, NY, USA). CBCT findings Periodontitis was induced in all ligated groups, according to the CBCT findings. The results showed that the control group had no ABL. ABL was significantly lower in the DG-48 and DG-96 groups than in the P group ( p < 0.05; Figs. and ). Histopathological findings Histological examination showed that the control group had normal gingival tissue architecture and gingival epithelium; it showed no pathological findings. Hyperemia, ulcers in the gingival epithelial layer, inflammatory reactions in the gingival tissue and periodontal ligament, partial to severe cement destruction, and alveolar bone degradation were observed in the P group. Microscopic evaluations of the DG-48 and DG-96 groups revealed that the treatments ameliorated the pathological findings, compared with the P group. Furthermore, cellular infiltration, ABL, and cement destruction were reduced in the DG-96 group, compared with the DG-48 group (Fig. ). Immunohistochemical findings The expression patterns of ALP, Bcl-2, BAX, Col-1, BMP-2, OCN, and RANKL in mesenchymal cells in all groups were observed immunohistochemically. Positive immunoexpression was indicated by a brown color. During examinations of the ALP, BAX, Bcl-2, BMP-2, Col-1, OCN, and RANKL immunostained sections, slight to negative immunoexpression findings were observed in the control group. ALP, Bcl-2, BMP-2, Col-1, and OCN expression levels were significantly lower in the P group than in the control group ( p < 0.05). Treatment significantly increased the expression levels of ALP, Bcl-2, BMP-2, Col-1, and OCN in the DG groups, compared with the P group ( p < 0.05). Additionally, DG-96 was more effective than DG-48 for normalizing immunoexpression (Fig. ). Statistical analysis results of the immunohistochemical scores are shown in Fig. . BAX and RANKL expression levels increased in the P group, compared with the control group ( p < 0.05). Treatment significantly decreased RANKL and BAX levels in the DG groups, compared with the P group ( p < 0.05). Finally, DG-96 significantly decreased the expression levels of RANKL and BAX, compared with DG-48 ( p < 0.05; Figs. and ). Periodontitis was induced in all ligated groups, according to the CBCT findings. The results showed that the control group had no ABL. ABL was significantly lower in the DG-48 and DG-96 groups than in the P group ( p < 0.05; Figs. and ). Histological examination showed that the control group had normal gingival tissue architecture and gingival epithelium; it showed no pathological findings. Hyperemia, ulcers in the gingival epithelial layer, inflammatory reactions in the gingival tissue and periodontal ligament, partial to severe cement destruction, and alveolar bone degradation were observed in the P group. Microscopic evaluations of the DG-48 and DG-96 groups revealed that the treatments ameliorated the pathological findings, compared with the P group. Furthermore, cellular infiltration, ABL, and cement destruction were reduced in the DG-96 group, compared with the DG-48 group (Fig. ). The expression patterns of ALP, Bcl-2, BAX, Col-1, BMP-2, OCN, and RANKL in mesenchymal cells in all groups were observed immunohistochemically. Positive immunoexpression was indicated by a brown color. During examinations of the ALP, BAX, Bcl-2, BMP-2, Col-1, OCN, and RANKL immunostained sections, slight to negative immunoexpression findings were observed in the control group. ALP, Bcl-2, BMP-2, Col-1, and OCN expression levels were significantly lower in the P group than in the control group ( p < 0.05). Treatment significantly increased the expression levels of ALP, Bcl-2, BMP-2, Col-1, and OCN in the DG groups, compared with the P group ( p < 0.05). Additionally, DG-96 was more effective than DG-48 for normalizing immunoexpression (Fig. ). Statistical analysis results of the immunohistochemical scores are shown in Fig. . BAX and RANKL expression levels increased in the P group, compared with the control group ( p < 0.05). Treatment significantly decreased RANKL and BAX levels in the DG groups, compared with the P group ( p < 0.05). Finally, DG-96 significantly decreased the expression levels of RANKL and BAX, compared with DG-48 ( p < 0.05; Figs. and ). In the present study, we used histomorphometry, immunohistochemistry, and CBCT to evaluate the effects of DG dose on ABL in experimental periodontitis. To our knowledge, this is the first study regarding the effects of DG in healthy rats with experimental periodontitis. The doses of DG were determined on the basis of previous findings . There are three methods that are frequently used to induce periodontal disease, which are: ligature application, oral bacterial inoculation, and the lipopolysaccharide injection technique. In the ligature model, sterile non-absorbable sutures or orthodontic wires are widely used to induce local irritation and bacterial plaque accumulation. Secondly, mono and mixed cultures of periodontal bacteria are inoculated orally by gavage or topical application. Lastly, lipopolysaccharide extracted from pathogenic bacteria can be directly injected into the gingival sulcus to induce inflammation and stimulate osteoclastogenesis and alveolar bone loss. Among these methods, ligature application induces inflammation and alveolar bone resorption more promptly compared to other methods . Hence, we preferred this method for inducing bone loss. Micro-computerized tomography is regarded as the “gold standard” method for analyzing trabecular bone and tooth microstructure, evaluating the development of the skull bones, and assessing tissue engineering . However, several studies have evaluated the efficacy of CBCT as an alternative for assessing periodontal defects because micro-computerized tomography involves ultra-high radiation doses and is not routinely used in clinical settings . Thus, Tayman et al. investigated the use of CBCT to measure periodontal defects; they concluded that it provides useful linear and volumetric measurements of such defects in vitro . Other studies have suggested that CBCT can be used to evaluate periodontal defects and the structures and trabecular microarchitecture of alveolar bone . In an experimental study, Lektemur Alpan et al. demonstrated that CBCT measurements of ABL levels were accurate . Thus, we measured ABL using CBCT in this study. RANKL is the primary regulator of osteoclastogenesis; it has a critical role in osteoclast-associated diseases . Several studies have demonstrated that ABL is associated with high RANKL levels . Zhang et al. performed 24 mg/kg body weight/day, 48 mg/kg body weight/day and 96 mg/kg body weight/day as dosages of DG and they reported that a high dose of DG decreased bone loss by modulating the RANKL and osteoprotegerin levels in an ovariectomized rat model . In an another study, Zhang et al. evaluated protective effects of DG on ABL in ovariectomized rats and they indicated that DG inhibited osteogenesis and osteoclastogenesis by regulating the releasing of important molecules in the Wnt, RANKL or osteoclastogenic cytokine pathways . In previous study, we evaluated the effects of DG on RANKL in diabetic rats with periodontitis and our results showed that 96 mg/kg DG treatment significantly decreased in RANKL levels and ABL. In the present study, DG treatments significantly downregulated the RANKL levels and inhibited RANKL-induced osteoclastogenesis in rats in a dose-dependent manner, compared with the untreated periodontitis group. Furthermore, DG significantly decreased ABL in a dose-dependent manner, compared with the untreated group. These results suggested that DG prevents ABL by inhibiting RANKL expression and RANKL-induced osteoclastogenesis, consistent with the findings of earlier reports . Inflammation can increase oxidative stress, thus, worsening DNA damage and tissue apoptosis . Moreover, periodontal disease reportedly leads to an imbalance between pro- and anti‐apoptotic processes . Therefore, we evaluated apoptotic marker levels in our study. BAX is a member of the Bcl-2 family; expression levels of BAX and Bcl-2 are considered indicators of apoptosis or survival in cells . Wu et al. applied the 10, 50, or 100 mg/kg DG daily in ovariectomized rats and they reported that DG treatment decreases BAX and BAX/Bcl-2 levels and it has a therapeutic potential for ovariectomy-induced cardiac apoptosis . In vitro study demonstrated that 2, 6, and 8 µM doses of DG alleviates the apoptosis by maintaining the Bcl-2 expression . Additionally, 96 mg/kg DG treatment significantly reduced Bax and increased Bcl-2 levels in our previous experimental study . In the present study, we evaluated BAX and Bcl-2 levels to identify the effects of DG on apoptosis signaling pathways. The results showed that experimental periodontitis upregulated and downregulated the expression levels of BAX and Bcl-2, respectively. In contrast, DG treatment upregulated and downregulated the expression levels of Bcl-2 and BAX, respectively, in our experimental periodontitis model. Particularly, DG significantly increased the Bcl-2 levels in higher dose group than low dose group. These results indicate that dose depending DG treatment decreases periodontitis–induced apoptosis by suppressing the expression of BAX and inducing the expression of Bcl-2; these results are also consistent with previous findings . Additionally, apoptosis is a complex process, and additional markers or assays might be needed to confirm this effect comprehensively. Several biochemical markers have been used to evaluate bone metabolic activity, including ALP, Col-1, OCN, and BMP-2 . ALP is released by osteoblast cells; measurements of ALP level are used to evaluate osteoblastic activity. OCN controls mineral deposition; thus, it has critical roles in bone formation and remodeling . Furthermore, BMP-2 mediates the differentiation of osteoblastic cells and induces the release of ALP, OCN, and Col-1 . Zhao et al. performed 10 mg/kg, 30 mg/kg, and 90 mg/kg DG in retinoic acid-induced osteoporosis in rats and they indicated DG significantly reduced the ALP levels and increased OCN levels in 30 mg/kg, and 90 mg/kg DG groups and promoted bone formation and inhibits bone absorption by regulating bone metabolism and mineralization . Another study applied DG via oral gavage at a dosage of 100 mg/kg body weight daily and they found that DG could enhance the bone formation process through increased Wnt and BMP signaling activity; these pathways regulate the osteogenic differentiation of mesenchymal stem cells and preosteoblasts . Liao et al. found that the arginyl–DG conjugate stimulates BMP-2-induced osteoblastic differentiation with synergistic effects on ALP activity and mineralization . Additionally, Zhang et al. treated the DG group rats by oral gavage with 100 mg/kg body weight DG and they showed that DG has anti-bone loss efficiacy on rat alveolar bone by alleviating the OCN levels . In diabetic rats with experimental periodontitis, we used the 96 mg/kg DG treatment and previous results reported that DG treatment significantly improved the expression of ALP, OCN and BMP-2 in test group . In the present study, DG significantly promoted the expression of ALP, OCN and BMP-2 in test groups than P group. Also, high dose DG treatment significantly promoted BMP-2 and ALP levels compared the low dose group. These findings suggest that DG treatment, especially high dose of DG, enhances bone formation by increasing new bone activity through enhanced expression of ALP, OCN, and BMP-2; this is also consistent with previous findings . Col-1 is an important factor that stimulates osteoblast differentiation and mineral matrix deposition . The increasing of Col-1 supports the ABL formation in the experimental periodontitis . A few studies investigated the association between DG and Col-1 level. In our previous study, 96 mg/kg DG treatment significantly increased the Col-1 levels in diabetic rats with periodontitis . The present study showed periodontitis decreased the Col-1 level and both dose of DG treatment significantly increased the Col-1 level and confirmed the previous study the association between the periodontitis and Col-1 . However, further studies are needed to evaluating DG on Col-1 levels. This study have several limitation. We did not evaluate the effect of DG on the Wnt pathways or osteoprotegerin levels or other relevant markers of bone metabolism, inflammation and did not compare the CBCT findings with micro-computerized tomography; these were limitations of the present study. Therefore, further studies are needed to investigate the effects of DG on the other bone metabolic pathways and relevant markers expression in periodontal disease. Another limitation of our study is the inability to examine DG in humans by histological examination due to ethical barriers and potential side effects. Ligature induced periodontitis causes acute inflammation in rats however periodontitis is a chronic course in humans in terms of proinflammatory, anti-inflammatory cytokine activities and oxidant/antioxidant balance and that is a limitation. Additionally, we preferred DG doses according to previous studies nonetheless different doses of DG could be evaluated further studies. The present study indicated that both doses of DG—particularly the higher dose—regulate bone activity, prevent RANKL-induced osteoclastogenesis and improve new bone activity and bone formation. Although limitations, our results indicate that DG administration can prevent alveolar bone damage in periodontal disease. |
Environmental predictors impact microbial-based postmortem interval (PMI) estimation models within human decomposition soils | 6a5d5209-19e8-4cf5-bcc9-6ed20d22380a | 11469530 | Microbiology[mh] | Microbial communities undergo succession in response to disturbance events . Vertebrate death and subsequent decomposition represent one such event, where microbial community composition is altered in response to nutrient deposition and altered environmental conditions . Microbial succession has been studied in various carcass/cadaver decomposition microhabitats, including internal organs , skin , bone , and soils . These studies suggest that these successional changes may be robust and universal enough to be used to predict the postmortem interval (PMI), or time elapsed since death (or beginning of decomposition). PMI estimations can be important evidence for death investigations, allowing law enforcement to establish a timeline of events . The work developing microbial-based PMI models has been inspired by forensic entomology methods, which link insect succession or development to PMI. Entomological PMI estimation methods are widely used, but limited to forensic cases where insects are present, the species is identifiable, and temperature data can be collected. Unlike insects, microbes do not have a pre-colonization phase where they must be exposed to, detect, and accept a carcass , as they are host-associated and present in the surrounding environment . Together, this makes microbes advantageous for developing a forensic application estimating time since death. Thus far researchers have assessed microbial abundance-based PMI prediction models in all major microhabitats and three different mammalian species. This includes internal , external/skin , and soil microbial communities during pig ( Sus scrofa ) , mouse ( Mus musculus ) , and human ( Homo sapiens ) decomposition. These models suggest some level of predictability in microbial succession; however, PMI estimations differ based on the microhabitat, taxonomic level considered, and algorithm used for model construction. To date, most studies apply supervised machine learning algorithms to taxonomic abundance data derived from amplicon sequencing of conserved markers of a few taxa or whole microbial communities . Within these studies, random forest regression is the most frequently used supervised machine learning algorithm. These microbial PMI models report error ranging from ~15 hours for mouse intestine samples to ~58 hours in mouse brain samples , up to 138 accumulated degree days for human skin samples , and two to six days for soil samples below decomposing mice and humans . Of the three decomposition microhabitats in terrestrial decomposition systems ( i.e ., internal, external, and soil), soils have received the least attention. While there have been multiple studies assessing internal and external/skin succession , only two studies have included swabs of the soil surface ( i.e ., O horizon) during mouse and human decomposition, demonstrating repeatable succession . It is unknown if these PMI models could be applied to microbiological changes within mineral soil horizons. Further, it is unclear how inter-individual variation and potential species differences may impact model performance, and thus what the predictability would be across a large population of humans. Recent work suggests differences in decomposition patterns may exist between species and even within species, due to intrinsic carcass properties ( e.g ., body composition) . Consequently, there is a need to investigate the predictability of soil microbial succession within larger human sample sizes in order to assess applicability of these models for forensic science. Additionally, current microbial-based PMI estimation models are trained using relative abundance of microbial taxa as model features. However, observed changes in soil environmental parameters over time may also be used as indicators of decomposition time. For example, soil electrical conductivity (correlates to salinity), ammonium, and nitrate concentrations have been shown to undergo predictable changes in human decomposition soils . Thus, it is possible that the inclusion of environmental predictors along with taxon relative abundance may help to improve model estimation. The goals of this study were to 1) determine the utility of soil microbial communities for predicting decomposition time during human decomposition using 19 replicate human donors at a single location (East Tennessee); 2) determine which biological marker (16S rRNA, Internal Transcribed Spacer (ITS), or both) and taxonomic level (phylum, class, order, or OTU) results in the most accurate model predictions; and 3) assess how inclusion of soil environmental parameters (e.g., moisture, temperature, pH, conductivity, and enzyme rates) as model features affects model accuracy. Our first aim was to investigate microbial-based model performance across a human sample set (n = 19) collected in East Tenneessee to validate previously developed models. Our second aim was to evaluate model performance using different biological markers, i.e ., genes commonly used as sequencing targets. Previous work reported that 16S rRNA gene-based models performed better then 18S rRNA gene or ITS models from organic residues collected on soil surfaces . However, we hypothesized that ITS-based models would be more accurate than 16S-based models due to our previous obervations showing that fungal community composition between individuals became more similar over decomposition time, whereas bacterial communities did not, suggesting less noise and better predictability in the fungal communities . We also aimed to assess which taxonomic level(s) resulted in the greatest model accuracy. Based on previous results ), we hypothesized that higher taxonomic levels, such as phylum and class would provide better PMI prediction. Our third aim was to probe the impact of environmental features on model prediction. While no previous studies have addressed this question, we hypothesized that inclusion of environmental predictors known to change in decomposition soils would help to improve PMI model predictions. Soil pH and conductivity are known drivers of microbial community dynamics , while enzyme activities provide insight into functionality of the microbial community, so we chose to evaluate these parameters. We addressed our study aims using sequencing (16S rRNA and ITS2 amplicon) and soil physicochemical data from 19 deceased human individuals decomposed on the soil surface at the University of Tennessee’s Anthropology Research Facility (ARF) in Tennessee, USA. Random forest regressions were applied to datasets with different combinations of biological markers (16S only, ITS only, 16S and ITS combined) and taxonomic levels (phylum, class, order, OTU), both with and without environmental predictors. Model performance was then compared by calculating the mean absolute error (MAE) to determine the influence of different combinations of features on PMI estimation.
Study design This work uses datasets generated from our previous study , which revealed the influence of intrinsic, or cadaver-related factors, on explaining variation in soil microbial communities during human decomposition. The current study, however, uses these datasets to assess the effects of environmental factors on predictability of this succession to estimate PMI. Full experimental details are reported in . Briefly, decomposition of 19 deceased whole body human donors took place at the Anthropology Research Facility (ARF), located at the University of Tennessee in Knoxville, TN, USA (35°56’ 28” N, 83°56’ 25” W). The ARF is a forested outdoor facility consisting of clay loam and channery clay loam soils of the Coghill-Corryton complex (CcE) . Adult individuals with no open wounds or had not been autopsied were chosen for this study, as this could alter microbial decomposers prior to and during our study. Individuals were selected independent of demographic categories, however all individuals self-identified as White and ranged in age from 40 to 91 years . All individuals were whole body donors to the Forensic Anthropology Center ( https://fac.utk.edu/body-donation/ ) specifically for the purpose of decomposition research. No living human subjects were involved and only donors who consent to decomposition research on their donation paperwork were enrolled in this study. The University of Tennessee, Knoxville, Human Research Protections Program (HRPP) reviewed this project and determined that research with human donors is exempt under 45 CFR 46.101. Individuals were placed supine unclothed on the soil surface between February 2019 and March 2020 . Hourly temperatures were recorded using TinyTag temperature and humidity loggers (Gemini Data Loggers, UK) until un-enrollment at the end of active decomposition, characterized by collapse of the abdomen and cessation of fluid leaking from the trunk . Accumulated degree hours (ADH) were calculated using hourly temperature readings: 0 ADH was defined as time of placement within ARF, and a baseline temperature of 10°C was used for ADH calculations to keep our results comparable with entomology-based methods . Soil sampling and analysis Five-cm soil cores were collected from the decomposition-impacted area surrounding each individual (within ~ 7.6 cm of the body), as well as from control sites located at least 1 m away from the donor (either upslope or at the same elevation) at predetermined accumulated degree hour (ADH) intervals until the end of active decomposition . ADH intervals included 0 (prior to placement), 100, 250, 500, 750, and 1000 ADH, and thereafter at 500 ADH intervals until un-enrollment. For each respective sample, cores were homogenized and debris ( e.g ., roots, insect larvae, rocks, etc.) removed by hand. A subset of soils (~ 20 g) were stored in a 4 oz. Whirl-Pak bag (Nasco), flash frozen in liquid nitrogen and stored at -80°C prior to DNA extraction and extracellular enzyme assays. The remaining soil was was stored in a 7 oz. Whirl-Pak bag (Nasco) at 4°C for soil physiochemical measurements . Soil slurries were prepared as a 1:2 ratio of soil to deionized water, allowed to come to room temperature for 30 minutes, and soil pH and electrical conductivity (EC) were measured using an Orion Star ™ A329 pH/ISE/Conductivity/Dissolved Oxygen portable multiparameter meter (ThermoFisher). Gravimetric soil moisture was measured in duplicate by oven drying 2 to 3 g soil aliquots at 105°C for 72 hours. Enzyme activities of β -glucosidase (BG), N-acetyl- β -D-glucosaminidase (NAG), leucine amino peptidase (LAP), and alkaline phosphatase (PHOS) were measured according to a modified procedure by Bell et al. (2013) . Breifly, 2.75 g of soil was weighed from soils stored at -80°C and held at -20°C until assays. Soils were thawed at room temperature prior to slurrying in 50 mM Tris buffer at pH 6.7 in a blender (Waring commercial blender, model WF2212114). Assays were conducted in triplicate using 800 μl of slurry and 200 μl of enzyme substrate (1,500 μM). Standard curves (MUB or MUC) were evaluted for each plate with conentrations ranging from 0 μM to 200 μM . DNA extraction, sequencing, and amplicon sequence analysis DNA was extracted from soils stored at -80°C . Briefly, 0.25 g of soil was extracted with the DNeasy Powerlyzer PowerSoil kit (QIAGEN Inc.) following manufacturer’s instructions with modifications for our soil texture (clay loam) and condition (high organic content). Specifically, soils were homogenized under parameters suggested for high organic soils (2,500 RPM for 45 s). DNA concentration was determined using a fluorometric assay (Quant-iT ™ PicoGreen ® dsDNA Assay Kit, Invitrogen) with a total volume of 200 μ l and 1 μ l of DNA. All DNA extracts were sent to the University of Tennessee Sequencing Core Facility (Knoxville, TN) for 16S rRNA and ITS2 region amplicon sequencing on the Illumina MiSeq platform (2 x 150 bp). The primer set 515F /806R was used to amplify the V4 region of the 16S rRNA gene, while the ITS2 region in fungi was amplified using a mixture of primers (6 forward and 2 reverse: ITS3NGS1, ITS3NGS2, ITS3NGS3, ITS3NGS4, ITS3NGS5, ITS3NGS10, ITS4NGR, and ARCH-ITS4) described previously . All raw sequences have been deposited in the National Center for Biotechnology Information’s Sequence Read Archive under the BioProject PRJNA817528 . Raw sequences were processed in Mothur (v.1.43.0) to cluster into 97% similarity operational taxonomic units (OTUs) and generate OTU count tables for both 16S and ITS datasets as described in . Briefly, paired-end reads were combined into contings, removing low-quality sequences (16S: Q > 20, bp ≤ 50; ITS Q > 20, bp < 200), sequences with ambigious bases (≥ 1), and primers/adpaters. Chimeras were removed using VSEARCH. Remaining sequences were classified using the SILVA non-redundant database (v132) or UNITE RefS database (version 02.02.2020) for 16S and ITS sequences, respectively. Bacterial sequences were then clustered into OTUs based on ≥ 97% sequence similarity and the Mothur default method, opticlust, while fungal sequences were clustered using abundance-based greedy clustering. We chose to cluster our sequences into OTUs rather than ASVs to reduce dimensionality in our dataset and the probability of splitting single genomes across multiple ASVs , especially when considering the diversity expected across soil microbial genomes. Count tables were then exported for analysis in R (version 4.4.0). Control samples ( e.g ., those not exposed to decomposition) and samples greater than 5000 ADH were removed using phyloseq (v1.44.0). Samples were cut off at 5000 ADH to capture the linear response of soil parameters and account for variation in decomposition timeframes between individuals . This resulted in 78 samples from 19 individuals (mean = 4.1 samples per individual) for model construction. Machine learning models Read counts were total sum scaled (TSS) by determining the relative abundance of each OTU and normalizing to a standard library size (10,000 for all samples) using phyloseq (v1.44.0). This allowed for comparison of reads across samples and between biomarkers. We also removed OTUs with less than 10 reads across all samples in TSS normalized count tables to reduce noise in the datasets. 16S and ITS TSS read count tables were generated at the phylum, order, and class levels by summing the corresponding OTU table at each respective taxonomic level and then applying the TSS normalization as described above. Taxonomic levels were chosen to represent a subset which covered the full range from phylum to OTU. One of our goals was to compare predictability of bacterial (16S only), fungal (ITS only), or both (16S-ITS) communities; therefore, after TSS normalization, 16S-ITS combined datasets for each taxonomic level (phylum, order, class, OTU) were generated by merging respective 16S and ITS TSS count tables. As a result, 12 datasets were created and used for random forest models. We chose to apply random forest regression to datasets to predict PMI in ADH. This kept our study similar to those previously conducted on decomposition residues collected from soil surfaces , while also assessing predictability of soil microbial succession during human decomposition in our geographical region (Knoxville, TN). Model construction was completed in R using the Ranger (v0.16.0) package. First, samples were assigned to testing or training datasets. This was completed by randomly assigning 6 donors (~1/3) to the test set, while the remaining 13 were grouped into the training set. This approach was conducted following Belk et al. (2018) , to ensure that all samples from a single individual were in either the testing or training set, respectively, to prevent overfitting. Next, random forest regressions were applied to microbial taxa TSS normalized count tables in Ranger. First, random forest model parameters node size (3, 5, 7, 9) and sample size (0.55, 0.632, 0.70, 0.80) were hyper-tuned by comparing models with different combinations of the parameters listed. The optimal model was chosen by assessing the out-of-bag mean square error (OOB MSE) of each model and choosing the set of parameters with the lowest OOB MSE. The optimum model for each biomarker and taxonomic level was assessed by calculating the OOB MSE of the model and the root mean square error (RMSE) and mean absolute error (MAE) for predictions of the testing set in 100 runs of the optimum model. RMSE and MAE were calculated using rmse and mae functions from the R package Metrics (v 0.1.4). This process was repeated for models including measured environmental parameters, with values for ambient temperature (°C), pH, electrical conductivity (EC), moisture, β -glucosidase (BG) activity, N-acetyl- β -D-glucosaminidase (NAG) activity, leucine amino peptidase (LAP) activity, and alkaline phosphatase (PHOS) activity included as model features . For all environmental parameters, aside from temperature, log response ratio normalized values were used to account for natural seasonal differences in these parameters. The top 25 most influential model features were extracted from each optimum model to assess taxa/environmental factors influencing model predictions. To evaluate the potential differences in model predication between biomarkers and taxonomic levels, linear regression was applied to the average MAE values (mean of 100 runs per model). Variation in MAE due to treatment variables was assessed with ANOVA, while differences between treatment groups were determined with post-hoc t-tests in R. Code for generating all feature tables and random forest model development can be found at https://github.com/jdebruyn/TOX-microbiology .
This work uses datasets generated from our previous study , which revealed the influence of intrinsic, or cadaver-related factors, on explaining variation in soil microbial communities during human decomposition. The current study, however, uses these datasets to assess the effects of environmental factors on predictability of this succession to estimate PMI. Full experimental details are reported in . Briefly, decomposition of 19 deceased whole body human donors took place at the Anthropology Research Facility (ARF), located at the University of Tennessee in Knoxville, TN, USA (35°56’ 28” N, 83°56’ 25” W). The ARF is a forested outdoor facility consisting of clay loam and channery clay loam soils of the Coghill-Corryton complex (CcE) . Adult individuals with no open wounds or had not been autopsied were chosen for this study, as this could alter microbial decomposers prior to and during our study. Individuals were selected independent of demographic categories, however all individuals self-identified as White and ranged in age from 40 to 91 years . All individuals were whole body donors to the Forensic Anthropology Center ( https://fac.utk.edu/body-donation/ ) specifically for the purpose of decomposition research. No living human subjects were involved and only donors who consent to decomposition research on their donation paperwork were enrolled in this study. The University of Tennessee, Knoxville, Human Research Protections Program (HRPP) reviewed this project and determined that research with human donors is exempt under 45 CFR 46.101. Individuals were placed supine unclothed on the soil surface between February 2019 and March 2020 . Hourly temperatures were recorded using TinyTag temperature and humidity loggers (Gemini Data Loggers, UK) until un-enrollment at the end of active decomposition, characterized by collapse of the abdomen and cessation of fluid leaking from the trunk . Accumulated degree hours (ADH) were calculated using hourly temperature readings: 0 ADH was defined as time of placement within ARF, and a baseline temperature of 10°C was used for ADH calculations to keep our results comparable with entomology-based methods .
Five-cm soil cores were collected from the decomposition-impacted area surrounding each individual (within ~ 7.6 cm of the body), as well as from control sites located at least 1 m away from the donor (either upslope or at the same elevation) at predetermined accumulated degree hour (ADH) intervals until the end of active decomposition . ADH intervals included 0 (prior to placement), 100, 250, 500, 750, and 1000 ADH, and thereafter at 500 ADH intervals until un-enrollment. For each respective sample, cores were homogenized and debris ( e.g ., roots, insect larvae, rocks, etc.) removed by hand. A subset of soils (~ 20 g) were stored in a 4 oz. Whirl-Pak bag (Nasco), flash frozen in liquid nitrogen and stored at -80°C prior to DNA extraction and extracellular enzyme assays. The remaining soil was was stored in a 7 oz. Whirl-Pak bag (Nasco) at 4°C for soil physiochemical measurements . Soil slurries were prepared as a 1:2 ratio of soil to deionized water, allowed to come to room temperature for 30 minutes, and soil pH and electrical conductivity (EC) were measured using an Orion Star ™ A329 pH/ISE/Conductivity/Dissolved Oxygen portable multiparameter meter (ThermoFisher). Gravimetric soil moisture was measured in duplicate by oven drying 2 to 3 g soil aliquots at 105°C for 72 hours. Enzyme activities of β -glucosidase (BG), N-acetyl- β -D-glucosaminidase (NAG), leucine amino peptidase (LAP), and alkaline phosphatase (PHOS) were measured according to a modified procedure by Bell et al. (2013) . Breifly, 2.75 g of soil was weighed from soils stored at -80°C and held at -20°C until assays. Soils were thawed at room temperature prior to slurrying in 50 mM Tris buffer at pH 6.7 in a blender (Waring commercial blender, model WF2212114). Assays were conducted in triplicate using 800 μl of slurry and 200 μl of enzyme substrate (1,500 μM). Standard curves (MUB or MUC) were evaluted for each plate with conentrations ranging from 0 μM to 200 μM .
DNA was extracted from soils stored at -80°C . Briefly, 0.25 g of soil was extracted with the DNeasy Powerlyzer PowerSoil kit (QIAGEN Inc.) following manufacturer’s instructions with modifications for our soil texture (clay loam) and condition (high organic content). Specifically, soils were homogenized under parameters suggested for high organic soils (2,500 RPM for 45 s). DNA concentration was determined using a fluorometric assay (Quant-iT ™ PicoGreen ® dsDNA Assay Kit, Invitrogen) with a total volume of 200 μ l and 1 μ l of DNA. All DNA extracts were sent to the University of Tennessee Sequencing Core Facility (Knoxville, TN) for 16S rRNA and ITS2 region amplicon sequencing on the Illumina MiSeq platform (2 x 150 bp). The primer set 515F /806R was used to amplify the V4 region of the 16S rRNA gene, while the ITS2 region in fungi was amplified using a mixture of primers (6 forward and 2 reverse: ITS3NGS1, ITS3NGS2, ITS3NGS3, ITS3NGS4, ITS3NGS5, ITS3NGS10, ITS4NGR, and ARCH-ITS4) described previously . All raw sequences have been deposited in the National Center for Biotechnology Information’s Sequence Read Archive under the BioProject PRJNA817528 . Raw sequences were processed in Mothur (v.1.43.0) to cluster into 97% similarity operational taxonomic units (OTUs) and generate OTU count tables for both 16S and ITS datasets as described in . Briefly, paired-end reads were combined into contings, removing low-quality sequences (16S: Q > 20, bp ≤ 50; ITS Q > 20, bp < 200), sequences with ambigious bases (≥ 1), and primers/adpaters. Chimeras were removed using VSEARCH. Remaining sequences were classified using the SILVA non-redundant database (v132) or UNITE RefS database (version 02.02.2020) for 16S and ITS sequences, respectively. Bacterial sequences were then clustered into OTUs based on ≥ 97% sequence similarity and the Mothur default method, opticlust, while fungal sequences were clustered using abundance-based greedy clustering. We chose to cluster our sequences into OTUs rather than ASVs to reduce dimensionality in our dataset and the probability of splitting single genomes across multiple ASVs , especially when considering the diversity expected across soil microbial genomes. Count tables were then exported for analysis in R (version 4.4.0). Control samples ( e.g ., those not exposed to decomposition) and samples greater than 5000 ADH were removed using phyloseq (v1.44.0). Samples were cut off at 5000 ADH to capture the linear response of soil parameters and account for variation in decomposition timeframes between individuals . This resulted in 78 samples from 19 individuals (mean = 4.1 samples per individual) for model construction.
Read counts were total sum scaled (TSS) by determining the relative abundance of each OTU and normalizing to a standard library size (10,000 for all samples) using phyloseq (v1.44.0). This allowed for comparison of reads across samples and between biomarkers. We also removed OTUs with less than 10 reads across all samples in TSS normalized count tables to reduce noise in the datasets. 16S and ITS TSS read count tables were generated at the phylum, order, and class levels by summing the corresponding OTU table at each respective taxonomic level and then applying the TSS normalization as described above. Taxonomic levels were chosen to represent a subset which covered the full range from phylum to OTU. One of our goals was to compare predictability of bacterial (16S only), fungal (ITS only), or both (16S-ITS) communities; therefore, after TSS normalization, 16S-ITS combined datasets for each taxonomic level (phylum, order, class, OTU) were generated by merging respective 16S and ITS TSS count tables. As a result, 12 datasets were created and used for random forest models. We chose to apply random forest regression to datasets to predict PMI in ADH. This kept our study similar to those previously conducted on decomposition residues collected from soil surfaces , while also assessing predictability of soil microbial succession during human decomposition in our geographical region (Knoxville, TN). Model construction was completed in R using the Ranger (v0.16.0) package. First, samples were assigned to testing or training datasets. This was completed by randomly assigning 6 donors (~1/3) to the test set, while the remaining 13 were grouped into the training set. This approach was conducted following Belk et al. (2018) , to ensure that all samples from a single individual were in either the testing or training set, respectively, to prevent overfitting. Next, random forest regressions were applied to microbial taxa TSS normalized count tables in Ranger. First, random forest model parameters node size (3, 5, 7, 9) and sample size (0.55, 0.632, 0.70, 0.80) were hyper-tuned by comparing models with different combinations of the parameters listed. The optimal model was chosen by assessing the out-of-bag mean square error (OOB MSE) of each model and choosing the set of parameters with the lowest OOB MSE. The optimum model for each biomarker and taxonomic level was assessed by calculating the OOB MSE of the model and the root mean square error (RMSE) and mean absolute error (MAE) for predictions of the testing set in 100 runs of the optimum model. RMSE and MAE were calculated using rmse and mae functions from the R package Metrics (v 0.1.4). This process was repeated for models including measured environmental parameters, with values for ambient temperature (°C), pH, electrical conductivity (EC), moisture, β -glucosidase (BG) activity, N-acetyl- β -D-glucosaminidase (NAG) activity, leucine amino peptidase (LAP) activity, and alkaline phosphatase (PHOS) activity included as model features . For all environmental parameters, aside from temperature, log response ratio normalized values were used to account for natural seasonal differences in these parameters. The top 25 most influential model features were extracted from each optimum model to assess taxa/environmental factors influencing model predictions. To evaluate the potential differences in model predication between biomarkers and taxonomic levels, linear regression was applied to the average MAE values (mean of 100 runs per model). Variation in MAE due to treatment variables was assessed with ANOVA, while differences between treatment groups were determined with post-hoc t-tests in R. Code for generating all feature tables and random forest model development can be found at https://github.com/jdebruyn/TOX-microbiology .
Soil environmental parameters We previously reported how the measured soil parameters were altered in response to human decomposition . In summary, soil EC increased with progression of decomposition in soils surrounding all decomposing individuals. Soil pH was variable between individuals, with pH increasing (n = 5 individuals), decreasing (n = 12), or displaying minimal change relative to the controls (n = 2) . Extracellular enzyme activities were also variable between individuals, however general trends included increased NAG and PHOS over time. BG and LAP were variable over time;LAP activity correlated to soil pH . General model statistics In total, 24 models were built in R. Twelve of the models contained environmental features and the other twelve did not. The number of taxa included as features in models without environmental data are reported in . Bacterial (16S) and fungal (ITS) features ranged from 35 to 5195 and 16 to 2219, respectively, depending on taxonomic level. For all models, MAE ranged from 804.18 to 996.8 ADH . Across all variables considered, the best performing model was the 16S phylum level model with environmental predictors (MAE 804.18) . In contrast, the worst performing model was the ITS phylum level without environmental data (MAE 996.8) . Predictability, assessed by the linear relationship between predicted and observed values, for the training and testing datasets for the best 16S (phylum + environmental data), ITS only (order + environmental data), and 16S-ITS (order) models are shown in . R 2 for all models ranged from 0.869 to 0.962 when predicting PMI for the training set, however these values were reduced when making predictions for the testing dataset (r 2 = 0.369–0.741) . Model comparison: Biological marker Ability of random forest regressions to predict ADH varied depending on the biological marker used to build models (ANOVA F = 9.655, p = 0.001) . ITS models were generally less accurate in predicting ADH compared to 16S or 16S-ITS models independent of taxonomic level and environmental data . ITS models ranged in MAE from 872.16 to 996.8 ADH, with a mean MAE of 909.59 ADH. Post-hoc t-tests show that ITS models, in general, had higher MAE than both 16S (t-test p = 0.006) and 16S-ITS ( p = 0.012) models . ITS models represented seven of the 10 worst performing models. In comparison, 16S and 16S-ITS models performed similarly ( p = 0.466) . 16S models ranged in MAE from 804.18 to 889.81 ADH, with an average MAE of 841.73 ADH, while 16S-ITS models ranged in MAE from 812.35 to 890.94 ADH (average MAE = 852.82 ADH). This can also be observed among the best and worst performing models, where no ITS-only model was in the top 10 best performing models and combined 16S-ITS models were dispersed among the best and worst models. For example, the 16S-ITS order level model without environmental data had the third lowest MAE (MAE 812.35) overall, but also the 16S-ITS class level model with environmental data had the sixth highest MAE (MAE 890.94). Model comparison: Taxonomic level Some variation was observed in MAE due to taxonomic level considered for model development, however these differences were not significant (ANOVA F = 1.538; p = 0.24) . When considering the potential influence of taxonomic level within biomarkers, no significant difference in MAE by taxonomic level was observed for 16S ( p = 0.141) or ITS ( p = 0.609) models, while 16S-ITS models was significant ( p = 0.048) , likely driven by a difference in MAE between order and class level models for this biological marker . While most results were not significant, some trends were observed. First, order level models had the lowest MAE for all three biological markers assessed. This was also observed in , where order level models had the lowest MAE for all models without environmental data. Trends for the other taxonomic levels varied depending on the biological marker in consideration. Phylum and class level models performed similarly within 16S models, with OTU models generating the highest MAE. Within 16S-ITS models, class and OTU level models performed similarly, displaying the first and second highest MAE, respectively. For ITS models, phylum level models had the highest MAE, followed by OTU and class. Model comparison: Environmental parameters Overall, inclusion of environmental parameters in random forest models to predict ADH from soil microbial taxa impacted model accuracy. The direction of effect ( i.e ., increase or decrease in MAE) was dependent on biological marker and taxonomic level considered . For ITS models, inclusion of environmental factors reduced MAE irrespective of the taxonomic level. This reduction was most pronounced for the phylum level model, in which MAE was reduced by 116.007 ADH. In models containing 16S sequencing data (16S and 16S-ITS), effect of environmental features differed by taxonomic level. Specifically, for 16S models, phylum, class, and order level models performed better and OTU level models performed worse when environmental data was included. This was similar for the combined 16S-ITS datasets at the phylum and OTU levels; however, class and order level models performed worse ( i.e ., increased MAE) when environmental factors were included. Model features: Top models In addition to assessing the predictability of different random forest models, we also looked at important model features to observe which taxa and/or environmental parameters impacted model performance. Here we highlight the top 25 features of best performing 16S (phylum + environmental), 16S-ITS (order), and ITS (order + environmental) models , determined by lowest MAE. Both 16S and ITS best models included environmental predictors, while the combined 16S-ITS model did not . For the 16S phylum model with environmental data, the most important model feature, as assessed by decrease in MSE, was the phylum Firmicutes . The remaining important features included soil electrical conductivity (EC), Acidobacteria , Epsilonbacteraeota , and Proteobacteria , respectively. Other features of interest for this model included Nitrospirae , leucine aminopeptidase activity, pH, and soil moisture . For the ITS order model with environmental predictors, the most important model features were Pleosporales , soil EC, Unclassified fungi, Rhizophydiales , Unclassified Glomeromycota , Unclassified Basidiomycota , and Auriculariales . In this model, no other environmental parameters were among the top 25 important features. Other top taxonomic features of interest included Saccharomycetales and Unclassified Sordariomycetes , as their members are present in the human mycobiome and feces, respectively . The best performing 16S-ITS model was the order level model without environmental features. Top features for this model were the bacterial order Lactobacillales and the fungal order Pleosporales . Bacterial orders Bacteroidales , Cardiobacteriales , and Clostridiales were third, fourth, and eleventh most important features, respectively . The fungal order Saccharomycetales was also observed in the top 25. Relative abundance of anaerobic bacterial taxa identified in random forest models, including Firmicutes , Bacteroidales , Clostridiales and Lactobacillales , increased as decomposition progressed. In contrast, relative abundance of the aerobic nitrifying organisms of the phylum Nitrospirae decreased . Acidobacteria , among the top phyla in the 16S model, decreased in relative abundance during decomposition . The phylum Epsilonbacteraeota , containg many gut-related taxa, displayed increased relative abundance over time . Relative abundance of the bacterial orders Cardiobacteriales and Pseudomonadales and fungal order Pleosporales , identified in the mixed 16S-ITS order model, increased and decreased, respectively.
We previously reported how the measured soil parameters were altered in response to human decomposition . In summary, soil EC increased with progression of decomposition in soils surrounding all decomposing individuals. Soil pH was variable between individuals, with pH increasing (n = 5 individuals), decreasing (n = 12), or displaying minimal change relative to the controls (n = 2) . Extracellular enzyme activities were also variable between individuals, however general trends included increased NAG and PHOS over time. BG and LAP were variable over time;LAP activity correlated to soil pH .
In total, 24 models were built in R. Twelve of the models contained environmental features and the other twelve did not. The number of taxa included as features in models without environmental data are reported in . Bacterial (16S) and fungal (ITS) features ranged from 35 to 5195 and 16 to 2219, respectively, depending on taxonomic level. For all models, MAE ranged from 804.18 to 996.8 ADH . Across all variables considered, the best performing model was the 16S phylum level model with environmental predictors (MAE 804.18) . In contrast, the worst performing model was the ITS phylum level without environmental data (MAE 996.8) . Predictability, assessed by the linear relationship between predicted and observed values, for the training and testing datasets for the best 16S (phylum + environmental data), ITS only (order + environmental data), and 16S-ITS (order) models are shown in . R 2 for all models ranged from 0.869 to 0.962 when predicting PMI for the training set, however these values were reduced when making predictions for the testing dataset (r 2 = 0.369–0.741) .
Ability of random forest regressions to predict ADH varied depending on the biological marker used to build models (ANOVA F = 9.655, p = 0.001) . ITS models were generally less accurate in predicting ADH compared to 16S or 16S-ITS models independent of taxonomic level and environmental data . ITS models ranged in MAE from 872.16 to 996.8 ADH, with a mean MAE of 909.59 ADH. Post-hoc t-tests show that ITS models, in general, had higher MAE than both 16S (t-test p = 0.006) and 16S-ITS ( p = 0.012) models . ITS models represented seven of the 10 worst performing models. In comparison, 16S and 16S-ITS models performed similarly ( p = 0.466) . 16S models ranged in MAE from 804.18 to 889.81 ADH, with an average MAE of 841.73 ADH, while 16S-ITS models ranged in MAE from 812.35 to 890.94 ADH (average MAE = 852.82 ADH). This can also be observed among the best and worst performing models, where no ITS-only model was in the top 10 best performing models and combined 16S-ITS models were dispersed among the best and worst models. For example, the 16S-ITS order level model without environmental data had the third lowest MAE (MAE 812.35) overall, but also the 16S-ITS class level model with environmental data had the sixth highest MAE (MAE 890.94).
Some variation was observed in MAE due to taxonomic level considered for model development, however these differences were not significant (ANOVA F = 1.538; p = 0.24) . When considering the potential influence of taxonomic level within biomarkers, no significant difference in MAE by taxonomic level was observed for 16S ( p = 0.141) or ITS ( p = 0.609) models, while 16S-ITS models was significant ( p = 0.048) , likely driven by a difference in MAE between order and class level models for this biological marker . While most results were not significant, some trends were observed. First, order level models had the lowest MAE for all three biological markers assessed. This was also observed in , where order level models had the lowest MAE for all models without environmental data. Trends for the other taxonomic levels varied depending on the biological marker in consideration. Phylum and class level models performed similarly within 16S models, with OTU models generating the highest MAE. Within 16S-ITS models, class and OTU level models performed similarly, displaying the first and second highest MAE, respectively. For ITS models, phylum level models had the highest MAE, followed by OTU and class.
Overall, inclusion of environmental parameters in random forest models to predict ADH from soil microbial taxa impacted model accuracy. The direction of effect ( i.e ., increase or decrease in MAE) was dependent on biological marker and taxonomic level considered . For ITS models, inclusion of environmental factors reduced MAE irrespective of the taxonomic level. This reduction was most pronounced for the phylum level model, in which MAE was reduced by 116.007 ADH. In models containing 16S sequencing data (16S and 16S-ITS), effect of environmental features differed by taxonomic level. Specifically, for 16S models, phylum, class, and order level models performed better and OTU level models performed worse when environmental data was included. This was similar for the combined 16S-ITS datasets at the phylum and OTU levels; however, class and order level models performed worse ( i.e ., increased MAE) when environmental factors were included.
In addition to assessing the predictability of different random forest models, we also looked at important model features to observe which taxa and/or environmental parameters impacted model performance. Here we highlight the top 25 features of best performing 16S (phylum + environmental), 16S-ITS (order), and ITS (order + environmental) models , determined by lowest MAE. Both 16S and ITS best models included environmental predictors, while the combined 16S-ITS model did not . For the 16S phylum model with environmental data, the most important model feature, as assessed by decrease in MSE, was the phylum Firmicutes . The remaining important features included soil electrical conductivity (EC), Acidobacteria , Epsilonbacteraeota , and Proteobacteria , respectively. Other features of interest for this model included Nitrospirae , leucine aminopeptidase activity, pH, and soil moisture . For the ITS order model with environmental predictors, the most important model features were Pleosporales , soil EC, Unclassified fungi, Rhizophydiales , Unclassified Glomeromycota , Unclassified Basidiomycota , and Auriculariales . In this model, no other environmental parameters were among the top 25 important features. Other top taxonomic features of interest included Saccharomycetales and Unclassified Sordariomycetes , as their members are present in the human mycobiome and feces, respectively . The best performing 16S-ITS model was the order level model without environmental features. Top features for this model were the bacterial order Lactobacillales and the fungal order Pleosporales . Bacterial orders Bacteroidales , Cardiobacteriales , and Clostridiales were third, fourth, and eleventh most important features, respectively . The fungal order Saccharomycetales was also observed in the top 25. Relative abundance of anaerobic bacterial taxa identified in random forest models, including Firmicutes , Bacteroidales , Clostridiales and Lactobacillales , increased as decomposition progressed. In contrast, relative abundance of the aerobic nitrifying organisms of the phylum Nitrospirae decreased . Acidobacteria , among the top phyla in the 16S model, decreased in relative abundance during decomposition . The phylum Epsilonbacteraeota , containg many gut-related taxa, displayed increased relative abundance over time . Relative abundance of the bacterial orders Cardiobacteriales and Pseudomonadales and fungal order Pleosporales , identified in the mixed 16S-ITS order model, increased and decreased, respectively.
The goal of this work was to assess the influence of biological marker, taxonomic level, and environmental parameters on model prediction of PMI from soil microbial communities. Model analysis revealed differences between model accuracy due to the biological marker, taxonomic level, and environmental parameters considered for model construction. Overall, models did not predict the test data well. R 2 dropped from 0.869—0.962 when predicting the training dataset to 0.369—0.741 for the test set. Additionally, models ranged in MAE from 804.18 to 996.8 ADH. In East Tennessee, these error rates would correspond to roughly 2.5 to 3.5 days in July and greater than 28 days in February, based on average seasonal temperatures for the region. Therefore, error rates in the summer would be comparable to those previously reported for microbial communities from organic residues collected from the soil surface (two to six days) but would be substantially higher if considering decomposition during cooler seasons. Further, considering our total decomposition time of 5000 ADH, errors of 804.18 to 996.8 ADH equates to 15.9% to 19.9% of the total decomposition time. The wide error range when including a greater number of subjects across multiple seasons suggests soil microbiome-based models may have low accuracy, particularly when considering individuals across the cline of human variation and through multiple seasons. Specifically, the decomposition systems were influenced by the starting resource, which is dictated by human variation at both the genetic and environmental levels. As a result, intrinsic factors have the capacity to alter both decomposer communities and decomposition rate, and therefore decomposer communities, leading to variation that can impact future models. One important source of variation in our study was the different rates of decomposition between individuals. While we attempted to correct for differences due to thermal energy input by using accumulated degree hours (ADH), there was still variability in terms of the morphological stage for a given ADH. For example, 5000 ADH represented the end of active decomposition for individual 009, but only about 25% of the active decomposition period for individual 010. Additionally, this time-period did not include decomposition past active decay for any individual in our study, including advanced decomposition or sustained mummification or skeletonization, which could further impact model accuracy. Both individuals (009 and 010) were placed within the facility in the summer, experiencing the same local environmental conditions and potential for insect and scavenger communities, suggesting there are additional factors leading to variation in microbial communities within decomposition-impacted soils. These may include additional environmental parameters not considered in our models, and/or intrinsic differences between the individuals themselves ( e.g ., age, weight, medications, medical conditions etc.) that directly or indirectly impacted microbial communities through interactions with other decomposers (insects and scavengers). Moving forward, we will need to employ a strategy to combine antemortem and environmental data in order to investigate which factors help improve model predictions. Influence of diversity and taxa succession on PMI estimations The trends we observed in model MAE between different biological markers, taxonomic levels, and inclusion of environmental data may be partly explained by differences in diversity between bacterial and fungal (16S vs. ITS) communities and the number of taxa ( i.e ., features), ultimately impacting resolution for predicting PMI. Overall, Chao1 richness and Inverse Simpson diversity were 10 and 15 times lower, respectively, in fungal communities compared to bacterial communities . This translated to differences in the total number of model features for 16S and ITS models: 16S models had roughly 1.7 to 2.3 times more features, depending on the taxonomic level considered. As a result, more features, or taxa, in the dataset with relationships to time ( i.e ., progression of decomposition) may help to distinguish between timepoints to improve model predictability. In our previous work, we observed that the fungal community composition became more similar as decomposition progressed, with only a few taxa driving fungal successional patterns . This was also observed in Fu et al. (2019) , in which only a few taxa ( e.g ., Ascomycota sp., Yarrowia lipolitica , etc.) displayed relationships with PMI. While we hypothesized that ITS-based models would be more accurate than 16S-based models because of these studies, our results revealed that 16S models generally outperformed ITS-based models and combining 16S and ITS did not improve 16S models alone. This result coincides with those reported by Belk et al. (2018) , in which 16S models (mean MAE 4.022 days) had lower error than ITS (mean MAE 4.452 days) or 18S models (mean MAE 4.195 days). The reduced number of taxa observed in fungal communities, in combination with relatively few taxa changing in abundance, may explain why ITS models had higher MAE than 16S models. This may also explain why combining 16S and ITS datasets, which would increase overall diversity, did not outperform either marker alone. With only a few fungal taxa displaying changes over time, their inclusion may not have added additional resolution to the bacterial model. Diversity differences between bacterial and fungal communities may also drive some of the trends observed between taxonomic levels and with or without environmental factors. In this study, order level models performed best for all biological markers when not considering environmental features. This contrasts with findings by Belk at al. (2018) , where lower error was reported for phylum and class level models. This may be linked to a balance between taxonomic resolution and noise for this timeperiod of decomposition. In our study, OTU level models displayed the highest MAE for all biological markers when not considering environmental data, which corresponds with previous decomposition studies reporting increased inter-individual variation at lower taxonomic levels . This may explain why OTU level models displayed the highest MAE for all biological markers when not considering environmental data. Within decomopsition studies, microbial taxonomic succession has mostly been characterized at higher taxonomic levels, at which general patterns are more similar between individuals. However, aggregating microbial abundances at coarse taxonomic levels, such as phyla and class, inherently reduces data dimensionality. It is possible that this decrease in features, in conjunction with trends in taxon abundance over time, reduces the ability of the random forest regression to resolve timepoints at the highest taxonomic levels. This balance between diversity and features with resolution over time may also explain the effect of environmental features effect on model MAE. We hypothesized that inclusion of environmental parameters would improve all model predictions, by combining soil chemical and microbial successional patterns. While inclusion of environmental predictors improved some models, it decreased performance of others. This effect appears to be linked to biological marker and taxonomic level considered for model creation. Specifically, inclusion of environmental parameters into the lower diversity fungal models added features that helped to improve overall resolution to predict PMI. In contrast, inclusion of environmental data may have added additional noise to high diversity bacterial datasets at lower taxonomic levels, overall leading to decreased model performance. Model features In addition to assessing model performance, we also investigated model features for each top performing 16S, 16S-ITS, and ITS model as determined by lowest MAE. This included the 16S phylum model with environmental data, the 16S-ITS order level model, and the ITS order level model with environmental data. Top model features for 16S phylum level models included taxa observed in previous human and animal decomposition studies, such as the bacterial phyla Firmicutes , Acidobacteria , and Proteobacteria . In our study, Firmicutes and Nitrospirae were shown to decrease as decomposition progressed, while Acidobacteria decreased and Proteobacteria remained consistent. These changes seem to be linked to differences in metabolism and environmental changes that occur when decomposition products are released into the surrounding soil. For example, it has been suggested heterotrophic microbial activity responding to the pulse of decomposition products results in depletion of soil oxygen . This would impact the presence of anaerobic gut and soil taxa. While we did not measure soil oxygen in this study, soil respiration was increased in these soils, and so oxygen depletion is to be expected . The increased presence of taxa containing facultative and obligate anaerobic members Firmicutes and Clostridiales in phylum and order models, respectively, and decrease in Nitrospirae , containing nitrifying bacteria that oxidize nitrogen under aerobic conditions, support this hypothesis. Increases in Firmicutes and Clostridiales follow successional trends observed in internal ( e.g ., organs) microbial communities. Specifically, increased relative abundance of Clostridium has been termed the “Clostridium Effect” by Javan et al. (2017) and observed in various organs and the rectum postmortem. Multiple studies, including this current work, have observed increased relative abundance in Firmicutes and Clostridiales in soils following deposition of decomposition fluid , suggesting some of these organisms may be host-derived. Decreased presence of Acidobacteria in decomposition-impacted soils is likely linked to their oligotrophic characteristics in response to high nutrient deposition . Succession of these taxa and other taxa past 5000 ADH and the potential implications for PMI models is unclear. Order level models also revealed some information about soil microbial succession during human decomposition. The 16S-ITS order level model had the lowest MAE among all 16S-ITS models. Within this model, important taxa were a combination of 16S and ITS features present in respective models. Among the top bacterial features, Lactobacillales , Bacteroidales , and Clostridiales were all shown to display general increases as decomposition progressed. This is consistent with previous literature . One interesting find was Cardiobacteriales as the fifth most important model feature. Cardiobacteriales is a bacterial order of gram-negative rods, whose members are generally capable of fermentation of various sugars . Within this order, only the genus Ignatzschineria was identified based on the SILVA non-redundant database (v132) . This genus has been identified in previous outdoor decomposition studies focusing on gut , skin , and soil microbial communities. Ignatzschineria are associated with insect species and first appear in the soil during release of fluid. We previously observed this taxon in bacterial decomposition fluid communities , suggesting decomposition fluids as potential vehicle for the transfer of both host- and insect- associated microbes into the surrounding soil. Their association with insects highlights the potential for decomposer insect and scavenger activity to impact microbial succession during decomposition and suggests that PMI estimation models specific to decomposition setting (indoor or outdoor) may be required. Within the ITS order level models, the fungal order Pleosporales was among the most influential taxon for PMI estimation. Pleosporales , a member of Ascomycota fungi, decreased as decomposition progressed. This was similar to observations by Fu et al. (2019) , where Pleosporales sp. was shown to be associated with non-decomposition soils by LEfSe (linear discriminant analysis effect size). Pleosporales are often associated with plants, found as endophytes, epiphytes, and the rhizosphere . Reduced relative abundance of these fungi in decomposition soils is interesting considering Pleosporales have been shown to positively respond to nitrogen amendments . Their response to decomposition products may suggest sensitivity to highly concentrated nitrogen amendments and/or other soil changes, such as osmotic stress in response to high EC, or intolerance to hypoxia typically observed in decomposition soils. Of the environmental predictors assessed, electrical conductivity (EC) appeared to be most influential. EC was recorded as the top and second most important feature for the 16S phylum and ITS order models with environmental data, respectively. This is likely due to patterns in soil EC being more consistent between individuals over time. Specifically, EC was shown to increase within decomposition soil over time for all 19 individuals. Increases in EC observed in decomposition soils has been shown to positively correlate with increased ammonium concentrations , suggesting ammonium would also be a valuable predictor of microbial community dynamics. The other measured environmental parameters (pH, enzyme rates) were not identified as a top predictive features in the models. This is likely because these parameters were more variable both over time as well as between individuals, displaying both increases and decreases in response to human decomposition . While we did not consider all possible environmental parameters in this study, these results suggest that feature selection may help to identify relevant environmental parameters for model construction. Limitations and considerations While there are intriguing investigations suggesting that microbial succession could be used to predict PMI, validation is critical prior to forensic application. Variation between decomposition studies, including vertebrate species observed, and experimental design, along with small sample sizes have limited model development to date. Additionally, most decomposition studies focus on bloat and active decay stages, when decomposers are most active in degrading soft tissues . While informative for initial compositional shifts, this timeframe does not allow us to assess for how long these communities may be impacted or if they return to pre-decomposition conditions . This study starts to address factors that influence PMI estimations from soil microbial succession during human decomposition, however many foundational questions remain. Below we discuss multiple areas to be expanded upon with future investigation. First, we did not include all possible environmental and soil data as model predictors, nor account for interactions with other decomposer communities. Other factors, such as respiration rates, oxygen concentration, ammonium, nitrate, dissolved organic carbon and nitrogen, sulfur, among others may be relevant features for models predicting PMI within the soil environment as they have been shown to change during decomposition and have the capacity to structure microbial communities. In addition, it is possible that changes in soil parameters during human decomposition differ based on region due to soil type and climatic differences impacting decomposition progression or presence of microbial taxa , as well as the insect and scavenging species present across ecosystems. Lines of inquiry should include, but are not limited to, regional and seasonal (both within and between regions) soil microbial successional patterns in response to carcass decomposition and microbial-insect interactions, including effects of local insect species on microbial community dynamics. For example, Chrysomya megacephala , an invasive fly species that has a proclivity for feces, has only recently been documented colonizing human remains in Tennessee, USA . This species carries up to 10 times the pathogenic bacterial load compared to the house fly, Musca domestica , potentially introducing microbes that could alter the progression of decomposition and result in different microbial community succession between regions with and without this fly species . Second, we chose to implement the random forest regression algorithm as it is not as sensitive to non-linear data and has high interpretability compared to other forms of supervised and unsupervised machine learning algorithms . This allowed us to assess prediction of PMI and identify taxa and environmental features that correlate with PMI, as well as kept our results similar to previous decomposition studies within the soil environment . However, recent studies have compared multiple machine learning algorithms in other decomposition microhabitats ( i.e ., skin, organs), showing variation both between internal organs and within the same habitat . Both Liu et al. (2020) and Johnson et al. (2016) observed other machine learning algorithms performed better than random forest in higher diversity microhabitats such as the skin and caecum. As the soil environment is among the most diverse microbial habitat on the planet, it is necessary to assess different machine learning approaches when predicting PMI within this microhabitat . Third, total PMI (5000 ADH) considered for model construction will likely impact the performance and applicability of these models . Here we showed that order level models had the lowest MAE in models that do not include environmental features. This contrasts with findings by Belk at al. (2018) and our hypothesis, in which we expected lower error for phylum and class level models, suggesting differences between studies, such as region, sampling strategy, number of individuals, species, study timeframe or intrinsic differences between donor populations may impact model performance. For example, our study went through 5000 ADH, while Belk et al. (2018) presented data up to 25 days. In our study, 5000 ADH corresponded to 13 to 115 days depending on the individual and time of year, suggesting the unit of time chosen for PMI estimates may impact model interpretation. While out of the scope of this paper, a comparison of model performance trained with different units of time would be informative. Additionally, Belk et al. (2018) observed decreased model error when only using data points from the first 25 days of decomposition compared to the first 50 days, suggesting microbial-based models may not be as accurate at higher PMIs. Therefore, future work is needed to determine the PMI range for which microbial-based PMI estimations are most accurate. Fourth, we chose to use operational taxonomic units (OTUs) and ADH calculated with a baseline of 10°C, as opposed to amplicon sequence variants (ASVs) and/or ADH with a baseline of 0°C or 4°C. The recent application of denoising methods to generate ASVs has become popular in microbial studies using amplicon sequencing. However, we chose to cluster sequences into OTUs to reduce dimensionality in our raw dataset and the probability of splitting single genomes across multiple ASVs . While Glassman and Martiny (2021) observed similar results for alpha and beta diversity from OTUs and ASVs in leaf litter communities, other studies have shown differences in diversity when comparing the two methods . Thus, it is unclear if using OTUs or ASVs will impact machine learning algorithms such as random forest to predict PMI. Future work should investigate differences in PMI estimations from models constructed with OTUs as well as ASVs. Additionally, we chose to use a baseline of 10°C, which is commonly used for entomological methods due to the developmental threshold of regional (east Tennessee) insects . However, other decomposition studies within the soil environment have also used 0°C or 4°C as thresholds for ADH or accumulated degree day (ADD) calculations. These differences may impact PMI estimates; however, no one has addressed effects of different thresholds for ADH/ADD calculation on PMI estimates from microbial successional patterns within the soil. Therefore, a comprehensive comparison between different thermal energy unit ( i.e ., ADH, ADD and baseline) calculations is necessary.
The trends we observed in model MAE between different biological markers, taxonomic levels, and inclusion of environmental data may be partly explained by differences in diversity between bacterial and fungal (16S vs. ITS) communities and the number of taxa ( i.e ., features), ultimately impacting resolution for predicting PMI. Overall, Chao1 richness and Inverse Simpson diversity were 10 and 15 times lower, respectively, in fungal communities compared to bacterial communities . This translated to differences in the total number of model features for 16S and ITS models: 16S models had roughly 1.7 to 2.3 times more features, depending on the taxonomic level considered. As a result, more features, or taxa, in the dataset with relationships to time ( i.e ., progression of decomposition) may help to distinguish between timepoints to improve model predictability. In our previous work, we observed that the fungal community composition became more similar as decomposition progressed, with only a few taxa driving fungal successional patterns . This was also observed in Fu et al. (2019) , in which only a few taxa ( e.g ., Ascomycota sp., Yarrowia lipolitica , etc.) displayed relationships with PMI. While we hypothesized that ITS-based models would be more accurate than 16S-based models because of these studies, our results revealed that 16S models generally outperformed ITS-based models and combining 16S and ITS did not improve 16S models alone. This result coincides with those reported by Belk et al. (2018) , in which 16S models (mean MAE 4.022 days) had lower error than ITS (mean MAE 4.452 days) or 18S models (mean MAE 4.195 days). The reduced number of taxa observed in fungal communities, in combination with relatively few taxa changing in abundance, may explain why ITS models had higher MAE than 16S models. This may also explain why combining 16S and ITS datasets, which would increase overall diversity, did not outperform either marker alone. With only a few fungal taxa displaying changes over time, their inclusion may not have added additional resolution to the bacterial model. Diversity differences between bacterial and fungal communities may also drive some of the trends observed between taxonomic levels and with or without environmental factors. In this study, order level models performed best for all biological markers when not considering environmental features. This contrasts with findings by Belk at al. (2018) , where lower error was reported for phylum and class level models. This may be linked to a balance between taxonomic resolution and noise for this timeperiod of decomposition. In our study, OTU level models displayed the highest MAE for all biological markers when not considering environmental data, which corresponds with previous decomposition studies reporting increased inter-individual variation at lower taxonomic levels . This may explain why OTU level models displayed the highest MAE for all biological markers when not considering environmental data. Within decomopsition studies, microbial taxonomic succession has mostly been characterized at higher taxonomic levels, at which general patterns are more similar between individuals. However, aggregating microbial abundances at coarse taxonomic levels, such as phyla and class, inherently reduces data dimensionality. It is possible that this decrease in features, in conjunction with trends in taxon abundance over time, reduces the ability of the random forest regression to resolve timepoints at the highest taxonomic levels. This balance between diversity and features with resolution over time may also explain the effect of environmental features effect on model MAE. We hypothesized that inclusion of environmental parameters would improve all model predictions, by combining soil chemical and microbial successional patterns. While inclusion of environmental predictors improved some models, it decreased performance of others. This effect appears to be linked to biological marker and taxonomic level considered for model creation. Specifically, inclusion of environmental parameters into the lower diversity fungal models added features that helped to improve overall resolution to predict PMI. In contrast, inclusion of environmental data may have added additional noise to high diversity bacterial datasets at lower taxonomic levels, overall leading to decreased model performance.
In addition to assessing model performance, we also investigated model features for each top performing 16S, 16S-ITS, and ITS model as determined by lowest MAE. This included the 16S phylum model with environmental data, the 16S-ITS order level model, and the ITS order level model with environmental data. Top model features for 16S phylum level models included taxa observed in previous human and animal decomposition studies, such as the bacterial phyla Firmicutes , Acidobacteria , and Proteobacteria . In our study, Firmicutes and Nitrospirae were shown to decrease as decomposition progressed, while Acidobacteria decreased and Proteobacteria remained consistent. These changes seem to be linked to differences in metabolism and environmental changes that occur when decomposition products are released into the surrounding soil. For example, it has been suggested heterotrophic microbial activity responding to the pulse of decomposition products results in depletion of soil oxygen . This would impact the presence of anaerobic gut and soil taxa. While we did not measure soil oxygen in this study, soil respiration was increased in these soils, and so oxygen depletion is to be expected . The increased presence of taxa containing facultative and obligate anaerobic members Firmicutes and Clostridiales in phylum and order models, respectively, and decrease in Nitrospirae , containing nitrifying bacteria that oxidize nitrogen under aerobic conditions, support this hypothesis. Increases in Firmicutes and Clostridiales follow successional trends observed in internal ( e.g ., organs) microbial communities. Specifically, increased relative abundance of Clostridium has been termed the “Clostridium Effect” by Javan et al. (2017) and observed in various organs and the rectum postmortem. Multiple studies, including this current work, have observed increased relative abundance in Firmicutes and Clostridiales in soils following deposition of decomposition fluid , suggesting some of these organisms may be host-derived. Decreased presence of Acidobacteria in decomposition-impacted soils is likely linked to their oligotrophic characteristics in response to high nutrient deposition . Succession of these taxa and other taxa past 5000 ADH and the potential implications for PMI models is unclear. Order level models also revealed some information about soil microbial succession during human decomposition. The 16S-ITS order level model had the lowest MAE among all 16S-ITS models. Within this model, important taxa were a combination of 16S and ITS features present in respective models. Among the top bacterial features, Lactobacillales , Bacteroidales , and Clostridiales were all shown to display general increases as decomposition progressed. This is consistent with previous literature . One interesting find was Cardiobacteriales as the fifth most important model feature. Cardiobacteriales is a bacterial order of gram-negative rods, whose members are generally capable of fermentation of various sugars . Within this order, only the genus Ignatzschineria was identified based on the SILVA non-redundant database (v132) . This genus has been identified in previous outdoor decomposition studies focusing on gut , skin , and soil microbial communities. Ignatzschineria are associated with insect species and first appear in the soil during release of fluid. We previously observed this taxon in bacterial decomposition fluid communities , suggesting decomposition fluids as potential vehicle for the transfer of both host- and insect- associated microbes into the surrounding soil. Their association with insects highlights the potential for decomposer insect and scavenger activity to impact microbial succession during decomposition and suggests that PMI estimation models specific to decomposition setting (indoor or outdoor) may be required. Within the ITS order level models, the fungal order Pleosporales was among the most influential taxon for PMI estimation. Pleosporales , a member of Ascomycota fungi, decreased as decomposition progressed. This was similar to observations by Fu et al. (2019) , where Pleosporales sp. was shown to be associated with non-decomposition soils by LEfSe (linear discriminant analysis effect size). Pleosporales are often associated with plants, found as endophytes, epiphytes, and the rhizosphere . Reduced relative abundance of these fungi in decomposition soils is interesting considering Pleosporales have been shown to positively respond to nitrogen amendments . Their response to decomposition products may suggest sensitivity to highly concentrated nitrogen amendments and/or other soil changes, such as osmotic stress in response to high EC, or intolerance to hypoxia typically observed in decomposition soils. Of the environmental predictors assessed, electrical conductivity (EC) appeared to be most influential. EC was recorded as the top and second most important feature for the 16S phylum and ITS order models with environmental data, respectively. This is likely due to patterns in soil EC being more consistent between individuals over time. Specifically, EC was shown to increase within decomposition soil over time for all 19 individuals. Increases in EC observed in decomposition soils has been shown to positively correlate with increased ammonium concentrations , suggesting ammonium would also be a valuable predictor of microbial community dynamics. The other measured environmental parameters (pH, enzyme rates) were not identified as a top predictive features in the models. This is likely because these parameters were more variable both over time as well as between individuals, displaying both increases and decreases in response to human decomposition . While we did not consider all possible environmental parameters in this study, these results suggest that feature selection may help to identify relevant environmental parameters for model construction.
While there are intriguing investigations suggesting that microbial succession could be used to predict PMI, validation is critical prior to forensic application. Variation between decomposition studies, including vertebrate species observed, and experimental design, along with small sample sizes have limited model development to date. Additionally, most decomposition studies focus on bloat and active decay stages, when decomposers are most active in degrading soft tissues . While informative for initial compositional shifts, this timeframe does not allow us to assess for how long these communities may be impacted or if they return to pre-decomposition conditions . This study starts to address factors that influence PMI estimations from soil microbial succession during human decomposition, however many foundational questions remain. Below we discuss multiple areas to be expanded upon with future investigation. First, we did not include all possible environmental and soil data as model predictors, nor account for interactions with other decomposer communities. Other factors, such as respiration rates, oxygen concentration, ammonium, nitrate, dissolved organic carbon and nitrogen, sulfur, among others may be relevant features for models predicting PMI within the soil environment as they have been shown to change during decomposition and have the capacity to structure microbial communities. In addition, it is possible that changes in soil parameters during human decomposition differ based on region due to soil type and climatic differences impacting decomposition progression or presence of microbial taxa , as well as the insect and scavenging species present across ecosystems. Lines of inquiry should include, but are not limited to, regional and seasonal (both within and between regions) soil microbial successional patterns in response to carcass decomposition and microbial-insect interactions, including effects of local insect species on microbial community dynamics. For example, Chrysomya megacephala , an invasive fly species that has a proclivity for feces, has only recently been documented colonizing human remains in Tennessee, USA . This species carries up to 10 times the pathogenic bacterial load compared to the house fly, Musca domestica , potentially introducing microbes that could alter the progression of decomposition and result in different microbial community succession between regions with and without this fly species . Second, we chose to implement the random forest regression algorithm as it is not as sensitive to non-linear data and has high interpretability compared to other forms of supervised and unsupervised machine learning algorithms . This allowed us to assess prediction of PMI and identify taxa and environmental features that correlate with PMI, as well as kept our results similar to previous decomposition studies within the soil environment . However, recent studies have compared multiple machine learning algorithms in other decomposition microhabitats ( i.e ., skin, organs), showing variation both between internal organs and within the same habitat . Both Liu et al. (2020) and Johnson et al. (2016) observed other machine learning algorithms performed better than random forest in higher diversity microhabitats such as the skin and caecum. As the soil environment is among the most diverse microbial habitat on the planet, it is necessary to assess different machine learning approaches when predicting PMI within this microhabitat . Third, total PMI (5000 ADH) considered for model construction will likely impact the performance and applicability of these models . Here we showed that order level models had the lowest MAE in models that do not include environmental features. This contrasts with findings by Belk at al. (2018) and our hypothesis, in which we expected lower error for phylum and class level models, suggesting differences between studies, such as region, sampling strategy, number of individuals, species, study timeframe or intrinsic differences between donor populations may impact model performance. For example, our study went through 5000 ADH, while Belk et al. (2018) presented data up to 25 days. In our study, 5000 ADH corresponded to 13 to 115 days depending on the individual and time of year, suggesting the unit of time chosen for PMI estimates may impact model interpretation. While out of the scope of this paper, a comparison of model performance trained with different units of time would be informative. Additionally, Belk et al. (2018) observed decreased model error when only using data points from the first 25 days of decomposition compared to the first 50 days, suggesting microbial-based models may not be as accurate at higher PMIs. Therefore, future work is needed to determine the PMI range for which microbial-based PMI estimations are most accurate. Fourth, we chose to use operational taxonomic units (OTUs) and ADH calculated with a baseline of 10°C, as opposed to amplicon sequence variants (ASVs) and/or ADH with a baseline of 0°C or 4°C. The recent application of denoising methods to generate ASVs has become popular in microbial studies using amplicon sequencing. However, we chose to cluster sequences into OTUs to reduce dimensionality in our raw dataset and the probability of splitting single genomes across multiple ASVs . While Glassman and Martiny (2021) observed similar results for alpha and beta diversity from OTUs and ASVs in leaf litter communities, other studies have shown differences in diversity when comparing the two methods . Thus, it is unclear if using OTUs or ASVs will impact machine learning algorithms such as random forest to predict PMI. Future work should investigate differences in PMI estimations from models constructed with OTUs as well as ASVs. Additionally, we chose to use a baseline of 10°C, which is commonly used for entomological methods due to the developmental threshold of regional (east Tennessee) insects . However, other decomposition studies within the soil environment have also used 0°C or 4°C as thresholds for ADH or accumulated degree day (ADD) calculations. These differences may impact PMI estimates; however, no one has addressed effects of different thresholds for ADH/ADD calculation on PMI estimates from microbial successional patterns within the soil. Therefore, a comprehensive comparison between different thermal energy unit ( i.e ., ADH, ADD and baseline) calculations is necessary.
This study aimed to assess microbial abundance-based prediction of PMI from soil microbial communities. We compared models with different biological markers, taxonomic levels, and presence/absence of environmental variables to expand upon previous estimations of PMI from machine learning algorithms. From this dataset of 19 individuals across multiple seasons, we observed higher error rates and decreased model precision compared to previously published models based on small datasets. Our results show that 16S and 16S-ITS models performed similarly and outperformed ITS models. Further, order level models have the lowest MAE when not considering environmental parameters. We also show that the addition of other factors, such as environmental parameters, have the potential to impact PMI estimations. We observed some level of predictability in soil microbial succession, however high error rates were seen across 19 individuals and across seasons. While our the number of individuals in our study is one of the largest to date, it was demographically limited, and we certainly did not capture all antemortem conditions which could influence decomposition rates. Together this means microbial-based PMI models would need considerable validation and refinement across a diverse population and geographical regions prior to implementing in a forensic context.
S1 Fig Total decomposition time differs for each donor. Soil samples (black points) were collected at predetermined intervals through the end of active decomposition. Endpoints differed between donors, therefore a cutoff of 5000 ADH (dashed blue line) was chosen to capture the most timepoints across all donors. (TIF) S2 Fig Relative abundance of the 5 most important bacterial orders in the top 16S-ITS random forest model (16S-ITS order). Relative abundance of the orders Lactobacillales, Bacteroidales, Cardiobacteriales, Clostridiales, and Pseudomonadales change over time, here accumulated degree hours (ADH), within decomposition-impacted soils. Trends for each of the 19 individuals (named “TOX###”) are delineated by color. (TIF) S1 Table Demographics of study individuals. ‘Timepoints for Models’ is the number of samples included in model creation for respective individuals. (PDF) S2 Table Summary statistics for all random forest models. Values are means for 100 runs of each model. OOB MSE = out-of-bag mean squared error, RMSE = root mean squared error, MAE = mean absolute error, OTU = Operational taxonomic unit. (PDF) S3 Table Cross validation results for all random forest models. Random forest models perform better (r 2 ) on training set than the testing set. (PDF) S4 Table Analysis of variance (ANOVA) results from linear model testing for the effect of biological marker ( e.g ., 16S, ITS, or 16S-ITS) on random forest model mean absolute error (MAE). (PDF) S5 Table Post-hoc t test results for testing differences between biological marker groups ( e.g ., 16S, ITS, or 16S-ITS). P values were adjusted for multiple comparison (Adjusted p ) using the Holm method. (PDF) S6 Table Analysis of variance (ANOVA) results from linear model testing for the the effect of taxonomic level ( e.g ., phylum, class, order, OTU) on random forest model mean absolute error (MAE). (PDF) S7 Table Analysis of variance (ANOVA) results from linear models testing for the effect of taxonomic level ( e.g ., phylum, class, order, OTU) within biological marker groups ( e.g . 16S, ITS, or 16S-ITS) on random forest model mean absolute error (MAE). (PDF) S8 Table Top 25 most important model features within the top performing model for each biological marker (16S phylum + env, 16S-ITS order, and ITS order + env). Features are 16S OTU (Otu####), ITS OTU (ITS####), or environmetal predictors, depending on the model. Importance reports the the decrease in mean square error (MSE) for each feature. For 16S and ITS features, taxonomy is report to the lowest taxonomic level for each respective model. (PDF)
|
ICMR National Virtual Centre for Clinical Pharmacology with Network of Rational Use of Medicines & Product Development Centres | cdcc7b69-8067-41be-96ae-87d5257efd84 | 9210526 | Pharmacology[mh] | Clinical Pharmacology involves the development of new drugs; their application as therapeutic agents, and study of adverse effects in individuals and society . The Indian Council of Medical Research (ICMR) has supported the development of clinical pharmacology in India over the last 50 yr through its extramural and intramural programmes by way of training programmes for capacity building and advanced research activities . The training programmes also imparted training to the participants from other countries. The Centres of Advanced Research (CAR) were set up in Mumbai, Chandigarh, Puducherry and Hyderabad for research on pharmacokinetics, therapeutic drug monitoring (TDM), pharmacovigilance, clinical trials, pharmacodynamics, pharmacogenetics, traditional medicines, relevant to public health, development of national policies, drug development and education, attracting grants from agencies such as WHO . In 2010, a brain storming session recommended creation of an Institute of Clinical Pharmacology, with public health orientation for safe, effective and economic products and rational use of medicines for Indian population . However, it could not be started due to paucity of funds. The need for an institute was again reiterated in a review of clinical pharmacology research in India for developing products for Indian population and rational use of medicines. Furthermore, it was noted that drug development in academia and Government funded institutions is hampered by inadequate trained manpower, lack of interaction between industry and academia/public research institutions , and between basic sciences and clinical researchers. In view of above and current scientific developments and to further strengthen clinical pharmacology towards healthcare needs of the country, the National virtual Centre for Clinical Pharmacology (NvCCP) with a network of Product Development Centres (PDCs) to promote drug development in line with the New Drugs and Clinical Trials Rules 2019 notified on 19 March, 2019 and Rational Use of Medicines Centres (RUMCs) for cost-effective use, with a Technical Advisory Group (TAG) (deemed virtual centre) of experts from different disciplines for guidance and monitoring progress of these activities was set up in 2019.
The envisaged objectives and output of the PDCs were as follows: ( i ) to evaluate (20/yr) completed research projects, for suitability to develop products for human use; ( ii ) recommend suitable products for further validation, studies for investigational new drug (IND) and to develop IND application; ( iii ) carry out Phase I, II, and III clinical trials (two/year); ( iv ) carry out studies for evidence/provide evidence based recommendations for safe and effective use of marketed products using TDM, biomarkers and genetic tests; and (two/year); ( v ) carry out studies for evidence/provide evidence-based recommendation for standard treatment guidelines for public health/Government programmes (two/year). The primary impact will be the development of national asset for conducting clinical trials, publications, training, capacity building, development of guidelines for minimum/optimum requirements for conducting clinical trials, standard operating procedures (SOPs) for clinical trial related activities. Data from clinical trials of marketed drugs will provide evidence base for policies, practice, cost-saving strategies and evaluated recommended projects/products, if successful, will lead to safe and effective products. Eleven institutions and investigators were identified for PDCs, based on their prior work, initiatives taken, publications, departmental infrastructure, faculty that could contribute, availability of collaborating institutes, previous grants received and were approved in 2019. There were four PDCs in Mumbai, Maharashtra, two in Telangana, and one each in Chandigarh, New Delhi, Lucknow, Kolkata and Patna. During the first year, these PDCs evaluated completed research projects funded by ICMR and shortlisted five projects for further development. Guidelines for infrastructure and facilities and SOPs required for Phase I studies were prepared. The PDCs also conducted studies for population pharmacokinetics of hydroxychloroquine (HCQ) in healthcare workers and COVID vaccine trials.
Widespread overuse, inappropriate selection of antimicrobials and high level of polypharmacy leading to adverse drug reactions (ADR), antimicrobial resistance, lack of effect, increasing cost to patient and society have been noted . Globally, half of the medicines use has been found to be inappropriate . In UK, 7-10 per cent of prescriptions of newly graduated doctors were found to contain errors . Inadequate education, ineffective, insufficient regulation around appropriate medication use were found to be the important reasons . Hence, prescribing competency for medical graduates was included in the curriculum . In view of this, in 2019, the ICMR set up a network of RUMCs (rational use of medicine centres) in the departments of Pharmacology of various teaching medical institutions located in different parts of the country with the following envisaged objectives and output: ( i ) Prescription audit/research, evaluate, analyze, interpret, for WHO indicators, inappropriateness, use of irrational fixed-dose drug combinations (FDCs), non-national list of essential medicines (NLEMs), identify gaps and errors, (for 1000 prescriptions per year), contribute to national database, recommend corrective steps; ( ii ) develop online training course for prescribing skills (PSC) (for interns, Government medical officers, private general practitioners); ( iii ) based on the Medical Council of India (MCI) curriculum, university curriculum, with prioritization based on published literature, experience, develop curriculum (for 2 modules, review two modules developed by other centre); ( iv ) develop training modules for the course based on standard treatment guidelines, standard treatment workflows and other resources, (for 2 modules, review 2 modules developed by other centre); ( v ) develop assessment questions, validate for two modules, review two modules developed by the other centre. The envisaged impact of these centres was as follows: the online PSC made available to all interns, practitioners in the country. Pre-test and post-test assignments will add to the training experience and assess change in knowledge. Prescription research will evaluate approximately 10,000 prescriptions. Data of all centres will be aggregated and published and will also be used in revising the content of the online course and provide inputs for NLEM revisions. Fifteen non-ICMR institutions and investigators were identified based on their prior work, initiatives taken, publications, departmental infrastructure, faculty that could contribute, link with collaborating institutes, and previous grant received. Five RUMCs were also set up at the ICMR institutes. These centres were approved in 2019 and set up in the same. There were two centres each in Kolkata, Ludhiana and New Delhi, and one each in Mumbai, Puducherry, Ahmedabad, Chandigarh, Patna, Bhopal, Vadodra, Vellore and Bangaluru. The six centres in ICMR Institutes were at National Institute for Research in Reproductive and Child Health (NIRRCH) and National Institute of Immunohematology (NIIH), Mumbai, National Institute of Cholera and Enteric Diseases (NICED), Kolkata, National Institute for Research in Tuberculosis (NIRT), Chennai and The Rajendra Memorial Research Institute of Medical Sciences (RMRIMS), Patna, National institute of Epidemiology (NIE), Chennai. During the first year, RUMCs constituted RUMC committees of clinicians from clinical departments and community medicine, and developed curriculum, training modules and assessment questions for prescribing skill course (PSC). The online course for PSC was launched in September 2020 by the Director General, ICMR, with the ICMR-National Institute of Epidemiology through Government of India SWAYAM portal. Approximately 5000 prescriptions were captured by the RUMCs and analyzed. Safety and efficacy of hydroxychloroquine for prophylaxis against COVID-19 in healthcare workers was also studied.
This initiative of the ICMR has a vision to create a national platform to promote new therapeutic products as an outcome of research from Indian institutions, to create competency for rational use of medicines and has a goal of providing cost-effective healthcare. The research activities are undertaken under a virtual center with network of various centres, funded for five years. Subsequently, there will be a need to establish a permanent centre with physical infrastructure that will enable a robust mechanism for catering to the research on different aspect of product development and other areas of clinical pharmacology with translational potential for the benefit of the Indian population.
|
Posterior approach unilateral laminectomy, debridement, and preshaped titanium mesh bone grafting with internal fixation for treatment of lumbar tuberculosis | b063e767-b7b7-4ccf-a908-ae67bd5656d8 | 11861922 | Surgical Procedures, Operative[mh] | Tuberculosis (TB) is one of the oldest known human infectious diseases , which remains prevalent in many regions of developing countries presently. Approximately 10% of patients with active pulmonary disease also have skeletal TB, with the spine being the most common site of skeletal involvement . Spinal TB can lead to intervertebral space and vertebral body destruction and collapse, causing spinal kyphosis and even a certain probability of paralysis . Currently, spinal chemotherapy is considered the most important method for treating spinal TB . However, for patients with severe spinal destruction and accompanying neurological symptoms, mere drug treatment may not reverse the extensive bone damage and neural compression. Such patients require surgical intervention . The thoracic and lumbar vertebrae are the most common sites for spinal TB occurrence. Surgical approaches for lumbar TB that have been documented include anterior, posterior, and combined anterior-posterior procedures. In recent years, many spinal surgeons have achieved satisfactory clinical outcomes by performing single-stage posterior approach procedure . The objective of surgical treatment is to achieve thorough lesion debridement while minimizing surgical trauma, expediting postoperative recovery, and ensuring the enduring efficacy of the intervention. Thus, we endeavored to preserve the spinous process and one side of the vertebral lamina during the posterior approach surgery in patients with single-segment lumbar TB. Simultaneously, lesion clearance was conducted, and structural bone grafting was performed using titanium mesh. This procedure aimed to strike a balance between treatment effectiveness and the preservation of posterior spinal structures, thereby reducing surgical trauma and its impact on spinal stability. A comparative evaluation of this approach, involving lesion clearance and titanium mesh fusion fixation while preserving the spinous process, was conducted against the traditional posterior approach without spinous process preservation, to assess its therapeutic effectiveness.
Inclusion and exclusion criteria Inclusion criteria: (1) Single-segment lumbar spine tuberculosis (SSLTB) with the affected segment located between L1 and S1, resulting in severe bone destruction, leading to spinal instability or kyphotic deformity. (2) Presence of nerve compression or functional impairment. (3) Severe back pain or neurological symptoms, leading to walking difficulty, with poor response to conservative treatment. (4) Treatment through a posterior approach surgery. Exclusion criteria: (1) Patients with cervical or thoracic spine tuberculosis, or those with multi-segmental infections. (2) Spinal bone destruction caused by other types of spine infections or tumors. (3) Treatment through other surgical approaches such as anterior or combined anterior-posterior approaches. (4) Presence of other spinal disorders or a history of previous spinal surgeries. General information A total of 65 cases of lumbar spine TB patients who underwent surgical treatment at our hospital were collected from the period of 2016 to 2020. Among them, there were 33 male patients and 32 female patients. The age range was 28 to 67 years. All patients presented with varying degrees of back pain and lower limb neurological symptoms. Blood tests and imaging examinations showed changes consistent with characteristics of TB infection. All patients were divided into two groups, Group A and Group B, based on the different surgical procedure. Group A underwent posterior approach unilateral laminectomy with debridement, titanium mesh bone grafting and internal fixation (ULT). Group B underwent traditional posterior approach bilateral laminectomy, spinus process removal, lesion debridement with titanium mesh bone grafting and internal fixation (BLT). Among them, Group A consisted of 34 patients, including 16 males and 18 females, while Group B consisted of 31 patients, including 17 males and 14 females. Preoperative management All patients underwent preoperative blood tests including complete blood count (CBC), liver and kidney function, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), tuberculosis T-spot, and other relevant examinations. Additionally, spinal X-rays, CT scans, and MRI scans were performed for detailed imaging assessment. Prior to surgery, all patients received preoperative antituberculosis drug treatment and supportive care, including pain management, neural nourishment, and nutritional supplementation. All patients received at least 2 weeks of quadruple antituberculosis therapy (isoniazid, rifampicin, ethambutol, pyrazinamide, HREZ) before surgery. Surgical method In group A, all 34 patients were treated with ULT procedure. The patient is positioned prone, and a midline incision is made over the affected segment to fully expose the posterior structures of spine. Pedicle screws are inserted into the normal pedicle above and below the lesion, and their positions are confirmed using C-arm X-ray. Remove a portion of the facet joint and lamina on the side with more pronounced bone destruction, exposing the spinal canal, and clearing necrotic and purulent tissue within the spinal canal. Temporary fixation rods are placed on the opposite side, and the lesion gap is appropriately expanded, or an intervertebral spreader is used to temporarily expand the gap to enlarge the surgical field and the operating space. The dura mater and nerve roots are identified and protected with nerve root hooks. Various tools such as bone nibblers, curettes, and bone knives of different sizes and orientations are used in conjunction with a suction device to thoroughly debride the lesion and sclerotic bone until normal bone tissue is visible. The excised lesion tissue is collected for pathological examination and bacterial culture. The area is then flushed with hydrogen peroxide and copious amounts of normal saline. The resected laminae bone tissue that remains undamaged by tuberculous infection is used as autologous bone graft. When the amount of autologous bone tissue is insufficient, allogeneic bone is used as a supplement. A properly sized titanium mesh is shaped based on the extent of the resected lamina. Autologous or allogeneic bone particles are packed into the pre-shaped titanium mesh. The titanium mesh is inserted into the anterior space, and the remaining gap in the vertebral body lesion is filled with bone particles. A fixation rod is placed on the debrided side. Bilateral compression of screws is applied for titanium mesh clamping and correction of kyphosis. C-arm fluoroscopy is used to confirm the proper placement of internal fixation and titanium mesh. The incision is thoroughly cleaned, isoniazid and streptomycin are placed, a drainage tube is inserted, and the incision is sutured in layers. For a simplified illustration of the procedure, refer to Fig. . In group B, all 31 patients were treated with BLT procedure. The patient is positioned in a prone posture, and a midline incision is made over the affected segment to adequately expose the posterior structures. Pedicle screws are inserted above and below the lesion into the normal pedicles, with their positions confirmed using C-arm X-ray. The spinous process and lamina on one side are removed, and a temporary fixation rod is placed on the opposite side. The lesion is adequately cleared on one side, and the excised lesion tissue is collected for pathological examination and bacterial culture. The temporary fixation rod is then repositioned on the opposite side, and the lamina on the other side is removed, and lesion clearance is performed in the same manner. Thorough debridement is followed by irrigation with hydrogen peroxide and copious amounts of normal saline. An appropriately sized titanium mesh is chosen and filled with healthy autologous bone graft particles sourced from the spinous process and lamina, or with allogeneic bone graft particles. The titanium mesh is inserted into the anterior column. Fixation rods are inserted and bilateral compression is applied. C-arm fluoroscopy is used to verify the proper positioning of the internal fixation and titanium mesh. Transverse connectors are added to further secure the fixation system. The incision is thoroughly cleaned, isoniazid and streptomycin are applied, a drainage tube is placed, and the incision is closed in layers. Postoperative management Intraoperative records include data on blood loss, surgical duration, transfusion events, and any other relevant information. Surgical and postoperative complications are also documented. During the hospital stay, regular follow-up blood tests including CBC, ESR, CRP, and others are conducted. Imaging examinations such as X-rays and CT scans are also performed. Routine antibiotic prophylaxis is administered for three days postoperatively to prevent incisional infection. Symptomatic supportive care, including pain management and nutritional supplementation, is provided. The drainage tube is removed once the drainage volume is less than 30 ml per day. After drainage tube removal, early ambulation is encouraged based on the wound healing progress and improvement in symptoms. The time of the patient’s first postoperative ambulation is recorded. Routine brace support is provided for 3–6 months postoperatively until satisfactory bone fusion is achieved. All patients continue to receive quadruple oral antituberculosis drug treatment after surgery. Pyrazinamide is discontinued after 6 months, and antituberculosis treatment continues for 9–12 months. During the follow-up period, assessments are conducted every 3 months within the first year and every 6 months after the first year. Each follow-up includes blood tests and X-ray/CT scans. Brantigan scoring system was chosen as the evaluation criterion for bone fusion. The Brantigan scoring criteria were as follows: 4 points for complete fusion with good contour and the appearance of continuous callus; 3 points for good fusion but with a still faintly visible translucent line; 2 points for continuous callus in the upper and lower parts (50%) but with a significant amount of translucent line remaining; 1 point for upper and lower parts not connected, but with bone volume greater than the postoperative intervertebral bone graft volume; and 0 points for intervertebral bone graft absorption, decreased intervertebral space height, and non-fusion of the vertebral body. A score of ≥ 3 points was considered indicative of intervertebral fusion. Patient data such as visual analogue scalescore (VAS), Oswestry disability index (ODI), and American spinal injury association (ASIA) grades are recorded during each follow-up. Ethical approval This retrospective chart review study was in accordance with the ethical standards of the institutional and national research committee and with the Helsinki Declaration and its later amendments. All protocols were approved by the Investigation Committee (Institutional Review Board) of Central-South University (ethics number: 202004137). We confirm that written informed consent was obtained from all subjects. Data analysis SPSS 22 software was used for data analysis. A paired sample t test was used to compare the preoperative, postoperative and follow-up data, including kyphosis Cobb angle, ESR, CRP, VAS, and ODI. An independent sample t test was used to compare the data between group A and group B. A p value < 0.05 was considered indicative of statistical significance.
Inclusion criteria: (1) Single-segment lumbar spine tuberculosis (SSLTB) with the affected segment located between L1 and S1, resulting in severe bone destruction, leading to spinal instability or kyphotic deformity. (2) Presence of nerve compression or functional impairment. (3) Severe back pain or neurological symptoms, leading to walking difficulty, with poor response to conservative treatment. (4) Treatment through a posterior approach surgery. Exclusion criteria: (1) Patients with cervical or thoracic spine tuberculosis, or those with multi-segmental infections. (2) Spinal bone destruction caused by other types of spine infections or tumors. (3) Treatment through other surgical approaches such as anterior or combined anterior-posterior approaches. (4) Presence of other spinal disorders or a history of previous spinal surgeries.
A total of 65 cases of lumbar spine TB patients who underwent surgical treatment at our hospital were collected from the period of 2016 to 2020. Among them, there were 33 male patients and 32 female patients. The age range was 28 to 67 years. All patients presented with varying degrees of back pain and lower limb neurological symptoms. Blood tests and imaging examinations showed changes consistent with characteristics of TB infection. All patients were divided into two groups, Group A and Group B, based on the different surgical procedure. Group A underwent posterior approach unilateral laminectomy with debridement, titanium mesh bone grafting and internal fixation (ULT). Group B underwent traditional posterior approach bilateral laminectomy, spinus process removal, lesion debridement with titanium mesh bone grafting and internal fixation (BLT). Among them, Group A consisted of 34 patients, including 16 males and 18 females, while Group B consisted of 31 patients, including 17 males and 14 females.
All patients underwent preoperative blood tests including complete blood count (CBC), liver and kidney function, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), tuberculosis T-spot, and other relevant examinations. Additionally, spinal X-rays, CT scans, and MRI scans were performed for detailed imaging assessment. Prior to surgery, all patients received preoperative antituberculosis drug treatment and supportive care, including pain management, neural nourishment, and nutritional supplementation. All patients received at least 2 weeks of quadruple antituberculosis therapy (isoniazid, rifampicin, ethambutol, pyrazinamide, HREZ) before surgery.
In group A, all 34 patients were treated with ULT procedure. The patient is positioned prone, and a midline incision is made over the affected segment to fully expose the posterior structures of spine. Pedicle screws are inserted into the normal pedicle above and below the lesion, and their positions are confirmed using C-arm X-ray. Remove a portion of the facet joint and lamina on the side with more pronounced bone destruction, exposing the spinal canal, and clearing necrotic and purulent tissue within the spinal canal. Temporary fixation rods are placed on the opposite side, and the lesion gap is appropriately expanded, or an intervertebral spreader is used to temporarily expand the gap to enlarge the surgical field and the operating space. The dura mater and nerve roots are identified and protected with nerve root hooks. Various tools such as bone nibblers, curettes, and bone knives of different sizes and orientations are used in conjunction with a suction device to thoroughly debride the lesion and sclerotic bone until normal bone tissue is visible. The excised lesion tissue is collected for pathological examination and bacterial culture. The area is then flushed with hydrogen peroxide and copious amounts of normal saline. The resected laminae bone tissue that remains undamaged by tuberculous infection is used as autologous bone graft. When the amount of autologous bone tissue is insufficient, allogeneic bone is used as a supplement. A properly sized titanium mesh is shaped based on the extent of the resected lamina. Autologous or allogeneic bone particles are packed into the pre-shaped titanium mesh. The titanium mesh is inserted into the anterior space, and the remaining gap in the vertebral body lesion is filled with bone particles. A fixation rod is placed on the debrided side. Bilateral compression of screws is applied for titanium mesh clamping and correction of kyphosis. C-arm fluoroscopy is used to confirm the proper placement of internal fixation and titanium mesh. The incision is thoroughly cleaned, isoniazid and streptomycin are placed, a drainage tube is inserted, and the incision is sutured in layers. For a simplified illustration of the procedure, refer to Fig. . In group B, all 31 patients were treated with BLT procedure. The patient is positioned in a prone posture, and a midline incision is made over the affected segment to adequately expose the posterior structures. Pedicle screws are inserted above and below the lesion into the normal pedicles, with their positions confirmed using C-arm X-ray. The spinous process and lamina on one side are removed, and a temporary fixation rod is placed on the opposite side. The lesion is adequately cleared on one side, and the excised lesion tissue is collected for pathological examination and bacterial culture. The temporary fixation rod is then repositioned on the opposite side, and the lamina on the other side is removed, and lesion clearance is performed in the same manner. Thorough debridement is followed by irrigation with hydrogen peroxide and copious amounts of normal saline. An appropriately sized titanium mesh is chosen and filled with healthy autologous bone graft particles sourced from the spinous process and lamina, or with allogeneic bone graft particles. The titanium mesh is inserted into the anterior column. Fixation rods are inserted and bilateral compression is applied. C-arm fluoroscopy is used to verify the proper positioning of the internal fixation and titanium mesh. Transverse connectors are added to further secure the fixation system. The incision is thoroughly cleaned, isoniazid and streptomycin are applied, a drainage tube is placed, and the incision is closed in layers.
Intraoperative records include data on blood loss, surgical duration, transfusion events, and any other relevant information. Surgical and postoperative complications are also documented. During the hospital stay, regular follow-up blood tests including CBC, ESR, CRP, and others are conducted. Imaging examinations such as X-rays and CT scans are also performed. Routine antibiotic prophylaxis is administered for three days postoperatively to prevent incisional infection. Symptomatic supportive care, including pain management and nutritional supplementation, is provided. The drainage tube is removed once the drainage volume is less than 30 ml per day. After drainage tube removal, early ambulation is encouraged based on the wound healing progress and improvement in symptoms. The time of the patient’s first postoperative ambulation is recorded. Routine brace support is provided for 3–6 months postoperatively until satisfactory bone fusion is achieved. All patients continue to receive quadruple oral antituberculosis drug treatment after surgery. Pyrazinamide is discontinued after 6 months, and antituberculosis treatment continues for 9–12 months. During the follow-up period, assessments are conducted every 3 months within the first year and every 6 months after the first year. Each follow-up includes blood tests and X-ray/CT scans. Brantigan scoring system was chosen as the evaluation criterion for bone fusion. The Brantigan scoring criteria were as follows: 4 points for complete fusion with good contour and the appearance of continuous callus; 3 points for good fusion but with a still faintly visible translucent line; 2 points for continuous callus in the upper and lower parts (50%) but with a significant amount of translucent line remaining; 1 point for upper and lower parts not connected, but with bone volume greater than the postoperative intervertebral bone graft volume; and 0 points for intervertebral bone graft absorption, decreased intervertebral space height, and non-fusion of the vertebral body. A score of ≥ 3 points was considered indicative of intervertebral fusion. Patient data such as visual analogue scalescore (VAS), Oswestry disability index (ODI), and American spinal injury association (ASIA) grades are recorded during each follow-up.
This retrospective chart review study was in accordance with the ethical standards of the institutional and national research committee and with the Helsinki Declaration and its later amendments. All protocols were approved by the Investigation Committee (Institutional Review Board) of Central-South University (ethics number: 202004137). We confirm that written informed consent was obtained from all subjects.
SPSS 22 software was used for data analysis. A paired sample t test was used to compare the preoperative, postoperative and follow-up data, including kyphosis Cobb angle, ESR, CRP, VAS, and ODI. An independent sample t test was used to compare the data between group A and group B. A p value < 0.05 was considered indicative of statistical significance.
General data All patients successfully underwent surgery, with 34 cases in Group A and 31 cases in Group B. The mean surgical duration for Group A was 137.14 ± 19.38 (CI95% 125.94–148.33) minutes, and for Group B, it was 157.05 ± 30.97 (CI95% 141.13–172.98) minutes. In Group A, 14 patients required intraoperative blood transfusion, with an average transfusion volume of 162.5 ± 25.00 (CI95% 122.71–202.28)ml. In Group B, 16 patients received intraoperative blood transfusions, with an average transfusion volume of 205.55 ± 39.08 (CI95% 175.51 to 235.60)ml. Additionally, we recorded the time from postoperative period to the patient’s first mobilization as an assessment of their recovery. The average time for Group A was 12.92 ± 0.99 (CI95% 12.35–13.50) days, and for Group B, it was 17.23 ± 2.65 (CI95% 15.86–18.60) days (Table ). Blood test results In the analysis of blood test results, in Group A, ESR decreased from preoperative 42.07 ± 7.77 (CI95% 37.58–46.55) mm/h to 19.07 ± 4.10 (CI95% 16.70–21.44) mm/h at 2 weeks postoperatively, and further to 6.57 ± 1.74 (CI95% 5.56–7.57) mm/h at the 6-month follow-up. In Group B, ESR decreased from preoperative 42.07 ± 7.77 (CI95% 39.84–51.68) mm/h to 19.07 ± 4.10 (CI95% 15.87–21.06) mm/h at 2 weeks postoperatively, and further to 6.57 ± 1.74 (CI95% 5.87–7.41) mm/h at the 6-month follow-up. In Group A, CRP decreased from preoperative 40.71 ± 6.97 (CI95% 36.68–44.74) mg/L to 16.71 ± 3.64 (CI95% 14.60–18.81) mg/L at 2 weeks postoperatively, and further to 5.28 ± 1.38 (CI95% 4.48–6.08) mg/L at the 6-month follow-up. In Group B, CRP decreased from preoperative 44.23 ± 10.43 (CI95% 38.86–49.60) mg/L to 16.88 ± 3.83 (CI95% 14.90–18.85) mg/L at 2 weeks postoperatively, and further to 5.47 ± 1.73 (CI95% 4.57–6.36) mg/L at the 6-month follow-up. In both groups, ESR and CRP levels were within the normal range at the 6-month follow-up (Table ; Fig. ). Symptomatic and neurological result In both groups, the VAS and ODI scores of patients improved significantly at 3 months postoperatively compared to preoperative scores, and further reductions were observed at the latest follow-up (Table ; Fig. ). The ASIA grades in both groups improved postoperatively compared to preoperative grades, and there were no instances of worsened ASIA grades in any patients during the postoperative and follow-up periods (Table ). Typical radiographic data of Group A cases are shown in Fig. , and typical radiographic data of Group B cases are shown in Fig. . Complications All patients were followed up for over 2 years, with the longest follow-up duration reaching 5 years. No cases of severe incisional infection, sinus tract formation, internal fixation failure, or worsened postoperative paralysis symptoms were observed during the follow-up period. There were no instances of TB recurrence. 4 patients experienced superficial wound infections postoperatively, all of which resolved after active wound care, intensified antibiotic treatment, and nutritional support. 3 patient experienced postoperative cerebrospinal fluid leakage, which was resolved after conservative treatments including maintaining a feet-elevated and head-lowered position and restricting positional activities, along with fluid supplementation (Table ).
All patients successfully underwent surgery, with 34 cases in Group A and 31 cases in Group B. The mean surgical duration for Group A was 137.14 ± 19.38 (CI95% 125.94–148.33) minutes, and for Group B, it was 157.05 ± 30.97 (CI95% 141.13–172.98) minutes. In Group A, 14 patients required intraoperative blood transfusion, with an average transfusion volume of 162.5 ± 25.00 (CI95% 122.71–202.28)ml. In Group B, 16 patients received intraoperative blood transfusions, with an average transfusion volume of 205.55 ± 39.08 (CI95% 175.51 to 235.60)ml. Additionally, we recorded the time from postoperative period to the patient’s first mobilization as an assessment of their recovery. The average time for Group A was 12.92 ± 0.99 (CI95% 12.35–13.50) days, and for Group B, it was 17.23 ± 2.65 (CI95% 15.86–18.60) days (Table ).
In the analysis of blood test results, in Group A, ESR decreased from preoperative 42.07 ± 7.77 (CI95% 37.58–46.55) mm/h to 19.07 ± 4.10 (CI95% 16.70–21.44) mm/h at 2 weeks postoperatively, and further to 6.57 ± 1.74 (CI95% 5.56–7.57) mm/h at the 6-month follow-up. In Group B, ESR decreased from preoperative 42.07 ± 7.77 (CI95% 39.84–51.68) mm/h to 19.07 ± 4.10 (CI95% 15.87–21.06) mm/h at 2 weeks postoperatively, and further to 6.57 ± 1.74 (CI95% 5.87–7.41) mm/h at the 6-month follow-up. In Group A, CRP decreased from preoperative 40.71 ± 6.97 (CI95% 36.68–44.74) mg/L to 16.71 ± 3.64 (CI95% 14.60–18.81) mg/L at 2 weeks postoperatively, and further to 5.28 ± 1.38 (CI95% 4.48–6.08) mg/L at the 6-month follow-up. In Group B, CRP decreased from preoperative 44.23 ± 10.43 (CI95% 38.86–49.60) mg/L to 16.88 ± 3.83 (CI95% 14.90–18.85) mg/L at 2 weeks postoperatively, and further to 5.47 ± 1.73 (CI95% 4.57–6.36) mg/L at the 6-month follow-up. In both groups, ESR and CRP levels were within the normal range at the 6-month follow-up (Table ; Fig. ).
In both groups, the VAS and ODI scores of patients improved significantly at 3 months postoperatively compared to preoperative scores, and further reductions were observed at the latest follow-up (Table ; Fig. ). The ASIA grades in both groups improved postoperatively compared to preoperative grades, and there were no instances of worsened ASIA grades in any patients during the postoperative and follow-up periods (Table ). Typical radiographic data of Group A cases are shown in Fig. , and typical radiographic data of Group B cases are shown in Fig. .
All patients were followed up for over 2 years, with the longest follow-up duration reaching 5 years. No cases of severe incisional infection, sinus tract formation, internal fixation failure, or worsened postoperative paralysis symptoms were observed during the follow-up period. There were no instances of TB recurrence. 4 patients experienced superficial wound infections postoperatively, all of which resolved after active wound care, intensified antibiotic treatment, and nutritional support. 3 patient experienced postoperative cerebrospinal fluid leakage, which was resolved after conservative treatments including maintaining a feet-elevated and head-lowered position and restricting positional activities, along with fluid supplementation (Table ).
Selection of surgical approach The surgical approach for lumbar TB has been a subject of ongoing debate. Currently, the employment of anterior-posterior combined procedures has become increasingly infrequent due to the substantial trauma. In a study by Wang, after comparing the therapeutic outcomes of three surgical methods, it was concluded that anterior approach surgery yielded favorable results and exhibited fewer complications compared to posterior and anterior-posterior combined approaches . However, in Wang’s meta-analysis , the posterior approach was found to be superior to the other two methods. Our perspective aligns with the notion that the anatomical complexity of the anterior approach surpasses that of the posterior approach. While the anterior approach effectively removes anterior lesions, it falls short in terms of internal fixation stability and the correction of kyphotic deformities compared to the posterior approach, particularly in cases involving long vertebral segments. Given these considerations, we are inclined to favor a single-stage posterior approach over the other two approach for our surgical interventions. Preserving spinus process The posterior column of the spine plays an indispensable role in upholding spinal stability and withstanding shear, rotational, and compressive forces , particularly when instability arises due to infection affecting the anterior column . In conventional posterior surgical procedures, comprehensive exposure of the operative field typically necessitates excision of spinous process and bilateral lamina – . In ULT procedure, we try to retain posterior structures including the spinous process, supraspinous ligament, interspinous ligament, and half lamina. This procedure significantly curtails disruption to the posterior column. Given the constrained field of view in unilateral lamina excision, proper extension of the temporary fixing rod or utilization of a unilateral spinal distractor effectively expands the surgical workspace. Employing a variety of sized and oriented curettes coupled with thorough irrigation enables effective lesion debridement. Additionally, meticulous removal of sclerotic bone is particularly significant, enhancing the penetration of anti-tuberculosis medication into the lesion site . In comparison to BLT, the ULT technique results in no marked discrepancy in postoperative declines of inflammatory markers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP). Conversely, surgery duration, blood loss, and transfusion requirements are appreciably reduced, indicative of diminished surgical trauma. Since the spinus process remained intact during the procedure, dura mater and nerve roots are well protected. Notably, there have been no instances of exacerbated paralysis or complications observed during the postoperative and follow-up period. Preshaped titanium mesh for grafting Prior reports have documented various methods of anterior structure grafting in posterior approach surgeries, such as spinous process autografts, rib grafts, and titanium mesh grafts – . In comparison to the former two, titanium mesh grafting boasts superior rigidity, facilitating the restoration of vertebral body and intervertebral height, while providing more ideal support for the anterior column . Grafting materials like spinous process are susceptible to complications such as bone absorption, whereas titanium mesh grafting circumvents such issues , . To ensure effective support and graft fusion, the dimensions of the titanium mesh should be maximized within the confines of the operating space. On the other hand, the stability of pedicle screws can be compromised due to varying degrees of osteoporosis resulting from TB infection – . At such times, employing a stabilizing titanium mesh alongside a relatively intact posterior column structure can reduce the risk of internal fixation failure. Given that the ULT procedure involves unilateral lamina removal, there are spatial limitations for placing the titanium mesh. Consequently, we have undertaken preliminary shaping of the titanium mesh. The use of pre-shaped titanium mesh effectively diminishes the required space for mesh insertion. Previously, other scholars have reported limitations in unilateral lamina removal due to the restricted operative space, often utilizing autogenous or allogeneic bone fragments for anterior grafting , . This might be inadequate for furnishing sufficient support to the anterior spine column compared to titanium mesh grafting. In ULT procedure, since only one side of lamina was removed, we chose pre-shaped titanium mesh as grafting container to provide ample support in the confined surgical space, thus ensuring adequate structural support. Benefitting from the robust support, in combination with the commendable corrective and fixation effects of pedicle screws, we have employed single-segment fusion in patients with relatively small infection zones, which further limits overall surgical trauma, retains as many functional units as possible, and enhances postoperative recovery and quality of life for patients. In cases of lumbar tuberculosis, vertebral body destruction often leads to vertebral collapse, resulting in varying degrees of lordosis decrease or kyphotic deformity. Among patients with such problems, a routine approach during surgery is to employ compression with fixation rod to restore physiological spinal curvature , , . The placement of customized titanium mesh acts as a robust lever fulcrum during posterior compression, and the relative integrity of the posterior structures enhances overall stability, further diminishing the risk of postoperative kyhosis. Limitations However, the ULT procedure does exhibit certain limitations. For patients with a broader scale of destruction involving multiple consecutive vertebrae necessitating surgical intervention, the constrained operative space resulting from unilateral lamina removal hampers effective lesion debridement and grafting. Currently, this approach is only applicable to single-segment spinal TB patients. Furthermore, this study is retrospective in nature, which introduces a degree of selection bias. The average follow-up time for all patients was 36 months, lacking a medium to long-term follow-up assessment. Moving forward, we will further expand the sample size and extend the follow-up period, incorporating additional indicators like quality of life and cost-effectiveness. Multicenter and prospective studies will also be pursued to enhance research persuasiveness.
The surgical approach for lumbar TB has been a subject of ongoing debate. Currently, the employment of anterior-posterior combined procedures has become increasingly infrequent due to the substantial trauma. In a study by Wang, after comparing the therapeutic outcomes of three surgical methods, it was concluded that anterior approach surgery yielded favorable results and exhibited fewer complications compared to posterior and anterior-posterior combined approaches . However, in Wang’s meta-analysis , the posterior approach was found to be superior to the other two methods. Our perspective aligns with the notion that the anatomical complexity of the anterior approach surpasses that of the posterior approach. While the anterior approach effectively removes anterior lesions, it falls short in terms of internal fixation stability and the correction of kyphotic deformities compared to the posterior approach, particularly in cases involving long vertebral segments. Given these considerations, we are inclined to favor a single-stage posterior approach over the other two approach for our surgical interventions.
The posterior column of the spine plays an indispensable role in upholding spinal stability and withstanding shear, rotational, and compressive forces , particularly when instability arises due to infection affecting the anterior column . In conventional posterior surgical procedures, comprehensive exposure of the operative field typically necessitates excision of spinous process and bilateral lamina – . In ULT procedure, we try to retain posterior structures including the spinous process, supraspinous ligament, interspinous ligament, and half lamina. This procedure significantly curtails disruption to the posterior column. Given the constrained field of view in unilateral lamina excision, proper extension of the temporary fixing rod or utilization of a unilateral spinal distractor effectively expands the surgical workspace. Employing a variety of sized and oriented curettes coupled with thorough irrigation enables effective lesion debridement. Additionally, meticulous removal of sclerotic bone is particularly significant, enhancing the penetration of anti-tuberculosis medication into the lesion site . In comparison to BLT, the ULT technique results in no marked discrepancy in postoperative declines of inflammatory markers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP). Conversely, surgery duration, blood loss, and transfusion requirements are appreciably reduced, indicative of diminished surgical trauma. Since the spinus process remained intact during the procedure, dura mater and nerve roots are well protected. Notably, there have been no instances of exacerbated paralysis or complications observed during the postoperative and follow-up period.
Prior reports have documented various methods of anterior structure grafting in posterior approach surgeries, such as spinous process autografts, rib grafts, and titanium mesh grafts – . In comparison to the former two, titanium mesh grafting boasts superior rigidity, facilitating the restoration of vertebral body and intervertebral height, while providing more ideal support for the anterior column . Grafting materials like spinous process are susceptible to complications such as bone absorption, whereas titanium mesh grafting circumvents such issues , . To ensure effective support and graft fusion, the dimensions of the titanium mesh should be maximized within the confines of the operating space. On the other hand, the stability of pedicle screws can be compromised due to varying degrees of osteoporosis resulting from TB infection – . At such times, employing a stabilizing titanium mesh alongside a relatively intact posterior column structure can reduce the risk of internal fixation failure. Given that the ULT procedure involves unilateral lamina removal, there are spatial limitations for placing the titanium mesh. Consequently, we have undertaken preliminary shaping of the titanium mesh. The use of pre-shaped titanium mesh effectively diminishes the required space for mesh insertion. Previously, other scholars have reported limitations in unilateral lamina removal due to the restricted operative space, often utilizing autogenous or allogeneic bone fragments for anterior grafting , . This might be inadequate for furnishing sufficient support to the anterior spine column compared to titanium mesh grafting. In ULT procedure, since only one side of lamina was removed, we chose pre-shaped titanium mesh as grafting container to provide ample support in the confined surgical space, thus ensuring adequate structural support. Benefitting from the robust support, in combination with the commendable corrective and fixation effects of pedicle screws, we have employed single-segment fusion in patients with relatively small infection zones, which further limits overall surgical trauma, retains as many functional units as possible, and enhances postoperative recovery and quality of life for patients. In cases of lumbar tuberculosis, vertebral body destruction often leads to vertebral collapse, resulting in varying degrees of lordosis decrease or kyphotic deformity. Among patients with such problems, a routine approach during surgery is to employ compression with fixation rod to restore physiological spinal curvature , , . The placement of customized titanium mesh acts as a robust lever fulcrum during posterior compression, and the relative integrity of the posterior structures enhances overall stability, further diminishing the risk of postoperative kyhosis.
However, the ULT procedure does exhibit certain limitations. For patients with a broader scale of destruction involving multiple consecutive vertebrae necessitating surgical intervention, the constrained operative space resulting from unilateral lamina removal hampers effective lesion debridement and grafting. Currently, this approach is only applicable to single-segment spinal TB patients. Furthermore, this study is retrospective in nature, which introduces a degree of selection bias. The average follow-up time for all patients was 36 months, lacking a medium to long-term follow-up assessment. Moving forward, we will further expand the sample size and extend the follow-up period, incorporating additional indicators like quality of life and cost-effectiveness. Multicenter and prospective studies will also be pursued to enhance research persuasiveness.
ULT procedure for single-segment lumbar tuberculosis can not only meet the requirements of removing lesions and stabilizing bone graft fusion but also reduce trauma and accelerate recovery after surgery.
|
International Multisite Study of Human-Induced Pluripotent Stem Cell-Derived Cardiomyocytes for Drug Proarrhythmic Potential Assessment | 2fe7ced6-e6f6-4507-b3e1-8f57ccb6fc98 | 6226030 | Physiology[mh] | Fourteen drugs have been removed from the market worldwide as a result of their potential to induce a rare but potentially fatal ventricular arrhythmia, torsades de pointes (TdP) ( ). The International Council on Harmonisation (ICH) adopted two guidelines on the assessment of drug-induced TdP (ICH S7B and ICH E14) that outline the assessment of the potential of new pharmaceuticals to delay ventricular repolarization in in vitro assays, including testing for their ability to block the human ether-a-go-go-related (hERG) potassium channel, and in vivo, to prolong the QT interval on the electrocardiogram. Adoption of these guidelines has been effective in preventing new drugs with unrecognized TdP risk from reaching the market; however, the current regulatory approach lacks specificity, because multiple drugs block hERG or prolong the QT interval but have a low risk of TdP. It is possible that overemphasis on hERG block and QT prolongation in proarrhythmic potential assessment has prevented some useful and safe drugs from reaching the market. The Comprehensive In Vitro Proarrhythmia Assay (CiPA) initiative represents a new paradigm to improve the specificity of proarrhythmic risk assessment ( ; ). The non-clinical aspects of CiPA rely on a mechanistic assessment of drug effects on cellular electrophysiology (EP) using (1) in silico reconstruction of human ventricular electrical activity based on drug effects on multiple human ionic currents, each expressed in heterologous expression systems, and (2) assessment of drug effects in human-induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) to detect any missed or unanticipated EP effects ( ). The use of hiPSC-CMs for cardiac safety evaluation of the new drug candidates continues to increase, as evidenced by numerous recent publications. Many of these studies demonstrate the ability of hiPSC-CMs as model systems to detect EP effects of drugs, including delayed or altered repolarization ( ; ; ). While encouraging, such studies typically use small test sets; different cellular preparations, protocols, and experimental endpoints; inconsistent criteria to interpret results; and different gold standards related to either delayed repolarization or proarrhythmic risk. Such differences hinder cross-site comparisons of data and recognition of sources of experimental variability. A significant step forward was made recently ( ; ), in which a large set of drugs was evaluated at multiple sites following a standardized experimental protocol; however, that study was limited to the evaluation of a single cell line and one EP platform used across sites with no statistical modeling of results. Comprehensive evaluations using multiple sites, interrogation techniques, and cell sources are necessary because all models have limitations that may appear under different circumstances. Despite possessing nearly identical underlying early after depolarization (EAD) properties as traditionally accepted models (e.g., mature canine ventricular cardiomyocytes ), hiPSCCMs are often described as having fetal or neonatal ion channel and ionic current stoichiometries ( ; ) that may interfere with the accurate prediction of proarrhythmic risk. To characterize the potential utility of hiPSC-CMs within the CiPA paradigm, the present study was conducted to characterize, in blinded fashion, the EP effects of 28 drugs with known clinical TdP risk on hiPSC-CMs using 2 commercially available hiPSC-CM lines tested across 10 experimental sites and 5 EP platforms. Specifically, this validation study focused on (1) characterization of site-to-site variability of the assessment of EP effects of the drugs using either microelectrode array (MEA) or voltage-sensing optical (VSO) techniques and standardized protocols to assess drug-induced altered repolarization, and (2) identification of important hiPSC-CM assay endpoints associated with high, intermediate, and low TdP risk using linear regression models. The present study builds upon on a previous smaller pilot study that evaluated the EP effects of 8 drugs using MEA approaches and 4 positive controls across a smaller number of sites ( ). Overall, the conceptual advance of this work is not in the discovery or the development of a new iPSC-CMs-based assay, but rather in performing a first large-scale multisite study combining MEA and VSO techniques to evaluate the current state of iPSC-CM-based assays in the assessment of drug-induced TdP.
Electrophysiological Effects Induced in hiPSC-CMs by Drugs with Known Risk Levels for Clinical TdP Risk Ten independent sites used a standardized protocol to evaluate the EP effects of 28 drugs with known levels of clinical risk categorized by expert consensus ( ) into 3 clinical TdP risk groups (high, intermediate, and low or no risk) ( ) (see for details on experimental sites, protocols, and platforms and for drug categories and concentrations). Fifteen complete datasets, including data from 5 sites that studied drug effects on both iCell cardiomyocytes 2 and Cor.4U cardiomyocytes (referred to here as iCell 2 and Cor.4U cells), were analyzed. Drug-induced repolarization prolongation (baseline and vehicle-controlled, rate-corrected action potential duration at 90% repolarization, ddAPD90c, or field potential duration, ddFPDc, depending on the EP platform used; see ) was measured along with drug-induced arrhythmias of different types ( ) or drug-induced cessation of hiPSC-CMs’ spontaneous beating (i.e., quiescence). Low TdP Risk Category The 9 drugs in the low TdP risk category were verapamil, diltiazem, loratadine, metoprolol, mexiletine, nifedipine, nitrendipine, ranolazine, and tamoxifen. hiPSC-CM responses to verapamil for all 10 sites is shown in , and corresponding figures for the rest of the low-risk drugs are shown in . No verapamil-induced repolarization prolongation or arrhythmias were observed at any concentrations studied. Diltiazem, loratadine, nifedipine, nitrendipine, and tamoxifen did not induce any arrhythmias or statistically significant repolarization prolongation at concentrations up to 20- to 140fold clinical Cmax. The remaining 3 drugs (ranolazine, metoprolol, and mexiletine) induced repolarization prolongation and arrhythmias at ≥1 of the concentrations in several datasets. Ranolazine is a known hERG blocker ( ) at clinical concentrations and produces QT prolongation. Consistent with this, 13 of 15 datasets show statistically significant ranolazine-induced repolarization prolongation at concentrations between 0.1- and 5.0-fold Cmax, but no ranolazine-induced arrhythmias. At the highest studied concentration, 100 μM, or >50-fold Cmax, 6 of 15 datasets show ranolazine-induced arrhythmia-like events or cessation of spontaneous beating (9 of 15 datasets) in at least 2 experimental wells. Metoprolol is a beta-1 blocker that slows heart rate clinically and is not associated with TdP risk. In hiPSC-CMs, where beta-blockade does not occur due to the absence of sympathetic innervation, metoprolol-induced arrhythmias occurred at 100 μM (55-fold Cmax) in 5 datasets and at 31.6 μM (~18-fold Cmax) in 1 dataset, which is consistent with metoprolol-induced hERG block at higher concentrations (drug concentration for 50% block [IC 50 ] = 145 μM) ( ). Finally, 10 μM mexiletine (~4-fold Cmax) induced arrhythmias in 3 of 15 datasets and mexiletine-induced cessation of spontaneous beating at the highest concentration (100 μM, ~40-fold Cmax) in 12 of 15 datasets. Intermediate TdP Risk Category The 11 drugs in the intermediate TdP risk category were terfenadine, astemizole, chlorpromazine, cisapride, clarithromycin, clozapine, domperidone, droperidol, ondansetron, pimozide, and risperidone. hiPSC-CM response to terfenadine for all 10 sites is shown in , and corresponding figures for the other intermediate risk drugs are shown in . None of the sites observed terfenadine-induced arrhythmia-like events in hiPSC-CMs, even at concentrations as high as 350-fold Cmax, but terfenadine-induced repolarization prolongation occurred in 11 of 15 datasets. Statistically significant repolarization prolongation at ≥1 studied concentrations was observed in a minimum of 10 of 15 datasets for all of the drugs in the intermediate-risk category but clozapine and chlorpromazine. Clozapine- and chlorpromazine-induced prolongation was reported in only 1 and 3 of 15 of the datasets, respectively. Drug-induced arrhythmia-like events at any concentration were observed in at least 10 of 15 datasets for all of the intermediate-risk drugs, except for chlorpromazine, clozapine, terfenadine, and risperidone (0–2 datasets of 15 contained arrhythmia events for these 4 drugs). High TdP Risk Category The 8 drugs in the high TdP risk category were dofetilide, azimilide, bepridil, D,L-sotalol, disopyramide, ibutilide, quinidine, and vandetanib. The hiPSC-CM response to dofetilide for all 10 sites is shown in , and corresponding figures for the other high-risk drugs are shown in . Statistically significant dofetilide-induced repolarization prolongation or dofetilide-induced arrhythmia-like events were consistently (14 of 15 datasets) observed in the studied drug concentration range (0.16- to 5-fold Cmax). All of the drugs in this category except bepridil induced statistically significant repolarization prolongation and/or arrhythmialike events in both hiPSC-CM lines in at least 10 of 15 datasets. While bepridil-induced statistically significant repolarization prolongation was reported in 8 of 15 datasets, only 2 datasets contained bepridil-induced arrhythmia-like events. Druginduced arrhythmia-like events were consistently observed at concentrations close to clinical Cmax for dofetilide, quinidine, and D,L-sotalol and at concentrations well below Cmax for ibutilide. Some of the drugs in this category were so potent and the chosen concentration escalation rate was so steep (i.e., logarithmic increase) that there were no detectable drug effects at one of the studied concentrations, and then at the next concentration, all of the hiPSC-CMs demonstrated arrhythmia-like events, preventing reliable measurement of repolarization duration. Minimal Effect of Site-to-Site Variability on Drug-Induced ddFPDc/APD90c Despite significant efforts to apply consistent experimental protocols across sites, minor deviations were noted (see for the experimental protocol deviations for each site). Site-to-site variability in drug-induced ddFPDc/APD90c averaged across all 28 drugs was compared to other sources of variability by treating site effects as either fixed or random effects ( ) and using the square root of the mean squared error (SR MSE) for each contribution. When site effect was treated as a fixed effect, SR MSE introduced by site (170 ms) was lower than variability induced by the hiPSC-CM line (245 ms). As expected, both values were lower than the contribution provided by drug concentration (482 ms). Similarly, if site effects were treated as random effects, the variability in drug-induced ddFPDc or ddAPD90c averaged over 28 drugs introduced by the site was lower than the total random variability from all of the other sources of random variability (36 versus 67 ms), including well-to-well variability, plate-to-plate variability, human error, and other sources of variability. Modeling of Drug Proarrhythmic Potential Based on Its hiPSC-CM Effects Data on the EP effects of 28 drugs with a known clinical risk of TdP obtained from all of the experimental sites were used to construct a model that would predict TdP risk category of a drug based on its effects on hiPSC-CMs. Seven endpoints from hiPSC-CM experiments were used as potential model predictors ( ). Predictors 1 and 2 describe the ability of a drug to induce arrhythmia-like events in hiPSC-CMs; predictors 3 and 4 reflect the amount of drug-induced repolarization prolongation (ddFPDc or ddAPD90c) at the lowest concentration at which statistically significant change from the baseline (predictor 3) or maximum prolongation at any of the studied concentrations (predictor 4) was observed; predictors 5 and 6 account for concentrations of a drug relative to its clinical Cmax when prolongation of FP/AP duration (predictor 5) or arrhythmia-like event (predictor 6) were first observed; predictor 7 is an estimated amount of prolongation that a drug would induce at the clinical Cmax ( ). Logistic regression models were used in the regression of risk group (high or intermediate risk versus low risk [model 1], and high risk versus low risk and intermediate risk versus low risk [model 2]) on all 7 risk predictors. Cluster analysis showed that the pairs of predictors 3 and 4 (Pearson correlation = 0.52) and predictors 5 and 6 (Pearson correlation = 0.65) are highly correlated, so 1 of each pair may be redundant (data not shown). The final fitted models included 3 significant predictors: predictor 1, predictor 4, and predictor 7. shows significant model predictors for all of the sites for each drug. hiPSC-CM type (iCell 2 or Cor.4U) was not significant (p = 0.089) and did not improve overall fitting for model 1, but it showed a slight improvement in fitting for model 2 by decreasing the Akaike information criterion (AIC) value from 705.3 to 703.2, where AIC is an estimator or the relative quality of statistical models (for a given set of data, smaller value indicates better fit): (1) L o g i t ( P 1 ) = ( P r e d i c t o r 1 ) + ( P r e d i c t o r 4 ) + ( P r e d i c t o r 7 ) (2) L o g i t ( P 2 a ) = ( C e l l T y p e ) + ( P r e d i c t o r 1 ) + ( P r e d i c t o r 4 ) + ( P r e d i c t o r 7 ) (3) L o g i t ( P 2 b ) = ( C e l l T y p e ) + ( P r e d i c t o r 1 ) + ( P r e d i c t o r 4 ) + ( P r e d i c t o r 7 ) where Logit(P) = log(P/(1-P)), P1 is a probability of a drug to be of high or intermediate TdP risk in model 1 and P2a and P2b are probabilities of a drug to be of high versus low or intermediate versus low TdP risk in model 2, respectively. Detailed model parameters are shown in Averaged across sites, risk probabilities predicted by models 1 and 2 are shown in Figures and , respectively. An example of model 1 and model 2 prediction is provided in Model 1 prediction fitted through the data of all of the sites had an area under the receiver operating characteristic (ROC) curve (AUC) value of 0.872 ( ). As expected for a model with 3 outcomes, model 2 AUC had a lower value of 0.826 ( ). Concordance indices (Somers’ deltas [Somers’ D], a measure of ordinal association between possibly dependent random variables, values from −1 to 1, with higher values indicating better quality of model prediction) calculated for models 1 and 2 were 0.74 and 0.65, respectively, showing good discriminating utility for both models. can be used to illustrate the potential role of hiPSC-CMs as a high-specificity preclinical assay under CiPA. By setting a threshold of low TdP risk versus high or intermediate TdP risk at 0.8 in model 1 ( ), the predicted TdP risk of all of the drugs in the low-risk category fall below the threshold, providing a user with reasonable confidence that no unanticipated effects were missed for a drug. The one exception is ranolazine, for which the upper confidence interval (CI) of its estimated risk crosses the 0.8 threshold. As has been shown before ( ), the TdP risk of drugs that have significant late sodium current effects (e.g., ranolazine) may not be adequately modeled by existing hiPSCCMs. Model 1 risk prediction fell below the 0.8 threshold for 1 drug from the high TdP risk category (bepridil) and 4 drugs from the intermediate drug risk category (risperidone, terfenadine, chlorpromazine, and clozapine), highlighting the limitations of the current hiPSC-CMs assays that are not developed to be used as a stand-alone assay, but can be useful when combined with other CiPA preclinical proarrhythmia assessment strategies. Model Validation The purpose of model validation is to estimate the performance of a model when applied for a new, independent dataset. One approach would be to split the data into training and validation sets, so one may use the training dataset to develop the model and then apply the model to the validation dataset to measure the performance. This approach usually requires a large sample size to avoid significant power loss for modeling. In the present study, due to the limited sample size, we have performed model validation and calibration using two alternative methods: cross-validation and bootstrapping. Both approaches allow for nearly unbiased estimates of future model performance, assuming that the present study sample represents a true random sampling of the population of interest. For cross-validation, the original data are randomly divided into k equally sized subsamples, then one subsample is used as the validation dataset, while the remaining subsamples are used as training data. The crossvalidation process is then repeated k times for each subset. The k results then are averaged to produce a single estimation. Here, k = 10 was used for the cross-validation process. Similarly, bootstrapping uses re-sampling with the replacement from the original dataset, so theoretically, an infinite number of samples from one set of data can be generated. For both methods, if the analysis from re-sampling produces results that are consistent with the original analysis, then the model is considered to be reliable and expected to perform for a new, independent dataset. For model 1, bootstrapping with 500 runs of re-sampling negligibly reduced the AUC, from 0.872 to 0.865. Similarly, cross-validation of model 1, omitting from the model 10% of all observations at a time, minimally reduced the AUC, from 0.872 to 0.862. Both results demonstrate the high reliability of model 1. For model 2, bootstrapping with 500 runs of re-sampling negligibly reduced the AUC value, from 0.819 to 0.808, demonstrating a robust model. Model validation results suggest that they would be practical, even when used by a single site applying one of the tested EP platform and cell type combinations. The experimental design used in the statistical models does not provide sufficient power to evaluate differences in the performance of the EP platforms or iPSCCMs lines.
Ten independent sites used a standardized protocol to evaluate the EP effects of 28 drugs with known levels of clinical risk categorized by expert consensus ( ) into 3 clinical TdP risk groups (high, intermediate, and low or no risk) ( ) (see for details on experimental sites, protocols, and platforms and for drug categories and concentrations). Fifteen complete datasets, including data from 5 sites that studied drug effects on both iCell cardiomyocytes 2 and Cor.4U cardiomyocytes (referred to here as iCell 2 and Cor.4U cells), were analyzed. Drug-induced repolarization prolongation (baseline and vehicle-controlled, rate-corrected action potential duration at 90% repolarization, ddAPD90c, or field potential duration, ddFPDc, depending on the EP platform used; see ) was measured along with drug-induced arrhythmias of different types ( ) or drug-induced cessation of hiPSC-CMs’ spontaneous beating (i.e., quiescence).
The 9 drugs in the low TdP risk category were verapamil, diltiazem, loratadine, metoprolol, mexiletine, nifedipine, nitrendipine, ranolazine, and tamoxifen. hiPSC-CM responses to verapamil for all 10 sites is shown in , and corresponding figures for the rest of the low-risk drugs are shown in . No verapamil-induced repolarization prolongation or arrhythmias were observed at any concentrations studied. Diltiazem, loratadine, nifedipine, nitrendipine, and tamoxifen did not induce any arrhythmias or statistically significant repolarization prolongation at concentrations up to 20- to 140fold clinical Cmax. The remaining 3 drugs (ranolazine, metoprolol, and mexiletine) induced repolarization prolongation and arrhythmias at ≥1 of the concentrations in several datasets. Ranolazine is a known hERG blocker ( ) at clinical concentrations and produces QT prolongation. Consistent with this, 13 of 15 datasets show statistically significant ranolazine-induced repolarization prolongation at concentrations between 0.1- and 5.0-fold Cmax, but no ranolazine-induced arrhythmias. At the highest studied concentration, 100 μM, or >50-fold Cmax, 6 of 15 datasets show ranolazine-induced arrhythmia-like events or cessation of spontaneous beating (9 of 15 datasets) in at least 2 experimental wells. Metoprolol is a beta-1 blocker that slows heart rate clinically and is not associated with TdP risk. In hiPSC-CMs, where beta-blockade does not occur due to the absence of sympathetic innervation, metoprolol-induced arrhythmias occurred at 100 μM (55-fold Cmax) in 5 datasets and at 31.6 μM (~18-fold Cmax) in 1 dataset, which is consistent with metoprolol-induced hERG block at higher concentrations (drug concentration for 50% block [IC 50 ] = 145 μM) ( ). Finally, 10 μM mexiletine (~4-fold Cmax) induced arrhythmias in 3 of 15 datasets and mexiletine-induced cessation of spontaneous beating at the highest concentration (100 μM, ~40-fold Cmax) in 12 of 15 datasets.
The 11 drugs in the intermediate TdP risk category were terfenadine, astemizole, chlorpromazine, cisapride, clarithromycin, clozapine, domperidone, droperidol, ondansetron, pimozide, and risperidone. hiPSC-CM response to terfenadine for all 10 sites is shown in , and corresponding figures for the other intermediate risk drugs are shown in . None of the sites observed terfenadine-induced arrhythmia-like events in hiPSC-CMs, even at concentrations as high as 350-fold Cmax, but terfenadine-induced repolarization prolongation occurred in 11 of 15 datasets. Statistically significant repolarization prolongation at ≥1 studied concentrations was observed in a minimum of 10 of 15 datasets for all of the drugs in the intermediate-risk category but clozapine and chlorpromazine. Clozapine- and chlorpromazine-induced prolongation was reported in only 1 and 3 of 15 of the datasets, respectively. Drug-induced arrhythmia-like events at any concentration were observed in at least 10 of 15 datasets for all of the intermediate-risk drugs, except for chlorpromazine, clozapine, terfenadine, and risperidone (0–2 datasets of 15 contained arrhythmia events for these 4 drugs).
The 8 drugs in the high TdP risk category were dofetilide, azimilide, bepridil, D,L-sotalol, disopyramide, ibutilide, quinidine, and vandetanib. The hiPSC-CM response to dofetilide for all 10 sites is shown in , and corresponding figures for the other high-risk drugs are shown in . Statistically significant dofetilide-induced repolarization prolongation or dofetilide-induced arrhythmia-like events were consistently (14 of 15 datasets) observed in the studied drug concentration range (0.16- to 5-fold Cmax). All of the drugs in this category except bepridil induced statistically significant repolarization prolongation and/or arrhythmialike events in both hiPSC-CM lines in at least 10 of 15 datasets. While bepridil-induced statistically significant repolarization prolongation was reported in 8 of 15 datasets, only 2 datasets contained bepridil-induced arrhythmia-like events. Druginduced arrhythmia-like events were consistently observed at concentrations close to clinical Cmax for dofetilide, quinidine, and D,L-sotalol and at concentrations well below Cmax for ibutilide. Some of the drugs in this category were so potent and the chosen concentration escalation rate was so steep (i.e., logarithmic increase) that there were no detectable drug effects at one of the studied concentrations, and then at the next concentration, all of the hiPSC-CMs demonstrated arrhythmia-like events, preventing reliable measurement of repolarization duration.
Despite significant efforts to apply consistent experimental protocols across sites, minor deviations were noted (see for the experimental protocol deviations for each site). Site-to-site variability in drug-induced ddFPDc/APD90c averaged across all 28 drugs was compared to other sources of variability by treating site effects as either fixed or random effects ( ) and using the square root of the mean squared error (SR MSE) for each contribution. When site effect was treated as a fixed effect, SR MSE introduced by site (170 ms) was lower than variability induced by the hiPSC-CM line (245 ms). As expected, both values were lower than the contribution provided by drug concentration (482 ms). Similarly, if site effects were treated as random effects, the variability in drug-induced ddFPDc or ddAPD90c averaged over 28 drugs introduced by the site was lower than the total random variability from all of the other sources of random variability (36 versus 67 ms), including well-to-well variability, plate-to-plate variability, human error, and other sources of variability.
Data on the EP effects of 28 drugs with a known clinical risk of TdP obtained from all of the experimental sites were used to construct a model that would predict TdP risk category of a drug based on its effects on hiPSC-CMs. Seven endpoints from hiPSC-CM experiments were used as potential model predictors ( ). Predictors 1 and 2 describe the ability of a drug to induce arrhythmia-like events in hiPSC-CMs; predictors 3 and 4 reflect the amount of drug-induced repolarization prolongation (ddFPDc or ddAPD90c) at the lowest concentration at which statistically significant change from the baseline (predictor 3) or maximum prolongation at any of the studied concentrations (predictor 4) was observed; predictors 5 and 6 account for concentrations of a drug relative to its clinical Cmax when prolongation of FP/AP duration (predictor 5) or arrhythmia-like event (predictor 6) were first observed; predictor 7 is an estimated amount of prolongation that a drug would induce at the clinical Cmax ( ). Logistic regression models were used in the regression of risk group (high or intermediate risk versus low risk [model 1], and high risk versus low risk and intermediate risk versus low risk [model 2]) on all 7 risk predictors. Cluster analysis showed that the pairs of predictors 3 and 4 (Pearson correlation = 0.52) and predictors 5 and 6 (Pearson correlation = 0.65) are highly correlated, so 1 of each pair may be redundant (data not shown). The final fitted models included 3 significant predictors: predictor 1, predictor 4, and predictor 7. shows significant model predictors for all of the sites for each drug. hiPSC-CM type (iCell 2 or Cor.4U) was not significant (p = 0.089) and did not improve overall fitting for model 1, but it showed a slight improvement in fitting for model 2 by decreasing the Akaike information criterion (AIC) value from 705.3 to 703.2, where AIC is an estimator or the relative quality of statistical models (for a given set of data, smaller value indicates better fit): (1) L o g i t ( P 1 ) = ( P r e d i c t o r 1 ) + ( P r e d i c t o r 4 ) + ( P r e d i c t o r 7 ) (2) L o g i t ( P 2 a ) = ( C e l l T y p e ) + ( P r e d i c t o r 1 ) + ( P r e d i c t o r 4 ) + ( P r e d i c t o r 7 ) (3) L o g i t ( P 2 b ) = ( C e l l T y p e ) + ( P r e d i c t o r 1 ) + ( P r e d i c t o r 4 ) + ( P r e d i c t o r 7 ) where Logit(P) = log(P/(1-P)), P1 is a probability of a drug to be of high or intermediate TdP risk in model 1 and P2a and P2b are probabilities of a drug to be of high versus low or intermediate versus low TdP risk in model 2, respectively. Detailed model parameters are shown in Averaged across sites, risk probabilities predicted by models 1 and 2 are shown in Figures and , respectively. An example of model 1 and model 2 prediction is provided in Model 1 prediction fitted through the data of all of the sites had an area under the receiver operating characteristic (ROC) curve (AUC) value of 0.872 ( ). As expected for a model with 3 outcomes, model 2 AUC had a lower value of 0.826 ( ). Concordance indices (Somers’ deltas [Somers’ D], a measure of ordinal association between possibly dependent random variables, values from −1 to 1, with higher values indicating better quality of model prediction) calculated for models 1 and 2 were 0.74 and 0.65, respectively, showing good discriminating utility for both models. can be used to illustrate the potential role of hiPSC-CMs as a high-specificity preclinical assay under CiPA. By setting a threshold of low TdP risk versus high or intermediate TdP risk at 0.8 in model 1 ( ), the predicted TdP risk of all of the drugs in the low-risk category fall below the threshold, providing a user with reasonable confidence that no unanticipated effects were missed for a drug. The one exception is ranolazine, for which the upper confidence interval (CI) of its estimated risk crosses the 0.8 threshold. As has been shown before ( ), the TdP risk of drugs that have significant late sodium current effects (e.g., ranolazine) may not be adequately modeled by existing hiPSCCMs. Model 1 risk prediction fell below the 0.8 threshold for 1 drug from the high TdP risk category (bepridil) and 4 drugs from the intermediate drug risk category (risperidone, terfenadine, chlorpromazine, and clozapine), highlighting the limitations of the current hiPSC-CMs assays that are not developed to be used as a stand-alone assay, but can be useful when combined with other CiPA preclinical proarrhythmia assessment strategies.
The purpose of model validation is to estimate the performance of a model when applied for a new, independent dataset. One approach would be to split the data into training and validation sets, so one may use the training dataset to develop the model and then apply the model to the validation dataset to measure the performance. This approach usually requires a large sample size to avoid significant power loss for modeling. In the present study, due to the limited sample size, we have performed model validation and calibration using two alternative methods: cross-validation and bootstrapping. Both approaches allow for nearly unbiased estimates of future model performance, assuming that the present study sample represents a true random sampling of the population of interest. For cross-validation, the original data are randomly divided into k equally sized subsamples, then one subsample is used as the validation dataset, while the remaining subsamples are used as training data. The crossvalidation process is then repeated k times for each subset. The k results then are averaged to produce a single estimation. Here, k = 10 was used for the cross-validation process. Similarly, bootstrapping uses re-sampling with the replacement from the original dataset, so theoretically, an infinite number of samples from one set of data can be generated. For both methods, if the analysis from re-sampling produces results that are consistent with the original analysis, then the model is considered to be reliable and expected to perform for a new, independent dataset. For model 1, bootstrapping with 500 runs of re-sampling negligibly reduced the AUC, from 0.872 to 0.865. Similarly, cross-validation of model 1, omitting from the model 10% of all observations at a time, minimally reduced the AUC, from 0.872 to 0.862. Both results demonstrate the high reliability of model 1. For model 2, bootstrapping with 500 runs of re-sampling negligibly reduced the AUC value, from 0.819 to 0.808, demonstrating a robust model. Model validation results suggest that they would be practical, even when used by a single site applying one of the tested EP platform and cell type combinations. The experimental design used in the statistical models does not provide sufficient power to evaluate differences in the performance of the EP platforms or iPSCCMs lines.
This study summarizes results of the first large multisite study assessing the potential of 2 commercially available hiPSC-CMs using in vitro -based MEA and VSO approaches to detect drug-induced repolarization abnormalities and predict the proarrhythmic potential of 28 drugs characterized for TdP risk under the CiPA initiative. Concentration-dependent effects from 7 EP responses were used to build 2 regression models that predict low, intermediate, or high clinical TdP risk categories. The most useful predictors were identified in the study: (1) the ability of a drug to induce “mild” (type A) or “severe” (all other) arrhythmia-like events at any concentration (predictor 1); (2) the extent of drug-induced repolarization prolongation at any concentration (predictor 4); and (3) the extent of drug-induced prolongation at the clinical Cmax (predictor 7). We found it interesting that the ability of a drug to inhibit hiPSC-CMs’ spontaneous beating or any of the other predictors did not further improve model prediction. Despite the variations in the experimental protocols, including intended range of the tested EP platforms (5 different platforms; both MEA and VSO were used) and some unintended variations in cell batch, recording medium composition, and other parameters ( ), the results for all 28 drugs were fairly consistent across 10 sites. shows the results of Pearson correlation analysis for drug-induced ddFPDc/APD90c change across 10 sites. Lower coefficient values for individual sites can be achieved by greater or lesser responses to drugs. Data from most of the sites were highly correlated (average Pearson coefficients of 78%–88%), while sites 2 and 4 had lower correlation coefficients (69% and 70%, respectively), potentially for different reasons because the Pearson coefficient between these 2 sites is low (37%). Differences observed for site 2 may be attributed to the differences in experimental protocols, because site 2 was the only test site that used the VSO platform instead of the MEA platforms and serum-free experimental medium instead of serum-containing medium. APD90 (VSO) and FPD (MEA) are equivalent measures, and site 2 showed appropriate APD changes and arrhythmia-like events in response to drugs, including dofetilide. However, the absence of serum in the assay media used by site 2 and the known potential of serum components (e.g., albumen) to modulate the bioavailability of some drugs in the serum-containing media of all of the other sites ( ; ) has the potential to explain the slightly lower average Pearson coefficient. In contrast, site 4 used experimental protocols that were largely consistent with other MEA sites, including the use of serum-containing media ( ), but overall correlation for that site was lower. Of note, shows that unlike all of the other sites, site 4 did not report large effects induced by dofetilide (no significant ddFPDc prolongation or drug-induced arrhythmia-like events). It will be critical to include positive drug controls (with known ion channel effects) on each plate to demonstrate suitable assay sensitivity based on the predominant mechanisms for affecting repolarization. It is important to recognize that the model demonstrates the ability of hiPSC-CMs across sites to detect delayed repolarization and predict TdP risk for 15 datasets for 28 drugs, but that an individual site may not be expected to detect the proarrhythmic risk for each drug. Furthermore, despite different reprogramming and differentiation protocols used to manufacture the two hiPSC-CM lines used in the study, they were similar in predicting intermediate versus low or low versus high- or intermediate-risk drugs, which is the current unmet need. However, it is important to note that this study was limited to two hiPSC-CM lines and that other lines will require their own validation. Furthermore, the drugs can be potentially tested in gender-specific or even subject-specific hiPSC-CMs when feasible for the intended drug target population, but for this study we focused on a general assessment of a molecule, so the choice was made to use well-characterized, commercially developed hiPSC-CMs lines. It is important to examine the outlier drugs that induced effects in hiPSC-CMs that are noticeably different from the other drugs in the same TdP risk category. Unlike other high-risk drugs and consistent with previous studies ( ), bepridil did not induce arrhythmias in hiPSC-CMs, even at 30-fold Cmax (except for 1 of 15 datasets). Bepridil is a potent hERG blocker that also blocks L-type calcium and peak and late sodium currents at higher concentrations ( ). High expression levels of calcium ion channels in hiPSC-CMs as compared to primary ventricular tissue ( ) may have contributed to more attenuated cellular proarrhythmic effects of the drug as compared to other drugs in the high TdP risk category. It is also possible that the known propensity of bepridil to induce cardiac arrhythmia in the clinic is at least partly related to the ability of bepridil to affect hERG surface expression ( ). hERG trafficking effects of drugs were not assessed in this study because of the short duration exposures of hERG. Another outlier drug was low TdP risk ranolazine, which induced significant repolarization prolongation and arrhythmias in hiPSC-CMs, uncharacteristic for this risk category. While ranolazine blocks the hERG potassium channel and prolongs QTc, it is not associated with TdP risk because hERG block is balanced by significant late sodium current block ( ). Lower expression levels of sodium channels and decreased late sodium current in hiPSC-CMs compared to primary human ventricular tissue ( ; ) may contribute to the apparent proarrhythmic effects of ranolazine in hiPSC-CMs. Similarly, lower densities of late sodium current in hiPSC-CMs may explain mexiletine-induced arrhythmia-like events. Finally, another low-risk drug that induced arrhythmia-like events in hiPSC-CMs was metoprolol, a beta-blocker, the effects of which may not be appropriately modeled in uninnervated hiPSC-CMs monocultures. The differences in cellular electrophysiology between native tissue and iPSC-cardiomyocytes has been well documented ( ; ; ), with iPSC-cardiomyocytes possessing, in general, spontaneous activity, depolarized membrane potential (Vm), slower AP upstroke, and longer APD and FPD. It is unclear how these differences would translate into systematic or class-specific misclassifications, but it does speak to the need for specific calibration controls demonstrating assay sensitivity for sodium, calcium, and potassium currents blockade. Relative differences in ion channel and current levels in iPSC-CMs as compared to adult ventricular myocytes are likely the most important factors to improve the accurate prediction of TdP risk, especially with multichannel blocking drugs. With the development of new biotechnologies aimed at the development of more adult-like hiPSC-CMs ( ; ), the predictivity of hiPSC-CMs assays is expected to further improve. Furthermore, other predictors of proarrhythmia risk may be added to the model based on their ability to differentiate drugs from the three categories examined here. For example, triangulation of the cardiac AP has previously been correlated with the ability to cause TdP experimentally ( ). Measurements of AP triangulation (corrected for AP duration) based on data from the VSO platform were correlated with TdP risk category ( ) and may prove to be a useful additional descriptor in the future. As demonstrated here, hiPSC-CMs are an important new human in vitro model for the assessment of TdP risk, and their role in CiPA should be considered along with the recent advances in in silico modeling to predict proarrhythmic cardiotoxicity ( ; ). Computer models of TdP risk based on experimentally measured multichannel drug effects show high predictivity and would be an important primary step in proarrhythmic risk assessment, at least until iPSC-CMs become even better representations of adult human cardiac myocytes. The advantages of using readily available human-derived cardiomyocyte preparations need to be considered along with comparisons of the accuracy of cardiomyocytes (versus ex vivo or in vivo animal models) in predicting proarrhythmic risk when defining the optimal role of hiPSC-CMs in drug discovery. Several experimental limitations of the study are worth noting. First, the free drug concentrations in hiPSC-CM experiments were not measured. As shown in , several drugs (e.g., disopyramide, azimilide, clarithromycin) were reported by multiple (but not all) sites as being poorly soluble in DMSO at the required concentrations. Thus, additional measures were taken, such as sonicating, warming at 37°C, or increasing the DMSO percentage. It has been shown ( ) that serum content in the cell culture medium used for drug dilution could affect drug solubility and availability. Although all of the sites followed the same nominal set of drug preparation instructions, measurements of drug concentrations in the experimental wells were not performed. Second, this study does not allow for the measurement of the effect of drug metabolites, which can in some cases be more toxic than the parent drug (e.g., the metabolite of astemizole, desmethylastemizole ). Third, measuring the effects of hERG blockers on FPD can be challenging for some drugs because of the decrease in repolarization T-wave amplitude, in addition to the drug-induced FPD prolongation. Fourth, the effects of only short exposures (30 min) of drugs were assessed in this study, while some nonacute proarrhythmic effects (not the emphasis of CiPA) may require longer exposures to affect channel expression. Finally, this study was not statistically powered to investigate the effect of the electrophysiological device on the hiPSC-CM assay’s predictivity of proarrhythmic drug potential. contains information on the fraction of drugs correctly characterized into a TdP risk category from the data stratified by the EP platform. However, these data should be interpreted with caution because the study design does not allow for distinguishing the effects of the specific device from other effects introduced by the cell type or by the experimental site itself. Further studies are needed to investigate whether device choice would be an important consideration in improving preclinical TdP risk assessment by hiPSC-CM-based assays. In summary, this study used statistical modeling to identify the most predictive endpoints of hiPSC-CMs assays in TdP risk assessment. Using only 3 endpoints, model 1 separated drugs into low-risk versus combined intermediate- and high-risk categories with an AUC value of 0.87 (87% predictivity) at the sample size we used, regardless of the type of hiPSC-CM used; model 2 separated drugs into 3 separate risk categories and showed a slightly lower AUC value of 0.82. Different thresholds with each model, which have associated sensitivity and specificity values, can be selected based on when the assay is being used in drug development. Because the goal of CiPA is to increase specificity and hiPSC-CMs will be used to check for missed or unanticipated effects, a threshold with a high specificity will be required. For example, a threshold of 0.8 in model 1 is associated with a specificity of 0.89 and a sensitivity of 0.63. If a drug is predicted to have a low risk in the in silico TdP risk metric, but is positive at this threshold, then it could be important to understand the reason for this discrepancy. If the drug has low proarrhythmic risk due to balanced multi-ion channel block, such as ranolazine with both hERG and late sodium current block, then this discrepancy would not be surprising. Such a result should not hinder progressing with clinical development, in which drug-induced QT prolongation and signs of balanced ion channel block (no J-Tpeak prolongation ) would still be assessed in first-in-human studies. Thus, it will be important to perform an integrated risk assessment, taking into account the different components of CiPA when implementing CiPA to improve specificity and provide more accurate predictions of clinical TdP risk, rather than solely focusing on hERG block and QT prolongation.
Study Sites and Platforms Ten independent laboratories participated in the study, using any 1 of the 4 MEA platforms: Maestro (Axion BioSystems, abbreviated to “AXN” in this paper), CardioECR (ACEA Biosciences, “ECR”), Multiwell (Multichannel Systems, “MCS”), and AlphaMED64 (Alpha MED Scientific, “AMD”), or the VSO platform: CellOPTIQ (Clyde Biosciences, “CLY”). hiPSC-CMs Two commercially available hiPSC-CM cell lines were used: iCell 2 (Cellular DynamicsInternational)andCor.4U(NCardia).iCell 2 arenormallycryopreserved at approximately day 30 of the differentiation (similar to Ma et al. [2011]); the production procedures for Cor.4U were not disclosed by the manufacturer. Sites were instructed to follow manufacturers’ recommendations for hiPSC-CM plating and maintenance, including cell culture plate coating, cell plating densities, and assay time window. Spontaneously beating, 100% confluent iPSC-CMs monolayers were used for drug testing. contains information on the specific cell lots, cell-handling details, and variations by experimental site. Drug Dilution and Addition BlindeddrugpowderwassenttoallofthesitesbytheChemotherapeuticAgents Repository of the National Cancer Institute, and stored at 20°C until the day of testing. Four concentrations of each drug were studied ( ). Four DMSO stocks for each drug concentration were prepared and either used on the same day or aliquoted and frozen. Concentrated (10×) testing solutions (50× for sequential dosing) for each concentration were prepared freshly on the day of testing by diluting DMSO stocks into experimental medium (serum-containing maintenancemediumforMEAexperimentsandserum-freemediumforVSOexperiments; see the section for more details). Ten-fold dilution was achieved when drugs were added to the experimental well to attain the targeted concentration. For sequential dosing, DMSO concentrations were adjusted sequentially up to 0.1% at the highest concentration to achieve the targeted concentration of each drug. If insoluble compound was observed in DMSO or 10× testing stock solutions, then warming to 37°C and sonicating for 20 min was recommended. contains information on when these measures were taken to improve drug solubility. MEA and VSO Recordings of Drug-Induced Effects in hiPSC-CMs All MEA and VSO recordings were performed at 37°C. Single concentrations of each drug were tested in each experimental well by all of the sites, except site 10, where sequential additions were used. A 100% media change was performed in hiPSC-CMs 2–24 hr before baseline recordings. Media compositions used for MEA and VSO recordings are shown in Concentration effects of each drug were recorded in ≥5 replicates for 97% of the collected data. Experimental points collected with <5 replicates are marked with a star in Figures and and in , , and . Vehicle (0.1% DMSO) control wells were included on each plate. After baseline recording and drug addition, the plates were left to re-equilibrate for at least 30 min before recordings. Data Analysis Data Exclusion Criteria The results were excluded from the analysis if baseline parameters for a specific well were outside the following pre-specified quality standards: (1) hiPSC-CMs baseline spontaneous beating rate had to be within the 20–90 beats per min range (i.e., 0.3–1.5 Hz), (2) the baseline beating rate had to be within 6 SDs calculated for the baseline beating rate on all of the wells on the given plate, (3) the coefficient of variation for the baseline beat period had to be <5%, and (4) the depolarization spike amplitude had to be >0.3 mV (MEA recordings only). Based on these criteria, no more than 3% of wells were excluded from analysis. Drug-Induced Changes in Repolarization and Arrhythmia-like Events Fridericia’s formula ( ) was used to correct hiPSC-CM action potential duration (APD) and field potential duration (FPD) dependence on beating rate (APDc, FPDc). While not thoroughly validated for hiPSC-CMs, this formula is widely used in these assays ( ; ; ; ). Baseline- and vehicle-controlled ddFPDc and ddAPD90c at 90% repolarization were calculated by averaging all DMSO-treated wells onthe plate for vehicle control.Drug-induced arrhythmia-like eventswere counted and classified in 1 of 4categories (A–D), as illustrated in . The relation between hiPSC-CMs action potential and field potential, including correspondence between different arrhythmia-like events recorded by MEA and VSO, have been described previously ( ). Combination of events (e.g., AB, AC, ABC, ABCD) was also observed and recorded. Several drugs inhibited spontaneous hiPSC-CMs contractions, leading to a quiescent state (Q). MEA and VSO instrument operators were blinded to the drug treatment during data collection and analysis. Statistical Methods Descriptive Analysis The primary measurement was the averaged baseline- and vehicle-controlled, Fridericia rate-corrected ddFPDc/ddAPD90c at each concentration. Drug concentrations were treated as ordinal variables, in which the order mattered but not the difference between concentration values. ddFPDc/APD90c was not calculated and was designated as missing for the drug concentrations in which ≥50% of the wells were arrhythmic after dosing. The concentration effects of each drug were recorded in ≥5 replicates for 97% of the collected data. Arrhythmia was a binary outcome and was designated as “Yes” if it occurred in at least one well at any concentration. Modeling and Model Validation Seven endpoints characterizing drug responses on hiPSC-CMs were used to build a linear regression model predicting the drug TdP risk category ( ). For the model development, drug-induced repolarization prolongation in hiPSC-CMs recorded with the MEA platform (ddFPDc) and the VSO platform (ddAPD90c) was considered equivalent. Cell type was treated as a fixed effect and experimental site was treated as a random effect in these models. The predictor selection procedure was based on model-fitting diagnostics of the AIC, the Bayesian information criterion (BIC), the AUC, and cluster analysis among continuous predictors. Model validation was achieved through cross-validation and bootstrapping. Statistical analysis was done using SAS (SAS Institute, Cary, NC) and R (RStudio, Boston, MA) software.
Ten independent laboratories participated in the study, using any 1 of the 4 MEA platforms: Maestro (Axion BioSystems, abbreviated to “AXN” in this paper), CardioECR (ACEA Biosciences, “ECR”), Multiwell (Multichannel Systems, “MCS”), and AlphaMED64 (Alpha MED Scientific, “AMD”), or the VSO platform: CellOPTIQ (Clyde Biosciences, “CLY”).
Two commercially available hiPSC-CM cell lines were used: iCell 2 (Cellular DynamicsInternational)andCor.4U(NCardia).iCell 2 arenormallycryopreserved at approximately day 30 of the differentiation (similar to Ma et al. [2011]); the production procedures for Cor.4U were not disclosed by the manufacturer. Sites were instructed to follow manufacturers’ recommendations for hiPSC-CM plating and maintenance, including cell culture plate coating, cell plating densities, and assay time window. Spontaneously beating, 100% confluent iPSC-CMs monolayers were used for drug testing. contains information on the specific cell lots, cell-handling details, and variations by experimental site.
BlindeddrugpowderwassenttoallofthesitesbytheChemotherapeuticAgents Repository of the National Cancer Institute, and stored at 20°C until the day of testing. Four concentrations of each drug were studied ( ). Four DMSO stocks for each drug concentration were prepared and either used on the same day or aliquoted and frozen. Concentrated (10×) testing solutions (50× for sequential dosing) for each concentration were prepared freshly on the day of testing by diluting DMSO stocks into experimental medium (serum-containing maintenancemediumforMEAexperimentsandserum-freemediumforVSOexperiments; see the section for more details). Ten-fold dilution was achieved when drugs were added to the experimental well to attain the targeted concentration. For sequential dosing, DMSO concentrations were adjusted sequentially up to 0.1% at the highest concentration to achieve the targeted concentration of each drug. If insoluble compound was observed in DMSO or 10× testing stock solutions, then warming to 37°C and sonicating for 20 min was recommended. contains information on when these measures were taken to improve drug solubility.
All MEA and VSO recordings were performed at 37°C. Single concentrations of each drug were tested in each experimental well by all of the sites, except site 10, where sequential additions were used. A 100% media change was performed in hiPSC-CMs 2–24 hr before baseline recordings. Media compositions used for MEA and VSO recordings are shown in Concentration effects of each drug were recorded in ≥5 replicates for 97% of the collected data. Experimental points collected with <5 replicates are marked with a star in Figures and and in , , and . Vehicle (0.1% DMSO) control wells were included on each plate. After baseline recording and drug addition, the plates were left to re-equilibrate for at least 30 min before recordings.
Data Exclusion Criteria The results were excluded from the analysis if baseline parameters for a specific well were outside the following pre-specified quality standards: (1) hiPSC-CMs baseline spontaneous beating rate had to be within the 20–90 beats per min range (i.e., 0.3–1.5 Hz), (2) the baseline beating rate had to be within 6 SDs calculated for the baseline beating rate on all of the wells on the given plate, (3) the coefficient of variation for the baseline beat period had to be <5%, and (4) the depolarization spike amplitude had to be >0.3 mV (MEA recordings only). Based on these criteria, no more than 3% of wells were excluded from analysis. Drug-Induced Changes in Repolarization and Arrhythmia-like Events Fridericia’s formula ( ) was used to correct hiPSC-CM action potential duration (APD) and field potential duration (FPD) dependence on beating rate (APDc, FPDc). While not thoroughly validated for hiPSC-CMs, this formula is widely used in these assays ( ; ; ; ). Baseline- and vehicle-controlled ddFPDc and ddAPD90c at 90% repolarization were calculated by averaging all DMSO-treated wells onthe plate for vehicle control.Drug-induced arrhythmia-like eventswere counted and classified in 1 of 4categories (A–D), as illustrated in . The relation between hiPSC-CMs action potential and field potential, including correspondence between different arrhythmia-like events recorded by MEA and VSO, have been described previously ( ). Combination of events (e.g., AB, AC, ABC, ABCD) was also observed and recorded. Several drugs inhibited spontaneous hiPSC-CMs contractions, leading to a quiescent state (Q). MEA and VSO instrument operators were blinded to the drug treatment during data collection and analysis.
The results were excluded from the analysis if baseline parameters for a specific well were outside the following pre-specified quality standards: (1) hiPSC-CMs baseline spontaneous beating rate had to be within the 20–90 beats per min range (i.e., 0.3–1.5 Hz), (2) the baseline beating rate had to be within 6 SDs calculated for the baseline beating rate on all of the wells on the given plate, (3) the coefficient of variation for the baseline beat period had to be <5%, and (4) the depolarization spike amplitude had to be >0.3 mV (MEA recordings only). Based on these criteria, no more than 3% of wells were excluded from analysis.
Fridericia’s formula ( ) was used to correct hiPSC-CM action potential duration (APD) and field potential duration (FPD) dependence on beating rate (APDc, FPDc). While not thoroughly validated for hiPSC-CMs, this formula is widely used in these assays ( ; ; ; ). Baseline- and vehicle-controlled ddFPDc and ddAPD90c at 90% repolarization were calculated by averaging all DMSO-treated wells onthe plate for vehicle control.Drug-induced arrhythmia-like eventswere counted and classified in 1 of 4categories (A–D), as illustrated in . The relation between hiPSC-CMs action potential and field potential, including correspondence between different arrhythmia-like events recorded by MEA and VSO, have been described previously ( ). Combination of events (e.g., AB, AC, ABC, ABCD) was also observed and recorded. Several drugs inhibited spontaneous hiPSC-CMs contractions, leading to a quiescent state (Q). MEA and VSO instrument operators were blinded to the drug treatment during data collection and analysis.
Descriptive Analysis The primary measurement was the averaged baseline- and vehicle-controlled, Fridericia rate-corrected ddFPDc/ddAPD90c at each concentration. Drug concentrations were treated as ordinal variables, in which the order mattered but not the difference between concentration values. ddFPDc/APD90c was not calculated and was designated as missing for the drug concentrations in which ≥50% of the wells were arrhythmic after dosing. The concentration effects of each drug were recorded in ≥5 replicates for 97% of the collected data. Arrhythmia was a binary outcome and was designated as “Yes” if it occurred in at least one well at any concentration. Modeling and Model Validation Seven endpoints characterizing drug responses on hiPSC-CMs were used to build a linear regression model predicting the drug TdP risk category ( ). For the model development, drug-induced repolarization prolongation in hiPSC-CMs recorded with the MEA platform (ddFPDc) and the VSO platform (ddAPD90c) was considered equivalent. Cell type was treated as a fixed effect and experimental site was treated as a random effect in these models. The predictor selection procedure was based on model-fitting diagnostics of the AIC, the Bayesian information criterion (BIC), the AUC, and cluster analysis among continuous predictors. Model validation was achieved through cross-validation and bootstrapping. Statistical analysis was done using SAS (SAS Institute, Cary, NC) and R (RStudio, Boston, MA) software.
The primary measurement was the averaged baseline- and vehicle-controlled, Fridericia rate-corrected ddFPDc/ddAPD90c at each concentration. Drug concentrations were treated as ordinal variables, in which the order mattered but not the difference between concentration values. ddFPDc/APD90c was not calculated and was designated as missing for the drug concentrations in which ≥50% of the wells were arrhythmic after dosing. The concentration effects of each drug were recorded in ≥5 replicates for 97% of the collected data. Arrhythmia was a binary outcome and was designated as “Yes” if it occurred in at least one well at any concentration.
Seven endpoints characterizing drug responses on hiPSC-CMs were used to build a linear regression model predicting the drug TdP risk category ( ). For the model development, drug-induced repolarization prolongation in hiPSC-CMs recorded with the MEA platform (ddFPDc) and the VSO platform (ddAPD90c) was considered equivalent. Cell type was treated as a fixed effect and experimental site was treated as a random effect in these models. The predictor selection procedure was based on model-fitting diagnostics of the AIC, the Bayesian information criterion (BIC), the AUC, and cluster analysis among continuous predictors. Model validation was achieved through cross-validation and bootstrapping. Statistical analysis was done using SAS (SAS Institute, Cary, NC) and R (RStudio, Boston, MA) software.
Data S1 Data S2 Data S3 Supplemental Figures and Tables
|
A novel PDX modeling strategy and its application in metabolomics study for malignant pleural mesothelioma | 15d34d13-194a-4066-8268-15f605915e71 | 8600931 | Anatomy[mh] | Malignant mesothelioma (MM) is a rare cancer originating from the mesothelial linings of the pleural or peritoneal cavities. Malignant pleural mesothelioma (MPM) is the predominant form of MM. The incidence of MPM is very low in China (1.5/1000,000) . Unfortunately, the prognosis of MM is extremely poor (survival time of 12-22 months) due to frequent late diagnosis that results from difficulties in early detection and lack in efficacy of current treatments . Thus, MPM has been a dismal disease troubling both patients and clinical doctors. MPM is normally induced by asbestos . Although this mineral fiber has been widely banned globally, it is still consumed in several countries, such as India, Russia, and China . Thus, MPM will continue to burden society due to previous and continuous use of asbestos as well as the long latency (20-40 years) of MPM . However, most doctors and pathologists in China are unfamiliar with MM due to its rarity, leading to a delay in treatment and even misdiagnosis, denoting an urgent need in identification of diagnostic biomarkers . Lack of specimens is the forefront obstacle in MPM research in China, even to the whole globe. Two studies suggested that the patient-derived xenograft (PDX) model, which implants tumor tissue from a patient into recipient mice, is suitable for testing both anti-cancer therapies and the biological functions of genes or proteins in MM . Compared to cell line-derived xenograft, PDX model better simulates clinical samples, because it greatly reserves heterogeneity of the primary tumor . Accordingly, PDX modeling is efficient and advantageous for biomarker detection and drug screening. Construction of PDX models relies on specimens from patients, which are commonly collected from surgery. However, in China, surgery in MPM is infrequent thus can rarely provide samples for modeling. Ultrasound-guided (US-guided) biopsy is often used for examination of unclear malignancies, and it is still able to provide samples under situations when MPM patients are not suitable for surgery (e.g., pleural adhesions) . Therefore, this study established a novel PDX modeling strategy based on an US-guided pleural biopsy, attributable to cooperation across three specializations, including US imaging, pathology, and lab research. Success in the establishment of two PDX models indicates the feasibility of our strategy. To date, many studies focus on genomes in MM [ – ], with relatively fewer investigating metabolomes. Nevertheless, metabolome has been viewed to utmost demonstrate phenotypes, indicating its importance and prominence for disease exploration through the discovery of metabolic biomarkers and therapeutic targets . Bononi et al. reported that MM initiation is linked to metabolic reprogramming , and Zhang et al. also revealed that metabolic enzymes, such as SLC7A11, might serve as treatment targets for MM . For the above reasons, we aimed to detect diagnostic metabolites and discover biological targets for MPM, through serum-based metabolomic profiling for the US-guided pleural biopsy-derived PDX model.
Ethical statement This study was performed in accordance with the Declaration of Helsinki (revised in 2013), and the protocol was approved by The Ethical Committee of Zhejiang Cancer Hospital (approval number: IRB-2018-9). Informed consent was obtained from each patient. Patients screening for PDX construction Patients who displayed pleural thickening in computed tomography (CT) imaging were reviewed by an experienced sonographer (Junping Liu) between March 1 st 2018 and December 30 th 2019 in Zhejiang Cancer Hospital. After exclusion, US-guided pleural biopsy was performed for the remaining patients. Those from patients who did not meet the following exclusion criteria were highly suspected (by the sonographer based on his past clinical experiences) to be MM: 1) pleural metastasis from lung cancer; 2) pleural metastasis from other cancers; 3) tuberculosis lesions; 4) unclear diagnosis; and 5) other reasons (according to imaging features and clinical history). Under the premise of not interfering with clinical diagnosis, these biopsies were collected and implanted into immunodeficient mice for constructing PDX models. At the same time, all biopsies were sent to a senior pathologist (Zhenying Guo) for precise diagnosis. The animal experiment was permitted by the Institutional Animal Care and Ethics Committee of Zhejiang Cancer Hospital (No. 2018–03-054), in compliance with national or institutional guidelines for the care and use of animals. All the animals were euthanized using isoflurane at the end of experiment. US-guided pleural biopsy Patients were fasted for at least 8 h before biopsy. The biopsy was performed according to Zhang et al. Briefly, Esaote MylabTMTwice ultrasound apparatus (Italy) with probes configuration convex transducer CA541 and linear LA524 was utilized. First, a low-frequency probe (4.0 MHz) was used to detect the pleural effusion, pleura and blood flow, and the thickest point of the pleura was selected for biopsy. Second, pleural biopsy was guided through alternatively using low-frequency (4.0 MHz) and high-frequency (9.0 MHz) probes conducted by two experienced sonographers. One sonographer provided the guidance, and the other used an 18 or 16 G automated cutting needle (MC1816, Bard Max. Core, Bard Inc., USA) to perform the biopsy under local anesthesia with 2% lidocaine. Specimens were immediately fixed in 10% formalin and sent to the Department of Pathology for examination. Without interfering clinical diagnosis, part of fresh specimens was sent for constructing PDX models. Immunohistochemistry (IHC) staining IHC staining was performed according to Wu et al. In brief, slides of formalin-fixed and paraffin-embedded tumors were deparaffinized and incubated in 3% hydrogen peroxidase. After inactivation of endogenous peroxidases, slices were treated with antigen retrieval by boiling at 100 °C for 90 s in citric acid repair solution (pH = 6). After blocking, slices were incubated with antibodies overnight at 4 °C. Slices were then incubated with HRP-labeled secondary antibody for 30 min. According to gender, histological features, and location of the tumor, a panel of biomarkers for IHC analysis was selected and the diagnosis was made by a senior pathologist (Zhenying Guo). Details of the antibodies used were listed in the . PDX model construction The procedure of PDX construction was similar to established protocol from Wu et al. In brief, 5-week-old female BALB/c immunodeficient mice (certificate number: 2017005004641) were purchased from Shanghai Slac Laboratory Animal Company (Shanghai, China). The mice adapted to the environment for 1 week. Every five of all mice were kept in one cage with free access to food and water, to a 12 h/12 h light/dark cycle, at temperature between 22 °C and 26 °C, with 55% relative humidity. Fresh tumor tissue was kept in a sterilized PBS buffer on ice, and was cut into blocks of 2 × 2 × 2 mm and then was engrafted subcutaneously into the flanks of BALB/c mice (P0) with a trocar needle. A PDX model which can be consecutively passed over twice is defined as a success (P2). Then P2 tumors were harvested and transplanted to 10 mice for the PDX model, and 7 mice without treatment were taken as a control group. When the average tumor size exceeded 200 mm 3 , blood was collected retro-orbitally under isoflurane anesthesia. Sera were separated with centrifugation for 10 min at 2400 g, 4 °C and were kept at -80 °C until analysis. Gas chromatography-mass spectrometry (GC-MS)-based metabolomics The GC-MS-based metabolomics was performed according to the previously published method from Zhao et al. The procedures including sample preparation and GC-MS analysis are described in the . Metabolomic data analysis After data formatting (to .abf) with Reifycs Abf Converter ( https://www.reifycs.com/AbfConverter/ ), MS-DIAL software was used for data processing, including peak picking, peak alignment, missing value interpolation, and so on. Metabolite annotation was performed through the untargeted database of GC-MS from Lumingbio. Finally, a data set with sample information and peak information was obtained. Principal component analysis (PCA) and partial least-squares-discriminant analysis (PLS-DA) were performed to visualize the metabolic shift among groups using R package ropls (version 1.18.8). Variable importance in the projection (VIP) was obtained from PLS-DA, which ranks the contribution of metabolite features in the PLS-DA model. Finally, features with VIP > 1.0 and P -value < 0.05 from two-tailed Student’s t-test were defined as differential metabolites. Heatmaps were plotted to illustrate the metabolic patterns using R package pheatmap (version1.0.12). Metabolic pathway analysis was performed using the online tool Metaboanalyst 5.0 ( https://www.metaboanalyst.ca/MetaboAnalyst ), and results were visualized using R package ggplot2 (version 3.3.3). Illustrations of metabolites and their corresponding pathways were created with BioRender ( https://biorender.com ).
This study was performed in accordance with the Declaration of Helsinki (revised in 2013), and the protocol was approved by The Ethical Committee of Zhejiang Cancer Hospital (approval number: IRB-2018-9). Informed consent was obtained from each patient.
Patients who displayed pleural thickening in computed tomography (CT) imaging were reviewed by an experienced sonographer (Junping Liu) between March 1 st 2018 and December 30 th 2019 in Zhejiang Cancer Hospital. After exclusion, US-guided pleural biopsy was performed for the remaining patients. Those from patients who did not meet the following exclusion criteria were highly suspected (by the sonographer based on his past clinical experiences) to be MM: 1) pleural metastasis from lung cancer; 2) pleural metastasis from other cancers; 3) tuberculosis lesions; 4) unclear diagnosis; and 5) other reasons (according to imaging features and clinical history). Under the premise of not interfering with clinical diagnosis, these biopsies were collected and implanted into immunodeficient mice for constructing PDX models. At the same time, all biopsies were sent to a senior pathologist (Zhenying Guo) for precise diagnosis. The animal experiment was permitted by the Institutional Animal Care and Ethics Committee of Zhejiang Cancer Hospital (No. 2018–03-054), in compliance with national or institutional guidelines for the care and use of animals. All the animals were euthanized using isoflurane at the end of experiment.
Patients were fasted for at least 8 h before biopsy. The biopsy was performed according to Zhang et al. Briefly, Esaote MylabTMTwice ultrasound apparatus (Italy) with probes configuration convex transducer CA541 and linear LA524 was utilized. First, a low-frequency probe (4.0 MHz) was used to detect the pleural effusion, pleura and blood flow, and the thickest point of the pleura was selected for biopsy. Second, pleural biopsy was guided through alternatively using low-frequency (4.0 MHz) and high-frequency (9.0 MHz) probes conducted by two experienced sonographers. One sonographer provided the guidance, and the other used an 18 or 16 G automated cutting needle (MC1816, Bard Max. Core, Bard Inc., USA) to perform the biopsy under local anesthesia with 2% lidocaine. Specimens were immediately fixed in 10% formalin and sent to the Department of Pathology for examination. Without interfering clinical diagnosis, part of fresh specimens was sent for constructing PDX models.
IHC staining was performed according to Wu et al. In brief, slides of formalin-fixed and paraffin-embedded tumors were deparaffinized and incubated in 3% hydrogen peroxidase. After inactivation of endogenous peroxidases, slices were treated with antigen retrieval by boiling at 100 °C for 90 s in citric acid repair solution (pH = 6). After blocking, slices were incubated with antibodies overnight at 4 °C. Slices were then incubated with HRP-labeled secondary antibody for 30 min. According to gender, histological features, and location of the tumor, a panel of biomarkers for IHC analysis was selected and the diagnosis was made by a senior pathologist (Zhenying Guo). Details of the antibodies used were listed in the .
The procedure of PDX construction was similar to established protocol from Wu et al. In brief, 5-week-old female BALB/c immunodeficient mice (certificate number: 2017005004641) were purchased from Shanghai Slac Laboratory Animal Company (Shanghai, China). The mice adapted to the environment for 1 week. Every five of all mice were kept in one cage with free access to food and water, to a 12 h/12 h light/dark cycle, at temperature between 22 °C and 26 °C, with 55% relative humidity. Fresh tumor tissue was kept in a sterilized PBS buffer on ice, and was cut into blocks of 2 × 2 × 2 mm and then was engrafted subcutaneously into the flanks of BALB/c mice (P0) with a trocar needle. A PDX model which can be consecutively passed over twice is defined as a success (P2). Then P2 tumors were harvested and transplanted to 10 mice for the PDX model, and 7 mice without treatment were taken as a control group. When the average tumor size exceeded 200 mm 3 , blood was collected retro-orbitally under isoflurane anesthesia. Sera were separated with centrifugation for 10 min at 2400 g, 4 °C and were kept at -80 °C until analysis.
The GC-MS-based metabolomics was performed according to the previously published method from Zhao et al. The procedures including sample preparation and GC-MS analysis are described in the .
After data formatting (to .abf) with Reifycs Abf Converter ( https://www.reifycs.com/AbfConverter/ ), MS-DIAL software was used for data processing, including peak picking, peak alignment, missing value interpolation, and so on. Metabolite annotation was performed through the untargeted database of GC-MS from Lumingbio. Finally, a data set with sample information and peak information was obtained. Principal component analysis (PCA) and partial least-squares-discriminant analysis (PLS-DA) were performed to visualize the metabolic shift among groups using R package ropls (version 1.18.8). Variable importance in the projection (VIP) was obtained from PLS-DA, which ranks the contribution of metabolite features in the PLS-DA model. Finally, features with VIP > 1.0 and P -value < 0.05 from two-tailed Student’s t-test were defined as differential metabolites. Heatmaps were plotted to illustrate the metabolic patterns using R package pheatmap (version1.0.12). Metabolic pathway analysis was performed using the online tool Metaboanalyst 5.0 ( https://www.metaboanalyst.ca/MetaboAnalyst ), and results were visualized using R package ggplot2 (version 3.3.3). Illustrations of metabolites and their corresponding pathways were created with BioRender ( https://biorender.com ).
Patient information and model construction From March 1 st 2018 to December 30 th 2019, 158 patients with pleural thickening were reviewed by a senior sonographer (Junping Liu) (Fig. ). A typical pleural thickening in CT scan was shown in Fig. A. A total of 42 individuals were excluded, and the remaining ( n = 116) patients were performed with US-guided pleural biopsy and pathological diagnosis (Fig. , Fig. B, Table S ). Fourteen of them were suspected to be MPM (Table S ), among which only 10 biopsies were accessible and were subsequently implanted into mice. Biopsies pathologically diagnosed with other diseases ( n = 5), including lung adenocarcinoma ( n = 3) and rhabdomyosarcoma ( n = 2), were further excluded (Fig. ). Among the 5 pathologically confirmed MM, 3 failed to grow on mice from P0 to P2, while 2 were successfully constructed. PDX1 was from a 55-year-old male with pleural thickening of annular nodules, lumps in the lungs and chest wall; while PDX2 was from an 85-year-old female, who showed pleural thickening of annular nodules, chest wall lumps. Their detailed information was shown in Table . PDX model confirmation by IHC In the metabolomics study, 6 out of 10 PDX1 models (P3) and 8 out of 10 PDX2 (P3) model grew and were used for GC-MS based metabolomics. Figures and demonstrated Haemotoxylin and Eosin stains (HE) and IHC results of the two primary tumors and the xenograft tumors, PDX1 (Fig. ) and PDX2 (Fig. ). HE stains presented the epithelioid (Fig. A, G) and sarcomatoid (Fig. A, H) features of MPM for the primary tumors and PDX1&2 tumors. Positive expression of CR, CK5/6, WT1, and D2-40 (Fig. B-E) were denoted in primary epithelioid tumor tissue, and PDX1 expressed the same pattern (Fig. H-K). TTF1 negative in primary tumor (Fig. F) and PAX8 negative in PDX1 (Fig. L) distinguish MPM from metastatic lung adenocarcinomas. Altogether, these results confirmed the pathological subtype of epithelioid mesothelioma, and suggested successful construction of a reliable PDX model. Similarly, expression of CR (Fig. B), and WT1 (Fig. C) confirmed the nature of the second model tumor being mesothelioma. CAM5.2 positive (Fig. E) distinguished this tumor from sarcoma. But CK5/6 (Fig. D) negative and VIM positive (Fig. F) can still discriminate its sarcomatoid identity from the epithelioid mesothelioma. In accordance, immunoprofiling for PDX2 showed CR (Fig. I), WT1 (Fig. J), and CAM5.2 (Fig. K) positive. In addition, DES (Fig. G) negative of primary tumor of PDX2 indicated poor differentiation ability of this tumor. Serum metabolic shift between PDX models and controls In total, 209 metabolites in serum samples were annotated. After multivariate analyses, samples in PDX1 group were successfully separated from control group by unsupervised PCA (Fig. S A), whereas PDX2 group cannot be separated from control group (Fig. S B). In supervised PLS-DA models, clearer separation trend for PDX 1 (Fig. S C) and 2 (Fig. S D) were present, though the latter remains ambiguous. Volcano plots revealed significant changes (VIP > 1.0, P -value < 0.05) in metabolites in between PDX1 and 2 (Fig. S ), and each in comparison to controls (Fig. S E, F). In PDX1 versus controls, 58 upregulated metabolites and 23 downregulated metabolites were obtained. And in PDX2, 21 upregulated metabolites and 3 downregulated metabolites were obtained. In addition, of the 12 overlapped metabolites, 10 were upregulated and 2 were downregulated, in both PDX1 and PDX2 versus controls. Between PDX1 and PDX2, 10 upregulated and 14 downregulated metabolites were presented. Detailed information of differential metabolites (including VIP, P -value, and fold change) was listed in Table S (PDX1 vs. control), Table S (PDX2 vs. control), Table S (overlapped in PDX1 and PDX2 compared to controls), and Table S (PDX1 vs. PDX2). Metabolic patterns and pathway enrichment Hierarchical clustering heatmaps were then conducted and showed distinctive differentially expressed metabolic patterns between PDX1 and control ( n = 82; Fig. ), PDX2 and control ( n = 25; Fig. ), overlapped between PDX1 and 2 ( n = 12; Fig. A), and between PDX1 and PDX2 (n = 25; Fig. B). Figure illustrated enriched pathways of the differential metabolites between PDX1 vs. control (Fig. A, Table S ), PDX2 vs. control (Fig. B, Table S ), overlapped in PDX1 and PDX2 (Fig. C , Table S ), and PDX1 vs. PDX2 (Fig. D, Table S ). Figure showed all annotated metabolites presented and their enrichments in amino acid metabolism, TCA cycle and glycolysis, and nucleotide metabolism in PDX1, PDX2, and control. With the knowledge of the enriched pathways and the metabolites involved, an overall illustration of amino acid metabolism, tryptophan degradation, glycolysis, and TCA cycle was pictured in Fig. . Dysregulated amino acid metabolism in MPM Among the 81 significant dysregulated metabolites presented in PDX1, 5 were amino acids with 4 downregulated and 1 upregulated. Within the 5 amino acids, upregulated kynurenine and downregulated serotonin were involved in tryptophan degradation. Meanwhile, 4 of top 10 most significantly enriched pathways in PDX1 were amino acid metabolism pathways, and according to rich factor and P -value, the aspartate and glutamine metabolism; and arginine biosynthesis were the top 2 enriched pathways. In PDX2, no significant changes were observed in amino acids. Nevertheless, similar to PDX1, kynurenine was also significantly upregulated, with a fold change of 1.99 compared to controls. However, tryptophan itself did not display significant change in circulating levels in both PDX models. Dysregulated glycolysis and TCA metabolites in MPM The top 3 most significantly enriched pathways in PDX2 were citrate cycle, pyruvate metabolism, and glycolysis/gluconeogenesis. In line, two metabolites in pyruvate metabolism, bisphosphoglycerol and pyruvic acid, were all upregulated, with fold changes of 1.46 and 1.37, respectively. Additionally, the end product of Warburg aerobic glycolysis, lactic acid, was found upregulated in PDX2 but not in PDX1. Succinic acid and pyruvic acid are overlapped metabolites of TCA cycle in both PDX models that showed significant changes. Succinic acid increased 2.17 folds in PDX1 and 1.49 folds in PDX2; and pyruvic acid increased 1.40 folds in PDX1 and 1.37 folds in PDX2. In addition, in PDX1, glutamine was downregulated to 0.49 folds, whereas fumaric acid was upregulated to 1.82 folds. But no significant alterations in other TCA compartments were detected in PDX2. Abnormality in nucleotide metabolism We detected uridine 5′-monophosphate (UMP), uridine, thymidine, and uracil enrichment in pyrimidine metabolism. In addition, hypoxanthine, inosine, allantoate, uric acid, and urea were enriched in purine metabolism. Consistently, uric acid, uridine, and inosine were significantly upregulated in both PDX1 and PDX2. Uracil, UMP, and hypoxanthine were upregulated in both models, whereas the trends were only significant in PDX1. Urea was only significantly increased in PDX2 with a fold change of 1.40, while thymidine was only downregulated in PDX1 with a fold change of 0.40. No change was detected in levels of allantoate in both models.
From March 1 st 2018 to December 30 th 2019, 158 patients with pleural thickening were reviewed by a senior sonographer (Junping Liu) (Fig. ). A typical pleural thickening in CT scan was shown in Fig. A. A total of 42 individuals were excluded, and the remaining ( n = 116) patients were performed with US-guided pleural biopsy and pathological diagnosis (Fig. , Fig. B, Table S ). Fourteen of them were suspected to be MPM (Table S ), among which only 10 biopsies were accessible and were subsequently implanted into mice. Biopsies pathologically diagnosed with other diseases ( n = 5), including lung adenocarcinoma ( n = 3) and rhabdomyosarcoma ( n = 2), were further excluded (Fig. ). Among the 5 pathologically confirmed MM, 3 failed to grow on mice from P0 to P2, while 2 were successfully constructed. PDX1 was from a 55-year-old male with pleural thickening of annular nodules, lumps in the lungs and chest wall; while PDX2 was from an 85-year-old female, who showed pleural thickening of annular nodules, chest wall lumps. Their detailed information was shown in Table .
In the metabolomics study, 6 out of 10 PDX1 models (P3) and 8 out of 10 PDX2 (P3) model grew and were used for GC-MS based metabolomics. Figures and demonstrated Haemotoxylin and Eosin stains (HE) and IHC results of the two primary tumors and the xenograft tumors, PDX1 (Fig. ) and PDX2 (Fig. ). HE stains presented the epithelioid (Fig. A, G) and sarcomatoid (Fig. A, H) features of MPM for the primary tumors and PDX1&2 tumors. Positive expression of CR, CK5/6, WT1, and D2-40 (Fig. B-E) were denoted in primary epithelioid tumor tissue, and PDX1 expressed the same pattern (Fig. H-K). TTF1 negative in primary tumor (Fig. F) and PAX8 negative in PDX1 (Fig. L) distinguish MPM from metastatic lung adenocarcinomas. Altogether, these results confirmed the pathological subtype of epithelioid mesothelioma, and suggested successful construction of a reliable PDX model. Similarly, expression of CR (Fig. B), and WT1 (Fig. C) confirmed the nature of the second model tumor being mesothelioma. CAM5.2 positive (Fig. E) distinguished this tumor from sarcoma. But CK5/6 (Fig. D) negative and VIM positive (Fig. F) can still discriminate its sarcomatoid identity from the epithelioid mesothelioma. In accordance, immunoprofiling for PDX2 showed CR (Fig. I), WT1 (Fig. J), and CAM5.2 (Fig. K) positive. In addition, DES (Fig. G) negative of primary tumor of PDX2 indicated poor differentiation ability of this tumor.
In total, 209 metabolites in serum samples were annotated. After multivariate analyses, samples in PDX1 group were successfully separated from control group by unsupervised PCA (Fig. S A), whereas PDX2 group cannot be separated from control group (Fig. S B). In supervised PLS-DA models, clearer separation trend for PDX 1 (Fig. S C) and 2 (Fig. S D) were present, though the latter remains ambiguous. Volcano plots revealed significant changes (VIP > 1.0, P -value < 0.05) in metabolites in between PDX1 and 2 (Fig. S ), and each in comparison to controls (Fig. S E, F). In PDX1 versus controls, 58 upregulated metabolites and 23 downregulated metabolites were obtained. And in PDX2, 21 upregulated metabolites and 3 downregulated metabolites were obtained. In addition, of the 12 overlapped metabolites, 10 were upregulated and 2 were downregulated, in both PDX1 and PDX2 versus controls. Between PDX1 and PDX2, 10 upregulated and 14 downregulated metabolites were presented. Detailed information of differential metabolites (including VIP, P -value, and fold change) was listed in Table S (PDX1 vs. control), Table S (PDX2 vs. control), Table S (overlapped in PDX1 and PDX2 compared to controls), and Table S (PDX1 vs. PDX2).
Hierarchical clustering heatmaps were then conducted and showed distinctive differentially expressed metabolic patterns between PDX1 and control ( n = 82; Fig. ), PDX2 and control ( n = 25; Fig. ), overlapped between PDX1 and 2 ( n = 12; Fig. A), and between PDX1 and PDX2 (n = 25; Fig. B). Figure illustrated enriched pathways of the differential metabolites between PDX1 vs. control (Fig. A, Table S ), PDX2 vs. control (Fig. B, Table S ), overlapped in PDX1 and PDX2 (Fig. C , Table S ), and PDX1 vs. PDX2 (Fig. D, Table S ). Figure showed all annotated metabolites presented and their enrichments in amino acid metabolism, TCA cycle and glycolysis, and nucleotide metabolism in PDX1, PDX2, and control. With the knowledge of the enriched pathways and the metabolites involved, an overall illustration of amino acid metabolism, tryptophan degradation, glycolysis, and TCA cycle was pictured in Fig. .
Among the 81 significant dysregulated metabolites presented in PDX1, 5 were amino acids with 4 downregulated and 1 upregulated. Within the 5 amino acids, upregulated kynurenine and downregulated serotonin were involved in tryptophan degradation. Meanwhile, 4 of top 10 most significantly enriched pathways in PDX1 were amino acid metabolism pathways, and according to rich factor and P -value, the aspartate and glutamine metabolism; and arginine biosynthesis were the top 2 enriched pathways. In PDX2, no significant changes were observed in amino acids. Nevertheless, similar to PDX1, kynurenine was also significantly upregulated, with a fold change of 1.99 compared to controls. However, tryptophan itself did not display significant change in circulating levels in both PDX models.
The top 3 most significantly enriched pathways in PDX2 were citrate cycle, pyruvate metabolism, and glycolysis/gluconeogenesis. In line, two metabolites in pyruvate metabolism, bisphosphoglycerol and pyruvic acid, were all upregulated, with fold changes of 1.46 and 1.37, respectively. Additionally, the end product of Warburg aerobic glycolysis, lactic acid, was found upregulated in PDX2 but not in PDX1. Succinic acid and pyruvic acid are overlapped metabolites of TCA cycle in both PDX models that showed significant changes. Succinic acid increased 2.17 folds in PDX1 and 1.49 folds in PDX2; and pyruvic acid increased 1.40 folds in PDX1 and 1.37 folds in PDX2. In addition, in PDX1, glutamine was downregulated to 0.49 folds, whereas fumaric acid was upregulated to 1.82 folds. But no significant alterations in other TCA compartments were detected in PDX2.
We detected uridine 5′-monophosphate (UMP), uridine, thymidine, and uracil enrichment in pyrimidine metabolism. In addition, hypoxanthine, inosine, allantoate, uric acid, and urea were enriched in purine metabolism. Consistently, uric acid, uridine, and inosine were significantly upregulated in both PDX1 and PDX2. Uracil, UMP, and hypoxanthine were upregulated in both models, whereas the trends were only significant in PDX1. Urea was only significantly increased in PDX2 with a fold change of 1.40, while thymidine was only downregulated in PDX1 with a fold change of 0.40. No change was detected in levels of allantoate in both models.
Sample scarcity has been an issue for research in MPM. Our methodology for PDX modeling from US-guided pleural biopsy in part removes this restraint, as biopsy samples are more accessible than samples from surgery. Although biopsy during video-assisted thoracoscopic surgery (VATS) was the gold standard for MPM diagnosis in current, it is more invasive than US-guided biopsy. Notably, dedicated examination and sampling used in our strategy helped improve efficiency as well as reduce the cost in modeling. For pleural lesions, a US-guided pleural biopsy was utilized, which outstands for its real-time multiplanar visualization that aids sampling . Then the senior sonographer helped narrow specimens down to MPM candidates for modeling through the aforementioned exclusion criteria. Precise diagnosis by an experienced pathologist then helped rule out non-MPM PDXs. Thereby, our modeling strategy can diminish the waste of resources and time while retaining the reliability of PDX models. In addition, PDX models offer a more manipulative environment for developing therapies and detecting biomarkers compared to human beings , as well as better retain heterogeneity of tumor compared to cell lines . Many cancer studies use PDX model as it highly resembles the primary tumor implanted into it [ – ]. Meanwhile, such model parallels the original tumor better than cell lines, and concurrent results are closely relevant to the clinic . Hence, for the above reasons, this modeling strategy is insightful and feasible, which can facilitate not only MPM-related research but also the research of other rare cancers limited by lack of specimens. By further applying metabolomics based on PDX sera, we detected a panel of dysregulated metabolites and enriched pathways. A significant proportion of metabolic changes denoted in our results was verified in other clinical studies, indicating the reliability of both our metabolomics method and the PDX model. Therefore, we encourage the use of US-guided pleural biopsy for PDX modeling in combination with metabolomics for investigation in rare cancers, including MPM, which brings an opportunity for the identification of predictive markers and treatment targets. Our metabolomics results revealed a dysregulated amino acid metabolism in MPM. Many studies had reported that cancers vastly demand amino acid for ATP yielding, growth, and progression, through overexpression of SLC7A5 and SLC1A5 [ – ]. Our results suggested MPM has same desire as other cancers for amino acids. Particularly, serum kynurenine, which is a downstream metabolite of tryptophan, increased significantly in both PDX1 and PDX2 in comparison to controls, suggesting an increased tryptophan metabolism. Some cancer research had opined immunosuppressive role of kynurenine . In line, many studies documented increased levels of IDO and TDO in cancers, which are two enzymes catalyzing anabolism of tryptophan to kynurenine [ – ]. Collectively, dysregulated circulating amino acid metabolism, especially kynurenine metabolism, was a significant metabolic feature of MPM. We also revealed dysregulations in TCA cycle and glycolysis in MPM. We detected an increase in pyruvic acid efflux, suggesting an elevated glycolysis rate. However, contradictory results were found in lactic acid, which was elevated in PDX2 but unchanged in PDX1, denoting a higher rate of aerobic glycolysis in PDX2 than PDX1. Increasing evidence has suggested that lactic acid secretion helps immune evasion of tumor by constructing a micro-environment with low pH that suppresses anti-tumor immune response . Therefore, the increased glycolysis rate in MPM may have dual purposes that increase energy fueling as well as help tumor escape from immunity. Dysregulations in metabolites of purine metabolism and pyrimidine metabolism were detected, indicating an imbalanced nucleotide metabolism. Based on our results, circulating uridine was significantly upregulated in both PDX models, which may be signs of a higher rate of uridine synthesis in MPM. Tumor synthesizes more nucleotides, increasing deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) pools that support proliferation . Further we found significant elevations in uric acid, which has been reported to be released from dying tumor cells . This also frequently occurs in other cancers, such as breast cancer , hepatocarcinoma , and head and neck carcinoma . Uric acid also promotes tumor immune rejection , and serves as a pro-oxidant that induces tumor growth . Herein, the excessively synthesized uridine and uric acid in MPM may be biomarkers for MPM. However, the changed metabolites were only phenotypes which lack clear molecular mechanisms to be elucidated. Also, it is well-known that BAP1 influences metabolism in MPM, so metabolic inconsistency between samples of different BAP1 status should be compared. It is a pity that we only have one patient’s information of genetic mutation, but future research with larger sample size and complete clinical information should fill in this gap.
In conclusion, our study developed a novel modeling technique to facilitate research in malignant mesothelioma, especially in China. In using a combination of CT scanning, pathological analysis and US-guided pleural biopsy for PDX modeling, we can remove barriers in MPM research that is caused by the scarcity in samples, thus improving availability of research in MPM or other infrequent cancers. By further coupling with metabolomics to screen for metabolic biomarkers, we can advance the current diagnostic method and treatment for MM. Nevertheless, studies with larger sample size are needed for this cancer, and molecular mechanisms as well as genetic predispositions should be further investigated to verify our results.
Additional file 1. Additional file 2. Additional file 3. Additional file 4.
|
Assessment of TP53 and CDKN2A status as predictive markers of malignant transformation of sinonasal inverted papilloma | 3a030776-4a74-4a71-a73d-0a0c6d058fbc | 11190283 | Anatomy[mh] | Sinonasal inverted papillomas (IP) are common benign mucosal neoplasms that occur in the sinonasal tract and are characterized by their inverted growth pattern . Although IPs are classified as benign, they have the potential to progress into squamous cell carcinoma (SCC), with reported rates of malignant transformation ranging from 1.9 to 27% . The development of IP is a complex process that involves various genetic and environmental factors; however, their progression to SCC is poorly understood. Several studies investigated the molecular mechanisms underlying the tumorigenesis and malignant transformation of IP. Various genetic mutations, including EGFR, TP53 , CDKN2A and KRAS mutations, as well as human papillomavirus (HPV) infection have been reported as potential mechanisms of malignant transformation of IP – . Although one study used whole exome sequencing to find the genetic alterations related to malignant transformation , most other studies used targeted gene panels for next-generation sequencing , – , which limited the detection of genetic variants to only the genes included in the panel . Furthermore, targeted sequencing is often performed using only tumor tissue, making it difficult to distinguish between germline and somatic mutations. In addition, there is currently no reliable diagnostic method for predicting malignant transformation. A better understanding of the genetic alterations that contribute to malignant transformation may enable the development of more accurate diagnostic methods and more effective treatment strategies. In this study, we aimed to investigate the genetic mutations involved in the stepwise progression of IP to SCC and explore potential biomarkers that could predict malignant transformation using whole exome sequencing with matched normal tissue. This approach has the potential to provide a more comprehensive understanding of the genetic alterations that contribute to malignant transformation and to identify new targets for early detection and prevention of IP progression to SCC. Sample selection and DNA extraction We included 14 patients who were diagnosed with and treated for SCC arising from IP (SCC-IP) at Seoul National University Bundang Hospital between 2004 and 2020. In addition, six patients who were diagnosed with IP without malignant transformation ("sIP") were included as a comparison group. The hematoxylin and eosin stained slides for each case were reviewed by two pathologists (S.K. and H.K.) to select the areas for sequencing and immunohistochemistry (IHC). In each case, we distinguished each component of normal mucosae, IP, IP with dysplasia, and invasive SCC for macro-dissection. DNA was extracted separately from each component. The list of patients and samples used for sequencing and IHC is shown in Fig. . The study protocol was approved by the Institutional Review Board of Seoul National University Bundang Hospital (IRB No. B-2008-630-307), and the study was performed in accordance with the Declaration of Helsinki. Informed consent was obtained from each patient, except for those who died. Whole exome sequencing DNA was extracted using the GeneRead DNA FFPE kit (Qiagen) following the manufacturer's protocol. The quality and quantity of purified DNA were assessed by fluorometry (Qubit, Invitrogen) and gel electrophoresis. Briefly, 200 ng of each sample was ligated to Illumina’s adapters and PCR-amplified. The samples were concentrated to < 1000 ng in 12 μL DW using a SpeedVac machine and hybridized with RNA probes, SureSelectXT Human All Exon V5 at 65 °C 1 min–37 °C 3 s, 60 cycles. After hybridization, the captured targets were pulled down by biotinylated probe/target hybrids using streptavidin-coated magnetic beads (Dynabeads My One Streptavidine T1; Life Technologies Ltd.) and buffers. The selected regions were then PCR-amplified using Illumina PCR primers. Libraries were quantified using the Agilent 4200 Bioanalyzer (Agilent) and KAPA Library Quantification Kit (Kapa Biosystems). The high quality-libraries were pooled and sequenced on the Illumina NovaSeq6000 platform (Illumina) with 150 bp paired-end by following the manufacturer’s protocols. Image analysis were performed using the NovaSeq6000 control Software version 1.3.1 and the output base calling data was de-multiplexed with bcl2fastq version v2.20.0.422 generating fastQC files. Sequencing reads were aligned to the human reference genome hg19 using Burrows Wheeler Aligner (BWA) (v.0.7.17) . After the alignment of the reads to reference genome, the duplicated reads were further removed using MarkDuplicates in Picard (v.2.20.7). Next, base quality score recalibration (BQSR) process was conducted to adjust the quality score using BaseRecalibrator in Genome Analysis Toolkit (GATK) (v.4.1.3) . For germline and somatic variants calling, GATK HaplotypeCaller and Mutect2 were utilized, respectively. Further, the LearnReadOrientationModel and FilterMutectCalls of GATK were employed to filter orientation bias, technical artifacts and sequencing error. In addition to matched normal samples, gnomAD database was utilized to further exclude germline variants. Only variants with a minimum of 10 supporting reads were included. All variants were then annotated using Ensembl VEP v100 considering the effects on transcripts, proteins, and regulatory regions. For known or overlapping variants, allele frequencies and disease or phenotype information were included. For downstream analysis, the variants call format (VCF) files were converted to mutation annotation format (MAF) files using vcf2maf. The variants annotated as PASS were summarized and visualized using R packages maftools . p53 and p16 immunohistochemistry Immunostaining for p53 and p16 were performed using monoclonal mouse anti-human p53 (clone DO-7, 1:1000, Dako, Carpinteria, CA, USA) primary antibody and monoclonal mouse p16 (clone E6H4, CINtec ® , Ventana Medical Systems, Inc., Tucson, AZ, USA) primary antibody on an automated platform (Benchmark Ultra; Ventana Medical Systems) according to the manufacturer’s instructions. The results were independently interpreted by two pathologists (S.K. and H.K.). P53 expression was classified as diffuse strong positive if there was a diffuse strong nuclear staining in > 80% of tumor cell nuclei, total loss if there was complete absence of staining, and patchy positive if there was variable nuclear staining in 1–80% of tumor cell nuclei . P16 expression was classified as diffuse strong positive if there was a diffuse strong nuclear and cytoplasmic staining in > 90% of tumor cells, total loss if there was complete absence of staining, and patchy positive if there was variable nuclear and/or cytoplasmic staining . Human papillomavirus genotyping HPV status was determined by HPV genotyping. HPV genotyping was performed using peptide nucleic acid probe-based fluorescence melting curve analysis in a real-time PCR system (PANA RealTyper™ HPV Kit, PANAGENE, Daejeon, Republic of Korea) according to the manufacturer’s instructions. It provides a qualitative detection of 40 HPV genotypes, including genotyping information of 20 high-risk types (16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 69, 70, 73, 82) and 2 low-risk types (6, 11), or the presence of 18 low-risk types (30, 32, 34, 40, 42, 43, 44, 54, 55, 61, 62, 67, 74, 81, 83, 84, 87, 90) without genotyping. We included 14 patients who were diagnosed with and treated for SCC arising from IP (SCC-IP) at Seoul National University Bundang Hospital between 2004 and 2020. In addition, six patients who were diagnosed with IP without malignant transformation ("sIP") were included as a comparison group. The hematoxylin and eosin stained slides for each case were reviewed by two pathologists (S.K. and H.K.) to select the areas for sequencing and immunohistochemistry (IHC). In each case, we distinguished each component of normal mucosae, IP, IP with dysplasia, and invasive SCC for macro-dissection. DNA was extracted separately from each component. The list of patients and samples used for sequencing and IHC is shown in Fig. . The study protocol was approved by the Institutional Review Board of Seoul National University Bundang Hospital (IRB No. B-2008-630-307), and the study was performed in accordance with the Declaration of Helsinki. Informed consent was obtained from each patient, except for those who died. DNA was extracted using the GeneRead DNA FFPE kit (Qiagen) following the manufacturer's protocol. The quality and quantity of purified DNA were assessed by fluorometry (Qubit, Invitrogen) and gel electrophoresis. Briefly, 200 ng of each sample was ligated to Illumina’s adapters and PCR-amplified. The samples were concentrated to < 1000 ng in 12 μL DW using a SpeedVac machine and hybridized with RNA probes, SureSelectXT Human All Exon V5 at 65 °C 1 min–37 °C 3 s, 60 cycles. After hybridization, the captured targets were pulled down by biotinylated probe/target hybrids using streptavidin-coated magnetic beads (Dynabeads My One Streptavidine T1; Life Technologies Ltd.) and buffers. The selected regions were then PCR-amplified using Illumina PCR primers. Libraries were quantified using the Agilent 4200 Bioanalyzer (Agilent) and KAPA Library Quantification Kit (Kapa Biosystems). The high quality-libraries were pooled and sequenced on the Illumina NovaSeq6000 platform (Illumina) with 150 bp paired-end by following the manufacturer’s protocols. Image analysis were performed using the NovaSeq6000 control Software version 1.3.1 and the output base calling data was de-multiplexed with bcl2fastq version v2.20.0.422 generating fastQC files. Sequencing reads were aligned to the human reference genome hg19 using Burrows Wheeler Aligner (BWA) (v.0.7.17) . After the alignment of the reads to reference genome, the duplicated reads were further removed using MarkDuplicates in Picard (v.2.20.7). Next, base quality score recalibration (BQSR) process was conducted to adjust the quality score using BaseRecalibrator in Genome Analysis Toolkit (GATK) (v.4.1.3) . For germline and somatic variants calling, GATK HaplotypeCaller and Mutect2 were utilized, respectively. Further, the LearnReadOrientationModel and FilterMutectCalls of GATK were employed to filter orientation bias, technical artifacts and sequencing error. In addition to matched normal samples, gnomAD database was utilized to further exclude germline variants. Only variants with a minimum of 10 supporting reads were included. All variants were then annotated using Ensembl VEP v100 considering the effects on transcripts, proteins, and regulatory regions. For known or overlapping variants, allele frequencies and disease or phenotype information were included. For downstream analysis, the variants call format (VCF) files were converted to mutation annotation format (MAF) files using vcf2maf. The variants annotated as PASS were summarized and visualized using R packages maftools . Immunostaining for p53 and p16 were performed using monoclonal mouse anti-human p53 (clone DO-7, 1:1000, Dako, Carpinteria, CA, USA) primary antibody and monoclonal mouse p16 (clone E6H4, CINtec ® , Ventana Medical Systems, Inc., Tucson, AZ, USA) primary antibody on an automated platform (Benchmark Ultra; Ventana Medical Systems) according to the manufacturer’s instructions. The results were independently interpreted by two pathologists (S.K. and H.K.). P53 expression was classified as diffuse strong positive if there was a diffuse strong nuclear staining in > 80% of tumor cell nuclei, total loss if there was complete absence of staining, and patchy positive if there was variable nuclear staining in 1–80% of tumor cell nuclei . P16 expression was classified as diffuse strong positive if there was a diffuse strong nuclear and cytoplasmic staining in > 90% of tumor cells, total loss if there was complete absence of staining, and patchy positive if there was variable nuclear and/or cytoplasmic staining . HPV status was determined by HPV genotyping. HPV genotyping was performed using peptide nucleic acid probe-based fluorescence melting curve analysis in a real-time PCR system (PANA RealTyper™ HPV Kit, PANAGENE, Daejeon, Republic of Korea) according to the manufacturer’s instructions. It provides a qualitative detection of 40 HPV genotypes, including genotyping information of 20 high-risk types (16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 69, 70, 73, 82) and 2 low-risk types (6, 11), or the presence of 18 low-risk types (30, 32, 34, 40, 42, 43, 44, 54, 55, 61, 62, 67, 74, 81, 83, 84, 87, 90) without genotyping. Clinicopathologic characteristics The clinicopathologic characteristics of the patients are summarized in Table . There was no significant difference in age (63.2 ± 12.1 vs. 62.9 ± 6.9 years), sex, and mean tumor size (4.1 cm vs. 3.4 cm) between the two groups ( p = 0.935, 0.573, and 0.191, respectively). The five-year survival rate was 71.4% in the SCC-IP group and 100% in the sIP group without statistically significant difference ( p = 0.763). Genomic alteration related with malignant transformation of inverted papilloma Various single nucleotide variants (SNVs) were identified in SCC-IP group. Top 50 genes that were frequently mutated are shown in Fig. . The most common mutated gene was TP53 (39%), followed by CDKN2A (27%), TTN (27%), PIK3CA (21%), and ARID1A (15%). When limited to SCC, the most frequently mutated genes were TP53 (43%), CDKN2A (36%), TTN (36%), ARID1A (21%), FAT1 (21%), KEAP1 (21%), and PIK3CA (21%). In contrast, rare mutations were identified in sIP group. The frequencies of commonly mutated genes in each tumor type subgroup are shown in Table . The entire list of the mutations can be found in Supplementary Table . The tumor mutational burden (TMB) was calculated as a number of non-synonymous SNVs and indels per mega base (Mb) (Fig. ). Mean TMB was higher in IP with malignant transformation (cIP) (0.64/Mb) than in sIP (0.3/Mb), and showed a tendency to gradually increase as cancer progressed within the SCC-IP group (0.64/Mb, 1.11/Mb, and 1.25 for IP, dysplasia, and SCC, respectively) (Fig. ). Multistep analysis of squamous cell carcinoma arising from inverted papilloma focusing on TP53 and CDKN2A There were six cases which had matched IP and SCC component available for sequencing (SCC-IP-4, 7, 14, 8, 10, and 12). In the case of TP53 mutations, there were 2/6 (33.3%) cases in which mutations identical to those observed in SCC were already present in the IP, 2/6 (33.3%) cases in which no TP53 mutation was observed in the IP while SCC had one, and 2/6 (33.3%) cases in which TP53 mutation was not observed in neither IP nor SCC. For CDKN2A mutations, 2/6 (33%) cases showed the same mutations in both IP and SCC, 1/6 (17%) case showed mutations in SCC but not in the IP, and 3/6 (50%) cases showed no mutation in neither IP nor SCC. Taken together, 3/6 (50%) of cIP had the same TP53 and/or CDKN2A mutation as SCC. In contrast, most of the observed mutations in dysplasia and in SCC were identical. There were seven cases which had matched dysplasia and SCC component available for sequencing (SCC-IP-3, 6, 11, 13, 8, 10, and 12). In all but one case, the mutational status of TP53 and CDKN2A in dysplasia and in SCC was the same. The exceptional case had nonsense TP53 mutation in dysplasia, but the SCC had no mutation (SCC-IP-10). p53 and p16 immunohistochemistry and their correlation with mutational status We first correlated the sequencing results with the IHC results in all samples to confirm the relationship between the presence or absence of TP53 and CDKN2A gene mutations and p53 and p16 protein expression (Table and Fig. ). Of the 17 samples which showed patchy positivity of p53 protein in IHC, all samples had wild-type TP53 . When p53 expression was diffuse strong positive in IHC, 10/11 (91%) had missense/indel mutation of TP53 and 1/11 (9%) had wild type TP53 . In the samples in which p53 expression showed total loss, nonsense mutation of TP53 was observed in 3/5 (60%) and wild type in 2/5 (40%). For CDKN2A and p16 expression, 12/14 (86%) samples had wild type forms and 2/14 (14%) samples had missense/indel mutation of CDKN2A in p16 patchy positive tumors. Among five samples which showed diffuse strong positive expression in p16 IHC, all had wild type form of CDKN2A , while three had RB1 frameshift insertion mutation and the other two had high-risk HPV (type 16) infection. The three samples with the RB1 mutation belong to one case (SCC-IP-12), and the two samples with HPV infection belong to another case (SCC-IP-11). When there was total loss p16 expression, 6/14 (43%) samples had missense/indel mutation, 1/14 (7%) samples had nonsense mutation, and 7/14 (50%) samples had wild type form of CDKN2A . Focusing on the six cases which had paired IP and SCC component available for sequencing, 4/6 (67%) cases showed aberrant expression (diffuse strong positive or total loss) of p53 and/or p16 in both IP and SCC. The other 2/6 (33%) cases, which showed patchy positive p53 and p16 expression in the IP, exhibited diffuse strong positive expression of p53 in the SCC, which acquired TP53 mutation during malignant transformation. In contrast, all sIP showed patchy positive p53 and p16 staining. The results of IHC and the mutational status of sIP and cIP are shown in Fig. . Human papillomavirus infection in squamous cell carcinoma arising from inverted papilloma High-risk HPV (type 16) was detected in two samples that belonged to one case (SCC-IP-11). Both dysplasia and SCC had HPV infection. As mentioned above, these samples showed diffuse strong positive p16 expression. The clinicopathologic characteristics of the patients are summarized in Table . There was no significant difference in age (63.2 ± 12.1 vs. 62.9 ± 6.9 years), sex, and mean tumor size (4.1 cm vs. 3.4 cm) between the two groups ( p = 0.935, 0.573, and 0.191, respectively). The five-year survival rate was 71.4% in the SCC-IP group and 100% in the sIP group without statistically significant difference ( p = 0.763). Various single nucleotide variants (SNVs) were identified in SCC-IP group. Top 50 genes that were frequently mutated are shown in Fig. . The most common mutated gene was TP53 (39%), followed by CDKN2A (27%), TTN (27%), PIK3CA (21%), and ARID1A (15%). When limited to SCC, the most frequently mutated genes were TP53 (43%), CDKN2A (36%), TTN (36%), ARID1A (21%), FAT1 (21%), KEAP1 (21%), and PIK3CA (21%). In contrast, rare mutations were identified in sIP group. The frequencies of commonly mutated genes in each tumor type subgroup are shown in Table . The entire list of the mutations can be found in Supplementary Table . The tumor mutational burden (TMB) was calculated as a number of non-synonymous SNVs and indels per mega base (Mb) (Fig. ). Mean TMB was higher in IP with malignant transformation (cIP) (0.64/Mb) than in sIP (0.3/Mb), and showed a tendency to gradually increase as cancer progressed within the SCC-IP group (0.64/Mb, 1.11/Mb, and 1.25 for IP, dysplasia, and SCC, respectively) (Fig. ). There were six cases which had matched IP and SCC component available for sequencing (SCC-IP-4, 7, 14, 8, 10, and 12). In the case of TP53 mutations, there were 2/6 (33.3%) cases in which mutations identical to those observed in SCC were already present in the IP, 2/6 (33.3%) cases in which no TP53 mutation was observed in the IP while SCC had one, and 2/6 (33.3%) cases in which TP53 mutation was not observed in neither IP nor SCC. For CDKN2A mutations, 2/6 (33%) cases showed the same mutations in both IP and SCC, 1/6 (17%) case showed mutations in SCC but not in the IP, and 3/6 (50%) cases showed no mutation in neither IP nor SCC. Taken together, 3/6 (50%) of cIP had the same TP53 and/or CDKN2A mutation as SCC. In contrast, most of the observed mutations in dysplasia and in SCC were identical. There were seven cases which had matched dysplasia and SCC component available for sequencing (SCC-IP-3, 6, 11, 13, 8, 10, and 12). In all but one case, the mutational status of TP53 and CDKN2A in dysplasia and in SCC was the same. The exceptional case had nonsense TP53 mutation in dysplasia, but the SCC had no mutation (SCC-IP-10). We first correlated the sequencing results with the IHC results in all samples to confirm the relationship between the presence or absence of TP53 and CDKN2A gene mutations and p53 and p16 protein expression (Table and Fig. ). Of the 17 samples which showed patchy positivity of p53 protein in IHC, all samples had wild-type TP53 . When p53 expression was diffuse strong positive in IHC, 10/11 (91%) had missense/indel mutation of TP53 and 1/11 (9%) had wild type TP53 . In the samples in which p53 expression showed total loss, nonsense mutation of TP53 was observed in 3/5 (60%) and wild type in 2/5 (40%). For CDKN2A and p16 expression, 12/14 (86%) samples had wild type forms and 2/14 (14%) samples had missense/indel mutation of CDKN2A in p16 patchy positive tumors. Among five samples which showed diffuse strong positive expression in p16 IHC, all had wild type form of CDKN2A , while three had RB1 frameshift insertion mutation and the other two had high-risk HPV (type 16) infection. The three samples with the RB1 mutation belong to one case (SCC-IP-12), and the two samples with HPV infection belong to another case (SCC-IP-11). When there was total loss p16 expression, 6/14 (43%) samples had missense/indel mutation, 1/14 (7%) samples had nonsense mutation, and 7/14 (50%) samples had wild type form of CDKN2A . Focusing on the six cases which had paired IP and SCC component available for sequencing, 4/6 (67%) cases showed aberrant expression (diffuse strong positive or total loss) of p53 and/or p16 in both IP and SCC. The other 2/6 (33%) cases, which showed patchy positive p53 and p16 expression in the IP, exhibited diffuse strong positive expression of p53 in the SCC, which acquired TP53 mutation during malignant transformation. In contrast, all sIP showed patchy positive p53 and p16 staining. The results of IHC and the mutational status of sIP and cIP are shown in Fig. . High-risk HPV (type 16) was detected in two samples that belonged to one case (SCC-IP-11). Both dysplasia and SCC had HPV infection. As mentioned above, these samples showed diffuse strong positive p16 expression. In this study, we found that TP53 and CDKN2A could be involved in the early stage of the stepwise progression of IP to SCC and that the assessment of TP53 and CDKN2A status could be a predictive marker of malignant transformation of IP. Moreover, using IHC, we found that p53 and p16 expression could be used as surrogate marker for TP53 and CDKN2A mutational status, respectively, and aberrant expression of p53 and/or p16 could be a predictive marker of malignant transformation of IP. Both TP53 and CDKN2A are tumor suppressor genes and are observed with high frequency in many tumors , . Recent studies have shown some conflicting results of TP53 mutation in cIP. Brown et al. reported that TP53 mutations and CDKN2A mutations/deletions were related to malignant transformation, based on the result that they were observed only in the carcinoma but not in the matched IP . In contrast, Yasukawa et al. reported that most of the TP53 mutations observed in dysplasia and SCC were already present in IP and there was little difference in mutations observed between IP and SCC . In this study, TP53 and CDKN2A mutations, which were identical to those present in SCC, were observed in 50% (3/6) of cIP, and dysplasia and SCC showed nearly identical mutations. Furthermore, TP53 and CDKN2A mutations were not observed in sIP. This suggests that TP53 and CDKN2A mutations are involved in the early stage of malignant transformation and can be used as biomarkers of early detection of cIP. There was a strong correlation between TP53 mutation and the aberrant expression of p53. It was concordant with previous studies on gastric and ovarian cancers . However, p16 expression did not show a correlation as strong as p53 expression did. When p16 was patchy positive, it was likely that CDKN2A was wild type. However, when p16 showed diffuse strong positive staining, there was no CDKN2A mutation, while when there was total loss of p16 expression, half of the cases had CDKN2A mutation. Traditionally, p16 IHC was used to differentiate high-grade squamous intraepithelial lesion, an HPV-associated squamous lesion of the lower anogenital tract , and as a surrogate marker for HPV testing in HPV-mediated oropharyngeal squamous cell carcinoma . In this context, it was meaningful if p16 expression was manifested as diffuse strong positive, and total loss of p16 would not play a role in the data interpretation. However, recently there has been some reports that the total loss of p16 expression is related to CDKN2A mutation, and it is argued that not only the diffuse strong positive expression but also the total loss of p16 expression should be regarded as an abnormal phenotype , . Moreover, it has been reported that total loss of p16 expression is more frequently seen in SCC-IP than in sIP and is a risk factor for the recurrence of sIP, although the mutational status of CDKN2A was not evaluated , . In this study, 7/9 (78%) samples with CDKN2A SNV showed total loss of p16. Conversely, when there was a total loss of p16, CDKN2A mutation was found in 7/14 (50%). Additionally, 3/6 (50%) of cIP showed total loss of p16 expression whereas none of the sIP did. Therefore, it is reasonable to consider the total loss of p16 as an aberrant expression and a predictive marker of malignant transformation. In addition, five samples that showed diffuse strong p16 expression had either high-risk HPV infection or RB1 mutation, which may explain the aberrant p16 expression without CDKN2A mutation. A recent meta-analysis demonstrated that high-risk HPV subtypes 16 and 18 infection was associated with increased risk of malignant transformation of IP . However, the prevalence of high-risk HPV in SCC-IP seems to be low, ranging from 0 to 25% , – . In this study, 1 out of 14 SCC-IP patients had high-risk HPV (type 16) and showed diffuse strong positive p16 expression, which implicated the role of high-risk HPV in the pathogenesis of SCC-IP. However, most other SCC-IP specimens did not express high-risk HPV infection. This result was similar to those shown in previous studies in Korea, which did not find HPV infection in any cIP specimen , . Further studies are needed to clarify the association between HPV infection and malignant transformation of IP. TMB is defined as the number of mutations per megabase , and whole exome sequencing is generally regarded as the gold standard for TMB measurement . The threshold for high tumor mutational burden (TMB-H) was 10/Mb in KEYNOTE-158 study, based on which the FDA has approved a PD-1 inhibitor, pembrolizumab, for all solid tumors with TMB greater than 10/Mb. Although it is still controversial whether the cut-off value of 10/Mb can be applied universally across all solid tumors , 1.25/Mb, the mean value of TMB of SCC in this study, is much lower than 10/Mb, the cut-off value. Even the highest TMB in this study was 2.66/Mb, which is still considerably low. Therefore, SCC-IP can be regarded as tumors with low TMB. As low-TMB tumors are not suitable candidates for immunotherapy, it is important to identify cIP before it transforms into SCC. Previous studies have reported frequent EGFR mutations in IP, especially exon 20 insertions – , , , . However, in this study, EGFR mutation was not found in any of the cases. This discrepancy can be explained from two points of view: the association with SCC, and geographical distribution. Sahnane et al. reported that EGFR mutation was less frequent in SCC-IP (30%) than in sIP (72), and EGFR -wild-type IP had higher tendency of malignant transformation than EGFR -mutated IP at 5-year follow-up . In this study, 27/33 (82%) samples are from SCC-IP, which might partially explain why all the samples were EGFR wild-type. In the aspect of geographical distribution, Yasukawa et al. reported the frequency of EGFR mutations to be 20%, 38%, and 0% in IP, dysplasia, and SCC-IP, respectively, in the samples from Hokkaido University Hospital, Japan , whereas Udager et. al reported the frequency of EGFR mutations to be 88% in the samples from University of Michigan, USA . As the EGFR mutation frequency differs significantly between Japan and USA, it can be assumed that there are difference in geographical distribution. However, Wang et al. reported a high frequency of EGFR mutations (78%) in Chinese patients, although the study included sIP only . Furthermore, Sasaki et al. reported that 90% of sIP and 88% of SCC-IP in Japanese patients harbored EGFR mutations , while Cabal et al. found EGFR exon 20 mutations in 38% of sIP and 50% of SCC-IP in Spanish patients . Therefore, the difference of the frequency of EGFR mutations cannot be explained by geographical distribution alone and further studies are needed. Nevertheless, this study has a few limitations. This was a retrospective study that included patients from a single institute; therefore, the number of patients was relatively small and we were unable to obtain peripheral blood lymphocytes. The normal mucosae that were used for sequencing were adjacent to IP or SCC and, therefore, may have already harbored some of the mutations of IP or SCC, potentially leading to false negative results. Moreover, the samples for separate sequencing of each component in SCC-IP were obtained synchronously, which may not directly reflect the time course of malignant transformation. However, we sought to compare the differences in genetic mutations between the regions of IP, dysplasia, and SCC tissue in the same patient. In malignant transformation, we suggest that there may be genetic evidence of the same spectrum. In addition, the synchronousness of each component may have some advantages in preoperative biopsy because performing p53 and p16 staining on the preoperative biopsy specimen can help determine the presence of coexisting SCC component and can be clinically helpful during surgical resection in deciding the extent of resection and the necessity of intraoperative frozen examination, etc. As mentioned above, there was no clinicopathologic difference between sIP and cIP; therefore, additional tests to differentiate between the two are of high importance. In conclusion, aberrant expression of p53 and/or p16 is indicative of genetic alterations of TP53 and CDKN2A , which could be used as a predictive marker of malignant transformation of IP to SCC. Supplementary Table S1. |
Knife wound or nosebleed—where does the blood at the crime scene come from? | a72db47c-c444-4ae4-ba5e-a0e7cb7724c1 | 10247842 | Forensic Medicine[mh] | Short tandem repeats (STRs) are commonly used in forensic casework for the identification of victims or perpetrators as well as the analysis of family relationships . In the majority of cases, STR analysis is sufficient to achieve the required results. Sometimes, however, additional information about the origin of the biological material is desired to reconstruct crime scenes and further elucidate course of events . So far, enzymatic or immunological methods, which are based on the presence of proteins, as well as microscopic detection methods have been routinely used for trace characterization, especially regarding blood and semen samples . However, these methods can neither identify other body fluids like vaginal secretion nor can they distinguish between venous/arterial blood and menstrual blood . In addition to proteins, RNA can be used to identify body fluids . However, ribonucleic acid is significantly less stable than DNA and easily degraded by several circumstances , whereas DNA is one of the most robust biological compounds that remains intact after long periods of exposure to light, heat, and humidity and still allows for genetic profiling . Therefore, an ideal method for identifying a type of secretion would be one that does not consume additional sample material and exploits the stability of the DNA . The analysis of cell-specific, differential methylation can be used as a method of secretion analysis, since cells can be distinguished from one another by their methylation pattern [ – ]. Specific CpGs in the context of body fluid analysis have already been described in the literature for the body secretions saliva, blood, semen, menstrual blood, and vaginal secretion [ – ]. The first forensic-based study to report differentially methylated genomic loci in venous blood, saliva, semen, skin epidermis, vaginal fluid, menstrual blood, and urine was done by Frumkin et al. . But the reproduction of their experiments failed in 2011 . In the following years, several different assays have been developed, comprising various CpGs for the discrimination of blood, saliva, semen, and vaginal fluid [ – ], later on additionally menstrual blood [ – ]. Kader et al. provide a good review of the body fluids that could have been identified by methylation analysis so far . But to the best of our knowledge, nasal discharge has never been investigated in this context. The proof of presence or absence of nose secretion or nose blood in a forensic trace can serve to confirm or refute a described crime scene scenario thus being of great help in reconstructing a crime scene scenario. This study aims to set up methylation assays for the identification of nasal secretions based on specific CpGs to distinguish not only nasal secretion but also blood derived from nose bleeding from other fluids including venous, arterial, and menstrual blood.
Samples The study included 182 samples of 67 adult individuals (age range 18–94 years) comprising 35 nasal mucus samples, 39 oral mucosa/saliva samples, 35 blood samples, 29 vaginal fluid samples, 21 menstruation blood samples, and 23 semen samples. No information was available about diseases or operations like vasectomy or hysterectomy. Samples were collected between 2021 and 2022 in the Institute of Legal Medicine, University Hospital Essen, Germany. Compliance with ethical standards All samples were obtained after informed consent and with approval of the Medical Ethics Committee at the University of Duisburg-Essen in accordance with the Declaration of Helsinki and national laws (ethic vote number: 21–9843-BO). Marker For a discrimination of nasal mucus, 27 CpGs associated to 19 different genes were chosen (Table ) which are described in context of air pollution or air pollution-induced asthma diseases in childhood [ – ]. Additionally, 27 CpG marker regions (several CpGs per amplicon) in genes associated with formation of tight junctions were selected (Table ) . DNA extraction, quantification, bisulfite conversion, amplification, and sequencing DNA extraction was performed using DNA IQ Casework Pro Kit and Casework Extraction Kit in the Maxwell 16® instrument according to the manufacturer’s instructions (Promega, Mannheim, Germany), resulting in an extraction volume of 50 μl. DNA concentration of samples was established by real-time PCR using the PowerQuant™ System (Promega) according to the manufacturer’s instructions providing a reproducible and reliable detection threshold at least down to 25 pg DNA . Using 2 μl DNA-containing solutions, each sample was analyzed in duplicates. Bisulfite conversion was performed applying MethylEdge Conversion System Kit (Promega) corresponding to the manufacturer’s instructions with an increased elution volume of 20 μl. An initial DNA amount of 50 ng was used in the conversion. DNA amplification of candidate CpGs for body fluid was done using PyroMark® PCR Kit following the manufacturer’s instructions, adapted to an increased number of 50 cycles (Qiagen, Hilden, Germany). One of the two PCR primers was biotinylated. Sequence analysis was established in a PyroMark® Q48 Autoprep instrument using the PyroMark® Q48 Advanced CpG Reagent Kit according to the manufacturer’s instructions (Qiagen) . In addition, strict attention was paid to the conditions during sequencing. For reliable results, the sequencer must be placed vibration-free and draught-free, the instrument has to be turned on at least half an hour before using, and the reagents must be at room temperature . Every sample and CpG site were analyzed at least twice.
The study included 182 samples of 67 adult individuals (age range 18–94 years) comprising 35 nasal mucus samples, 39 oral mucosa/saliva samples, 35 blood samples, 29 vaginal fluid samples, 21 menstruation blood samples, and 23 semen samples. No information was available about diseases or operations like vasectomy or hysterectomy. Samples were collected between 2021 and 2022 in the Institute of Legal Medicine, University Hospital Essen, Germany.
All samples were obtained after informed consent and with approval of the Medical Ethics Committee at the University of Duisburg-Essen in accordance with the Declaration of Helsinki and national laws (ethic vote number: 21–9843-BO).
For a discrimination of nasal mucus, 27 CpGs associated to 19 different genes were chosen (Table ) which are described in context of air pollution or air pollution-induced asthma diseases in childhood [ – ]. Additionally, 27 CpG marker regions (several CpGs per amplicon) in genes associated with formation of tight junctions were selected (Table ) .
DNA extraction was performed using DNA IQ Casework Pro Kit and Casework Extraction Kit in the Maxwell 16® instrument according to the manufacturer’s instructions (Promega, Mannheim, Germany), resulting in an extraction volume of 50 μl. DNA concentration of samples was established by real-time PCR using the PowerQuant™ System (Promega) according to the manufacturer’s instructions providing a reproducible and reliable detection threshold at least down to 25 pg DNA . Using 2 μl DNA-containing solutions, each sample was analyzed in duplicates. Bisulfite conversion was performed applying MethylEdge Conversion System Kit (Promega) corresponding to the manufacturer’s instructions with an increased elution volume of 20 μl. An initial DNA amount of 50 ng was used in the conversion. DNA amplification of candidate CpGs for body fluid was done using PyroMark® PCR Kit following the manufacturer’s instructions, adapted to an increased number of 50 cycles (Qiagen, Hilden, Germany). One of the two PCR primers was biotinylated. Sequence analysis was established in a PyroMark® Q48 Autoprep instrument using the PyroMark® Q48 Advanced CpG Reagent Kit according to the manufacturer’s instructions (Qiagen) . In addition, strict attention was paid to the conditions during sequencing. For reliable results, the sequencer must be placed vibration-free and draught-free, the instrument has to be turned on at least half an hour before using, and the reagents must be at room temperature . Every sample and CpG site were analyzed at least twice.
Marker selection In order to find nasal mucus markers, specifically regulated CpGs had to be found. Since no CpGs were mentioned in the literature in the context of body fluid identification and nasal mucus, it was decided to investigate CpGs in which methylation pattern changes have been described after NO x and air pollution exposure. In all industrialized countries all over the world, people’s nasal mucosa is more or less constantly exposed to exhaust gases . Therefore, changes due to this exposure could be a unique feature in the nasal mucosa leading to a distinguishable methylation pattern. Additionally, tight junctions forming cell–cell contacts are abundant in mucosa so that genes involved in the forming of these characteristic features may show different methylation patterns between tissues with and without tight junctions. Genes chosen for analysis are displayed in Table . Reliability of data Due to the demand of downstream methods, especially bisulfite conversion, all samples included in this study had a DNA concentration between 2.5 ng/μl and 50 ng/μl. For all markers, identically prepared samples with regard to extraction method or bisulfite treatment were used so that an impact of incomplete bisulfite conversion problems can be excluded. Amplification and pyrosequencing could be successfully demonstrated for every locus included in this study. Duplicate analysis of samples showed a maximum deviation in methylation rate of 5%. Nasal sample identification A DNA methylation marker that allows traces to be assigned to specific cell or tissue types should ideally show hypermethylation (> 90%) in the target and hypomethylation (< 10%) in the nontarget or vice versa . To determine the suitability of the markers chosen for this study, all 54 CpG markers were analyzed in saliva, blood, and nasal secretion samples. Here, twelve of the 54 CpGs showed no amplicon after amplification (N15, N18, N25, N35, N36, N40, N41, N43, N45, N46, N49, and N54), for two of the 54 CpG markers it was not possible to design a working assay (N13 and N24), and for one of the 54 CpG markers sequencing of the desired fragment was not possible (N38). Additionally, 35 of the 54 CpG markers showed no difference in methylation percentage between saliva and nasal secretion or blood and nasal secretion. Strikingly, the associated genes of 18 of these 35 markers are often involved in signal transduction. Consequently, all 50 CpG markers mentioned above were omitted from further studies. In the four remaining markers N2 (cg23602092), N10 (cg09080874), N21 (cg16518142), and N27 (cg20864568), DNA methylation percentage in nasal secretion varied between 11 and 26% (N2), 61% and 88% (N10), 38% and 99% (N21), and 18% and 52% (N27), respectively (Table ). In addition to the determination of methylation levels in saliva and blood, experiments with these four markers in vaginal secretion, menstrual blood, and semen samples were conducted. Methylation range of marker N2 demonstrated a small overlap with methylation results of blood and semen samples and a total overlap with vaginal secretion (Fig. A). Similarly, a small overlap of methylation results of nasal secretion to methylation range of blood samples and a total overlap to results of menstrual blood could be seen in CpG marker N10 (Fig. B). Therefore, no distinct cut-off value clearly discriminating nasal secretion from other body fluids could be determined for markers N2 and N10. CpG marker N21 showed the greatest variance for methylation in nasal secretion/blood (38%–99%; mean 64%, standard deviation 18%) (Fig. C). Regarding saliva, blood, and semen samples, this marker demonstrated hypermethylation with mean values > 90%, whereas methylation results from vaginal secretion and menstrual blood varied between 68 and 94%. These results enabled us to set a cut-off value at 65%; every unknown sample with a N21 methylation rate lower than 65% can be clearly identified as nasal secretion/blood and discriminated from other secretions. In the samples included in this study, such an identification was possible for 22 samples out of a total of 35 analyzed nasal samples (regardless of whether they were secretions or blood) corresponding to 63% of all nasal samples. In CpG marker N27, nasal secretions/blood showed a methylation range between 18 and 52% (mean 33%, standard deviation 9%). Overlaps to saliva, blood, vaginal secretion, and menstrual blood methylation values could be seen (Fig. D). Therefore, by drawing two cut-off limits > 40% and < 70%, about 26% of all tested nasal samples could be identified and discriminated from other secretions. Workflow for unknown samples In unknown samples from a crime scene, it is very important to determine the sample’s origin. Usually, starting with a blood pretest (human) which is highly specific and sensitive is very useful. A positive result would confirm the presence of human blood cells, but could not distinguish between menstrual blood, nasal blood, and other sources. The application of the CpG assays N21 and N27 established in this study then determines the presence or absence of nasal blood. If no nasal epithelial cells could be found, further methylation analyses must be done to identify another source of blood cells. A negative result of the blood test excludes the presence of nasal blood, blood, and menstrual blood. Then, our CpG assays N10, N21, and N27 could be able to identify nasal secretion if present. Here, in N21, the amount of identifiable samples does not change regardless of the pretest result. Regarding CpG marker N10, all blood negative samples with < 75% methylation include nasal cells. In this study, this allowed identification of 53% of all analyzed nasal samples (Fig. B dashed line). For the CpG marker N27, a negative blood pretest increases the proportion of nasal secretion samples that can be clearly differentiated from saliva, vaginal secretion, and semen by reducing the lower threshold from 40 to 30% (Fig. D dashed line). As a result, the percentage of clearly identifiable nasal samples could be raised from 26 to 68% in this study. Moreover, the presence of seminal fluid could be determined directly, because its methylation values did not overlap with any other fluid in CpG marker N27. If nasal secretion and seminal fluid were excluded, a saliva pretest as well as further methylation analysis to identify vaginal secretion should be done. In summary, our workflow allows several outcomes in that an unknown sample may be identified directly (Fig. ), e.g., a sample with a negative blood pretest and a methylation value of 47% in CpG marker N21 definitely identifies nasal secretion. On the other hand, an unknown sample with a positive blood pretest and a methylation value of 35% in the CpG marker N27 could still be blood from any possible source and requires further analysis. An even greater problem is the identification of mixture samples of several body fluids . Since a blood pretest only detects the presence of blood cells, a positive test does not exclude the presence of cells from other sources. For example, a menstrual blood sample may also contain sperm. In order to be able to determine such mixtures as well, artificial composite samples must be created and analyzed in the next step. So further analyses (CpG marker assays and pretests) must be carried out in order to identify the composition of an unknown sample thus establishing a more complete workflow to identify and discriminate all seven body fluids and mixtures thereof in a forensic genetic context.
In order to find nasal mucus markers, specifically regulated CpGs had to be found. Since no CpGs were mentioned in the literature in the context of body fluid identification and nasal mucus, it was decided to investigate CpGs in which methylation pattern changes have been described after NO x and air pollution exposure. In all industrialized countries all over the world, people’s nasal mucosa is more or less constantly exposed to exhaust gases . Therefore, changes due to this exposure could be a unique feature in the nasal mucosa leading to a distinguishable methylation pattern. Additionally, tight junctions forming cell–cell contacts are abundant in mucosa so that genes involved in the forming of these characteristic features may show different methylation patterns between tissues with and without tight junctions. Genes chosen for analysis are displayed in Table .
Due to the demand of downstream methods, especially bisulfite conversion, all samples included in this study had a DNA concentration between 2.5 ng/μl and 50 ng/μl. For all markers, identically prepared samples with regard to extraction method or bisulfite treatment were used so that an impact of incomplete bisulfite conversion problems can be excluded. Amplification and pyrosequencing could be successfully demonstrated for every locus included in this study. Duplicate analysis of samples showed a maximum deviation in methylation rate of 5%.
A DNA methylation marker that allows traces to be assigned to specific cell or tissue types should ideally show hypermethylation (> 90%) in the target and hypomethylation (< 10%) in the nontarget or vice versa . To determine the suitability of the markers chosen for this study, all 54 CpG markers were analyzed in saliva, blood, and nasal secretion samples. Here, twelve of the 54 CpGs showed no amplicon after amplification (N15, N18, N25, N35, N36, N40, N41, N43, N45, N46, N49, and N54), for two of the 54 CpG markers it was not possible to design a working assay (N13 and N24), and for one of the 54 CpG markers sequencing of the desired fragment was not possible (N38). Additionally, 35 of the 54 CpG markers showed no difference in methylation percentage between saliva and nasal secretion or blood and nasal secretion. Strikingly, the associated genes of 18 of these 35 markers are often involved in signal transduction. Consequently, all 50 CpG markers mentioned above were omitted from further studies. In the four remaining markers N2 (cg23602092), N10 (cg09080874), N21 (cg16518142), and N27 (cg20864568), DNA methylation percentage in nasal secretion varied between 11 and 26% (N2), 61% and 88% (N10), 38% and 99% (N21), and 18% and 52% (N27), respectively (Table ). In addition to the determination of methylation levels in saliva and blood, experiments with these four markers in vaginal secretion, menstrual blood, and semen samples were conducted. Methylation range of marker N2 demonstrated a small overlap with methylation results of blood and semen samples and a total overlap with vaginal secretion (Fig. A). Similarly, a small overlap of methylation results of nasal secretion to methylation range of blood samples and a total overlap to results of menstrual blood could be seen in CpG marker N10 (Fig. B). Therefore, no distinct cut-off value clearly discriminating nasal secretion from other body fluids could be determined for markers N2 and N10. CpG marker N21 showed the greatest variance for methylation in nasal secretion/blood (38%–99%; mean 64%, standard deviation 18%) (Fig. C). Regarding saliva, blood, and semen samples, this marker demonstrated hypermethylation with mean values > 90%, whereas methylation results from vaginal secretion and menstrual blood varied between 68 and 94%. These results enabled us to set a cut-off value at 65%; every unknown sample with a N21 methylation rate lower than 65% can be clearly identified as nasal secretion/blood and discriminated from other secretions. In the samples included in this study, such an identification was possible for 22 samples out of a total of 35 analyzed nasal samples (regardless of whether they were secretions or blood) corresponding to 63% of all nasal samples. In CpG marker N27, nasal secretions/blood showed a methylation range between 18 and 52% (mean 33%, standard deviation 9%). Overlaps to saliva, blood, vaginal secretion, and menstrual blood methylation values could be seen (Fig. D). Therefore, by drawing two cut-off limits > 40% and < 70%, about 26% of all tested nasal samples could be identified and discriminated from other secretions.
In unknown samples from a crime scene, it is very important to determine the sample’s origin. Usually, starting with a blood pretest (human) which is highly specific and sensitive is very useful. A positive result would confirm the presence of human blood cells, but could not distinguish between menstrual blood, nasal blood, and other sources. The application of the CpG assays N21 and N27 established in this study then determines the presence or absence of nasal blood. If no nasal epithelial cells could be found, further methylation analyses must be done to identify another source of blood cells. A negative result of the blood test excludes the presence of nasal blood, blood, and menstrual blood. Then, our CpG assays N10, N21, and N27 could be able to identify nasal secretion if present. Here, in N21, the amount of identifiable samples does not change regardless of the pretest result. Regarding CpG marker N10, all blood negative samples with < 75% methylation include nasal cells. In this study, this allowed identification of 53% of all analyzed nasal samples (Fig. B dashed line). For the CpG marker N27, a negative blood pretest increases the proportion of nasal secretion samples that can be clearly differentiated from saliva, vaginal secretion, and semen by reducing the lower threshold from 40 to 30% (Fig. D dashed line). As a result, the percentage of clearly identifiable nasal samples could be raised from 26 to 68% in this study. Moreover, the presence of seminal fluid could be determined directly, because its methylation values did not overlap with any other fluid in CpG marker N27. If nasal secretion and seminal fluid were excluded, a saliva pretest as well as further methylation analysis to identify vaginal secretion should be done. In summary, our workflow allows several outcomes in that an unknown sample may be identified directly (Fig. ), e.g., a sample with a negative blood pretest and a methylation value of 47% in CpG marker N21 definitely identifies nasal secretion. On the other hand, an unknown sample with a positive blood pretest and a methylation value of 35% in the CpG marker N27 could still be blood from any possible source and requires further analysis. An even greater problem is the identification of mixture samples of several body fluids . Since a blood pretest only detects the presence of blood cells, a positive test does not exclude the presence of cells from other sources. For example, a menstrual blood sample may also contain sperm. In order to be able to determine such mixtures as well, artificial composite samples must be created and analyzed in the next step. So further analyses (CpG marker assays and pretests) must be carried out in order to identify the composition of an unknown sample thus establishing a more complete workflow to identify and discriminate all seven body fluids and mixtures thereof in a forensic genetic context.
In this study, it was possible to identify nasal mucosa-specific CpG markers and to set up methylation assays for the identification or discrimination of nasal samples. Even if an unambiguous determination of nasal secretion is not possible in 100% of samples, the results obtained so far are applicable to legally relevant questions in many cases. By optimizing or extending our workflow with additional CpG markers specific for other secretions, the unambiguously determinable proportion of unknown secretion samples can be increased.
|
Long-term outcomes of a paediatric quality improvement project in Central Asia: changes take time, time for a change | 7976c33a-44c5-4e2f-b4d0-940f6a289ad5 | 11950901 | Pediatrics[mh] | Study design We conducted an interrupted time series analysis. Between September 2021 and March 2023, two research teams travelled to the hospitals that participated in the WHO 2012–2014 quality improvement (QI) project (hereafter referred to as intervention hospitals) and to hospitals not included in the project (hereafter referred to as control hospitals) for retrospective data collection from medical records. The researchers reviewed medical records of children hospitalised during three time periods: 2012 (prior to the start of the QI project), 2015 (at the end of the QI project), and 2021 (for assessment of long-term benefits of the QI project). None of the researchers in this study had participated in the implementation of the QI project. Hospital selection The hospital and medical records selection process is summarised in Figure S1 in the . In the Kyrgyz Republic, 10 hospitals were selected in 2012 to be part of the QI project and 10 hospitals with no intervention were identified in the context of a cluster randomised controlled trial . Among those 20 hospitals, we collected data from five intervention hospitals and five control hospitals, which were purposely selected for geographical representation (Figure S2 in the ). In Tajikistan, data were collected from eight out of the 10 hospitals that were part of the 2012–2014 QI project. No control hospitals had been previously identified. Therefore, we collected data from the 10 hospitals selected for the upcoming (at the time of study proposal) QI project as control hospitals, as these hospitals had no interventions prior to 2021 (Figure S3 in the ). In both countries, the hospitals are district or regional public hospitals. Case record selection In each hospital, we randomly selected medical records of children hospitalised in 2012, 2015, and 2021. We selected the medical records by picking one out of every three (or every ten in the biggest hospitals) from the piles of paper-based records, until obtaining the required number of records. We chose the number of 40 medical records to review based on previous similar work . However, this number had to be reduced to 20 in some hospitals in Tajikistan due to logistical and time constraints. From the control hospitals in Tajikistan, data were collected from patients hospitalised in 2021 in the context of a health system evaluation . Data from 2012 and 2015 were not collected in these hospitals initially and could not be collected later due to logistical constraints. Inclusion criteria We reviewed medical records of children 2–59 months of age hospitalised with a primary diagnosis of an acute respiratory infection (upper respiratory infection, pneumonia, acute bronchitis, acute bronchiolitis or other acute lower respiratory tract infection) or diarrhoea (acute gastroenteritis), as they are the most common causes of paediatric hospitalisation in both countries . Indicators, standards of care, and determination of unnecessary and unnecessarily prolonged hospitalisations Indicators were selected based on the key findings reported in the baseline and endline assessments of the WHO QI project : unnecessary hospitalisations, unnecessarily prolonged hospitalisations, use of pulse oximeter, and prescription of antimicrobials, oral rehydration salts (ORS), zinc, theophylline, and calcium gluconate. For the classification of each hospitalisation into necessary or unnecessary, the reference for standards of care was the WHO pocket book of hospital care for children as it has been adapted and adopted by the two countries as a reference manual for inpatient care, is broadly used in both countries and has previously been utilised in similar assessments . To determine whether hospitalisation was necessary or not, we reviewed medical records for clinical characteristics present at the time of admission for the primary condition leading to hospitalisation and compared them against standard of care. Children with necessary hospitalisation were further classified as having unnecessarily prolonged hospitalisation if all discharge criteria were met for more than 24 hours before discharge, without any new hospitalisation criteria ( ; Table S1 in the ). Data collection, management and analysis We extracted data from medical records including general patient characteristics, hospitalisation and discharge dates, primary and secondary diagnoses, oxygen saturation measurement, and all medications received during hospitalisation. These data together with the classification of necessary or unnecessary hospitalisation and prolonged hospitalisation were directly entered into a digital database using predefined answers in prepopulated dropdown menus where applicable (Excel file). Prior to data collection, data collectors were trained by the same researcher for both countries to ensure consistency and comparability between hospitals and countries. We calculated cluster-level summary statistics by study arm (intervention and control hospitals), country, and year. We analysed data using Stata 18.0 (Stata Corporation LLC, College Station, USA) and used Microsoft Excel for graphs.
We conducted an interrupted time series analysis. Between September 2021 and March 2023, two research teams travelled to the hospitals that participated in the WHO 2012–2014 quality improvement (QI) project (hereafter referred to as intervention hospitals) and to hospitals not included in the project (hereafter referred to as control hospitals) for retrospective data collection from medical records. The researchers reviewed medical records of children hospitalised during three time periods: 2012 (prior to the start of the QI project), 2015 (at the end of the QI project), and 2021 (for assessment of long-term benefits of the QI project). None of the researchers in this study had participated in the implementation of the QI project.
The hospital and medical records selection process is summarised in Figure S1 in the . In the Kyrgyz Republic, 10 hospitals were selected in 2012 to be part of the QI project and 10 hospitals with no intervention were identified in the context of a cluster randomised controlled trial . Among those 20 hospitals, we collected data from five intervention hospitals and five control hospitals, which were purposely selected for geographical representation (Figure S2 in the ). In Tajikistan, data were collected from eight out of the 10 hospitals that were part of the 2012–2014 QI project. No control hospitals had been previously identified. Therefore, we collected data from the 10 hospitals selected for the upcoming (at the time of study proposal) QI project as control hospitals, as these hospitals had no interventions prior to 2021 (Figure S3 in the ). In both countries, the hospitals are district or regional public hospitals.
In each hospital, we randomly selected medical records of children hospitalised in 2012, 2015, and 2021. We selected the medical records by picking one out of every three (or every ten in the biggest hospitals) from the piles of paper-based records, until obtaining the required number of records. We chose the number of 40 medical records to review based on previous similar work . However, this number had to be reduced to 20 in some hospitals in Tajikistan due to logistical and time constraints. From the control hospitals in Tajikistan, data were collected from patients hospitalised in 2021 in the context of a health system evaluation . Data from 2012 and 2015 were not collected in these hospitals initially and could not be collected later due to logistical constraints.
We reviewed medical records of children 2–59 months of age hospitalised with a primary diagnosis of an acute respiratory infection (upper respiratory infection, pneumonia, acute bronchitis, acute bronchiolitis or other acute lower respiratory tract infection) or diarrhoea (acute gastroenteritis), as they are the most common causes of paediatric hospitalisation in both countries .
Indicators were selected based on the key findings reported in the baseline and endline assessments of the WHO QI project : unnecessary hospitalisations, unnecessarily prolonged hospitalisations, use of pulse oximeter, and prescription of antimicrobials, oral rehydration salts (ORS), zinc, theophylline, and calcium gluconate. For the classification of each hospitalisation into necessary or unnecessary, the reference for standards of care was the WHO pocket book of hospital care for children as it has been adapted and adopted by the two countries as a reference manual for inpatient care, is broadly used in both countries and has previously been utilised in similar assessments . To determine whether hospitalisation was necessary or not, we reviewed medical records for clinical characteristics present at the time of admission for the primary condition leading to hospitalisation and compared them against standard of care. Children with necessary hospitalisation were further classified as having unnecessarily prolonged hospitalisation if all discharge criteria were met for more than 24 hours before discharge, without any new hospitalisation criteria ( ; Table S1 in the ).
We extracted data from medical records including general patient characteristics, hospitalisation and discharge dates, primary and secondary diagnoses, oxygen saturation measurement, and all medications received during hospitalisation. These data together with the classification of necessary or unnecessary hospitalisation and prolonged hospitalisation were directly entered into a digital database using predefined answers in prepopulated dropdown menus where applicable (Excel file). Prior to data collection, data collectors were trained by the same researcher for both countries to ensure consistency and comparability between hospitals and countries. We calculated cluster-level summary statistics by study arm (intervention and control hospitals), country, and year. We analysed data using Stata 18.0 (Stata Corporation LLC, College Station, USA) and used Microsoft Excel for graphs.
Overall, 2095 medical records were reviewed and included in the analysis. Their general characteristics are summarised in . In the Kyrgyz Republic, infants (2–11 months) accounted for 45.0 and 41.1% of the children across the intervention and control hospitals, respectively. In Tajikistan, 56.5 and 46.9% were infants across the intervention and control hospitals, respectively. In the Kyrgyz Republic, 60.9% of children in the intervention hospitals and 62.5% in the control hospitals were referred from primary healthcare or other hospitals. In Tajikistan, this was the case for 45.3 and 53.1% of children, respectively. In the Kyrgyz Republic, among the selected medical records, the most common primary diagnosis was diarrhoea, followed by pneumonia, acute bronchitis and bronchiolitis, and acute respiratory infections, as recorded by health care providers. In Tajikistan, the most common primary diagnosis at admission was acute respiratory infection, followed by diarrhoea, pneumonia, and acute bronchitis and acute bronchiolitis. Among other diagnoses documented during hospitalisation, anaemia was by far the most common in both countries, followed by ‘neurotoxicosis’ and malnutrition in Tajikistan. The main findings are summarised in Table S2 in the . Unnecessary and unnecessarily prolonged hospitalisations Standards of care: Children should be hospitalised only when and for the time that is strictly required . Findings: The proportion of unnecessary hospitalisations in the Kyrgyz Republic decreased from 43.7 to 21.6% in the intervention hospitals between the start and the end of the QI project, with sustained improvements in 2021 (24.1%) ( ; Table S3 in the ). In Tajikistan, the proportion of unnecessary hospitalisations slightly decreased from 48.8 to 44.9% over the QI project in intervention hospitals, remaining at 41.1% in 2021, similar to the situation in control hospitals (41.6% in 2021) ( ; Table S4–5 in the ). Unnecessarily prolonged hospitalisations decreased in both countries in intervention hospitals during the QI project, with sustainable improvements in 2021, which was not the case in control hospitals ( ; Table S2–5 in the ). Use of pulse oximeter Standards of care: Oxygen saturation must be checked with a pulse oximeter and recorded upon admission of all children with a respiratory condition, to guide the possible use of oxygen . Findings: The use of pulse oximeter among children with a primary diagnosis of a respiratory condition at admission increased in all the hospitals in the Kyrgyz Republic, while its use in 2021 remains very low in Tajikistan ( ; Table S2–5 in the ). Antimicrobials prescription during hospitalisation Standards of care: Antibiotic prescription is justified in children with pneumonia, but antibiotics are not needed for all children with acute bronchitis, acute bronchiolitis and acute respiratory infections . Children with diarrhoea (dysentery excluded) do not need antibiotics . Findings: Most children hospitalised with a respiratory infection received at least one antimicrobial during hospitalisation ( ; Table S2–5 in the ). In 2012, around 95% of children with diarrhoea (dysentery excluded) were prescribed antibiotics. By the end of the QI project, the proportion of unjustified antibiotic prescriptions in children with diarrhoea reduced in intervention hospitals to 59.0% (Kyrgyz Republic) and 41.0% (Tajikistan). Improved results were sustained in 2021 in both countries while unjustified prescription remained high (over 85%) in control hospitals ( ; Table S2–5 in the ). Oral rehydration salts (ORS) and zinc prescription in children with diarrhoea Standards of care: Children hospitalised with diarrhoea should receive ORS and zinc . Findings: The proportion of children with diarrhoea who were prescribed ORS increased between 2012 and 2015 in intervention hospitals in both countries but was only sustained in the Kyrgyz Republic. However, the proportion of ORS prescription in 2021 was higher in intervention hospitals than in control hospitals in the two countries ( ; Table S2–5 in the ). Prescription of zinc among children with diarrhoea increased considerably among intervention hospitals in Tajikistan, from 52,2% in 2012 to 83.5% in 2021, while zinc prescription remains almost nil in all the hospitals in the Kyrgyz Republic ( ; Table S2–5 in the ). Theophylline and calcium gluconate prescription in children with a respiratory condition Standards of care: Medications should be prescribed only when indicated. In both countries, the baseline assessment showed that children were commonly prescribed medications when they were not indicated or with no evidence of benefits, such as intravenous (IV) theophylline or calcium gluconate in children with a respiratory infection . Findings: This practice improved in both countries by the end of the QI project, both in intervention and control hospitals ( ; Table S2–5 in the ).
Standards of care: Children should be hospitalised only when and for the time that is strictly required . Findings: The proportion of unnecessary hospitalisations in the Kyrgyz Republic decreased from 43.7 to 21.6% in the intervention hospitals between the start and the end of the QI project, with sustained improvements in 2021 (24.1%) ( ; Table S3 in the ). In Tajikistan, the proportion of unnecessary hospitalisations slightly decreased from 48.8 to 44.9% over the QI project in intervention hospitals, remaining at 41.1% in 2021, similar to the situation in control hospitals (41.6% in 2021) ( ; Table S4–5 in the ). Unnecessarily prolonged hospitalisations decreased in both countries in intervention hospitals during the QI project, with sustainable improvements in 2021, which was not the case in control hospitals ( ; Table S2–5 in the ).
Standards of care: Oxygen saturation must be checked with a pulse oximeter and recorded upon admission of all children with a respiratory condition, to guide the possible use of oxygen . Findings: The use of pulse oximeter among children with a primary diagnosis of a respiratory condition at admission increased in all the hospitals in the Kyrgyz Republic, while its use in 2021 remains very low in Tajikistan ( ; Table S2–5 in the ).
Standards of care: Antibiotic prescription is justified in children with pneumonia, but antibiotics are not needed for all children with acute bronchitis, acute bronchiolitis and acute respiratory infections . Children with diarrhoea (dysentery excluded) do not need antibiotics . Findings: Most children hospitalised with a respiratory infection received at least one antimicrobial during hospitalisation ( ; Table S2–5 in the ). In 2012, around 95% of children with diarrhoea (dysentery excluded) were prescribed antibiotics. By the end of the QI project, the proportion of unjustified antibiotic prescriptions in children with diarrhoea reduced in intervention hospitals to 59.0% (Kyrgyz Republic) and 41.0% (Tajikistan). Improved results were sustained in 2021 in both countries while unjustified prescription remained high (over 85%) in control hospitals ( ; Table S2–5 in the ).
Standards of care: Children hospitalised with diarrhoea should receive ORS and zinc . Findings: The proportion of children with diarrhoea who were prescribed ORS increased between 2012 and 2015 in intervention hospitals in both countries but was only sustained in the Kyrgyz Republic. However, the proportion of ORS prescription in 2021 was higher in intervention hospitals than in control hospitals in the two countries ( ; Table S2–5 in the ). Prescription of zinc among children with diarrhoea increased considerably among intervention hospitals in Tajikistan, from 52,2% in 2012 to 83.5% in 2021, while zinc prescription remains almost nil in all the hospitals in the Kyrgyz Republic ( ; Table S2–5 in the ).
Standards of care: Medications should be prescribed only when indicated. In both countries, the baseline assessment showed that children were commonly prescribed medications when they were not indicated or with no evidence of benefits, such as intravenous (IV) theophylline or calcium gluconate in children with a respiratory infection . Findings: This practice improved in both countries by the end of the QI project, both in intervention and control hospitals ( ; Table S2–5 in the ).
Sustained improvement in paediatric care in the Kyrgyz Republic Findings from the Kyrgyz Republic show that focused training, supportive supervision, and provision of medicines, supplies, and equipment over a two-year period improved the quality of paediatric care in hospitals participating in the QI project, based on selected indicators. This improvement was sustained until 2021, seven years after the project ended. In the control hospitals in the Kyrgyz Republic, an improvement was also observed between 2012 and 2015. However, this was not sustained until 2021 for some of the indicators, including unnecessary hospitalisations and unnecessarily prolonged hospitalisations. The improvement in the control hospitals could be attributed to the Hawthorne effect, where doctors performed better knowing they were part of a study. It may also be due to a ripple effect, possibly caused by the partial replication of quality improvement approaches by other organisations in other hospitals, or by national supportive supervisors incorporating these methods into their on-the-job training curricula. Alternatively, it could be due to a trend that had reasons beyond the project, in which case however the regression to old practices after the end of the project could not be explained . Factors affecting quality of care The inadequate quality of care arises from multiple factors, including limited health worker knowledge, lack of motivation and support systems, shortages of essential medicines, vaccines, and equipment, as well as insufficient financing and leadership . These factors directly impeded the project's broader impact, as many improvements relied on robust infrastructure and consistent resource availability. For example, the provision of zinc and antibiotics aligned with international guidelines depends on supply chains, their prescription relies on health worker knowledge and context acceptance, and patient intake depends on the population’s trust in the health system as well as the affordability and accessibility of health care facilities. Evidence from large systematic reviews suggests that multifaceted interventions, which include infrastructure strengthening, training, supportive supervision, and various management techniques, can effectively improve health care providers' practices [ - ]. Determining whether effects diminish over time is essential to improving sustainability. However, little is known about the effectiveness in the long term. A systematic review of 37 studies conducted in LMICs found that effects varied according to the type of interventions chosen to improve health care providers’ practices . The median follow-up time was six months, which limits the ability to extrapolate these findings to longer term benefits. For training alone, effects tended to decline over time, while effects were maintained for group problem solving plus training and increased when the intervention consisted of group problem-solving alone. Group problem-solving strategies involve continuous quality improvement through ongoing cycles of planning (Plan), implementing (Do), monitoring (Check), and revising (Act) the strategy as it is executed. This approach, widely recognised as the PDCA or quality cycle, was the strategy employed in the QI project in Tajikistan and the Kyrgyz Republic . Contextual variations in LMICs Other research has pointed out that the effectiveness of quality improvement interventions varies considerably in LMICs depending on the context . In Tajikistan, the same interventions had less effect in improving quality of hospital care than in its neighbouring country. The quality of care improved in the hospitals that participated in the QI project for some indicators such as unnecessary hospitalisation (minor improvement), unnecessarily prolonged hospitalisation, antibiotic prescription in children with diarrhoea, and zinc prescription in children with diarrhoea. However, improvements were not sustained until 2021 for some of the indicators and there was no improvement for other indicators ( e.g . use of pulse oximeter in children with a respiratory condition). No control hospitals were studied in Tajikistan before (2012) and at the end (2015) of the QI project. Data collected in 2021 from hospitals not previously involved in the QI project show performance trends that do not consistently align with those of the intervention hospitals. Persistent challenges in Central Asia The indicators were selected based on the key findings from the baseline and endline assessments conducted by the WHO and national teams in 2012 and 2014 . However, these issues of unnecessary and lengthy hospital stays as well as excessive and ineffective treatment including unjustified antibiotics are not new to the region. A similar assessment of the quality of hospital care for children conducted in Kazakhstan, the Republic of Moldova and the Russian Federation in 2002 already highlighted these critical areas . Despite the efforts dedicated to improving these issues, these are persistent challenges in Central Asia . Findings showed sustainable improvements linked to the QI project in the Kyrgyz Republic and Tajikistan for some indicators. However, in 2021, intervention hospitals continued to have between one in four and one in two children with unnecessary hospitalisations, about half of the children with unnecessarily prolonged hospitalisations, and over half of the children with diarrhoea receiving unjustified antibiotics; the situation was worse in control hospitals. If improved practices were truly embedded in the health system, further improvements and outcomes at national scale, including the control hospitals, would be expected in the long term. Why are sustainable and better outcomes not achieved for these persistent challenges, despite several projects aimed at addressing them? System-wide actions for high-quality health systems Health care is delivered through health systems, which are complex adaptive systems functioning across multiple interconnected levels . This study highlights that sustainable quality improvement requires system-wide actions to address core structural issues. Micro-level improvement interventions, such as training of health care providers or quality improvement at facility level, tend to struggle to improve the underlying performance of the whole system. Achieving high-quality health systems requires structural actions at all levels of the system . First, sound governance with a quality-of-care vision is needed, supported by updated and enforced laws and regulations including for the private sector. Second, adequate investment in the health system is key, both in absolute and relative terms to GDP and government budget, with low out-of-pocket spending. Third, defining service delivery between hospitals and primary health care with clear referral pathways is crucial for achieving optimal health outcomes. Fourth, ensuring a well-trained, well-remunerated and well-respected workforce. Fifth, the population needs to be engaged, health-educated, and empowered for a responsible use of the health system, seeking evidence-based high-quality of care. Therefore, unless development projects include actions at all these levels or are embedded within a network of partners so that all levels are addressed, improving quality of care is unlikely to be efficient and sustainable. For example, if health professionals are trained on the management of children with diarrhoea, we could expect a decrease in the unjustified prescription of antibiotics for these children. However, if the health professionals in the private sector continue to prescribe antibiotics, if pharmaceutical companies incentivise antibiotic prescription, if there are no regulations preventing the over-the-counter sale of antibiotics, and if the population perceives antibiotic prescription as good practice, then training health professionals on the appropriate indications for antibiotics as a single intervention is unlikely to result in improved quality care. In this way, among the indicators that we selected for this study, a sustainable decrease in unnecessary hospitalisations and unnecessarily prolonged hospitalisations at the national level would require, at least, transformation of the health financing, strengthening of primary health care, and increased trust from the population in seeking care in PHC. In addition, enforcement of legislation around selling antibiotics and other medicines would be needed to decrease polypharmacy practices and antibiotics overuse. Zinc prescription is another clear example of the need for system-wide actions. In the Kyrgyz Republic, zinc is included in the national list of essential medicines and is produced by a pharmaceutical company in the capital since 2016. However, it is not available in most hospitals and pharmacies in the country. In Tajikistan, zinc is free of charge for children hospitalised with diarrhoea. According to the Ministry of Health of the Republic of Tajikistan, zinc is available in the pharmacies throughout the country since 2015 and has been distributed in hospitals and primary health care centres since before the start of the QI project, with no interruption in its administration (personal communication). Increasing the use of pulse oximeters and ORS prescriptions for children with diarrhoea, however, likely depends on improving health professionals' practices, provided these tools are available in health facilities and their use is acceptable to the population. Designing development projects While designing development projects aimed at improving quality of care, all stakeholders must be fully aware of the complexity of the health system and understand how it performs. The health system performance assessment (HSPA) framework for universal health coverage illustrates the already aforementioned relevant health system areas, the so-called functions (governance, financing, resource generation and service delivery), and the influence they have on each other as well as to intermediate objectives (effectiveness, safety, user experience, access, efficiency and equity of service delivery) and final goals (people-centredness, health improvement, financial protection, efficiency and equity of the health system) . High quality (intermediate objectives) is required for achieving the final goals of the health system but achieving high quality depends on the functionality of all the interlinked functions. Shift towards system-based solutions For a long time, most efforts have been focused on the intermediate objectives for improving the final goals. It is time to shift from small scale (micro-level) quality improvement interventions towards systems-based solutions. Changes take time. The challenge is that changes in health systems are often driven by independent development projects that are based on time-bound activities. These projects prioritise short-term outcomes that must be reported upon the completion of the (too short) projects to satisfy the donors. Longer timeframes in funding projects are surely needed for implementing effective and lasting results. In addition, consistent leadership with a long-term vision and commitment is key in the coordination and implementation of the different activities, under an overall strategic plan . Frequent turnover in leadership is likely to hamper continuity in the support required for this process . Finally, better coordination and transparency between international partners and the multiple actors are urgently needed to allow for the shift from independent micro-level interventions towards a comprehensive package of interventions at all levels. Limitations and strengths of the study The main limitation of this research is that it is based on quantitative data collected retrospectively from medical reports. As such, the findings rely on the veracity of the information recorded by health workers in those documents. There is a strong punitive culture both in the Kyrgyz Republic and Tajikistan, reflected in the medical practice by regular control visits in the hospitals with review of the medical records. Such practices are likely to lead, in some cases, to information in the medical records that do not reflect the reality, as health workers may record data to meet the criteria known to be controlled for avoiding punishment. This could result in an overestimation of the improvements linked to the QI project. A major strength of this study is the evaluation of outcomes seven years after the completion of the project, providing important long-term data. This significantly exceeds the longest follow-up period of 34 months reported in the 37 studies included in Arsenault's systematic review . The use of the same methods and data collection tools for both countries is another major strength of this study, allowing analysis of trends and comparison between countries. The endline assessment of the QI project was conducted in 2014 with rigorous methodology, but by the same team involved in the implementation of the interventions, leading to potential detection and reporting bias. In this study, data collection, analysis and interpretation were performed by an independent team.
Findings from the Kyrgyz Republic show that focused training, supportive supervision, and provision of medicines, supplies, and equipment over a two-year period improved the quality of paediatric care in hospitals participating in the QI project, based on selected indicators. This improvement was sustained until 2021, seven years after the project ended. In the control hospitals in the Kyrgyz Republic, an improvement was also observed between 2012 and 2015. However, this was not sustained until 2021 for some of the indicators, including unnecessary hospitalisations and unnecessarily prolonged hospitalisations. The improvement in the control hospitals could be attributed to the Hawthorne effect, where doctors performed better knowing they were part of a study. It may also be due to a ripple effect, possibly caused by the partial replication of quality improvement approaches by other organisations in other hospitals, or by national supportive supervisors incorporating these methods into their on-the-job training curricula. Alternatively, it could be due to a trend that had reasons beyond the project, in which case however the regression to old practices after the end of the project could not be explained .
The inadequate quality of care arises from multiple factors, including limited health worker knowledge, lack of motivation and support systems, shortages of essential medicines, vaccines, and equipment, as well as insufficient financing and leadership . These factors directly impeded the project's broader impact, as many improvements relied on robust infrastructure and consistent resource availability. For example, the provision of zinc and antibiotics aligned with international guidelines depends on supply chains, their prescription relies on health worker knowledge and context acceptance, and patient intake depends on the population’s trust in the health system as well as the affordability and accessibility of health care facilities. Evidence from large systematic reviews suggests that multifaceted interventions, which include infrastructure strengthening, training, supportive supervision, and various management techniques, can effectively improve health care providers' practices [ - ]. Determining whether effects diminish over time is essential to improving sustainability. However, little is known about the effectiveness in the long term. A systematic review of 37 studies conducted in LMICs found that effects varied according to the type of interventions chosen to improve health care providers’ practices . The median follow-up time was six months, which limits the ability to extrapolate these findings to longer term benefits. For training alone, effects tended to decline over time, while effects were maintained for group problem solving plus training and increased when the intervention consisted of group problem-solving alone. Group problem-solving strategies involve continuous quality improvement through ongoing cycles of planning (Plan), implementing (Do), monitoring (Check), and revising (Act) the strategy as it is executed. This approach, widely recognised as the PDCA or quality cycle, was the strategy employed in the QI project in Tajikistan and the Kyrgyz Republic .
Other research has pointed out that the effectiveness of quality improvement interventions varies considerably in LMICs depending on the context . In Tajikistan, the same interventions had less effect in improving quality of hospital care than in its neighbouring country. The quality of care improved in the hospitals that participated in the QI project for some indicators such as unnecessary hospitalisation (minor improvement), unnecessarily prolonged hospitalisation, antibiotic prescription in children with diarrhoea, and zinc prescription in children with diarrhoea. However, improvements were not sustained until 2021 for some of the indicators and there was no improvement for other indicators ( e.g . use of pulse oximeter in children with a respiratory condition). No control hospitals were studied in Tajikistan before (2012) and at the end (2015) of the QI project. Data collected in 2021 from hospitals not previously involved in the QI project show performance trends that do not consistently align with those of the intervention hospitals.
The indicators were selected based on the key findings from the baseline and endline assessments conducted by the WHO and national teams in 2012 and 2014 . However, these issues of unnecessary and lengthy hospital stays as well as excessive and ineffective treatment including unjustified antibiotics are not new to the region. A similar assessment of the quality of hospital care for children conducted in Kazakhstan, the Republic of Moldova and the Russian Federation in 2002 already highlighted these critical areas . Despite the efforts dedicated to improving these issues, these are persistent challenges in Central Asia . Findings showed sustainable improvements linked to the QI project in the Kyrgyz Republic and Tajikistan for some indicators. However, in 2021, intervention hospitals continued to have between one in four and one in two children with unnecessary hospitalisations, about half of the children with unnecessarily prolonged hospitalisations, and over half of the children with diarrhoea receiving unjustified antibiotics; the situation was worse in control hospitals. If improved practices were truly embedded in the health system, further improvements and outcomes at national scale, including the control hospitals, would be expected in the long term. Why are sustainable and better outcomes not achieved for these persistent challenges, despite several projects aimed at addressing them?
Health care is delivered through health systems, which are complex adaptive systems functioning across multiple interconnected levels . This study highlights that sustainable quality improvement requires system-wide actions to address core structural issues. Micro-level improvement interventions, such as training of health care providers or quality improvement at facility level, tend to struggle to improve the underlying performance of the whole system. Achieving high-quality health systems requires structural actions at all levels of the system . First, sound governance with a quality-of-care vision is needed, supported by updated and enforced laws and regulations including for the private sector. Second, adequate investment in the health system is key, both in absolute and relative terms to GDP and government budget, with low out-of-pocket spending. Third, defining service delivery between hospitals and primary health care with clear referral pathways is crucial for achieving optimal health outcomes. Fourth, ensuring a well-trained, well-remunerated and well-respected workforce. Fifth, the population needs to be engaged, health-educated, and empowered for a responsible use of the health system, seeking evidence-based high-quality of care. Therefore, unless development projects include actions at all these levels or are embedded within a network of partners so that all levels are addressed, improving quality of care is unlikely to be efficient and sustainable. For example, if health professionals are trained on the management of children with diarrhoea, we could expect a decrease in the unjustified prescription of antibiotics for these children. However, if the health professionals in the private sector continue to prescribe antibiotics, if pharmaceutical companies incentivise antibiotic prescription, if there are no regulations preventing the over-the-counter sale of antibiotics, and if the population perceives antibiotic prescription as good practice, then training health professionals on the appropriate indications for antibiotics as a single intervention is unlikely to result in improved quality care. In this way, among the indicators that we selected for this study, a sustainable decrease in unnecessary hospitalisations and unnecessarily prolonged hospitalisations at the national level would require, at least, transformation of the health financing, strengthening of primary health care, and increased trust from the population in seeking care in PHC. In addition, enforcement of legislation around selling antibiotics and other medicines would be needed to decrease polypharmacy practices and antibiotics overuse. Zinc prescription is another clear example of the need for system-wide actions. In the Kyrgyz Republic, zinc is included in the national list of essential medicines and is produced by a pharmaceutical company in the capital since 2016. However, it is not available in most hospitals and pharmacies in the country. In Tajikistan, zinc is free of charge for children hospitalised with diarrhoea. According to the Ministry of Health of the Republic of Tajikistan, zinc is available in the pharmacies throughout the country since 2015 and has been distributed in hospitals and primary health care centres since before the start of the QI project, with no interruption in its administration (personal communication). Increasing the use of pulse oximeters and ORS prescriptions for children with diarrhoea, however, likely depends on improving health professionals' practices, provided these tools are available in health facilities and their use is acceptable to the population.
While designing development projects aimed at improving quality of care, all stakeholders must be fully aware of the complexity of the health system and understand how it performs. The health system performance assessment (HSPA) framework for universal health coverage illustrates the already aforementioned relevant health system areas, the so-called functions (governance, financing, resource generation and service delivery), and the influence they have on each other as well as to intermediate objectives (effectiveness, safety, user experience, access, efficiency and equity of service delivery) and final goals (people-centredness, health improvement, financial protection, efficiency and equity of the health system) . High quality (intermediate objectives) is required for achieving the final goals of the health system but achieving high quality depends on the functionality of all the interlinked functions.
For a long time, most efforts have been focused on the intermediate objectives for improving the final goals. It is time to shift from small scale (micro-level) quality improvement interventions towards systems-based solutions. Changes take time. The challenge is that changes in health systems are often driven by independent development projects that are based on time-bound activities. These projects prioritise short-term outcomes that must be reported upon the completion of the (too short) projects to satisfy the donors. Longer timeframes in funding projects are surely needed for implementing effective and lasting results. In addition, consistent leadership with a long-term vision and commitment is key in the coordination and implementation of the different activities, under an overall strategic plan . Frequent turnover in leadership is likely to hamper continuity in the support required for this process . Finally, better coordination and transparency between international partners and the multiple actors are urgently needed to allow for the shift from independent micro-level interventions towards a comprehensive package of interventions at all levels.
The main limitation of this research is that it is based on quantitative data collected retrospectively from medical reports. As such, the findings rely on the veracity of the information recorded by health workers in those documents. There is a strong punitive culture both in the Kyrgyz Republic and Tajikistan, reflected in the medical practice by regular control visits in the hospitals with review of the medical records. Such practices are likely to lead, in some cases, to information in the medical records that do not reflect the reality, as health workers may record data to meet the criteria known to be controlled for avoiding punishment. This could result in an overestimation of the improvements linked to the QI project. A major strength of this study is the evaluation of outcomes seven years after the completion of the project, providing important long-term data. This significantly exceeds the longest follow-up period of 34 months reported in the 37 studies included in Arsenault's systematic review . The use of the same methods and data collection tools for both countries is another major strength of this study, allowing analysis of trends and comparison between countries. The endline assessment of the QI project was conducted in 2014 with rigorous methodology, but by the same team involved in the implementation of the interventions, leading to potential detection and reporting bias. In this study, data collection, analysis and interpretation were performed by an independent team.
There is no quality of care without a functioning health system. Sustainable improvements in health care quality hinge on addressing foundational issues across governance, financing, resources and trust. This requires coordinated changes at all levels of the health system. But changes take time. Thus, it is time to change. The key message is clear: achieving high-quality health care systems demands a shift in focus from short-term outcomes to integrated, system-wide changes. Funding agencies and international partners must align their actions to achieve what truly matters: sustainable improvements towards high-quality health care systems.
Online Supplementary Document
|
Reliability of telemedicine for real-time paediatric ophthalmology consultations | d3911760-5ad0-4ab2-a641-6f6f6b1b9f8b | 9340009 | Ophthalmology[mh] | Worldwide, an estimated 19 million children aged 0–14 years are blind or visually impaired from mostly treatable or preventable conditions. Global prevalence of strabismus and amblyopia in children under 20 is 1.78% and 1.63%, respectively. The demand for paediatric eye care is immense, and even the wealthiest countries see geographic and socioeconomic disparities in access to care affecting timely diagnosis. This may result in lifetime visual impairment with economic costs to both the individual and society. Telemedicine may expand access to specialists, especially for underserved populations, and reduce the burden of disease. In ophthalmology, asynchronous telemedicine, in which images are sent to a remote specialist for interpretation, predominates. In paediatric populations, the use of asynchronous telemedicine for retinopathy of prematurity has been well validated. Until recently, real-time telemedicine had been less explored in ophthalmology, especially for a strictly paediatric population. At The Vision Center at Children’s Hospital Los Angeles (CHLA), 70%–80% of our patient population is underserved. We sought to leverage telemedicine in partnership with paediatric optometrists within our practice to provide timely ophthalmic consultations in community-based settings. Previously, we conducted prospective validation studies for each device used in our telemedicine examinations. In this investigation, we study agreement between telemedicine and in-person examinations for diagnosing and managing paediatric eye conditions.
Between February 2016 and April 2018, we conducted a prospective, non-inferiority study assessing agreement in diagnosis and management plan between telemedicine and in-person examinations in The Vision Center. A paediatric optometrist within our practice (JC-S) on the patient’s side conducted the telemedicine examination, while a paediatric ophthalmologist (SN) watched in real time and counselled the family on her findings and plan. To determine the accuracy of the telemedicine examination, she subsequently re-examined the patient in-person later the same day. Informed consent was obtained from a parent/legal guardian of all patient subjects. Assent was obtained from children aged ≥7 years, if cognitively able. Eligibility/study population Typically one (sometimes two) patient subject(s) and one parent/guardian were enrolled per family. Patient subjects were recruited from one of two sources: either they were newly referred to The Vision Center from an outside source (eg, paediatrician, other specialist or outside optometrist) (‘comprehensive examinations’) or they had been seen by our paediatric optometrist and required referral to a paediatric ophthalmologist for surgical consultation or diagnostic and management questions (‘consultation examinations’). Eligible patients included children <18 years who were able to participate in an age-appropriate manner. Platform The telemedicine system consisted of a Polycom RealPresence Group 500 video conferencing system (Polycom, San Jose, California, USA), Pivothead glasses (Pivothead, Denver, Colorado, USA), a Topcon SLD4 digital slit lamp with DC4 camera attachment (Topcon, Tokyo, Japan) and a Keeler Vantage Plus LED Digital Wireless Indirect Ophthalmoscope (Keeler, Windsor, UK). The Pivothead is a wireless, wearable device with a built-in high-definition camera at the nasal bridge and touch controls on the frame, allowing first person point-of-view video capture. Each device connects to a local computer to live-stream videos. To transmit videos to a remote examiner, the desktop was shared through an encrypted Polycom-to-Polycom video call over the hospital’s internal network at 4–6 Mbps. Videos were formatted most often at 720p (range 480–1080) and 29 frames/second for the Pivothead, 964p (range 900–964) and 20 frames/second (range 17–30) for the slit lamp, and always 576p and 25 frames/second for the indirect ophthalmoscope. Videos of the telemedicine system in use are provided in ( https://drive.google.com/file/d/1QBCs8i2HH9prxOGXHNGgNWJszGQmJxTZ/view?usp=sharing ). 10.1136/bjophthalmol-2020-318385.supp1 Supplementary data Design During recruitment, subjects were given the choice between a telemedicine (research) or in-person (standard) examination, on the premise that there was a shorter wait for research appointments by an order of months. Subjects were masked to the fact that if they opted for telemedicine, they would also see the ophthalmologist in-person. This was done to gauge willingness to participate in telemedicine examinations as well as trust in management plans (including surgical recommendations) attained via telemedicine. At the visit, the optometrist obtained history, performed preliminary testing (vision, stereoacuity, intraocular pressure) and conducted the telemedicine examination. At the outset of the call, the optometrist presented the patient, stating her specific question or the reason for referral from the outside provider. The optometrist directed the ophthalmologist to areas of interest, while the latter viewed and guided the examination, from a remote but nearby location, and recorded diagnoses with ICD-10 codes and management plans for each diagnosis. Non-medical diagnoses such as refractive errors were not included. There was opportunity for dialogue between the optometrist and ophthalmologist, but interpretation of the examination findings was left up to the ophthalmologist. Patient counselling and consenting for surgery were conducted via telemedicine. Afterward, the ophthalmologist performed an in-person examination and again recorded diagnoses and management plans. Examinations were modified depending on whether the patient required a dilated fundus examination. Patients referred by the optometrist for a specific question (consultation examinations) were not dilated unless clinically indicated. In these cases, strabismus measurements were verified in-person immediately following the telemedicine call. New referrals from an outside source underwent a comprehensive examination including sensorimotor, slit lamp if age appropriate and dilated indirect ophthalmoscopy (comprehensive examinations), all via telemedicine. Comprehensive examinations were conducted in two parts: at the completion of the undilated examination, the optometrist ended the call, dilated the patient and performed cycloplegic refraction. When finished, she called the ophthalmologist back for the dilated examination. Strabismus measurements, which cannot be accurately performed post-dilation, were not re-checked in-person following comprehensive telemedicine examinations. All strabismus measurements during consultation and follow-up telemedicine visits, including pre- and post-op examinations, were verified in-person. The ophthalmologist, optometrist and parent/guardian completed surveys at different time points capturing demographics, patient and provider satisfaction, optometrist’s knowledge gain, technical challenges, duration of examination, whether or not the patient consented to surgery via telemedicine, and duration of patient’s commute and time missed from school and work. Patient satisfaction was captured before subjects were unmasked to seeing the ophthalmologist in-person. This paper primarily focuses on clinical outcomes, while experience and access data will be reported separately. Outcome measures and data analysis The primary outcome measure was agreement in diagnoses and management plans between the telemedicine and in-person evaluations. Patients were classified as having no change, change in management plan or change in diagnosis. The threshold of non-inferiority was set at <1.5% for management plan or <15% for diagnosis discrepancies. In establishing these limits, an acceptable level of imprecision was balanced with the practicality of running this study in a realistic timeframe and with minimal disruption in a clinic already experiencing long waits for care. Although it would be ideal to establish conclusively that telemedicine was equal to in-person examinations, the number of cases required to compute a confidence interval (CI) around an outcome of zero management plan discrepancies grows exponentially as the upper bound approaches zero. For example, the upper limit of the CI would be <1% if 300 cases were evaluated and <0.5% for 600 cases. Given the practical challenges of evaluating 100–300 extra cases, an upper limit of 1.5% was chosen, which exceeds the upper limit of the CI (1.42%) if no discrepancies were found in 210 cases. This result would suggest that no more than 3 individuals out of 210 might result in a management plan discrepancy in 95% of study replications. If the current study identified a single case of management plan discrepancy, it would conclude that telemedicine is inferior, as the CI around the estimate of 1 case ranges from 0.01% to 2.62%. Diagnosis discrepancies were considered less important to the overall course of patient care, so a higher threshold of 15% was chosen, corresponding to 21 observed changes (CI=6.3%–14.9%). Secondary outcome measures were obtained from clinical evaluations and participant surveys, which encompassed the aforementioned metrics. For patients with strabismus, angle measurements and disease category were compared between telemedicine and in-person evaluations. For strabismus disease category, agreement was determined by weighted kappa (κ), where adjacent categorisations (eg, exotropia vs intermittent exotropia) were considered near matches, while distant categories were considered fair (eg, exotropia vs exophoria) or poor matches (eg, exotropia vs esotropia; see ). Interpretation of κ is based on the following scale: 0.0–0.20, no agreement; 0.21–0.39, minimal agreement; 0.40–0.59, weak; 0.60–0.79, moderate; 0.80–0.90, strong; >0.90, almost perfect. To quantify agreement of angle measurements, intraclass correlation coefficients (ICCs) were calculated. Because angle measurements were obtained from the same rater under two study conditions (in-person vs telemedicine), two-way mixed models evaluating absolute agreement of single scores were used (ICC(A,1) based on the McGraw and Wong (1996) nomenclature). Interpretation of ICC is as follows: <0.50, poor agreement; 0.50–0.75, moderate; 0.75–0.90, good;>0.90, excellent. All analyses were conducted using Stata/SE V.14.2 (StataCorp, College Station, Texas, USA). All p values <0.05 were considered statistically significant.
Typically one (sometimes two) patient subject(s) and one parent/guardian were enrolled per family. Patient subjects were recruited from one of two sources: either they were newly referred to The Vision Center from an outside source (eg, paediatrician, other specialist or outside optometrist) (‘comprehensive examinations’) or they had been seen by our paediatric optometrist and required referral to a paediatric ophthalmologist for surgical consultation or diagnostic and management questions (‘consultation examinations’). Eligible patients included children <18 years who were able to participate in an age-appropriate manner.
The telemedicine system consisted of a Polycom RealPresence Group 500 video conferencing system (Polycom, San Jose, California, USA), Pivothead glasses (Pivothead, Denver, Colorado, USA), a Topcon SLD4 digital slit lamp with DC4 camera attachment (Topcon, Tokyo, Japan) and a Keeler Vantage Plus LED Digital Wireless Indirect Ophthalmoscope (Keeler, Windsor, UK). The Pivothead is a wireless, wearable device with a built-in high-definition camera at the nasal bridge and touch controls on the frame, allowing first person point-of-view video capture. Each device connects to a local computer to live-stream videos. To transmit videos to a remote examiner, the desktop was shared through an encrypted Polycom-to-Polycom video call over the hospital’s internal network at 4–6 Mbps. Videos were formatted most often at 720p (range 480–1080) and 29 frames/second for the Pivothead, 964p (range 900–964) and 20 frames/second (range 17–30) for the slit lamp, and always 576p and 25 frames/second for the indirect ophthalmoscope. Videos of the telemedicine system in use are provided in ( https://drive.google.com/file/d/1QBCs8i2HH9prxOGXHNGgNWJszGQmJxTZ/view?usp=sharing ). 10.1136/bjophthalmol-2020-318385.supp1 Supplementary data
During recruitment, subjects were given the choice between a telemedicine (research) or in-person (standard) examination, on the premise that there was a shorter wait for research appointments by an order of months. Subjects were masked to the fact that if they opted for telemedicine, they would also see the ophthalmologist in-person. This was done to gauge willingness to participate in telemedicine examinations as well as trust in management plans (including surgical recommendations) attained via telemedicine. At the visit, the optometrist obtained history, performed preliminary testing (vision, stereoacuity, intraocular pressure) and conducted the telemedicine examination. At the outset of the call, the optometrist presented the patient, stating her specific question or the reason for referral from the outside provider. The optometrist directed the ophthalmologist to areas of interest, while the latter viewed and guided the examination, from a remote but nearby location, and recorded diagnoses with ICD-10 codes and management plans for each diagnosis. Non-medical diagnoses such as refractive errors were not included. There was opportunity for dialogue between the optometrist and ophthalmologist, but interpretation of the examination findings was left up to the ophthalmologist. Patient counselling and consenting for surgery were conducted via telemedicine. Afterward, the ophthalmologist performed an in-person examination and again recorded diagnoses and management plans. Examinations were modified depending on whether the patient required a dilated fundus examination. Patients referred by the optometrist for a specific question (consultation examinations) were not dilated unless clinically indicated. In these cases, strabismus measurements were verified in-person immediately following the telemedicine call. New referrals from an outside source underwent a comprehensive examination including sensorimotor, slit lamp if age appropriate and dilated indirect ophthalmoscopy (comprehensive examinations), all via telemedicine. Comprehensive examinations were conducted in two parts: at the completion of the undilated examination, the optometrist ended the call, dilated the patient and performed cycloplegic refraction. When finished, she called the ophthalmologist back for the dilated examination. Strabismus measurements, which cannot be accurately performed post-dilation, were not re-checked in-person following comprehensive telemedicine examinations. All strabismus measurements during consultation and follow-up telemedicine visits, including pre- and post-op examinations, were verified in-person. The ophthalmologist, optometrist and parent/guardian completed surveys at different time points capturing demographics, patient and provider satisfaction, optometrist’s knowledge gain, technical challenges, duration of examination, whether or not the patient consented to surgery via telemedicine, and duration of patient’s commute and time missed from school and work. Patient satisfaction was captured before subjects were unmasked to seeing the ophthalmologist in-person. This paper primarily focuses on clinical outcomes, while experience and access data will be reported separately.
The primary outcome measure was agreement in diagnoses and management plans between the telemedicine and in-person evaluations. Patients were classified as having no change, change in management plan or change in diagnosis. The threshold of non-inferiority was set at <1.5% for management plan or <15% for diagnosis discrepancies. In establishing these limits, an acceptable level of imprecision was balanced with the practicality of running this study in a realistic timeframe and with minimal disruption in a clinic already experiencing long waits for care. Although it would be ideal to establish conclusively that telemedicine was equal to in-person examinations, the number of cases required to compute a confidence interval (CI) around an outcome of zero management plan discrepancies grows exponentially as the upper bound approaches zero. For example, the upper limit of the CI would be <1% if 300 cases were evaluated and <0.5% for 600 cases. Given the practical challenges of evaluating 100–300 extra cases, an upper limit of 1.5% was chosen, which exceeds the upper limit of the CI (1.42%) if no discrepancies were found in 210 cases. This result would suggest that no more than 3 individuals out of 210 might result in a management plan discrepancy in 95% of study replications. If the current study identified a single case of management plan discrepancy, it would conclude that telemedicine is inferior, as the CI around the estimate of 1 case ranges from 0.01% to 2.62%. Diagnosis discrepancies were considered less important to the overall course of patient care, so a higher threshold of 15% was chosen, corresponding to 21 observed changes (CI=6.3%–14.9%). Secondary outcome measures were obtained from clinical evaluations and participant surveys, which encompassed the aforementioned metrics. For patients with strabismus, angle measurements and disease category were compared between telemedicine and in-person evaluations. For strabismus disease category, agreement was determined by weighted kappa (κ), where adjacent categorisations (eg, exotropia vs intermittent exotropia) were considered near matches, while distant categories were considered fair (eg, exotropia vs exophoria) or poor matches (eg, exotropia vs esotropia; see ). Interpretation of κ is based on the following scale: 0.0–0.20, no agreement; 0.21–0.39, minimal agreement; 0.40–0.59, weak; 0.60–0.79, moderate; 0.80–0.90, strong; >0.90, almost perfect. To quantify agreement of angle measurements, intraclass correlation coefficients (ICCs) were calculated. Because angle measurements were obtained from the same rater under two study conditions (in-person vs telemedicine), two-way mixed models evaluating absolute agreement of single scores were used (ICC(A,1) based on the McGraw and Wong (1996) nomenclature). Interpretation of ICC is as follows: <0.50, poor agreement; 0.50–0.75, moderate; 0.75–0.90, good;>0.90, excellent. All analyses were conducted using Stata/SE V.14.2 (StataCorp, College Station, Texas, USA). All p values <0.05 were considered statistically significant.
Clinical outcomes Two hundred ten patients (ages 0–17 years, median age=6 years, 3 sets of 2 siblings) and one parent/guardian per family participated. presents patient demographics. Of 210 initial encounters, 94 were comprehensive and 116 were consultation examinations. In total, 348 examinations were conducted. Sixty-six patients had at least one follow-up examination (median=2). Results are reported for initial visit only unless stated otherwise. The mean number of diagnoses per patient was 1.75 (range, 0–5). Sixty-two per cent of patients were primarily diagnosed with strabismus (n=131); other common primary diagnoses included eyelid abnormalities (n=12), glaucoma suspect (n=10) and conjunctival disorders (n=9) . Among primary diagnoses, we saw 135 (64.3%) motility findings, 22 (10.5%) oculoplastic, 21 (10.0%) posterior segment, 17 (8.1%) anterior segment and 15 (7.1%) ‘other’ or systemic disease. No primary diagnoses were changed between the telemedicine and in-person examinations, although two non-primary diagnoses were: a tiny non-visually significant lens opacity and a small intermittent vertical deviation. Both were noted in-person but not seen via telemedicine; neither affected management. No management plans, including surgical plans, were changed following in-person examination. Of 348 visits, we completed 310 Pivothead, 128 digital slit-lamp and 102 digital indirect ophthalmoscope examinations. The Pivothead was by far the most useful to the ophthalmologist in attaining diagnosis for sensorimotor findings and also nystagmus, nasolacrimal duct obstruction (NLDO) and eyelid findings. We performed slit-lamp biomicroscopy on 79 patients (59 dilated, 20 undilated; median age=9 years, age range=2–17 years) and gonioscopy once on an 8-year-old. All were successfully evaluated by the ophthalmologist via telemedicine. Comprehensive examinations versus consultation examinations The percentage of patients requiring surgery was slightly higher in the consultation compared with the comprehensive group . In the ophthalmologist’s estimation, more children in the consultation group had conditions warranting being seen by a paediatric ophthalmologist (78.4% vs 55.3% in the comprehensive group). The remaining patients either did not need to be seen at all or could have been seen by a qualified paediatric optometrist. Similarly, more children in the consultation group required follow-up care with ophthalmology (30.2% consultation vs 24.5% comprehensive) as opposed to follow-up exclusively with the optometrist or co-managed through telemedicine, although this difference may not be clinically significant. Patients with strabismus In pre-op and post-op patients with strabismus, excellent or almost perfect agreement between telemedicine and in-person examinations was observed for angle measurements (ICC=0.98–1.00) and disease categorisation (κ=0.94–1.00) . All agreement statistics were highly significant (p<0.0001). Measurement variations were minimal. Prism dioptres ranged from 1 to 70 in horizontal and 2–38 in vertical measurements . Surgical patients Sixty-two subjects (ages 0.5–17 years, median=6 years) had surgery, and three had more than one within 1 year of initial examination. Three surgeries were for NLDO; all others were for strabismus. Surgical measurements were obtained by Krimsky or Hirschberg in four patients, all others by alternate prism cover test. Almost all patients who consented for surgery at the initial visit (54/55) did so during the telemedicine examination, while masked to receiving an in-person examination. One subject who declined surgery during the telemedicine encounter changed his mind after seeing the surgeon in-person. Four patients declined entirely. Twenty patients consented for surgery at follow-ups; however, these subjects had already been unmasked to the study design. After the initial visit, 122 (58.1%) patients without active disease were deemed by the ophthalmologist to be appropriate to be seen in follow-up exclusively by the optometrist. Thirty (14.3%) with active disease requiring close follow-up could be co-managed through telemedicine. The remaining 58 (27.6%) required follow-up care with the ophthalmologist either for surgery (with pre-op and post-op visits via telemedicine, n=55) or for referral to another specialist (glaucoma or retina, n=3). Process and experience outcomes Forty of 348 (11.5%) examinations had some delay due to equipment challenges. Most delays lasted 5–10 min and involved the Pivothead or Polycom. Averaged across all encounters, this is approximately 1 min extra per encounter. In all 348 examinations, the ophthalmologist was able to hear and see the patient and visualise areas of interest. Nearly all parents felt comfortable with the quality of the telemedicine examination (98.5%) and reported they would participate in another one in the future (97.1%).
Two hundred ten patients (ages 0–17 years, median age=6 years, 3 sets of 2 siblings) and one parent/guardian per family participated. presents patient demographics. Of 210 initial encounters, 94 were comprehensive and 116 were consultation examinations. In total, 348 examinations were conducted. Sixty-six patients had at least one follow-up examination (median=2). Results are reported for initial visit only unless stated otherwise. The mean number of diagnoses per patient was 1.75 (range, 0–5). Sixty-two per cent of patients were primarily diagnosed with strabismus (n=131); other common primary diagnoses included eyelid abnormalities (n=12), glaucoma suspect (n=10) and conjunctival disorders (n=9) . Among primary diagnoses, we saw 135 (64.3%) motility findings, 22 (10.5%) oculoplastic, 21 (10.0%) posterior segment, 17 (8.1%) anterior segment and 15 (7.1%) ‘other’ or systemic disease. No primary diagnoses were changed between the telemedicine and in-person examinations, although two non-primary diagnoses were: a tiny non-visually significant lens opacity and a small intermittent vertical deviation. Both were noted in-person but not seen via telemedicine; neither affected management. No management plans, including surgical plans, were changed following in-person examination. Of 348 visits, we completed 310 Pivothead, 128 digital slit-lamp and 102 digital indirect ophthalmoscope examinations. The Pivothead was by far the most useful to the ophthalmologist in attaining diagnosis for sensorimotor findings and also nystagmus, nasolacrimal duct obstruction (NLDO) and eyelid findings. We performed slit-lamp biomicroscopy on 79 patients (59 dilated, 20 undilated; median age=9 years, age range=2–17 years) and gonioscopy once on an 8-year-old. All were successfully evaluated by the ophthalmologist via telemedicine.
The percentage of patients requiring surgery was slightly higher in the consultation compared with the comprehensive group . In the ophthalmologist’s estimation, more children in the consultation group had conditions warranting being seen by a paediatric ophthalmologist (78.4% vs 55.3% in the comprehensive group). The remaining patients either did not need to be seen at all or could have been seen by a qualified paediatric optometrist. Similarly, more children in the consultation group required follow-up care with ophthalmology (30.2% consultation vs 24.5% comprehensive) as opposed to follow-up exclusively with the optometrist or co-managed through telemedicine, although this difference may not be clinically significant.
In pre-op and post-op patients with strabismus, excellent or almost perfect agreement between telemedicine and in-person examinations was observed for angle measurements (ICC=0.98–1.00) and disease categorisation (κ=0.94–1.00) . All agreement statistics were highly significant (p<0.0001). Measurement variations were minimal. Prism dioptres ranged from 1 to 70 in horizontal and 2–38 in vertical measurements .
Sixty-two subjects (ages 0.5–17 years, median=6 years) had surgery, and three had more than one within 1 year of initial examination. Three surgeries were for NLDO; all others were for strabismus. Surgical measurements were obtained by Krimsky or Hirschberg in four patients, all others by alternate prism cover test. Almost all patients who consented for surgery at the initial visit (54/55) did so during the telemedicine examination, while masked to receiving an in-person examination. One subject who declined surgery during the telemedicine encounter changed his mind after seeing the surgeon in-person. Four patients declined entirely. Twenty patients consented for surgery at follow-ups; however, these subjects had already been unmasked to the study design. After the initial visit, 122 (58.1%) patients without active disease were deemed by the ophthalmologist to be appropriate to be seen in follow-up exclusively by the optometrist. Thirty (14.3%) with active disease requiring close follow-up could be co-managed through telemedicine. The remaining 58 (27.6%) required follow-up care with the ophthalmologist either for surgery (with pre-op and post-op visits via telemedicine, n=55) or for referral to another specialist (glaucoma or retina, n=3).
Forty of 348 (11.5%) examinations had some delay due to equipment challenges. Most delays lasted 5–10 min and involved the Pivothead or Polycom. Averaged across all encounters, this is approximately 1 min extra per encounter. In all 348 examinations, the ophthalmologist was able to hear and see the patient and visualise areas of interest. Nearly all parents felt comfortable with the quality of the telemedicine examination (98.5%) and reported they would participate in another one in the future (97.1%).
At The Vision Center, our motivation to study telemedicine was driven by an access problem, with the demand for paediatric ophthalmology care outstripping supply, and waits for a new appointment running over 5 months. Telemedicine offers considerable opportunity to address workforce shortages, links specialists with primary providers in the management of complex patients and speed access to care. Confidence with telemedicine among eye care providers is increasing, but at least one-third continue to feel “not at all confident” in remote screening for eye care. This underscores a need for research in clinical validation and also into participant experience. Our study demonstrates the non-inferiority of real-time telemedicine relative to in-person examinations for diagnosing and managing paediatric ophthalmic conditions. In contrast to earlier studies with older technology, the ophthalmologist was able to make accurate diagnoses, plans and measurements in virtually every telemedicine encounter. This held true even with video resolution as low as 480p with the Pivothead. It should be noted that we did not examine any children with anterior uveitis and cannot comment on the ability to detect anterior chamber inflammation by digital slit lamp. Also, only children who were able to participate in an age-appropriate manner were eligible to enrol, excluding many with developmental delays. As for technical difficulties, indirect ophthalmoscopy took longer via telemedicine, in part because the working distance must be adjusted due to a discrepancy between what is seen and what is streamed. The digital indirect also uses a brighter LED light, making it difficult for young children. Furthermore, equipment delays added slightly more time to telemedicine exams. The number of patients consenting for surgery during the telemedicine encounter indicates trust in the platform. Only five patients (8.1%) declined surgery at the initial telemedicine visit. For three of those, surgery was reconstructive and not for visual function (one changed his mind following the in-person examination). In another case, the patient already had surgery and the parent wanted to try a different treatment approach first. In the fifth case, the parents declined surgery recommended to improve binocularity, even after seeing the ophthalmologist in-person. Unexpectedly, the ophthalmologist’s surgical volume increased 25% from the same period the year prior, despite closing regular, much higher (average 4×) volume clinics to conduct the study. We suspect the increase was due in part to the study presenting the optometrist with a faster route to get surgical patients seen. Whatever the cause, this suggests possible improved access for surgical patients, although this should be evaluated in future studies. Real-time telemedicine is optimal for consultation examinations addressing a specific concern of the referring provider, rather than comprehensive examinations. While the ophthalmologist felt almost a third (31.9%) of patients overall did not need to be seen by an ophthalmologist, the difference between comprehensive (44.7%) and consultation (21.6%) subjects is striking. This—in combination with the fact that surgical volumes were similar between the comprehensive (28.7%) and consultation (30.2%) examinations—suggests the consultation group may have had a higher number of medical diagnoses requiring advanced care. This difference in complexity might also explain the higher rate of consultation examination patients who required follow-up care with an ophthalmologist after the initial visit (30.2% consultation vs 24.5% comprehensive). The fact that close to half (44.7%) of comprehensive examinations did not require ophthalmology may support a care model in which paediatric optometrists see a majority of new patients first, with telemedicine offering targeted ophthalmology consultations as needed, although further study testing this hypothesis is required, particularly in diverse practice settings. Such a model could enable high-volume clinics to shift stable, low-acuity and/or postoperative patients out to be managed by paediatric optometrists from the same practice—ideally in a setting closer to the patient’s home—thereby freeing ophthalmologists to focus on acute or surgical patients. At CHLA, our optometrists travel to remote locations where they can manage our more stable patients, while our ophthalmologists block time for telemedicine consults, which are usually surgical cases pre-screened by the optometrist. This telemedicine model is advantageous when patients live far from the surgical practice or when wait times are excessively long. The model holds less value in settings where patients face fewer barriers to care. The cost of equipment could present a challenge to scaling; however, the most useful piece of equipment in this study—the Pivothead—is relatively inexpensive (<US$1000). We also foresee that software video-conferencing capabilities will improve sufficiently to replace the need for the more expensive hard-wired conferencing system in the near future. The principal limitation of this study was that examinations were performed by one ophthalmologist and one optometrist who became adept at working together, operating the equipment and troubleshooting glitches. There is a learning curve, and less experienced providers may take time to become proficient, possibly impacting adoption. The optometrist must be trained in paediatric optometry and highly competent with cycloplegic retinoscopy, as this component is not repeated by the ophthalmologist and can greatly impact managment. Future research should examine the feasibility of someone other than a paediatric optometrist on the patient’s end, such as an orthoptist, technician or paediatrician, as availability of pediatrics-trained optometrists could be a limiting factor. Another limitation is confirmation bias, as the same ophthalmologist performed the telemedicine examination and the in-person examination. While the ophthalmologist was invested in study outcomes, this was outweighed by her interest in safe and correct patient care. Finally, this study was conducted entirely on our internal network. It would not be unreasonable to expect connectivity challenges at external sites with less robust broadband connection. This study demonstrates paediatric ophthalmic conditions can be reliably managed through real-time telemedicine. This model answers the specialist shortage, allows physicians to focus on surgical and medically complex patients, and helps ease access to care for underserved children.
|
PharmFrag: An Easy and Fast Multiplex Pharmacogenetics Assay to Simultaneously Analyze 9 Genetic Polymorphisms Involved in Response Variability of Anticancer Drugs | 3c51fbb5-af19-46c1-825f-102896b87df6 | 7766892 | Pharmacology[mh] | In the era of personalized medicine, cytotoxic anticancer drugs remain widely used to treat hematologic malignancies and solid tumors. An important interindividual variability in drug response can be observed with these therapeutic agents. Many chemotherapies have a narrow therapeutic range; therefore, a part of this suboptimal response can be explained by variations of drug blood concentrations. Regarding several cytotoxic agents, it was evidenced that genetic polymorphisms in genes encoding enzymes involved in their metabolism are associated with blood overexposure, leading to higher risk of toxicity . Genotyping these genes before treatment is a valuable strategy to prevent side effects and to predict individual response to drug therapy. Indeed, it allows the identification of patients who are carriers of allelic variants and who need dosage adjustment. This pharmacogenetic approach is strongly recommended by international clinical and pharmacological consortiums for chemotherapies, such as thiopurines (azathioprine, 6-mercaptopurine, thioguanine), irinotecan, and fluoropyrimidines (capecitabine and 5-fluorouracil) . Indeed, dosing algorithms are now available to help clinicians to individualize the prescription of these drugs, taking into account genotypes of TPMT , NUDT15, DPYD , and UGT1A1 , whose corresponding proteins metabolize thiopurines, fluoropyrimidines, and irinotecan, respectively. Thus, the implementation of pharmacogenetic assays may prevent the severe hematologic or digestive toxicities of these anticancer drugs. Routinely, laboratories only need to look for the most common and clinically relevant variants associated with the enzyme dysfunction. For TPMT, there are three main single-nucleotide polymorphisms (SNPs) of interest and one for NUDT15 . For DPYD , it is recommended that four SNPs be studied, and for UGT1A1 , a repetition in the TATA box of the promoter (allele *28) is the variant usually studied . The genotype–phenotype relationships of these genetic polymorphisms are presented in . Many genotyping methods have been reported to perform these pharmacogenetic analyses . First, low-throughput technologies or simplex methods are available to separately analyze the above-mentioned genetic polymorphisms (i.e., Sanger sequencing, PCR-RFLP, TaqMan TM genotyping assays). However, it can be challenging and cumbersome for some laboratories to use these methods to perform multiple gene analysis. To remove this hurdle, an alternative can be to use high-throughput technologies such as microarrays and next-generation sequencing (NGS). These methods allow multiplexing to analyze panels of many genes and many samples in a single experiment. Nevertheless, result interpretations require specialized skills (e.g., bioinformatics) and are not cost-effective below a certain threshold of samples to be analyzed. In this context, we aimed at developing and validating a fast, cost-effective, and easily implementable multiplex genotyping method suitable for analyzing a panel of nine variants involved in the pharmacogenetics of widely prescribed anticancer drugs.
2.1. Multiplex PCR Protocol Optimization The most important parameter of the multiplex PCR protocol is the primer concentration, which influences the intensity of the fluorescence and consequently the height of the peaks on fragment analysis. Initially, amplification reactions were performed using equal concentrations of the primers. Subsequently, the concentration of each primer was adjusted to give a more comparable peak height. Then, the primer concentrations were optimized to get signals at least 10 times higher than the background noise (raw intensity > 100). The number of PCR cycles and the DNA concentration were also adjusted. Eventually, the final procedure was chosen to obtain a ratio between intensities of the dyes FAM/HEX ≈ 1 for all polymorphism positions of interest. Moreover, the primers were designed with a substitution of the wild-type matrix at the -4 nucleotide of the 3′ flanking forward primers to reduce false positive genotyping and increase the assay specificity. A representative electropherogram of the migration of the fragments of interest is shown in . Besides, all the genetic polymorphisms analyzed are DNA substitutions (SNPs) except for UGT1A1 , which is a repetition of 6 TA nucleotides for the wild type or 7 TA for the variant. This implies that the variant forward primer UGT1A1 amplifies only the variant DNA and that the wild-type forward primer UGT1A1 amplifies the wild-type and variant DNA. Consequently, for the heterozygous genotype (TA6/TA7), high-resolution capillary electrophoresis was needed to distinguish the migration of two fragments associated with the same fluorescent dye and whose lengths were very close (only two base pairs different). Representative electropherograms of each genotype, for an SNP (e.g., TPMT rs1800460) or a repetition (e.g., UGT1A1 rs8175347), are illustrated in . 2.2. Validation of the Protocol 2.2.1. Repeatability Repeated analysis of the same internal control DNA samples (50 ng) gave the same genotype results on intraday experiments (duplicate) and interday experiments (6 days). 2.2.2. Accuracy A cohort of 187 DNA samples was screened by the multiplex protocol to check whether the genotyping results matched with the expected genotypes. All results were confirmed without false positive or false negative. The consistency of the genotyping results of these samples is reported in . 2.2.3. Robustness and Intersample Contamination The assay was designed to be performed on 50 ng of extracted DNA. However, the influence of lower and higher DNA amount on the performance of the analysis was assessed. The results are reported in . It appears that the genotypes can be accurately determined when DNA sample amounts range from 25 to 100 ng. Outside this range, the intensity of the signal does not meet the acceptance criteria to interpret migration fragment data (peak height < 100 units of intensity or saturation of the signal). No influence of the quality of the DNA was observed. Indeed, genotyping was successfully performed in our cohort, whose absorbance ratio ranged from 0.35 to 2.3 and 1.5 to 2.1 for ratios of 260/230 nm and 260/280 nm, respectively. Moreover, no intersample contamination was observed since no signal was detected at the position of the blank samples inserted between the DNA samples (data not shown). 2.2.4. Stability The stability of the ready-to-use pool of primer and fluorescent probes at the working conditions was validated for five freeze–thaw cycles. Indeed, the genotyping results were similar in an experiment performed with a freshly prepared pool and in another experiment performed with an aliquot frozen and thawed five times. Although the peak intensities of fragments were lower when a freeze-thawed reagent was used, they remained >100 intensity units . Thus, a pool of primers can be stored at −20 °C and used several times, which is very cost saving and convenient.
The most important parameter of the multiplex PCR protocol is the primer concentration, which influences the intensity of the fluorescence and consequently the height of the peaks on fragment analysis. Initially, amplification reactions were performed using equal concentrations of the primers. Subsequently, the concentration of each primer was adjusted to give a more comparable peak height. Then, the primer concentrations were optimized to get signals at least 10 times higher than the background noise (raw intensity > 100). The number of PCR cycles and the DNA concentration were also adjusted. Eventually, the final procedure was chosen to obtain a ratio between intensities of the dyes FAM/HEX ≈ 1 for all polymorphism positions of interest. Moreover, the primers were designed with a substitution of the wild-type matrix at the -4 nucleotide of the 3′ flanking forward primers to reduce false positive genotyping and increase the assay specificity. A representative electropherogram of the migration of the fragments of interest is shown in . Besides, all the genetic polymorphisms analyzed are DNA substitutions (SNPs) except for UGT1A1 , which is a repetition of 6 TA nucleotides for the wild type or 7 TA for the variant. This implies that the variant forward primer UGT1A1 amplifies only the variant DNA and that the wild-type forward primer UGT1A1 amplifies the wild-type and variant DNA. Consequently, for the heterozygous genotype (TA6/TA7), high-resolution capillary electrophoresis was needed to distinguish the migration of two fragments associated with the same fluorescent dye and whose lengths were very close (only two base pairs different). Representative electropherograms of each genotype, for an SNP (e.g., TPMT rs1800460) or a repetition (e.g., UGT1A1 rs8175347), are illustrated in .
2.2.1. Repeatability Repeated analysis of the same internal control DNA samples (50 ng) gave the same genotype results on intraday experiments (duplicate) and interday experiments (6 days). 2.2.2. Accuracy A cohort of 187 DNA samples was screened by the multiplex protocol to check whether the genotyping results matched with the expected genotypes. All results were confirmed without false positive or false negative. The consistency of the genotyping results of these samples is reported in . 2.2.3. Robustness and Intersample Contamination The assay was designed to be performed on 50 ng of extracted DNA. However, the influence of lower and higher DNA amount on the performance of the analysis was assessed. The results are reported in . It appears that the genotypes can be accurately determined when DNA sample amounts range from 25 to 100 ng. Outside this range, the intensity of the signal does not meet the acceptance criteria to interpret migration fragment data (peak height < 100 units of intensity or saturation of the signal). No influence of the quality of the DNA was observed. Indeed, genotyping was successfully performed in our cohort, whose absorbance ratio ranged from 0.35 to 2.3 and 1.5 to 2.1 for ratios of 260/230 nm and 260/280 nm, respectively. Moreover, no intersample contamination was observed since no signal was detected at the position of the blank samples inserted between the DNA samples (data not shown). 2.2.4. Stability The stability of the ready-to-use pool of primer and fluorescent probes at the working conditions was validated for five freeze–thaw cycles. Indeed, the genotyping results were similar in an experiment performed with a freshly prepared pool and in another experiment performed with an aliquot frozen and thawed five times. Although the peak intensities of fragments were lower when a freeze-thawed reagent was used, they remained >100 intensity units . Thus, a pool of primers can be stored at −20 °C and used several times, which is very cost saving and convenient.
Repeated analysis of the same internal control DNA samples (50 ng) gave the same genotype results on intraday experiments (duplicate) and interday experiments (6 days).
A cohort of 187 DNA samples was screened by the multiplex protocol to check whether the genotyping results matched with the expected genotypes. All results were confirmed without false positive or false negative. The consistency of the genotyping results of these samples is reported in .
The assay was designed to be performed on 50 ng of extracted DNA. However, the influence of lower and higher DNA amount on the performance of the analysis was assessed. The results are reported in . It appears that the genotypes can be accurately determined when DNA sample amounts range from 25 to 100 ng. Outside this range, the intensity of the signal does not meet the acceptance criteria to interpret migration fragment data (peak height < 100 units of intensity or saturation of the signal). No influence of the quality of the DNA was observed. Indeed, genotyping was successfully performed in our cohort, whose absorbance ratio ranged from 0.35 to 2.3 and 1.5 to 2.1 for ratios of 260/230 nm and 260/280 nm, respectively. Moreover, no intersample contamination was observed since no signal was detected at the position of the blank samples inserted between the DNA samples (data not shown).
The stability of the ready-to-use pool of primer and fluorescent probes at the working conditions was validated for five freeze–thaw cycles. Indeed, the genotyping results were similar in an experiment performed with a freshly prepared pool and in another experiment performed with an aliquot frozen and thawed five times. Although the peak intensities of fragments were lower when a freeze-thawed reagent was used, they remained >100 intensity units . Thus, a pool of primers can be stored at −20 °C and used several times, which is very cost saving and convenient.
3.1. Samples and DNA Extraction The samples analyzed in this study were anonymized DNA leftover samples from our center DNA bank. DNA was extracted from blood collected either on EDTA or heparinized tubes in humans who had given written consent for genetic research beforehand. The study was approved by a local ethical committee (authorization no. 20.131), approved on the 23 October 2020. DNA was extracted from blood with a Microlab STAR Liquid Handling System (Hamilton, Courtaboeuf, France) using a Macherey-Nagel ® (Hoerdt, France) Nucleospin Blood L kit as described in the manufacturer’s protocol. DNA concentration was measured using a NanoDrop One spectrophotometer (NanoDrop Technologies Inc., Wilmington, DE, USA). 3.2. Reference Method All the samples used to validate the assay were previously genotyped for SNPs of interest using a routine method based on Taqman TM allelic discrimination. Briefly, each SNP was analyzed using the appropriate reaction mix prepared with TaqMan TM Drug Metabolism Genotyping Assays (Thermo Fisher, Waltham, MA, USA). Analysis was performed on an ABI 7900HT instrument (Applied Biosystems, Foster City, CA, USA). For the UGT1A1 *28 allele, pyrosequencing was the reference method. 3.3. Principle of the Assay and Design of the Primers The assay was adapted from Schuelke’s work . We have designed and developed a multiplex allele-specific PCR where fragments are labeled by two different fluorescent markers (HEX/FAM) identifiable by fragment analysis. These two labels are used to characterize the genotype by discriminating bi-allelic variants, while the size of the fragment allows the identification of a particular SNP. The principle of the assay is illustrated in . The nine variant genomic regions were used as a reference for the selection of the primers using a Primer-BLAST tool. The primers were designed to obtain different fragment sizes for each genetic location (primer sizes varied in at least 20 nucleotides). The last nucleotide of the 3′ flanking forward primers was specific to the genotype (wild-type or variant allele). A second substitution of the wild-type matrix was applied to the -4 nucleotide of the 3′ flanking forward primers to enhance its specificity. A universal M13 sequence (−20) GTAAAACGACGGCCAGT was added to the 5′ flanking forward wild-type primers, and an M13 sequence (−40) GTTTTCCCAGTCACGAC was added to the 5′ flanking forward variant primers. A pigtail GTTTCTT was added to the 5′ flanking reverse primers to improve the amplicon migration on the capillary system and to avoid double peaks. Primer sequences are reported in . 3.4. PCR Multiplex Amplification The nine variants were amplified simultaneously using a Qiagen ® (Courtaboeuf, France) Multiplex PCR kit following the manufacturer’s protocol. Briefly, 50 ng of DNA was mixed with a pool solution including every primer . Fluorescent probes (HEX specific to M13 (−20) and FAM specific to M13 (−40), both at 0.7 pmol/µL in the final mix) were also added to the pool of primers (probes were provided by Eurofins MWG Operon, Les Ulis, France). Each primer concentration was optimized to get optimal signal detection . The PCR was run on a thermocycler (Veriti™ 96-Well Thermal Cycler, Thermo Fisher Scientific, Illkirch, France) and started with an activation cycle of 95 °C for 15 min, and then 30 cycles of amplifications were run with the following sequence: 94 °C for 30 s, 58 °C for 90 s, and 72 °C for 60 s. The last cycle was set at 72 °C for 30 min. 3.5. Genotyping and Fragment Analysis The PCR products were denatured with deionized formamide (Thermo Fisher Scientific, Illkirch, France) and separated on an ABI 3130 Genetic Analyzer (Applied Biosystems TM /Thermo Fisher Scientific, Illkirch, France). GeneScan™ 500 ROX™ (Applied Biosystems TM /Thermo Fisher Scientific, Illkirch, France) was used as a dye size internal standard. Data were processed using the GeneMapper ® 4.0 software (Applied Biosystems TM ). We defined the acceptance criterion of the genotyping results after migration on the sequencer as follows: the height of the signal of each fragment should be higher than 100 (intensity unit) and lower than 7500 (intensity unit) to avoid saturation, and the length of each fragment should not differ from a +/− 1 base pair from the theoretical length of the fragment. 3.6. Internal Control Samples In order to check the success of each set of experiments during method development and validation, DNA samples of known genotypes were used as homozygous wild-type, heterozygous, and homozygous variant internal controls. These samples came from a DNA bank of human DNA whose genotypes were previously confirmed by Taqman TM and Sanger sequencing. Homozygous variant control samples were not available for TPMT rs1800462, NUDT15 rs116855232, and the four DPYD SNPs since these genotypes are very rare in the population. Moreover, at the time of the method development, we did not have enough amount of DNA sample in our bank that matched with the genotype heterozygous for DPYD rs55886062 (low-frequency variant). Thus, this control was synthetized by performing a subcloning experiment as described in . The DNA samples used as internal control samples were chosen to get a balance between the lowest number of control samples to be analyzed and the need to be representative of each allelic combination of interest. Thus, we found a combination of only eight DNA samples suitable as internal controls of the 21 genotypes we had to discriminate in the patient samples. The genotypes of internal controls are shown in , and their corresponding electropherograms are illustrated in . 3.7. Validation of the Assay Several parameters were assessed in order to validate the assay. Repeatability was assessed by genotyping internal control samples in duplicate six times on six independent experiments (6 different days). Accuracy was evaluated by the concordance of the genotype results of a cohort of human DNA analyzed with the new method in comparison with the results obtained from a previous analysis with the reference methods. Depending on the number of known genotype results available, the accuracy was assessed in at least 20 samples per gene of interest. Robustness was checked by testing the influence of variation in the amount of DNA analyzed (from 1 to 500 ng) on the assay specificity and sensitivity. The influence of the quality of the DNA extracted was assessed by looking at the absorbance ratio (260/230 nm and 260/280 nm) of the DNA samples in our cohort. Intersample contamination was evaluated by inserting blank samples (water without DNA) between DNA samples. Besides, the stability of the mixed pool of diluted primers and fluorescent probes was checked for five freeze–thaw cycles (−20 °C ambient temperature). The genotyping results from six DNA samples were compared between the samples analyzed with a freshly prepared pool of primers and the same samples analyzed with the same pool frozen and thawed five times.
The samples analyzed in this study were anonymized DNA leftover samples from our center DNA bank. DNA was extracted from blood collected either on EDTA or heparinized tubes in humans who had given written consent for genetic research beforehand. The study was approved by a local ethical committee (authorization no. 20.131), approved on the 23 October 2020. DNA was extracted from blood with a Microlab STAR Liquid Handling System (Hamilton, Courtaboeuf, France) using a Macherey-Nagel ® (Hoerdt, France) Nucleospin Blood L kit as described in the manufacturer’s protocol. DNA concentration was measured using a NanoDrop One spectrophotometer (NanoDrop Technologies Inc., Wilmington, DE, USA).
All the samples used to validate the assay were previously genotyped for SNPs of interest using a routine method based on Taqman TM allelic discrimination. Briefly, each SNP was analyzed using the appropriate reaction mix prepared with TaqMan TM Drug Metabolism Genotyping Assays (Thermo Fisher, Waltham, MA, USA). Analysis was performed on an ABI 7900HT instrument (Applied Biosystems, Foster City, CA, USA). For the UGT1A1 *28 allele, pyrosequencing was the reference method.
The assay was adapted from Schuelke’s work . We have designed and developed a multiplex allele-specific PCR where fragments are labeled by two different fluorescent markers (HEX/FAM) identifiable by fragment analysis. These two labels are used to characterize the genotype by discriminating bi-allelic variants, while the size of the fragment allows the identification of a particular SNP. The principle of the assay is illustrated in . The nine variant genomic regions were used as a reference for the selection of the primers using a Primer-BLAST tool. The primers were designed to obtain different fragment sizes for each genetic location (primer sizes varied in at least 20 nucleotides). The last nucleotide of the 3′ flanking forward primers was specific to the genotype (wild-type or variant allele). A second substitution of the wild-type matrix was applied to the -4 nucleotide of the 3′ flanking forward primers to enhance its specificity. A universal M13 sequence (−20) GTAAAACGACGGCCAGT was added to the 5′ flanking forward wild-type primers, and an M13 sequence (−40) GTTTTCCCAGTCACGAC was added to the 5′ flanking forward variant primers. A pigtail GTTTCTT was added to the 5′ flanking reverse primers to improve the amplicon migration on the capillary system and to avoid double peaks. Primer sequences are reported in .
The nine variants were amplified simultaneously using a Qiagen ® (Courtaboeuf, France) Multiplex PCR kit following the manufacturer’s protocol. Briefly, 50 ng of DNA was mixed with a pool solution including every primer . Fluorescent probes (HEX specific to M13 (−20) and FAM specific to M13 (−40), both at 0.7 pmol/µL in the final mix) were also added to the pool of primers (probes were provided by Eurofins MWG Operon, Les Ulis, France). Each primer concentration was optimized to get optimal signal detection . The PCR was run on a thermocycler (Veriti™ 96-Well Thermal Cycler, Thermo Fisher Scientific, Illkirch, France) and started with an activation cycle of 95 °C for 15 min, and then 30 cycles of amplifications were run with the following sequence: 94 °C for 30 s, 58 °C for 90 s, and 72 °C for 60 s. The last cycle was set at 72 °C for 30 min.
The PCR products were denatured with deionized formamide (Thermo Fisher Scientific, Illkirch, France) and separated on an ABI 3130 Genetic Analyzer (Applied Biosystems TM /Thermo Fisher Scientific, Illkirch, France). GeneScan™ 500 ROX™ (Applied Biosystems TM /Thermo Fisher Scientific, Illkirch, France) was used as a dye size internal standard. Data were processed using the GeneMapper ® 4.0 software (Applied Biosystems TM ). We defined the acceptance criterion of the genotyping results after migration on the sequencer as follows: the height of the signal of each fragment should be higher than 100 (intensity unit) and lower than 7500 (intensity unit) to avoid saturation, and the length of each fragment should not differ from a +/− 1 base pair from the theoretical length of the fragment.
In order to check the success of each set of experiments during method development and validation, DNA samples of known genotypes were used as homozygous wild-type, heterozygous, and homozygous variant internal controls. These samples came from a DNA bank of human DNA whose genotypes were previously confirmed by Taqman TM and Sanger sequencing. Homozygous variant control samples were not available for TPMT rs1800462, NUDT15 rs116855232, and the four DPYD SNPs since these genotypes are very rare in the population. Moreover, at the time of the method development, we did not have enough amount of DNA sample in our bank that matched with the genotype heterozygous for DPYD rs55886062 (low-frequency variant). Thus, this control was synthetized by performing a subcloning experiment as described in . The DNA samples used as internal control samples were chosen to get a balance between the lowest number of control samples to be analyzed and the need to be representative of each allelic combination of interest. Thus, we found a combination of only eight DNA samples suitable as internal controls of the 21 genotypes we had to discriminate in the patient samples. The genotypes of internal controls are shown in , and their corresponding electropherograms are illustrated in .
Several parameters were assessed in order to validate the assay. Repeatability was assessed by genotyping internal control samples in duplicate six times on six independent experiments (6 different days). Accuracy was evaluated by the concordance of the genotype results of a cohort of human DNA analyzed with the new method in comparison with the results obtained from a previous analysis with the reference methods. Depending on the number of known genotype results available, the accuracy was assessed in at least 20 samples per gene of interest. Robustness was checked by testing the influence of variation in the amount of DNA analyzed (from 1 to 500 ng) on the assay specificity and sensitivity. The influence of the quality of the DNA extracted was assessed by looking at the absorbance ratio (260/230 nm and 260/280 nm) of the DNA samples in our cohort. Intersample contamination was evaluated by inserting blank samples (water without DNA) between DNA samples. Besides, the stability of the mixed pool of diluted primers and fluorescent probes was checked for five freeze–thaw cycles (−20 °C ambient temperature). The genotyping results from six DNA samples were compared between the samples analyzed with a freshly prepared pool of primers and the same samples analyzed with the same pool frozen and thawed five times.
Here, we propose a fast and cost-effective method to genotype in a multiplex assay of eight SNPs and one repetition of four genes involved in the pharmacological response of anticancer drugs. Instead of using different fluorescent primers to genotype each genetic polymorphism, we used universal primers as adapters so that we only had to order a single one couple of expensive fluorescent probes. Furthermore, the genotyping protocol can easily be extended with new SNPs by just ordering new standard primers. The method is therefore particularly suitable for low-throughput genotyping of well-known variants. This is a major advantage compared with similar multiplex assays such as SNaPshot (R) , which use fluorescent dNTP and more expensive reagents . The performances of our assay were rigorously assessed in particular to confirm its accuracy and robustness. Accuracy was studied on 187 samples, all of which were correctly genotyped; therefore, the method is now routinely used in clinical practice in our center. However, more patients should be tested to get a complete measure of the sensitivity and specificity of the assay. We will collect further data on a prospective basis. A downside of the assay is that it is not designed to genotype tri-allelic variants since we use two fluorescent probes. Thus, we cannot exclude that some subjects are carriers of rare variants that would lead to false negative results. As in every PCR assay, we must also be aware that a polymorphism located in the 3′ region of primers could create a mismatch, leading to a failure of the amplification of the fragment of interest. Besides, we showed that a minimal amount of DNA of 25 ng was needed in our assay to get interpretable results. This means that the sensitivity of this assay is lower than that of other existing multiplexing approaches (NGS, digital droplet PCR, etc.). However, in the case of the pharmacogenetic assay performed on constitutional DNA, we are not limited by the amount of DNA available in the samples since DNA is extracted from several millimeters of whole blood collected by venipuncture. Therefore, this assay is sensitive enough to the conditions of clinical use intended. A limitation of the protocol reported in the present study could be the number of variants that can be simultaneously studied. We multiplexed the analysis of 9 genetic variants. The instructions of the manufacturer of the multiplex PCR kit reagent used in our assay suggested that this approach is enough discriminant to genotype a panel of up to 16 variants. To be able to analyze a pharmacogenetic panel with a greater number of SNPs and samples, targeted sequencing would be more attractive. A comparison of usual pharmacogenetic methods regarding the criterion of cost and analytical performances is reported in . In conclusion, we developed a multiplex genotyping method that should make the analysis of pharmacogenes accessible to a large number of labs with a capillary sequencer by using inexpensive reagents and materials. Genotyping results can be returned in 1 day, which should be notably useful to optimize and individualize the treatment of many patients receiving anticancer drug therapy.
|
Significance of the immunofluorescence staining patterns and titres of the antinuclear antibody test in paediatric rheumatology setting | 11419a39-5369-4c43-aaa7-8a9d83e528bd | 10387863 | Internal Medicine[mh] | The term of antinuclear antibody (ANA) refers to any of a large group of autoantibodies that recognise predominantly, but not always specifically, cellular antigens in the cell nucleus. They are antibodies that develop against structures such as DNA, histones, and centromeres . Presence of antinuclear antibodies (ANAs) is associated with various systemic rheumatic diseases, including systemic lupus erythematosus (SLE), systemic sclerosis, primary Sjögren syndrome, mixed connective tissue disease and idiopathic inflammatory myopathies (such as polymyositis and dermatomyositis). These diseases are collectively referred to as ANA associated rheumatic diseases, and several autoantibodies that are specific to each disease have been identified . However, without any signs of disease, ANA could be detected in healthy people and observed 4%–15% of healthy children . Besides, a positive ANA test might be observed in malignant and infectious diseases and has a high rate of false-positive results for rheumatic diagnoses . The American College of Rheumatology ANA Task Force position statement recommended the indirect immunofluorescence assay (IFA) using HEp-2 substrate as the “gold standard” for primary ANA detection . However, some clinical laboratories use solid-phase immunoassays, in some cases as a reflex test to supplement HEp-2 IFA screening test, or even replace HEp-2 IFA testing. Nevertheless, most clinical laboratories worldwide depend heavily on HEp-2 IFA as the primary screening method. The ANA staining pattern raises suspicion for various diseases and helps clinicians to perform confirmatory tests with clinical basis . Positive ANA and high titres have been reported to be associated with a diagnosis of SLE in children but, no diagnostic utility has been shown in children with JIA . Besides, positive ANA alone was suggested as a poor indicator for a rheumatic diagnosis in children . In this study, we aimed to investigate the frequency of the positive ANA test in paediatric rheumatology setting and the association of the immunofluorescence staining patterns and titres of ANA with rheumatic diseases.
Medical charts of children, evaluated in the paediatric rheumatology clinic between January 2016 and December 2021, in whom at least one ANA test was ordered, were reviewed. Patients with a positive ANA at least in one occasion were included. Positive ANA was defined as ≥1/80 titre in indirect IFA on Hep-2 cell substrates. Patients with a positive ANA, with immunofluorescence staining patterns of ANA not reported according to The International Consensus on ANA-staining Patterns (ICAP) recommendations , were excluded. Age, sex, and diagnosis of the patients were collected from the medical charts. Among patients, association of the titres and patterns of ANA were investigated in patients with JIA, ANA associated diseases and nonrheumatic conditions. Immunofluorescence staining patterns of ANA were investigated under 4 major groups, according to the ICAP recommendations . While homogenous and dense fine speckled (DFS) patterns were assessed as sole patterns, fine and coarse speckled patterns were classified under speckled. Other nuclear staining patterns, such as centromere, nucleolar and discrete nuclear dots, were classified under the other nuclear group. Cytoplasmic and mitotic staining patterns were not investigated due to the rare detection rate. Among patients with a positive ANA more than one occasion, staining pattern of the highest titre were taken into account. ANA titres were classified as 1+ in titres between 1/80 and 1/320, 2+ in titres between 1/320 and 1/1000, 3+ in titres between 1/1000 and 1/3200 and 4+ in titres > 1/3200 due to the laboratory preference of the ANA results. In statistical evaluation, data obtained by measurement are shown as mean ± standard deviation, and data obtained by counting are shown as percentage. The normal distribution of data was analysed by using the Kolmogorov-Smirnov test. One-way Anova or Kruskal-Wallis test were used for the analyses of quantitative data between three groups according to the distribution of the data. The Kruskal-Wallis test followed by Dunn’s posthoc test was used to compare the groups according to age at positive ANA. Chi-square test was used for the comparison of qualitative data. Posthoc analysis for the significant chi-square values was carried out by calculating the significant adjusted residuals. The level of significance was set at p value < 0.05.
Between January 2016 and December 2021, ANA test was ordered in 2477 patients and a positive ANA was reported in 697 (28.1%) of them. Among ANA positive patients, only 273 (39.2%) were diagnosed with a rheumatic disease. Most common diagnosed rheumatic disease was JIA in 120 (43.8%) of the patients followed by SLE in 50 (18.2%) and vasculitis in 46 (16.8%). Among patients with JIA, most common subtype was persistent oligoarticular JIA in 91 (75.8%) followed by polyarticular JIA in 16 (13.3%), extended oligoarticular in 14 (11.7) patients. Enthesitis related arthritis, psoriatic arthritis and systemic onset JIA were the least frequent diagnoses in ANA positive patients with JIA. ANA associated diseases including SLE, juvenile dermatomyositis, Sjögren syndrome and scleroderma was observed in 67 (24.5%) of the patients. Among patients with vasculitis most frequent diagnosis was immunoglobulin A vasculitis in 36 (78.3%) of the patients. Behçet’s disease, Takayasu arteritis, polyarteritis nodosa and hypocomplementemic urticarial vasculitis were the other diagnoses in patients with positive ANA. Since positive ANA is not implicated in pathogenesis and clinical findings in patients with vasculitis and heterogeneity of the pathogenesis among different vasculitis types, we did not include patients with vasculitis in comparative analyses. Flow chart of the study population and distribution of the rheumatic diseases are shown in and , respectively. Comparison of the age and sex according to the diagnosis of the patients revealed a higher age at ANA testing (mean age in ANA associated diseases: 12.5 ± 3.5, JIA: 9.4 ± 4.6 and nonrheumatic conditions: 9.7 ± 3.9, p: <0.0001) and a higher frequency of female sex (ANA associated diseases 85.1%, JIA 68.4% and nonrheumatic conditions 62.2%, p: 0.001) in ANA associated diseases than patients with JIA and nonrheumatic conditions. In patients with ANA associated diseases, the most reported ANA pattern was homogenous in 34.3% of the patients. While spotted ANA patterns were the most observed pattern in patients with JIA (28.3%), DFS staining pattern was the most frequently reported pattern in patients with nonrheumatic conditions (34.8%). Comparison of the staining patterns among patients revealed a significant trend towards increased frequency of DFS pattern (ANA associated diseases 10.4%, JIA 21.7% and nonrheumatic conditions 34.8%, p < 0.0001) and decreased frequency of homogenous staining (ANA associated diseases 34.3%, JIA 26.7% and nonrheumatic conditions 8.9%, p < 0.0001) in nonrheumatic conditions. Assessment of the ANA titres revealed a significantly more frequent high ANA titres (>1/1000) in ANA associated diseases compared to the patients with JIA and nonrheumatic conditions (ANA associated diseases 53.7% vs. JIA 16.7% and nonrheumatic conditions 11.6%, p < 0.0001). However, distribution of the titres between patients with JIA and nonrheumatic conditions did not significantly differ among patients (p > 0.05). Comparison of patient characteristics, ANA staining pattern and titres among patients with ANA associated diseases, JIA and nonrheumatic diseases is given in .
In our study, the majority of the children who tested positive for ANA did not have a rheumatic diagnosis and JIA was the most common rheumatic disease in children with a positive ANA result. Similar to our results, JIA was the most common rheumatic diagnosis in children with a positive ANA in another study . In contrast, an earlier study found nonrheumatic conditions in 27% of the children with a positive ANA and showed that majority of the children who have positive ANA test without any autoimmune diagnosis at initial diagnosis will not develop an autoimmune condition . This might be associated with increased referral of patients to the rheumatology departments and increased usage of ANA testing without solid indications. A similar observation was reported by Haslak et al. , and in their study majority of the patients (94.1%) referred to paediatric rheumatology clinic for a positive ANA had no underlying disease and none of them developed any autoimmune conditions or ANA associated rheumatic diseases. In our study, homogenous staining pattern and higher titres were more frequently detected in ANA associated diseases and majority of the patients with ANA associated diseases diagnosed with SLE. SLE is a prototypic autoimmune disease, and immunological hallmark is the production of ANA . Higher titres, presence of multiple autoantibodies and homogenous staining pattern were shown to be associated with a diagnosis of SLE . ANA testing is often considered in paediatric patients presenting with joint pain to determine the possibility of an alternative diagnosis to JIA or systemic autoimmune related diseases such as SLE. A positive-ANA has been reported in 30%–50% of JIA patients in varying proportions across the JIA subtypes . An elevated ANA titre has been reported in all JIA subtypes although is most prevalent in oligoarticular JIA (persistent and extended) . The detection of ANA is important because presence determines the frequency of ocular assessment for asymptomatic uveitis . Traditionally, ANA has not been used as an aid in the diagnosis of JIA, but as a risk biomarker for developing uveitis . Uveitis is the most common of extraarticular manifestations in JIA and can have a significant impact on morbidity if detection and treatment of uveitis is delayed . The most commonly reported and most sensitive cut-off titre for JIA was 1/80, although there was a large variation in published ANA immunofluorescence serum dilutions (1/40–1/320) used for laboratory investigation . Although previous studies have shown that ANA titres were not significantly different from in patients with JIA than nonrheumatic conditions , no study investigated the staining patterns between JIA and healthy controls. In our study, except for higher homogenous pattern in patients with JIA, none of the staining patterns significantly differed between children with nonrheumatic conditions and JIA. In a Nordic study, antihistone antibodies were found to be significantly associated with JIA uveitis and antihistone antibodies were expected to be stained as homogenous which might partially explain the observation of higher homogenous staining pattern in patients with JIA in our study. Also, a recent study reported a more frequent homogenous pattern of ANA in children with uveitis . The staining pattern of ANA may provide clues about diseases. For example, the homogeneous nuclear pattern frequently associated with SLE while fine granular mottled pattern observed more common in Sjögren’s disease . It has been reported that some patterns of ANA staining are associated with certain nuclear antigens that are related to particular manifestations of specific diseases . As there are many possible nuclear antigens, ANA are classified into specific autoantibodies using different techniques such as immunoblotting or enzyme-linked immunosorbent assay such as anti-dsDNA, anti-Sm, anti-SSA/Ro and myositis specific antibodies. In our study, DFS staining pattern was more frequently reported in children with nonrheumatic conditions. In the absence of any disease specific antibodies, a positive DFS pattern with a positive anti-DFS70 antibody is unlikely to be associated with systemic autoimmune disease . In a recent study, only half of the children with a positive ANA with DFS pattern exhibited positive anti-DFS70 antibodies. In addition, anti-DFS70 antibodies were less likely to be found positive in children with autoimmune diseases and in all children with a positive anti-DFS70 antibody with an autoimmune disease, a disease specific antibody was observed . Thus, evaluation of disease specific and anti-DFS70 antibodies along with ANA test might be more accurate than the assessment of the immunofluorescence pattern of ANA alone in children with suspected autoimmune disease. Retrospective design is the main limitation of our study. Besides, this study did not include the follow-up data of patients with nonrheumatic conditions which might overestimate the prevalence of nonrheumatic conditions in ANA positive children. Despite immunofluorescence patterns of ANA was reported in a single laboratory, interpretation of the staining patterns might not be standardised. Another point to consider is the pretest probability of rheumatic disease in ANA tested children. Also presence of uveitis and association with pattern and titre of ANA was not investigated which might be regarded as a limitation. A higher pretest probability results with a higher predictive value . And that might result in different predictive value of ANA test in a different rheumatology setting. In conclusion, the majority of the children with a positive ANA did not have a rheumatic disorder. Ordering ANA test with more solid indications might result in an increased sensitivity for rheumatic diseases. Despite homogenous staining pattern and higher titres of ANA were associated with ANA associated diseases, presence of autoimmune diseases in patients with DFS pattern ANA suggests that interpretation of ANA test might be more accurate in the presence of specific antibody panels.
|
Improved genetic algorithm based on greedy and simulated annealing ideas for vascular robot ordering strategy | 92f90b51-abaa-4b34-a03a-19419779230a | 11841910 | Surgical Procedures, Operative[mh] | Medical robotics is a rapidly evolving field that leverages advanced algorithms to unlock the full potential of cutting-edge technologies . One of these technologies is the vascular robot, which can perform precise and minimally invasive procedures within the human vasculature. Among the various types of vascular robots, the ABLVR vascular robot is a novel and promising technology that consists of a robotic vessel and four operators who can navigate the bloodstream autonomously . However, this technology poses a unique challenge in terms of resource allocation and optimization, as the operators require a week-long biological learning process before they can be fully operational, and the robotic vessel needs to be periodically removed for maintenance . Vascular robots need to be fully trained in the vessel boat before they can work. Maintenance of medical robots is a healthcare resource allocation problem. Navaz et al. conducted a comprehensive review, highlighting various approaches to optimizing resource allocation in this context. Their work serves as a foundational reference, summarizing existing methodologies and identifying research gaps. Faccincani et al. . investigated adaptive resource allocation strategies, emphasizing the need for dynamic models that can adapt to changing conditions. Their research underscores the importance of flexibility in resource allocation to meet the evolving demands of healthcare environments. Zouri et al. delved into cost-effective resource allocation models, emphasizing the the importance of cost optimization in rehabilitation hospitals. Their study provides insights into the trade-offs between cost and treatment efficiency, a critical consideration in healthcare robotics. Guo et al. focused on system control models for vascular robots and used robust controllers to optimize their performance and improve system stability. Their research contributes to understanding the dynamics of acquiring and maintaining robotic assets for vascular treatments. However, their study does not directly relate to optimizing the birth of robotic assets for vascular therapy, and it is difficult for hospitals to go directly through their methodology to design and optimize acquisition strategies for robots for vascular therapy. In recent years, the rapid development of computer technology has allowed it to be used in a wide range of applications in the healthcare industry. Pashaei et al. used a hybrid binary COOT algorithm with simulated annealing to search for targeted genes. Pashaei et al. proposed a simulated annealing-based mRMR search method for feature selection in high-dimensional biomedical data. Yu et al. explored the role of reinforcement learning in the healthcare domain for health resource allocation and scheduling and health management problems. These applications show that the application of computer technology in the healthcare industry promotes the development of the healthcare industry. This study aims to address this challenge by developing a comprehensive and adaptive long-term strategy for the acquisition, utilization, and maintenance of ABLVR vascular robots and operators . This study aims to address this challenge by developing a comprehensive, adaptable, and long-term strategy for the acquisition, use, and maintenance of the ABLVR vascular robot and operator. We designed a genetic algorithm based on improved greed and simulated annealing, along with an ARIMA time-series model optimized by the genetic algorithm, for optimizing the ordering and scheduling of the ABLVR vascular robot’s capacity boats and operators to ensure efficient treatment while minimizing costs and addressing potential damage to the robot. Specifically, with the known number of vascular robot uses required per week, the greedy algorithm is first embedded into a genetic algorithm to solve for the optimal number of vessel boats and operators to be purchased per week as an initial solution to the genetic algorithm. Then the genetic algorithm optimized based on the simulated annealing idea is built to solve the final result. The ARIMA model, optimized by the genetic algorithm, forecasts the unknown demand for vascular robot uses in a time series, and determines the optimal number of vessel boats and operators to purchase accordingly. The key contributions and highlights of our research include: Comprehensive Resource Allocation Model: We have developed a robust resource allocation model that optimizes the procurement of both robotic vessels and operators, considering the dynamic nature of healthcare environments. Incorporating Adaptive Learning: Our model accounts for the adaptive learning process required for operators, as well as the maintenance and disposal of robotic components. Hybrid Genetic Algorithm: We introduce a hybrid genetic algorithm that incorporates simulated annealing and greedy approaches to efficiently solve the optimization problem. Time Series Forecasting: We use an ARIMA time series model to predict the demand for vascular robots, enhancing the adaptability of our procurement strategy. We compare our proposed method with traditional heuristic approaches and machine learning-based methods, highlighting the advantages of our approach in terms of optimization and transparency. This study proposes an innovative approach based on an improved genetic algorithm that can optimize the ordering and scheduling of the ABLVR vascular robots and operators. By applying computational optimization techniques, this study seeks to enhance the efficiency and cost-effectiveness of this groundbreaking medical technology. 2.1 Assumptions This research employs a quantitative approach grounded in computational modeling and optimization techniques to address the multifaceted challenges posed by the acquisition and utilization of ABLVR vascular robots in healthcare settings. To maintain the reasonableness and accuracy of the model solution in alignment with the specific problem, it is necessary to introduce the following assumptions: Cost-Based Part Disposal: The model adopts a cost-centric approach, whereby any component of the robot is discarded if the cost of maintaining it surpasses the combined cost of purchasing a new part and facilitating the learning process for both new and used parts . Additionally, parts are considered for disposal if they are no longer viable for reuse. Rounding in Calculations: Throughout the calculation process in this paper, rounding is performed at each step. It is assumed that rounding does not introduce significant deviations or impact the overall reliability of the model’s results. This assumption is made to ensure the feasibility of computational solutions . Predictable Part Reliability: It is assumed that each component of the robot will not experience unexpected failures due to internal problems during use. This assumption simplifies our model, allowing focus on external factors and optimizing the ordering strategy . 2.2 Mathematical modeling 2.2.1 Optimal decision making based on the single-objective genetic algorithm First, there is need to establish a robot purchase strategy model to make the lowest cost of purchased operators and container boats under the premise of meeting the hospital treatment demand, and the purchase cost is linearly related to the quantity . In order to better control costs, this paper takes into account that there may be cases where the cost of maintenance to the robot vessel boat or operator exceeds the cost of purchase of that part and learning of new and used parts, i.e W O × P O m > P O + 2 P O t (1) W C × P C m > P C (2) Where W O and W C indicate the current number of weeks of maintenance for either operator and vessel, respectively. This document chooses to discard this part when the vessel maintenance cost exceeds the purchase price of a new vessel and the operator maintenance cost exceeds the purchase cost of a new operator and the learning cost of a new or old operator, or when the part was last used. To determine the objective function, the current total number of operators and vessel boats can be expressed as follows,respectively. N C i = ∑ j = 1 i ( C B i - C D i ) + N C o (3) N O i = ∑ j = 1 i ( C O i - C O i ) + N O 0 (4) Where N Ci indicates the total number of vessels owned in week i, N O i u indicates the total number of operators owned in week i, C Bi indicates the number of vessels purchased in week i, and C Di indicates the number of vessels discarded in week i. N C 0 indicates the number of vessels owned at the beginning of week 1, and N O 0 indicates the number of operators owned at the beginning of week 1. Hence the total purchase cost can be expressed as: m i n P ( C B i , O B i , N O i g , N O i t , N C i m , N O i m ) = ( ∑ C B i ) × P C + ( ∑ O B i ) × P O + ( ∑ N O i g + N O i t ) × P O t + ( ∑ N C i m ) × P O m + ( ∑ N O i m ) × P C m (5) Where P C denotes the unit price per vessel boat, and P O denotes the unit price per operator, the P Ot denotes the price of one operator training, and P Om denotes the price of one operator maintenance, and P Cm denotes the total cost required to complete hospital treatment operations. To determine the constraint conditions the relationship between the total number of operators and vessel boats in week 1 and the number in maintenance is shown below: N C i = N C i m + N C i u (6) N C i u = R i (7) N O i = N O i m + N O i u + N O i g + N O i t (8) N O i u = 4 R i (9) N O i t = O B i (10) N ∂ i g = ⌈ O B i G ⌉ (11) Where G indicates the number of new operators each skilled operator can instruct, and the total number of operators owned in week i is equal to the sum of the number of operators in maintenance, use, training instruction, and training at that point. Since the vascular robot must be dismantled after one week of work in the vasculature, the robot’s operators within it cannot work again until after 7 days of maintenance . Therefore, the number of operators under maintenance in week i is equal to four times the number of robots in the hospital’s operational requirements in week i-1 minus the number of operators discarded in that week, i.e. N O i m ≥ 4 R i - 1 - O D i (12) R i −1 denotes the number of robots in demand for hospital operations in week i-1. Since the newly purchased vessel boats cannot start working until after a week of commissioning, there is: the number of vascular robots required by the hospital in week 1 must be less than the total number of vessel boats owned in week 1 minus the number of vessel boats discarded in week 1. This can be expressed mathematically as: { R i < N C i - 1 - C D i , i > 1 R i < N C 0 - C D i , i = 1 (13) where N C 0 indicates the original number of container boats. Therefore, the robot buying strategy can be modelled as follows: m i n P ( C B i , O B i , N O i g , N O i t , N C i m , N O i m ) (14) Assuming that 20% of the vascular robots in the human body are destroyed each week, so that the total number of vessel boats and operators changes each week, the total number of vessel boats and operators by changing the composition of a set of functions can be expressed recursively as: N C i = N C i - 1 + C B i - C D i - K × N C i - 1 u (15) N O i = N O i - 1 + O B i - O D i - K × N O i - 1 u (16) where K indicates the percentage of vascular robots destroyed in the human body at this time, here K = 20%. Since the robot’s operators do not work again until after 7 days of maintenance . Therefore, the number of operators under maintenance in week 1 is greater than equal to four times the number of machines in the hospital’s operational demand in week 1 minus the number of operators discarded in that week minus the number of macrophages hit, i.e. N O i m ⩾ R i - 1 - O D i - K × N O i - 1 u (17) Since the newly purchased vessel boats cannot start working until after a week of commissioning, there is: the number of vascular robots required by the hospital in week i must be less than the sum of the total number of vessel boats owned in week i − 1 minus the number of vessel boats discarded in week i and the number of vessel boats destroyed in week i − 1, which can be expressed mathematically as: { R i < N C i - 1 - C D i - K × N C i - 1 u , i > 1 R i < N C 0 - C D i , i = 1 (18) where N C 0 indicates the number of original container boats. Now we need to consider the probability of hitting a macrophage resulting in the complete destruction of the vascular robot, while changing the upper limit of how much each skilled operator can instruct a new operator to learn from G to 20. At this time, 10% of the vascular robots in the human body are destroyed, and the total number of vessel boats and operators changes each week as the amount of destruction changes, and the total number of vessel boats and operators can be expressed recursively as: N C i = N C i - 1 + C B i - C D i - K × N C i - 1 u (19) N O i = N O i - 1 + O B i - O D i - K × N O i - 1 u (20) The number of operators in maintenance is equal to four times the number of machines in the hospital’s operational requirements in week i-1 minus the number of operators discarded in that week minus the number of macrophages hit, i.e. N O i m ⩾ 4 R i - 1 - O D i - K × N O i - 1 u (21) Where K indicates the percentage of vascular robots destroyed in the human body at this time, here K = 10%, while the upper limit of what each skilled operator can instruct new operators to learn G changes to 20 . And for vessel boats, the number of vascular robots needed to have week i hospitals must be less than the sum of the total number of vessel boats owned in week i − 1 minus the number of vessel boats discarded in week i and the number of vessel boats destroyed in week i − 1. The inequality can be expressed as: { R i < N C i - 1 - C D i - K × N C i - 1 u , i > 1 R i < N C 0 - C D i , i = 1 (22) Where N C 0 denotes the number of original vessel boats and K denotes the percentage of vascular robots destroyed in the human body at this time, here K = 10%. The specific analytical thought process is shown in the following . 2.2.2 ARIMA sequences and seasonal sequence forecasting In this section, this paper predicts the demand for the use of vascular robots from 105–112 weeks by ARIMA time series . Time series analysis refers to a set of random variables ordered by time. The main idea of the model is to make dynamic predictions of unknown data based on inter-observations by dependence and correlation. This time model is used to forecast the 105–112 week demand by building this time model, and the implementation of the ARIMA model consists of the following five main steps. Step 1 : Perform data processing, for the mean value of the number of vascular robots used in weeks 1–104 in Annex II y ( t ) can be expressed as : Y ¯ = 1 n ∑ t = 1 n y ( t ) (23) The sample value of the new sequence obtained after its differential processing can be expressed as: x ( t ) = y ( t ) - Y ¯ (24) Generally for d-order difference can be expressed as follows: (∇ d is called the d-order difference operator) . ∇ d X t = ( 1 - B ) d X t (25) The number of differences is determined by the parameter d in the ARIMA (p,d,0) model, i.e., one difference is made, d = 1, i.e., two differences are made, d = 2, i.e., no difference is made, at which point the model structure is changed to ARIMA (p). Step 2 : Parameter estimation: Recursive least squares with forgetting factors can be used for parameter estimation. The forgetting factor enhances the effect of current observations on parameter estimation while weakening the effect of previous observations. The inclusion of the forgetting factor in recursion can take into account the time-varying nature of the model parameters, and the ARIMA (p) model for the sequence y ( t ) can be expressed as : y ( t ) = φ T ( t ) θ + e ( t ) (26) φ T ( t ) = [ y ( t - 1 ) , y ( t - 2 ) , ⋯ , y ( t - p ) ] (27) θ = [ a 1 , a 2 , ⋯ , a p ] T (28) Recursive parameter estimation by substituting φ T , θ into a recursive least squares formulation with a forgetting factor. Step 3 : Forecasting algorithm: the Astrom forecasting method based on the linear minimum variance forecasting principle which can better solve the random geodesic problem in forecasting is used for forecasting, and the ARIMA (p,d,q) process can be expressed as: A ( B ) ∇ d y ( t ) = C ( B ) e ( t ) (29) where y(t), e(t) denote the original sequence and the white noise sequence, respectively. A ( B ) = 1 - a 1 B - a 2 B 2 - ⋯ - a p B p (30) C ( B ) = 1 - c 1 B - c 2 B 2 - ⋯ - c p B p (31) B denotes the back-shift operator is: B n y ( t ) = y ( t - n ) , n = 1 , 2 , ⋯ (32) Minimum variance predictor is: Y ^ ( t + k t ) = G ( B ) C ( B ) y ( t ) (33) Step 4 : model check: this is achieved by checking whether the error series between the original time series and the established model is stochastic; if the model check fails, the model is rebuilt. Step 5 : Export the appropriate prediction model and perform the actual prediction analysis. This research employs a quantitative approach grounded in computational modeling and optimization techniques to address the multifaceted challenges posed by the acquisition and utilization of ABLVR vascular robots in healthcare settings. To maintain the reasonableness and accuracy of the model solution in alignment with the specific problem, it is necessary to introduce the following assumptions: Cost-Based Part Disposal: The model adopts a cost-centric approach, whereby any component of the robot is discarded if the cost of maintaining it surpasses the combined cost of purchasing a new part and facilitating the learning process for both new and used parts . Additionally, parts are considered for disposal if they are no longer viable for reuse. Rounding in Calculations: Throughout the calculation process in this paper, rounding is performed at each step. It is assumed that rounding does not introduce significant deviations or impact the overall reliability of the model’s results. This assumption is made to ensure the feasibility of computational solutions . Predictable Part Reliability: It is assumed that each component of the robot will not experience unexpected failures due to internal problems during use. This assumption simplifies our model, allowing focus on external factors and optimizing the ordering strategy . 2.2.1 Optimal decision making based on the single-objective genetic algorithm First, there is need to establish a robot purchase strategy model to make the lowest cost of purchased operators and container boats under the premise of meeting the hospital treatment demand, and the purchase cost is linearly related to the quantity . In order to better control costs, this paper takes into account that there may be cases where the cost of maintenance to the robot vessel boat or operator exceeds the cost of purchase of that part and learning of new and used parts, i.e W O × P O m > P O + 2 P O t (1) W C × P C m > P C (2) Where W O and W C indicate the current number of weeks of maintenance for either operator and vessel, respectively. This document chooses to discard this part when the vessel maintenance cost exceeds the purchase price of a new vessel and the operator maintenance cost exceeds the purchase cost of a new operator and the learning cost of a new or old operator, or when the part was last used. To determine the objective function, the current total number of operators and vessel boats can be expressed as follows,respectively. N C i = ∑ j = 1 i ( C B i - C D i ) + N C o (3) N O i = ∑ j = 1 i ( C O i - C O i ) + N O 0 (4) Where N Ci indicates the total number of vessels owned in week i, N O i u indicates the total number of operators owned in week i, C Bi indicates the number of vessels purchased in week i, and C Di indicates the number of vessels discarded in week i. N C 0 indicates the number of vessels owned at the beginning of week 1, and N O 0 indicates the number of operators owned at the beginning of week 1. Hence the total purchase cost can be expressed as: m i n P ( C B i , O B i , N O i g , N O i t , N C i m , N O i m ) = ( ∑ C B i ) × P C + ( ∑ O B i ) × P O + ( ∑ N O i g + N O i t ) × P O t + ( ∑ N C i m ) × P O m + ( ∑ N O i m ) × P C m (5) Where P C denotes the unit price per vessel boat, and P O denotes the unit price per operator, the P Ot denotes the price of one operator training, and P Om denotes the price of one operator maintenance, and P Cm denotes the total cost required to complete hospital treatment operations. To determine the constraint conditions the relationship between the total number of operators and vessel boats in week 1 and the number in maintenance is shown below: N C i = N C i m + N C i u (6) N C i u = R i (7) N O i = N O i m + N O i u + N O i g + N O i t (8) N O i u = 4 R i (9) N O i t = O B i (10) N ∂ i g = ⌈ O B i G ⌉ (11) Where G indicates the number of new operators each skilled operator can instruct, and the total number of operators owned in week i is equal to the sum of the number of operators in maintenance, use, training instruction, and training at that point. Since the vascular robot must be dismantled after one week of work in the vasculature, the robot’s operators within it cannot work again until after 7 days of maintenance . Therefore, the number of operators under maintenance in week i is equal to four times the number of robots in the hospital’s operational requirements in week i-1 minus the number of operators discarded in that week, i.e. N O i m ≥ 4 R i - 1 - O D i (12) R i −1 denotes the number of robots in demand for hospital operations in week i-1. Since the newly purchased vessel boats cannot start working until after a week of commissioning, there is: the number of vascular robots required by the hospital in week 1 must be less than the total number of vessel boats owned in week 1 minus the number of vessel boats discarded in week 1. This can be expressed mathematically as: { R i < N C i - 1 - C D i , i > 1 R i < N C 0 - C D i , i = 1 (13) where N C 0 indicates the original number of container boats. Therefore, the robot buying strategy can be modelled as follows: m i n P ( C B i , O B i , N O i g , N O i t , N C i m , N O i m ) (14) Assuming that 20% of the vascular robots in the human body are destroyed each week, so that the total number of vessel boats and operators changes each week, the total number of vessel boats and operators by changing the composition of a set of functions can be expressed recursively as: N C i = N C i - 1 + C B i - C D i - K × N C i - 1 u (15) N O i = N O i - 1 + O B i - O D i - K × N O i - 1 u (16) where K indicates the percentage of vascular robots destroyed in the human body at this time, here K = 20%. Since the robot’s operators do not work again until after 7 days of maintenance . Therefore, the number of operators under maintenance in week 1 is greater than equal to four times the number of machines in the hospital’s operational demand in week 1 minus the number of operators discarded in that week minus the number of macrophages hit, i.e. N O i m ⩾ R i - 1 - O D i - K × N O i - 1 u (17) Since the newly purchased vessel boats cannot start working until after a week of commissioning, there is: the number of vascular robots required by the hospital in week i must be less than the sum of the total number of vessel boats owned in week i − 1 minus the number of vessel boats discarded in week i and the number of vessel boats destroyed in week i − 1, which can be expressed mathematically as: { R i < N C i - 1 - C D i - K × N C i - 1 u , i > 1 R i < N C 0 - C D i , i = 1 (18) where N C 0 indicates the number of original container boats. Now we need to consider the probability of hitting a macrophage resulting in the complete destruction of the vascular robot, while changing the upper limit of how much each skilled operator can instruct a new operator to learn from G to 20. At this time, 10% of the vascular robots in the human body are destroyed, and the total number of vessel boats and operators changes each week as the amount of destruction changes, and the total number of vessel boats and operators can be expressed recursively as: N C i = N C i - 1 + C B i - C D i - K × N C i - 1 u (19) N O i = N O i - 1 + O B i - O D i - K × N O i - 1 u (20) The number of operators in maintenance is equal to four times the number of machines in the hospital’s operational requirements in week i-1 minus the number of operators discarded in that week minus the number of macrophages hit, i.e. N O i m ⩾ 4 R i - 1 - O D i - K × N O i - 1 u (21) Where K indicates the percentage of vascular robots destroyed in the human body at this time, here K = 10%, while the upper limit of what each skilled operator can instruct new operators to learn G changes to 20 . And for vessel boats, the number of vascular robots needed to have week i hospitals must be less than the sum of the total number of vessel boats owned in week i − 1 minus the number of vessel boats discarded in week i and the number of vessel boats destroyed in week i − 1. The inequality can be expressed as: { R i < N C i - 1 - C D i - K × N C i - 1 u , i > 1 R i < N C 0 - C D i , i = 1 (22) Where N C 0 denotes the number of original vessel boats and K denotes the percentage of vascular robots destroyed in the human body at this time, here K = 10%. The specific analytical thought process is shown in the following . 2.2.2 ARIMA sequences and seasonal sequence forecasting In this section, this paper predicts the demand for the use of vascular robots from 105–112 weeks by ARIMA time series . Time series analysis refers to a set of random variables ordered by time. The main idea of the model is to make dynamic predictions of unknown data based on inter-observations by dependence and correlation. This time model is used to forecast the 105–112 week demand by building this time model, and the implementation of the ARIMA model consists of the following five main steps. Step 1 : Perform data processing, for the mean value of the number of vascular robots used in weeks 1–104 in Annex II y ( t ) can be expressed as : Y ¯ = 1 n ∑ t = 1 n y ( t ) (23) The sample value of the new sequence obtained after its differential processing can be expressed as: x ( t ) = y ( t ) - Y ¯ (24) Generally for d-order difference can be expressed as follows: (∇ d is called the d-order difference operator) . ∇ d X t = ( 1 - B ) d X t (25) The number of differences is determined by the parameter d in the ARIMA (p,d,0) model, i.e., one difference is made, d = 1, i.e., two differences are made, d = 2, i.e., no difference is made, at which point the model structure is changed to ARIMA (p). Step 2 : Parameter estimation: Recursive least squares with forgetting factors can be used for parameter estimation. The forgetting factor enhances the effect of current observations on parameter estimation while weakening the effect of previous observations. The inclusion of the forgetting factor in recursion can take into account the time-varying nature of the model parameters, and the ARIMA (p) model for the sequence y ( t ) can be expressed as : y ( t ) = φ T ( t ) θ + e ( t ) (26) φ T ( t ) = [ y ( t - 1 ) , y ( t - 2 ) , ⋯ , y ( t - p ) ] (27) θ = [ a 1 , a 2 , ⋯ , a p ] T (28) Recursive parameter estimation by substituting φ T , θ into a recursive least squares formulation with a forgetting factor. Step 3 : Forecasting algorithm: the Astrom forecasting method based on the linear minimum variance forecasting principle which can better solve the random geodesic problem in forecasting is used for forecasting, and the ARIMA (p,d,q) process can be expressed as: A ( B ) ∇ d y ( t ) = C ( B ) e ( t ) (29) where y(t), e(t) denote the original sequence and the white noise sequence, respectively. A ( B ) = 1 - a 1 B - a 2 B 2 - ⋯ - a p B p (30) C ( B ) = 1 - c 1 B - c 2 B 2 - ⋯ - c p B p (31) B denotes the back-shift operator is: B n y ( t ) = y ( t - n ) , n = 1 , 2 , ⋯ (32) Minimum variance predictor is: Y ^ ( t + k t ) = G ( B ) C ( B ) y ( t ) (33) Step 4 : model check: this is achieved by checking whether the error series between the original time series and the established model is stochastic; if the model check fails, the model is rebuilt. Step 5 : Export the appropriate prediction model and perform the actual prediction analysis. First, there is need to establish a robot purchase strategy model to make the lowest cost of purchased operators and container boats under the premise of meeting the hospital treatment demand, and the purchase cost is linearly related to the quantity . In order to better control costs, this paper takes into account that there may be cases where the cost of maintenance to the robot vessel boat or operator exceeds the cost of purchase of that part and learning of new and used parts, i.e W O × P O m > P O + 2 P O t (1) W C × P C m > P C (2) Where W O and W C indicate the current number of weeks of maintenance for either operator and vessel, respectively. This document chooses to discard this part when the vessel maintenance cost exceeds the purchase price of a new vessel and the operator maintenance cost exceeds the purchase cost of a new operator and the learning cost of a new or old operator, or when the part was last used. To determine the objective function, the current total number of operators and vessel boats can be expressed as follows,respectively. N C i = ∑ j = 1 i ( C B i - C D i ) + N C o (3) N O i = ∑ j = 1 i ( C O i - C O i ) + N O 0 (4) Where N Ci indicates the total number of vessels owned in week i, N O i u indicates the total number of operators owned in week i, C Bi indicates the number of vessels purchased in week i, and C Di indicates the number of vessels discarded in week i. N C 0 indicates the number of vessels owned at the beginning of week 1, and N O 0 indicates the number of operators owned at the beginning of week 1. Hence the total purchase cost can be expressed as: m i n P ( C B i , O B i , N O i g , N O i t , N C i m , N O i m ) = ( ∑ C B i ) × P C + ( ∑ O B i ) × P O + ( ∑ N O i g + N O i t ) × P O t + ( ∑ N C i m ) × P O m + ( ∑ N O i m ) × P C m (5) Where P C denotes the unit price per vessel boat, and P O denotes the unit price per operator, the P Ot denotes the price of one operator training, and P Om denotes the price of one operator maintenance, and P Cm denotes the total cost required to complete hospital treatment operations. To determine the constraint conditions the relationship between the total number of operators and vessel boats in week 1 and the number in maintenance is shown below: N C i = N C i m + N C i u (6) N C i u = R i (7) N O i = N O i m + N O i u + N O i g + N O i t (8) N O i u = 4 R i (9) N O i t = O B i (10) N ∂ i g = ⌈ O B i G ⌉ (11) Where G indicates the number of new operators each skilled operator can instruct, and the total number of operators owned in week i is equal to the sum of the number of operators in maintenance, use, training instruction, and training at that point. Since the vascular robot must be dismantled after one week of work in the vasculature, the robot’s operators within it cannot work again until after 7 days of maintenance . Therefore, the number of operators under maintenance in week i is equal to four times the number of robots in the hospital’s operational requirements in week i-1 minus the number of operators discarded in that week, i.e. N O i m ≥ 4 R i - 1 - O D i (12) R i −1 denotes the number of robots in demand for hospital operations in week i-1. Since the newly purchased vessel boats cannot start working until after a week of commissioning, there is: the number of vascular robots required by the hospital in week 1 must be less than the total number of vessel boats owned in week 1 minus the number of vessel boats discarded in week 1. This can be expressed mathematically as: { R i < N C i - 1 - C D i , i > 1 R i < N C 0 - C D i , i = 1 (13) where N C 0 indicates the original number of container boats. Therefore, the robot buying strategy can be modelled as follows: m i n P ( C B i , O B i , N O i g , N O i t , N C i m , N O i m ) (14) Assuming that 20% of the vascular robots in the human body are destroyed each week, so that the total number of vessel boats and operators changes each week, the total number of vessel boats and operators by changing the composition of a set of functions can be expressed recursively as: N C i = N C i - 1 + C B i - C D i - K × N C i - 1 u (15) N O i = N O i - 1 + O B i - O D i - K × N O i - 1 u (16) where K indicates the percentage of vascular robots destroyed in the human body at this time, here K = 20%. Since the robot’s operators do not work again until after 7 days of maintenance . Therefore, the number of operators under maintenance in week 1 is greater than equal to four times the number of machines in the hospital’s operational demand in week 1 minus the number of operators discarded in that week minus the number of macrophages hit, i.e. N O i m ⩾ R i - 1 - O D i - K × N O i - 1 u (17) Since the newly purchased vessel boats cannot start working until after a week of commissioning, there is: the number of vascular robots required by the hospital in week i must be less than the sum of the total number of vessel boats owned in week i − 1 minus the number of vessel boats discarded in week i and the number of vessel boats destroyed in week i − 1, which can be expressed mathematically as: { R i < N C i - 1 - C D i - K × N C i - 1 u , i > 1 R i < N C 0 - C D i , i = 1 (18) where N C 0 indicates the number of original container boats. Now we need to consider the probability of hitting a macrophage resulting in the complete destruction of the vascular robot, while changing the upper limit of how much each skilled operator can instruct a new operator to learn from G to 20. At this time, 10% of the vascular robots in the human body are destroyed, and the total number of vessel boats and operators changes each week as the amount of destruction changes, and the total number of vessel boats and operators can be expressed recursively as: N C i = N C i - 1 + C B i - C D i - K × N C i - 1 u (19) N O i = N O i - 1 + O B i - O D i - K × N O i - 1 u (20) The number of operators in maintenance is equal to four times the number of machines in the hospital’s operational requirements in week i-1 minus the number of operators discarded in that week minus the number of macrophages hit, i.e. N O i m ⩾ 4 R i - 1 - O D i - K × N O i - 1 u (21) Where K indicates the percentage of vascular robots destroyed in the human body at this time, here K = 10%, while the upper limit of what each skilled operator can instruct new operators to learn G changes to 20 . And for vessel boats, the number of vascular robots needed to have week i hospitals must be less than the sum of the total number of vessel boats owned in week i − 1 minus the number of vessel boats discarded in week i and the number of vessel boats destroyed in week i − 1. The inequality can be expressed as: { R i < N C i - 1 - C D i - K × N C i - 1 u , i > 1 R i < N C 0 - C D i , i = 1 (22) Where N C 0 denotes the number of original vessel boats and K denotes the percentage of vascular robots destroyed in the human body at this time, here K = 10%. The specific analytical thought process is shown in the following . In this section, this paper predicts the demand for the use of vascular robots from 105–112 weeks by ARIMA time series . Time series analysis refers to a set of random variables ordered by time. The main idea of the model is to make dynamic predictions of unknown data based on inter-observations by dependence and correlation. This time model is used to forecast the 105–112 week demand by building this time model, and the implementation of the ARIMA model consists of the following five main steps. Step 1 : Perform data processing, for the mean value of the number of vascular robots used in weeks 1–104 in Annex II y ( t ) can be expressed as : Y ¯ = 1 n ∑ t = 1 n y ( t ) (23) The sample value of the new sequence obtained after its differential processing can be expressed as: x ( t ) = y ( t ) - Y ¯ (24) Generally for d-order difference can be expressed as follows: (∇ d is called the d-order difference operator) . ∇ d X t = ( 1 - B ) d X t (25) The number of differences is determined by the parameter d in the ARIMA (p,d,0) model, i.e., one difference is made, d = 1, i.e., two differences are made, d = 2, i.e., no difference is made, at which point the model structure is changed to ARIMA (p). Step 2 : Parameter estimation: Recursive least squares with forgetting factors can be used for parameter estimation. The forgetting factor enhances the effect of current observations on parameter estimation while weakening the effect of previous observations. The inclusion of the forgetting factor in recursion can take into account the time-varying nature of the model parameters, and the ARIMA (p) model for the sequence y ( t ) can be expressed as : y ( t ) = φ T ( t ) θ + e ( t ) (26) φ T ( t ) = [ y ( t - 1 ) , y ( t - 2 ) , ⋯ , y ( t - p ) ] (27) θ = [ a 1 , a 2 , ⋯ , a p ] T (28) Recursive parameter estimation by substituting φ T , θ into a recursive least squares formulation with a forgetting factor. Step 3 : Forecasting algorithm: the Astrom forecasting method based on the linear minimum variance forecasting principle which can better solve the random geodesic problem in forecasting is used for forecasting, and the ARIMA (p,d,q) process can be expressed as: A ( B ) ∇ d y ( t ) = C ( B ) e ( t ) (29) where y(t), e(t) denote the original sequence and the white noise sequence, respectively. A ( B ) = 1 - a 1 B - a 2 B 2 - ⋯ - a p B p (30) C ( B ) = 1 - c 1 B - c 2 B 2 - ⋯ - c p B p (31) B denotes the back-shift operator is: B n y ( t ) = y ( t - n ) , n = 1 , 2 , ⋯ (32) Minimum variance predictor is: Y ^ ( t + k t ) = G ( B ) C ( B ) y ( t ) (33) Step 4 : model check: this is achieved by checking whether the error series between the original time series and the established model is stochastic; if the model check fails, the model is rebuilt. Step 5 : Export the appropriate prediction model and perform the actual prediction analysis. In order to solve the difficult problem of training and working with vascular robots, this paper builds a model and then formulates a comprehensive algorithm based on the greedy idea of using simulated annealing to improve the genetic algorithm to solve this problem. Genetic algorithm is a computational model that simulates the mechanism of natural selection and inheritance in biological evolution according to the law of biological evolution. This algorithm seeks the optimal solution by simulating the natural evolutionary process. That is, the process of seeking optimal solutions is transformed into the process of chromosome crossover and gene mutation in biological evolution. When solving more complex combinatorial optimization problems, better optimization results can often be obtained faster than traditional optimization algorithms. Simulated annealing algorithm is a generalized stochastic algorithm whose idea is a probabilistic algorithm derived from the process of annealing a solid. A solid at a sufficiently high temperature is slowly cooled and the particles become ordered from disorder. The approximate solution is continuously optimized as the temperature is reduced. The cooling process is controlled in this problem by simulated annealing ideas combined with a genetic algorithm in order to facilitate global optimization. The specific steps of the annealing idea in the genetic algorithm are as follows. The flowchart of the hybrid genetic algorithm combining simulated annealing and greedy methods is shown in . Step 1 : Set the initial temperature, T, the temperature change coefficient s, and the equilibrium temperature t. Step 2 : Calculate the objective function values and individual fitness and competitive superiority, and perform optimal solution substitution when a solution that better meets the requirements appears. To converge to the global optimum as much as possible, whenever a more optimal solution appears, the current temperature is warmed up to increase the number of iterations and search breadth . The warming function is T = T + 0.5ln( T − 1). Step 3 : Multiple mating and mutation of the parents after roulette selection to produce offspring, with the range of mutation depending on the current temperature. Step 4 : Add the offspring to the current optimal solution and reduce the current temperature, T = T × s , where s is the cooling coefficient. If the current temperature is not reduced to the termination temperature, repeat Step 2 to Step 4 . The termination temperature set in this paper is 0.01. When the experimental temperature is reached, the experiment is terminated. Step 5 : Output the global optimal solution at this point after the current temperature is lower than the termination temperature. Algorithm 1 : Improved Genetic Algorithm Based on Greedy and Simulated Annealing Input : v , number of operators and boats purchased, cumulative price impacts from these purchases Output : Optimal solution 1 sum ← 0 // Total cost of the current solution 2 v ← 0 // Total volume of operations for the current solution 3 label ← 0 // Label of the current solution 4 Function Greedy( v, number of operators, number of boats, price impacts ): 5 for i ← 1 to m do 6 label ← k // k is the number of robots 7 if sum > sumb then 8 while sum > sumb do 9 min ← inf // Minimum cost 10 for j ← 1 to n do // Adjust the number of operators and boats // Update the cost and the number of operations 11 if sum < min then 12 min ← sum m ← k 13 Function Heat( T, best, f ): 14 if f < best then 15 T ← T × 1.1 // Warm up 16 best ← f 17 Function Crossover( parent, T ): // Select two parents and perform crossover and mutation 18 return child 19 Function Update( T, child, best ): // Update solution and temperature 20 while T > 0.01 do // Roulette selection, mating, and mutation // Warm up, update solution // Output the optimal solution The two algorithms are combined in this problem in order to solve the global optimal solution more accurately and with higher confidence. The flow of the genetic algorithm based on the simulated annealing idea after optimization is shown below: Greedy algorithms refer to solving a problem by always making the choice that seems best at the moment. That is, the algorithm obtains a locally optimal solution under the relevant constraints without considering the overall optimality. In this paper, we adopt the greedy idea of minimum cost, from a feasible solution that satisfies the hospital’s treatment volume, we continuously reduce the number of robot parts purchased, in order to achieve the purpose of cost reduction, and we hope that we can finally obtain the optimal solution of the problem through the greedy choice of the number of robots purchased each time. In addition, considering that the data is too cumbersome and belongs to the NP problem, the time complexity of linear programming is too high, so it is considered that the initial optimal solution can be obtained through the greedy idea first, and the preliminary optimal solution obtained by this greediness is used as the initial solution in the genetic algorithm based on the simulated annealing idea. Algorithm 1 embodies the ideas and solutions of the greedy algorithm and the genetic algorithm improved by simulated annealing. In this paper, we adopt a greedy algorithm that aims to minimize costs by continuously reducing the number of robot parts purchased from a feasible solution that satisfies the hospital’s treatment volume. The greedy algorithm terminates when a feasible solution is obtained. This preliminary optimal solution serves as the initial solution for the genetic algorithm based on the simulated annealing idea, which further optimizes the solution. By restructuring and optimizing the algorithm, we conduct a series of experiments on the improved algorithm. Ultimately, it can be concluded that our algorithm has superior performance and accuracy for such application scenarios. 4.1 Experiments The initial optimal solution is first solved by Matlab using the greedy algorithm, and then the initial optimal solution is passed to the genetic algorithm based on the idea of simulated annealing as its initial solution, and then the solution is solved . Specific purchase data are shown in Tables and . When only the constants corresponding to the prices of the purchased robot parts are changed in Matlab, and the same genetic algorithm based on the idea of simulated annealing after optimization is used to solve the problem, the solution results are the 498 container boats and 2308 operators purchased in weeks 1–104 will both satisfy the treatment and minimize the cost, with the minimum cost being: 409,360 yuan. On the genetic algorithm based on the idea of simulated annealing, instead of solving the preliminary optimal solution by the greedy algorithm and then solving it by the genetic algorithm, the algorithm can be used to solve the problem directly as stated earlier. This approach is cost effective and it prepares for the subsequent supply of skilled operators at week 104. is a schematic diagram of the relevant curves in the genetic algorithm based on the idea of simulated annealing after optimization, including the way the objective function change curve, the competitiveness curve of the offspring in the genetic algorithm, and the quenching temperature curve in the idea of simulated annealing to help iterate the genetic algorithm. shows the end iterations of the objective function for solving the problem with the greedy algorithm followed by the genetic algorithm. Obviously, the number of solution iterations of the genetic algorithm is greatly reduced by performing greedy processing first and then performing the genetic algorithm. shows the images of changes in the variables associated with vessel boats, operators, and weekly costs as a function of. When changed the purchase price and the strategy of judging whether to “throw away”, that is, the value of the constant and greedy strategy, the main constraints and objective function did not change significantly. shows a schematic diagram of the relevant curves in the genetic algorithm based on the idea of simulated annealing after optimization. shows that the initial optimal solution is first obtained by the greedy algorithm, and this temporary optimal solution is used as the initial solution in the genetic algorithm model, and then the end iteration of the objective function for solving the problem by improving the genetic algorithm based on the idea of simulated annealing is schematically shown below. Obviously, greedy processing before the initial optimal solution, and then this solution further into the genetic algorithm model for calculation will reduce the number of iterations will be greatly reduced. The above graphs reflect the difference in time complexity between calculating the initial optimal solution by the greedy algorithm and solving it directly by the genetic algorithm, which reflects the necessity and importance of the greedy algorithm in solving the model in this paper. Then, a time series analysis of the demand for the use of vascular robots from 1–104 weeks was conducted, and the final model type selected was ARIMA(3,1,4) . The ARIMA(3,1,4) model built on the original data was run, and the results are shown in . From the above table, the coefficients of AR and MA are -1.016, -0.877, -0.860 versus -1.323, -.718, 0.324, respectively. Since the ACF and PACF plots are fluctuating, it is reasonable to choose ARIMA(3,1,4) in this paper. The results of the model are: X t = ∑ i = 1 p γ i y t - 1 + ϵ t + ∑ i = 1 q θ i ϵ t - 1 (34) Where, γ = [−1.016, −0.877, −0.860], θ = [−1.323, −0.718, 0.324]. Finally, the goodness-of-fit analysis was performed as on ACF and PACF plots of the residuals. The model fit R 2 =0.987, which is close to 1, indicating a good fit and a good fitting effect, and the fitted prediction graph is shown as . 4.2 Discussions In the field of healthcare robotics and resource allocation, several algorithms and approaches have been proposed to address similar optimization problems. This section provides a comparative analysis of our proposed algorithm with respect to existing methods, highlighting the advantages and contributions of our approach. 4.2.1 Traditional methods To demonstrate our method’s efficacy, we conducted comparative analyses with several state-of-the-art approaches, including those by Zhang et al. , Yu et al. , and Deng et al. . For accuracy, we retrained the models from these studies using the same dataset as ours, evaluating both the number of iterations and total cost. indicates that our algorithm significantly surpasses these methods in convergence speed and final cost efficiency. In healthcare resource allocation, traditional heuristics, such as greedy algorithms and rule-based policies, have been prevalent. While these methods offer intuitive and computationally efficient solutions, they generally fall short in identifying globally optimal solutions. By contrast, our method employs systematic exploration of the solution space through mathematical optimization techniques, yielding robust, near-optimal results. Therefore, to validate the accuracy and robustness of our method, we compare it with other swarm optimization algorithms, including PSO , CGWO , GA , and AOA . Since they require simultaneous iterations, this experiment will compare their convergence speed and accuracy. The results of the comparison are shown in . Analysis of the images reveals that the enhanced genetic algorithm, which integrates greedy search and simulated annealing, surpasses competing algorithms in convergence speed and accuracy. This superiority stems from the initial application of the greedy search to rapidly identify a near-optimal solution, which is subsequently refined by the genetic algorithm. This strategy effectively circumvents the issue of entrapment in local optima, a common pitfall in swarm algorithms (e.g., in PSO, AOA curves). Additionally, the greedy search’s expedited discovery of an approximate optimal solution, followed by the precision-enhancing simulated annealing, results in the improved genetic algorithm’s superior performance in terms of both convergence speed and accuracy. 4.2.2 Machine learning approaches Machine learning techniques, including neural networks and reinforcement learning, have been applied to resource allocation problems in healthcare . While these methods can adapt to complex patterns and dynamic environments, they often require substantial amounts of data for training and may not provide interpretable solutions. In contrast, our approach relies on transparent mathematical models, allowing for better understanding and control of the optimization process. Some recent approaches combine optimization and machine learning components to address resource allocation challenges. While hybrid methods can leverage the strengths of both paradigms, they may introduce additional complexity and computational overhead . Our algorithm offers a balance by providing efficient optimization while maintaining transparency and ease of implementation. However, we also believe that machine-learning approaches have tremendous potential and room for optimization in the medical field. The initial optimal solution is first solved by Matlab using the greedy algorithm, and then the initial optimal solution is passed to the genetic algorithm based on the idea of simulated annealing as its initial solution, and then the solution is solved . Specific purchase data are shown in Tables and . When only the constants corresponding to the prices of the purchased robot parts are changed in Matlab, and the same genetic algorithm based on the idea of simulated annealing after optimization is used to solve the problem, the solution results are the 498 container boats and 2308 operators purchased in weeks 1–104 will both satisfy the treatment and minimize the cost, with the minimum cost being: 409,360 yuan. On the genetic algorithm based on the idea of simulated annealing, instead of solving the preliminary optimal solution by the greedy algorithm and then solving it by the genetic algorithm, the algorithm can be used to solve the problem directly as stated earlier. This approach is cost effective and it prepares for the subsequent supply of skilled operators at week 104. is a schematic diagram of the relevant curves in the genetic algorithm based on the idea of simulated annealing after optimization, including the way the objective function change curve, the competitiveness curve of the offspring in the genetic algorithm, and the quenching temperature curve in the idea of simulated annealing to help iterate the genetic algorithm. shows the end iterations of the objective function for solving the problem with the greedy algorithm followed by the genetic algorithm. Obviously, the number of solution iterations of the genetic algorithm is greatly reduced by performing greedy processing first and then performing the genetic algorithm. shows the images of changes in the variables associated with vessel boats, operators, and weekly costs as a function of. When changed the purchase price and the strategy of judging whether to “throw away”, that is, the value of the constant and greedy strategy, the main constraints and objective function did not change significantly. shows a schematic diagram of the relevant curves in the genetic algorithm based on the idea of simulated annealing after optimization. shows that the initial optimal solution is first obtained by the greedy algorithm, and this temporary optimal solution is used as the initial solution in the genetic algorithm model, and then the end iteration of the objective function for solving the problem by improving the genetic algorithm based on the idea of simulated annealing is schematically shown below. Obviously, greedy processing before the initial optimal solution, and then this solution further into the genetic algorithm model for calculation will reduce the number of iterations will be greatly reduced. The above graphs reflect the difference in time complexity between calculating the initial optimal solution by the greedy algorithm and solving it directly by the genetic algorithm, which reflects the necessity and importance of the greedy algorithm in solving the model in this paper. Then, a time series analysis of the demand for the use of vascular robots from 1–104 weeks was conducted, and the final model type selected was ARIMA(3,1,4) . The ARIMA(3,1,4) model built on the original data was run, and the results are shown in . From the above table, the coefficients of AR and MA are -1.016, -0.877, -0.860 versus -1.323, -.718, 0.324, respectively. Since the ACF and PACF plots are fluctuating, it is reasonable to choose ARIMA(3,1,4) in this paper. The results of the model are: X t = ∑ i = 1 p γ i y t - 1 + ϵ t + ∑ i = 1 q θ i ϵ t - 1 (34) Where, γ = [−1.016, −0.877, −0.860], θ = [−1.323, −0.718, 0.324]. Finally, the goodness-of-fit analysis was performed as on ACF and PACF plots of the residuals. The model fit R 2 =0.987, which is close to 1, indicating a good fit and a good fitting effect, and the fitted prediction graph is shown as . In the field of healthcare robotics and resource allocation, several algorithms and approaches have been proposed to address similar optimization problems. This section provides a comparative analysis of our proposed algorithm with respect to existing methods, highlighting the advantages and contributions of our approach. 4.2.1 Traditional methods To demonstrate our method’s efficacy, we conducted comparative analyses with several state-of-the-art approaches, including those by Zhang et al. , Yu et al. , and Deng et al. . For accuracy, we retrained the models from these studies using the same dataset as ours, evaluating both the number of iterations and total cost. indicates that our algorithm significantly surpasses these methods in convergence speed and final cost efficiency. In healthcare resource allocation, traditional heuristics, such as greedy algorithms and rule-based policies, have been prevalent. While these methods offer intuitive and computationally efficient solutions, they generally fall short in identifying globally optimal solutions. By contrast, our method employs systematic exploration of the solution space through mathematical optimization techniques, yielding robust, near-optimal results. Therefore, to validate the accuracy and robustness of our method, we compare it with other swarm optimization algorithms, including PSO , CGWO , GA , and AOA . Since they require simultaneous iterations, this experiment will compare their convergence speed and accuracy. The results of the comparison are shown in . Analysis of the images reveals that the enhanced genetic algorithm, which integrates greedy search and simulated annealing, surpasses competing algorithms in convergence speed and accuracy. This superiority stems from the initial application of the greedy search to rapidly identify a near-optimal solution, which is subsequently refined by the genetic algorithm. This strategy effectively circumvents the issue of entrapment in local optima, a common pitfall in swarm algorithms (e.g., in PSO, AOA curves). Additionally, the greedy search’s expedited discovery of an approximate optimal solution, followed by the precision-enhancing simulated annealing, results in the improved genetic algorithm’s superior performance in terms of both convergence speed and accuracy. 4.2.2 Machine learning approaches Machine learning techniques, including neural networks and reinforcement learning, have been applied to resource allocation problems in healthcare . While these methods can adapt to complex patterns and dynamic environments, they often require substantial amounts of data for training and may not provide interpretable solutions. In contrast, our approach relies on transparent mathematical models, allowing for better understanding and control of the optimization process. Some recent approaches combine optimization and machine learning components to address resource allocation challenges. While hybrid methods can leverage the strengths of both paradigms, they may introduce additional complexity and computational overhead . Our algorithm offers a balance by providing efficient optimization while maintaining transparency and ease of implementation. However, we also believe that machine-learning approaches have tremendous potential and room for optimization in the medical field. To demonstrate our method’s efficacy, we conducted comparative analyses with several state-of-the-art approaches, including those by Zhang et al. , Yu et al. , and Deng et al. . For accuracy, we retrained the models from these studies using the same dataset as ours, evaluating both the number of iterations and total cost. indicates that our algorithm significantly surpasses these methods in convergence speed and final cost efficiency. In healthcare resource allocation, traditional heuristics, such as greedy algorithms and rule-based policies, have been prevalent. While these methods offer intuitive and computationally efficient solutions, they generally fall short in identifying globally optimal solutions. By contrast, our method employs systematic exploration of the solution space through mathematical optimization techniques, yielding robust, near-optimal results. Therefore, to validate the accuracy and robustness of our method, we compare it with other swarm optimization algorithms, including PSO , CGWO , GA , and AOA . Since they require simultaneous iterations, this experiment will compare their convergence speed and accuracy. The results of the comparison are shown in . Analysis of the images reveals that the enhanced genetic algorithm, which integrates greedy search and simulated annealing, surpasses competing algorithms in convergence speed and accuracy. This superiority stems from the initial application of the greedy search to rapidly identify a near-optimal solution, which is subsequently refined by the genetic algorithm. This strategy effectively circumvents the issue of entrapment in local optima, a common pitfall in swarm algorithms (e.g., in PSO, AOA curves). Additionally, the greedy search’s expedited discovery of an approximate optimal solution, followed by the precision-enhancing simulated annealing, results in the improved genetic algorithm’s superior performance in terms of both convergence speed and accuracy. Machine learning techniques, including neural networks and reinforcement learning, have been applied to resource allocation problems in healthcare . While these methods can adapt to complex patterns and dynamic environments, they often require substantial amounts of data for training and may not provide interpretable solutions. In contrast, our approach relies on transparent mathematical models, allowing for better understanding and control of the optimization process. Some recent approaches combine optimization and machine learning components to address resource allocation challenges. While hybrid methods can leverage the strengths of both paradigms, they may introduce additional complexity and computational overhead . Our algorithm offers a balance by providing efficient optimization while maintaining transparency and ease of implementation. However, we also believe that machine-learning approaches have tremendous potential and room for optimization in the medical field. This research have addressed the pressing challenges of optimizing robotic operator and vessel boat acquisition strategies within the dynamic healthcare environment. Our research objectives, including the development of a robust buying strategy model, adaptability to macrophage attacks, consideration of skilled operator variations, and the creation of a comprehensive framework, have all been successfully achieved. This study significantly contributes to healthcare robotics by bridging existing gaps in the literature and offering practical solutions that enhance cost-effectiveness, treatment efficiency, and resource allocation. The potential impact of our research on medical treatment, particularly in vascular diseases and virus removal, is substantial. While this work represents a significant advancement, future research can explore real-world implementation and further incorporate advanced technologies for even greater adaptability and prediction accuracy. Overall, our study marks a pivotal step in the evolution of healthcare robotics, with far-reaching implications for patient care and well-being. S1 Dataset (XLSX) S1 File (DOCX) |
Changing knowledge, attitudes and behaviours towards cytomegalovirus in pregnancy through film-based antenatal education: a feasibility randomised controlled trial of a digital educational intervention | ae46f84a-10aa-464f-9133-354021586708 | 8375137 | Patient Education as Topic[mh] | Congenital cytomegalovirus (CMV) is the commonest congenital infection globally and has a birth prevalence of 0.3–1% [ – ]. Congenital CMV (cCMV) can occur following the first infection with CMV during pregnancy (primary CMV infection), after reactivation of CMV acquired previously or following infection with a different strain of CMV (secondary CMV infection). The risk of transmission to the fetus is significantly higher in primary infection than in secondary infection . Despite this, globally more infants with cCMV are born to mothers with secondary infection than with primary infection due to the high CMV seroprevalence in many parts of the world . CMV is transmitted through contact with infected bodily fluids and those people who have a child, or children, already are at increased risk of acquiring the infection, primarily through contact with infected saliva or urine from their young child . The clinical spectrum of cCMV at birth is wide: around 85% of infants will be ‘asymptomatic’ and 15% will have symptoms at birth . Long term sequelae occur in about 40–60% of babies who are symptomatic at birth, and 10–15% of babies who are asymptomatic . The most common long-term effect of cCMV is sensorineural hearing loss, with cCMV being the most frequent non-genetic cause of sensorineural hearing loss and the only preventable cause . cCMV represents a significant public health problem, but there are currently no licensed vaccines and no routinely recommended treatments for antenatal CMV infection. The United Kingdom (UK) currently has no national screening programme for CMV for pregnant women or infants, and women are not routinely counselled about CMV risk reduction measures. Antenatal education about CMV risk reduction may provide a significant opportunity to reduce CMV infection in pregnancy and consequently reduce the incidence of cCMV. In a recent systematic review, seven studies were identified which investigated preventative hygiene-based interventions in pregnancy or in women of child-bearing age . This concluded that hygiene-based interventions in pregnancy could play a useful role in primary prevention of CMV infection in pregnancy, however the studies were too heterogeneous in terms of study population, intervention and outcome to form firm conclusions on the relative impact of such interventions. Additionally, the majority of interventions would not be easily translatable to routine antenatal care, without the provision of significant additional resources. A randomised controlled trial (RCT) using an acceptable educational intervention—which can be subsequently integrated into routine care in the UK—is urgently needed. RACE FIT (Reducing Acquisition of CMV through antenatal Education) was designed to inform the feasibility and design of a large-scale RCT in a UK setting to investigate the efficacy of the educational intervention on the risk of acquiring CMV infection in pregnancy. It was designed in two phases, the first of these involved in-depth interviews with pregnant women and the families of children affected by cCMV. These interviews explored their knowledge and attitudes about CMV, and perspectives on infection prevention in pregnancy, in order to prioritise themes to include in the intervention . From these findings, a script was produced and the digital intervention developed as a short educational film through an iterative process involving review by pregnant or recently pregnant women, families affected by CMV, and knowledge experts. The aim of this second phase of RACE FIT was to test the digital intervention in a feasibility study where women were randomised to the intervention or treatment as usual groups, in order to provide information about recruitment and conduct of a future trial, assess the acceptability of the educational intervention and explore changes in knowledge, attitudes and behaviours in the two groups. We also determined CMV seroconversion in both groups. The overarching aim was to inform the feasibility and design of a large-scale randomised controlled trial (RCT) in a UK setting to investigate the efficacy of the digital, antenatal educational intervention on the risk of acquiring CMV infection in pregnancy.
Study setting and screening We recruited all participants from a single teaching hospital in an ethnically diverse area of South-West London. We approached women in their first trimester of pregnancy who were attending antenatal clinics between September 2018 and September 2019; women who lived with a child or children less than four years of age were asked for their consent for CMV serology to be undertaken on an additional blood sample. All women in the study were tested for both CMV IgG and IgM antibodies. Women who were seronegative (no evidence of previous CMV infection; IgG negative) were invited to take part in the RCT; those who were CMV IgG positive and CMV IgM negative were not eligible to take part and those who were CMV IgM positive had additional serology undertaken including CMV IgG avidity testing. Women with serological evidence of recent CMV infection were referred for counselling and further investigation under an established routine clinical pathway. Eligibility Women were considered to be eligible for the RCT if they were aged over 18, pregnant, willing and able to provide informed consent, seronegative for CMV, having no documented immunodeficiency, living with at least one child aged less than four and willing to be followed up until delivery. Randomisation After providing informed consent, participants were randomised in a 1:1 ratio to the intervention or treatment as usual group using the randomisation service provided by the King’s College Clinical Trials Unit. The randomisation sequence was computer generated. Neither the participant nor the researcher was blinded to group allocation. Study materials Participants who were randomised to the intervention group watched the educational film—developed in phase one—at their first study visit. The film was made up of three parts: a presentation of facts about CMV, including prevalence and routes of transmission; families of affected children telling their stories; and advice provided about how the risk of infection could be reduced (Supplementary material ). Participants in the treatment as usual group viewed a series of slides about influenza vaccination in pregnancy. Influenza vaccination is routinely recommended in pregnancy in the UK and all pregnant women receive information about this as part of routine care. Study design The study was approved by the NHS Health Research Authority and South-Central Oxford Research Ethics Committee (16/SC/0683). Women had their first study visit at home or in clinic before 16 gestational weeks. Following informed written consent, all participants completed a questionnaire (Supplementary material ) and were then randomised into either the intervention or treatment as usual groups. Participants then either watched the digital educational intervention (intervention group) or reviewed a series of slides about influenza vaccination in pregnancy (treatment as usual group) and then immediately completed a second questionnaire about the materials they had been presented with (Supplementary material and ). At 34 gestational weeks, participants completed a final online questionnaire (Supplementary material and ). Within two weeks of delivery a blood sample was obtained from all participating mothers. This was tested for CMV specific IgG and IgM antibody to assess for seroconversion over the study period. Clinical follow up was organised for those participants and their infants who were found to have seroconverted since initial screening. Measures. Participant demographics Information was collected about age, marital status, ethnicity, length of residence in the UK, qualifications, number of previous pregnancies, number of children under four years of age and whether participants worked regularly with children as part of their job. Familiarity with CMV At baseline, participants indicated how familiar they were with a range of conditions affecting newborns, including CMV, and about how common they thought these conditions were . Response to materials At the first study visit, the intervention group provided their responses to the educational film by indicating their level of agreement with a range of statements. For the following domains, participants in both groups were asked for their responses at baseline and at 34 weeks: Knowledge of CMV: Participants were asked to specify their level of agreement with 12 statements about CMV . These included both true and false statements. Perceived severity and susceptibility Participants were asked to indicate their level of agreement with statements about the severity of CMV and their perceived susceptibility to CMV. Anxiety and depression scores Participants were asked to indicate how they had been feeling recently using the Kessler Psychological Distress Scale and the Edinburgh Postnatal Depression Scale . Daily activities Participants were asked how often they engaged in a range of behaviours relating to contact with a child’s saliva, urine or faeces. At 34 gestational weeks, participants were asked to indicate how hard it had been to make the suggested behavioural changes. Laboratory methods CMV IgM and IgG were measured using the Roche Elecsys assay (Roche, Switzerland), according to manufacturer’s instructions. For individuals who were found to be CMV IgM positive further testing was performed for IgG avidity using the VIDAS CMV IgG avidity 11 assay (Biomerieux, France). Data collection and analysis Study data were collected and managed using REDCap electronic data capture tools hosted at St George’s, University of London. Statistical analyses Data were graphically explored and summarised. Anxiety and depression scores, which exhibited a wide range of values (additive scores), were treated as continuous data. Outcomes reflecting measurements for familiarity, attitudes, behaviour and knowledge were of ordinal type. Missing responses were assessed for each variable of interest. Both per-protocol (PP) and intention-to-treat (ITT) analyses were conducted . Given the randomisation, permutation tests have been conducted for between groups comparisons assuming that the missing observations were completely at random [ – ]. Similar assumptions were considered for within groups’ comparisons. The PP and ITT analyses did not show markable qualitative differences for any of the outcomes. This study aimed to detect potentially important signals to be investigated in a larger trial and was not designed as a hypotheses testing study. Given the exploratory phase of this research, classical Bonferroni corrections for multiple outcome testing were not applied. All analyses and graphics have been produced using STATA 16 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC).
We recruited all participants from a single teaching hospital in an ethnically diverse area of South-West London. We approached women in their first trimester of pregnancy who were attending antenatal clinics between September 2018 and September 2019; women who lived with a child or children less than four years of age were asked for their consent for CMV serology to be undertaken on an additional blood sample. All women in the study were tested for both CMV IgG and IgM antibodies. Women who were seronegative (no evidence of previous CMV infection; IgG negative) were invited to take part in the RCT; those who were CMV IgG positive and CMV IgM negative were not eligible to take part and those who were CMV IgM positive had additional serology undertaken including CMV IgG avidity testing. Women with serological evidence of recent CMV infection were referred for counselling and further investigation under an established routine clinical pathway.
Women were considered to be eligible for the RCT if they were aged over 18, pregnant, willing and able to provide informed consent, seronegative for CMV, having no documented immunodeficiency, living with at least one child aged less than four and willing to be followed up until delivery.
After providing informed consent, participants were randomised in a 1:1 ratio to the intervention or treatment as usual group using the randomisation service provided by the King’s College Clinical Trials Unit. The randomisation sequence was computer generated. Neither the participant nor the researcher was blinded to group allocation.
Participants who were randomised to the intervention group watched the educational film—developed in phase one—at their first study visit. The film was made up of three parts: a presentation of facts about CMV, including prevalence and routes of transmission; families of affected children telling their stories; and advice provided about how the risk of infection could be reduced (Supplementary material ). Participants in the treatment as usual group viewed a series of slides about influenza vaccination in pregnancy. Influenza vaccination is routinely recommended in pregnancy in the UK and all pregnant women receive information about this as part of routine care.
The study was approved by the NHS Health Research Authority and South-Central Oxford Research Ethics Committee (16/SC/0683). Women had their first study visit at home or in clinic before 16 gestational weeks. Following informed written consent, all participants completed a questionnaire (Supplementary material ) and were then randomised into either the intervention or treatment as usual groups. Participants then either watched the digital educational intervention (intervention group) or reviewed a series of slides about influenza vaccination in pregnancy (treatment as usual group) and then immediately completed a second questionnaire about the materials they had been presented with (Supplementary material and ). At 34 gestational weeks, participants completed a final online questionnaire (Supplementary material and ). Within two weeks of delivery a blood sample was obtained from all participating mothers. This was tested for CMV specific IgG and IgM antibody to assess for seroconversion over the study period. Clinical follow up was organised for those participants and their infants who were found to have seroconverted since initial screening.
Participant demographics Information was collected about age, marital status, ethnicity, length of residence in the UK, qualifications, number of previous pregnancies, number of children under four years of age and whether participants worked regularly with children as part of their job. Familiarity with CMV At baseline, participants indicated how familiar they were with a range of conditions affecting newborns, including CMV, and about how common they thought these conditions were . Response to materials At the first study visit, the intervention group provided their responses to the educational film by indicating their level of agreement with a range of statements. For the following domains, participants in both groups were asked for their responses at baseline and at 34 weeks: Knowledge of CMV: Participants were asked to specify their level of agreement with 12 statements about CMV . These included both true and false statements. Perceived severity and susceptibility Participants were asked to indicate their level of agreement with statements about the severity of CMV and their perceived susceptibility to CMV. Anxiety and depression scores Participants were asked to indicate how they had been feeling recently using the Kessler Psychological Distress Scale and the Edinburgh Postnatal Depression Scale . Daily activities Participants were asked how often they engaged in a range of behaviours relating to contact with a child’s saliva, urine or faeces. At 34 gestational weeks, participants were asked to indicate how hard it had been to make the suggested behavioural changes.
Information was collected about age, marital status, ethnicity, length of residence in the UK, qualifications, number of previous pregnancies, number of children under four years of age and whether participants worked regularly with children as part of their job.
At baseline, participants indicated how familiar they were with a range of conditions affecting newborns, including CMV, and about how common they thought these conditions were .
At the first study visit, the intervention group provided their responses to the educational film by indicating their level of agreement with a range of statements. For the following domains, participants in both groups were asked for their responses at baseline and at 34 weeks:
Participants were asked to specify their level of agreement with 12 statements about CMV . These included both true and false statements.
Participants were asked to indicate their level of agreement with statements about the severity of CMV and their perceived susceptibility to CMV.
Participants were asked to indicate how they had been feeling recently using the Kessler Psychological Distress Scale and the Edinburgh Postnatal Depression Scale .
Participants were asked how often they engaged in a range of behaviours relating to contact with a child’s saliva, urine or faeces. At 34 gestational weeks, participants were asked to indicate how hard it had been to make the suggested behavioural changes.
CMV IgM and IgG were measured using the Roche Elecsys assay (Roche, Switzerland), according to manufacturer’s instructions. For individuals who were found to be CMV IgM positive further testing was performed for IgG avidity using the VIDAS CMV IgG avidity 11 assay (Biomerieux, France).
Study data were collected and managed using REDCap electronic data capture tools hosted at St George’s, University of London.
Data were graphically explored and summarised. Anxiety and depression scores, which exhibited a wide range of values (additive scores), were treated as continuous data. Outcomes reflecting measurements for familiarity, attitudes, behaviour and knowledge were of ordinal type. Missing responses were assessed for each variable of interest. Both per-protocol (PP) and intention-to-treat (ITT) analyses were conducted . Given the randomisation, permutation tests have been conducted for between groups comparisons assuming that the missing observations were completely at random [ – ]. Similar assumptions were considered for within groups’ comparisons. The PP and ITT analyses did not show markable qualitative differences for any of the outcomes. This study aimed to detect potentially important signals to be investigated in a larger trial and was not designed as a hypotheses testing study. Given the exploratory phase of this research, classical Bonferroni corrections for multiple outcome testing were not applied. All analyses and graphics have been produced using STATA 16 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC).
Screening for participation A large number of women were approached about the study ( n = 3975), of whom 878 (22%) had a blood sample taken for CMV serology, Fig. . The most common reason for ineligibility for blood sampling was not living with a child aged less than four ( n = 2751; 88.8%). Overall, 43% ( n = 372) of participants were seronegative and eligible to be approached about participation in the RCT and 57% ( n = 493) of women were seropositive. Of all the women screened, ten (1.16%) had evidence indicating recent infection, within the last three months, and were referred to the Fetal Medicine Unit for further clinical investigation. Details of ethnicity were available for 532 women screened. The proportion of women who were seronegative varied by self-defined ethnicity: 61% White British ( n = 172), 39% White Other ( n = 32), 6% Black ( n = 3), 22% South Asian ( n = 21), 14% Asian Other ( n = 2), 46% Mixed ( n = 6). Feasibility randomized controlled trial Of the 372 women who were CMV seronegative, 103 women consented to participate in the RCT (27.7%), of whom 87 (84%) participants completed the study (Fig. ). Study completion was defined as collection of a final blood sample or completion of a 34-week questionnaire. Recruitment ended at the conclusion of the pre-defined recruitment period of 12 months. At that time, we had recruited about 25% of the initially planned recruitment number. Participant characteristics The demographic characteristics of the participants are shown in Table . Familiarity with CMV and other conditions On enrolment to the study, most participants who responded were unfamiliar with CMV; 64% ( n = 66) of participants reported that they were ‘not at all familiar’ with CMV compared with 1% ( n = 1) for Trisomy 21, 9% for rubella ( n = 9), 13% ( n = 13) for listeria and 32% ( n = 32) for toxoplasmosis. There was no evidence to suggest any difference between the distribution of the responses in the two randomisation groups (Supplementary material ; Fig. ). Participants' knowledge about CMV Knowledge about how CMV is transmitted and what effect congenital CMV can have on infants was consistent between randomisation groups, with no differences at baseline. At 34 gestational weeks, knowledge about CMV was significantly different between participants in the intervention group and participants in the treatment as usual group; a higher proportion of participants in the intervention group correctly agreed that CMV can be spread through saliva and urine, and could cause hearing loss and adverse neurodevelopmental outcomes, Table . Within the intervention group, there was a significant difference in knowledge about transmission of CMV and the potential consequences of congenital CMV for the infant or child, at baseline compared to 34 gestational weeks, Table . Knowledge about how CMV can be transmitted was also different at 34 gestational weeks compared to baseline in the treatment as usual group, however there was not a significant difference in knowledge about the impact of CMV on hearing and development in this group, suggesting the participants gained some knowledge about CMV during the study period, despite not being exposed to the intervention, Table . Perception of severity and susceptibility At baseline, participants’ perceptions about the severity of CMV and susceptibility to CMV were similar in the intervention and treatment as usual groups, Fig. . After the intervention (34 gestational weeks), a higher proportion of participants in the intervention group were likely to consider CMV to be serious and themselves personally susceptible to CMV, and to agree that advice about CMV should be given to pregnant women, compared to before the intervention (at baseline), Fig. . In contrast, the attitudes of pregnant women towards CMV in the treatment as usual group were similar at baseline and 34 gestational weeks, Fig. . Risk behaviours for CMV At baseline, participants in the treatment as usual and intervention groups reported similar engagement with activities which might expose them to saliva or urine of children, for example commonly reporting eating left-over food from a child’s plate, Table . Within the intervention group, women reported eating left-over food, drinking from a child’s cup or kissing their child directly on the lips, less frequently after the intervention compared to before the intervention, Table . Differences in behaviours of women in the treatment as usual group were also observed between baseline and 34 gestational weeks, Table . Despite some differences in the frequency at which participants engaged in these activities in both groups, there was a difference between the two groups at 34 gestational weeks, with women in the intervention group reporting eating left-over food and kissing on the lips less frequently than women in the treatment as usual group, Table . Acceptability of educational intervention Participants in the intervention group responded positively to the educational film, reporting that they felt motivated to change activities and felt confident that they could do so and would recommend the film to friends, Table . Anxiety, depression There were no significant differences observed between scores on the Kessler Psychological Distress Scale or the Edinburgh Postnatal Depression Scale between the intervention and treatment as usual groups at baseline or at 34 weeks (Supplementary material ; table 1). Seroconversion Seroconversion between the end of the first trimester (baseline) and 34 gestational weeks was 4.55% in the intervention group and 4.65% in the treatment as usual group. There was one newborn infant, born to a mother in the intervention group who had seroconverted during pregnancy, who tested CMV PCR positive in urine at birth and therefore had congenital infection. The infant had no clinical features of cCMV and no treatment was required. The infant remained well with no clinical features of congenital CMV at 12 months of age.
A large number of women were approached about the study ( n = 3975), of whom 878 (22%) had a blood sample taken for CMV serology, Fig. . The most common reason for ineligibility for blood sampling was not living with a child aged less than four ( n = 2751; 88.8%). Overall, 43% ( n = 372) of participants were seronegative and eligible to be approached about participation in the RCT and 57% ( n = 493) of women were seropositive. Of all the women screened, ten (1.16%) had evidence indicating recent infection, within the last three months, and were referred to the Fetal Medicine Unit for further clinical investigation. Details of ethnicity were available for 532 women screened. The proportion of women who were seronegative varied by self-defined ethnicity: 61% White British ( n = 172), 39% White Other ( n = 32), 6% Black ( n = 3), 22% South Asian ( n = 21), 14% Asian Other ( n = 2), 46% Mixed ( n = 6).
Of the 372 women who were CMV seronegative, 103 women consented to participate in the RCT (27.7%), of whom 87 (84%) participants completed the study (Fig. ). Study completion was defined as collection of a final blood sample or completion of a 34-week questionnaire. Recruitment ended at the conclusion of the pre-defined recruitment period of 12 months. At that time, we had recruited about 25% of the initially planned recruitment number.
The demographic characteristics of the participants are shown in Table .
On enrolment to the study, most participants who responded were unfamiliar with CMV; 64% ( n = 66) of participants reported that they were ‘not at all familiar’ with CMV compared with 1% ( n = 1) for Trisomy 21, 9% for rubella ( n = 9), 13% ( n = 13) for listeria and 32% ( n = 32) for toxoplasmosis. There was no evidence to suggest any difference between the distribution of the responses in the two randomisation groups (Supplementary material ; Fig. ).
Knowledge about how CMV is transmitted and what effect congenital CMV can have on infants was consistent between randomisation groups, with no differences at baseline. At 34 gestational weeks, knowledge about CMV was significantly different between participants in the intervention group and participants in the treatment as usual group; a higher proportion of participants in the intervention group correctly agreed that CMV can be spread through saliva and urine, and could cause hearing loss and adverse neurodevelopmental outcomes, Table . Within the intervention group, there was a significant difference in knowledge about transmission of CMV and the potential consequences of congenital CMV for the infant or child, at baseline compared to 34 gestational weeks, Table . Knowledge about how CMV can be transmitted was also different at 34 gestational weeks compared to baseline in the treatment as usual group, however there was not a significant difference in knowledge about the impact of CMV on hearing and development in this group, suggesting the participants gained some knowledge about CMV during the study period, despite not being exposed to the intervention, Table .
At baseline, participants’ perceptions about the severity of CMV and susceptibility to CMV were similar in the intervention and treatment as usual groups, Fig. . After the intervention (34 gestational weeks), a higher proportion of participants in the intervention group were likely to consider CMV to be serious and themselves personally susceptible to CMV, and to agree that advice about CMV should be given to pregnant women, compared to before the intervention (at baseline), Fig. . In contrast, the attitudes of pregnant women towards CMV in the treatment as usual group were similar at baseline and 34 gestational weeks, Fig. .
At baseline, participants in the treatment as usual and intervention groups reported similar engagement with activities which might expose them to saliva or urine of children, for example commonly reporting eating left-over food from a child’s plate, Table . Within the intervention group, women reported eating left-over food, drinking from a child’s cup or kissing their child directly on the lips, less frequently after the intervention compared to before the intervention, Table . Differences in behaviours of women in the treatment as usual group were also observed between baseline and 34 gestational weeks, Table . Despite some differences in the frequency at which participants engaged in these activities in both groups, there was a difference between the two groups at 34 gestational weeks, with women in the intervention group reporting eating left-over food and kissing on the lips less frequently than women in the treatment as usual group, Table .
Participants in the intervention group responded positively to the educational film, reporting that they felt motivated to change activities and felt confident that they could do so and would recommend the film to friends, Table .
There were no significant differences observed between scores on the Kessler Psychological Distress Scale or the Edinburgh Postnatal Depression Scale between the intervention and treatment as usual groups at baseline or at 34 weeks (Supplementary material ; table 1).
Seroconversion between the end of the first trimester (baseline) and 34 gestational weeks was 4.55% in the intervention group and 4.65% in the treatment as usual group. There was one newborn infant, born to a mother in the intervention group who had seroconverted during pregnancy, who tested CMV PCR positive in urine at birth and therefore had congenital infection. The infant had no clinical features of cCMV and no treatment was required. The infant remained well with no clinical features of congenital CMV at 12 months of age.
This feasibility study demonstrates that recruitment to a future randomised controlled trial investigating the efficacy of a film-based educational intervention in reducing the risk of acquiring CMV infection in pregnancy would be feasible and has generated essential data upon which to design and power a larger RCT. This single-centre randomised controlled trial has shown that digital antenatal education about CMV is acceptable and accessible to pregnant women and does increase knowledge about CMV, change attitudes towards personal susceptibility and severity, and that pregnant women were willing to adopt risk-reducing behaviour change to reduce exposure to saliva and urine of young children. A future large multi-centre randomised controlled trial would be needed to determine whether such changes in knowledge, attitudes and behaviour would have an impact on seroconversion in pregnancy and therefore prevention of congenital CMV. In this feasibility study, we have been able to identify factors which would be crucial to the design of a multi-centre randomised controlled trial. To determine the efficacy of an educational intervention, it is necessary to identify and enrol seronegative women in order to demonstrate seroconversion – thus acquisition of infection. We have shown that testing for CMV serology is highly acceptable to pregnant women in the first trimester of pregnancy; 2.86% (n = 144) of women declined testing for CMV antibodies, suggesting that the vast majority of women would be willing to be screened for CMV infection in pregnancy, in the NHS setting, and is consistent with that reported in other studies . Multiparous seronegative women who have young children are at the highest risk of acquiring infection and transmitting this to their fetus, therefore these women would be the target population for future studies. We have demonstrated the challenges in identifying and enrolling this target population. A large number of women were ineligible for the study (n = 2320; 58.4%) because they were primiparous (this was their first pregnancy) and of those who were multiparous, a further 431 women were excluded because they did not have a child < 4 years of age. Together with the women who were ineligible for other reasons (n = 202) or for whom no sample was obtained (n = 13), only 878 (22%) of the 3975 women approached had a blood sample for CMV screening obtained. These factors are critical to take into account when designing and assessing the feasibility of future studies. Of the women who consented for CMV screening, 43% were seronegative and therefore at risk of primary CMV infection and eligible for the study, and 57% of women were seropositive. The proportion of women who were seropositive varied considerably with ethnicity. The seropositivity in white women of 39% is similar to that seen in previous studies (45.9% Tookey, 1992; 49% Pembrey, 2013) . However, we found lower seropositivity in women from South Asian ethnicity (78%) compared to that seen in the cohort of pregnant women in Bradford (89%—98%) and higher seropositivity in black women (94%) than has been observed in a population of women attending antenatal care in London in the 1990s (77%) . Both Tookey et al. and Pembrey et al . found place of birth, as well as ethnicity to be important in seroprevalence, with British born women less likely to be seropositive . We did not collect information about place of birth and so were unable to investigate this aspect. Because of the eligibility requirements of the studies being recruited for, we only screened women living with a child aged less than four years, which may mean that this population is not completely representative of the pregnant population as a whole, but does represent women who are likely to be at the highest risk of infection in pregnancy. A total of ten women (1.16%) had evidence indicating recent primary CMV infection within the first trimester of pregnancy, this is higher than that observed in an unselected population in a single centre in France (0.42% seroconversion) , but consistent with proportions seen in a population of women in Italy who had a young child or worked with young children (1.2%) . Although this is a small proportion of women, this results in a large number of infants born each year with CMV. Vertical transmission in the first trimester of pregnancy is estimated at 36.8% with nearly 20% of fetuses from these women showing evidence of being affected by CMV . Without interventions to reduce the risk of acquisition of CMV or transmission of CMV, these infants will continue to acquire CMV and a significant proportion of them continue to suffer long term adverse sequelae as a result of congenital CMV infection. As well as generating essential data to inform a future larger study, we have also been able to describe important differences in knowledge about CMV, perceived severity, susceptibility and CMV risk reducing behaviour of pregnant women in the two study groups before the intervention in early pregnancy and at 34 gestational weeks. By collecting post-intervention data at 34 gestational weeks, we are able to show that these differences were evident even at the end pregnancy, suggesting that women were able to sustain these changes throughout pregnancy. Before the intervention, most women were unfamiliar with CMV. Previous studies have also shown that only a minority of pregnant women have heard of CMV: 16% in an Australian study, 18% in a Japanese study and 20% in two separate studies in Singapore , and the US , and that the level of knowledge about CMV is less than for other conditions which affect newborn infants [ , , , ]. Despite the fact that CMV is the most common congenital infection in the UK, pregnant women in our study were also less knowledgeable about CMV than other conditions affecting newborns. In our study, 34.7% of women reported being ‘somewhat’ or ‘very’ familiar, a higher proportion than in other studies. This may reflect volunteer bias in which those individuals who are better informed about CMV are more likely to take part in research about it, or it may have been a product of the screening process in which it was necessary to provide some information about CMV in the process of obtaining consent for serological screening. Participants in the intervention group showed a greater awareness of the ways in which CMV can be transmitted and ways in which congenital CMV can affect children following the intervention, at 34 gestational weeks, compared to those women in the treatment as usual group. This is in agreement with the study by Price et al., who also included change in knowledge as an outcome following an antenatal educational intervention . The ultimate aim of a CMV educational intervention in pregnancy is not acquisition of facts, but rather to modify behaviours that would place a woman at increased risk of exposure to CMV. In agreement with other studies [ , , , – ], we found that an educational intervention in pregnancy was associated with a reduction in the frequency of activities which could expose women to saliva and urine of young children, compared to before the intervention and compared to the treatment as usual group, specifically a reduction in participants eating leftovers from their child’s plate and kissing their child on the lips. These behaviours have previously been identified as being most difficult to change . These changes in reported behaviours may relate to the change in the perception of severity and susceptibility which was seen in the intervention group; change in perception of severity of the condition and an individual’s susceptibility to it has been shown to be an important mediator of behaviour change . As far as possible, we wanted to have a single intervention early in pregnancy in order to create circumstances as similar as possible to clinical practice, and we therefore provided no reminders to participants about risk reduction, we did not ask them about their behaviours between the first appointment and the questionnaire at 34 weeks and we did not use any objective measures of adherence which is in contrast to some other studies [ , , ]. Whilst all of these measures were important to our ultimate goal of investigating an intervention which would have clinical utility in a routine setting, there are also limitations associated with this approach. Self-reported behaviour may not be the same as actual behaviour, especially when asking participants about their activities over a prolonged period. This may particularly be the case for those behaviours for which there is a perceived ‘right’ answer, for example washing hands after changing a nappy. We were unable to completely simulate real life conditions; in order to screen for the serostatus of potential participants it was necessary to provide some information about CMV which caused many of the participants to seek further information. This may have led to our whole study population being better informed about CMV than the general population and may have limited our ability to detect differences between the groups—although this would have led to an underestimation of the effect of the intervention and if such an intervention were used in routine care there might be an even greater impact on behaviours. In this study we used a film as our educational intervention that had been designed in partnership with pregnant women and families of affected children. The feedback we received from study participants suggests that this was highly accessible and acceptable to them. Importantly, participants in the intervention group had similar scores on a global measure of distress and on a screening tool designed to identify individuals at risk of perinatal depression compared to those in the treatment as usual group – both pre- and post-intervention. This study confirmed a finding which has been shown in repeated studies which is that pregnant women want to know about CMV and are often shocked that this has not been discussed with them before [ , , ]. This reinforces the importance of a future large trial to determine the efficacy of an educational intervention to reduce the risk of CMV acquisition in pregnancy and the optimal implementation strategy for CMV antenatal education in routine clinical practice.
We have demonstrated that a randomised controlled trial of a film-based educational intervention is feasible in the UK and generated essential data upon which to power such studies. This single-centre randomised controlled trial has also shown that the intervention was associated with differences in knowledge, attitudes and behaviours before and after the intervention. This gives confidence that it may be possible to reduce the risk of acquisition of CMV in pregnancy using a film-based educational intervention. The efficacy of this needs to be tested in a future multi-centre randomised controlled trial.
Additional file 1: Description of digital educational intervention. Table outlining the timings and content of the digital educational intervention. Additional file 2: Pre-intervention questionnaire (intervention and treatment as usual groups). Questionnaire completed by all participants at baseline. Additional file 3: Post-intervention questionnaire (intervention group). Questionnaire completed by participants in the intervention group after viewing the digital educational intervention. Additional file 4: Post-intervention questionnaire (treatment as usual group). Questionnaire completed by participants in the treatment as usual group after viewing the educational slides about influenza vaccination. Additional file 5: 34-week questionnaire intervention group. Questionnaire completed by participants in the intervention group at 34 weeks of gestation. Additional file 6: 34-week questionnaire TAU group. Questionnaire completed by participants in the treatment as usual group at 34 weeks of gestation. Additional file 7: Supplementary Figure S1 . Graphs showing familiarity of participants with conditions affecting newborn infants. Additional file 8: Supplementary Table S1 . Table of anxiety and depression scores for intervention and treatment as usual groups at baseline and 34 weeks. Additional file 9: CONSORT extension for pilot and feasibility studies. This is the CONSORT extension checklist for pilot and feasibility studies.
|
Important role of endoscope in tuberculum sellae meningioma resection via supraorbital keyhole approach | 7a6a8f56-3997-4340-a7c8-f6ab5eb82a0c | 11885384 | Surgical Procedures, Operative[mh] | The gold standard for the treatment of anterior skull base meningiomas is surgical resection. Since the extent of tumor resection correlates closely with recurrence rate and survival prognosis, the surgical objective is to maximize tumor removal. Within anterior skull base meningiomas, tuberculum sellae meningioma represents one of the most challenging subtypes which can invade the planum sphenoidale anteriorly and skull base inferiorly, extend laterally to involve the optic nerve and the carotid artery, and posteriorly may affect the pituitary stalk and diaphragm sellae. As the tumor enlarges, it can progressively compress the optic nerves or chiasm leading to visual disturbances. The tumors’ ability to invade these critical neural and vascular structures not only poses significant surgical challenges for dissection but also increases the risk of postoperative complications arising from injury to these structures including visual disturbances, hypopituitarism, diabetes insipidus, cerebrospinal fluid rhinorrhea, infection, hemorrhage, etc. It is reported that 57.1% of visual impairments can be improved, 7.5% will have diabetes insipidus, 3.6% will have hypopituitarism , with an overall complication rate of 23.9% . To safeguard these critical structures, tumor residue and insufficient dural tail sign removal are often inevitable. Literature reports that gross total resection (GTR) rate for tuberculum sellae meningioma is around 80% . However, GTR encompasses Simpson Grades I, II, and III resections. According to Simpson, recurrence rates vary across different Simpson Grade classifications . Therefore, Simpson grade II resection should be the ultimate goal for surgeons which means intraoperative assessment of the tumor residue and its dural tail signs assume paramount importance. To achieve this objective, this study elaborates on the application of endoscopy during tuberculum sellae meningioma resection via the supraorbital keyhole approach, renowned for its excellent cosmetic outcomes and broad applicability . By compensating for the limited view of the microscope, endoscope reveals the residual tumors and suspicious dural tail signs which were hard to be identified under the microscope. This enables further precise resection and coagulation to achieve minimally invasive Simpson Grade II tumor resection, and finally decrease the long-term recurrence rates. The immense value of endoscopy in such surgical procedure is well demonstrated.
Following general anesthesia, the patient was positioned in the supine position with the head fixed in a three-pin Mayfield headholder. The head was then retroflexed 15° and rotated 20° to the right. The incision was planned on the lateral aspect of the left eyebrow, extending medially until the supraorbital incisura. After making the initial skin incision, the subcutaneous tissue was meticulously dissected. The skin flap was gently retracted superiorly with spring hooks. Approximately 3 cm above the orbital rim, the frontal muscle was incised parallel to the orbital margin, extending medially to just the supraorbital incisura and laterally along the temporal line. The frontal muscle flap was mobilized and retracted downward with holding sutures and the temporal muscle was retracted laterally to exposure the MacCarty keyhole. After a single burr hole was made using a highspeed drill, a frontal keyhole craniotomy measuring about 3 × 2.5 cm was performed. Subsequently, the dura mater was opened in a C-shaped flap and retracted in a basal direction . During the microscopic procedure, cerebrospinal fluid (CSF) of the suprasellar cistern was successfully released, resulting in satisfactory control of intracranial pressure. After gently retracting left frontal lobe, optic chiasm and the tumor were exposed. The tumor was soft with moderate vascularity, primarily originating at the tuberculum sellae. After excising the base, the tumor was removed piecemeal until the complete removal of the tumor was achieved. Then the endoscope was introduced from the keyhole bone window to the first gap. Under direct visualization, a suspected thin layer of residual tumor tissue was identified near the right internal carotid artery (ICA). Additionally, a small amount of dural tail sign was noted lateral to the right optic nerve. Both residues were subsequently excised under endoscope and microscope. Final endoscopic inspection confirmed Simpson grade II gross total resection, with complete coagulation of the dural tail sign. From the endoscopic view, tumor base did not involve the diaphragm sellae or pituitary stalk, and both bilateral optic nerves, chiasm, and left olfactory nerve were preserved intact. Artificial dura mater was used to repair the dural defect, followed by meticulous and cosmetical closure of the incision.
Moderate tumor volume and controllable blood supply. The tumor can be expected to achieve a Simpson Grade II resection. The tumor invasion does not penetrate the anterior skull base and there is no need for reconstruction of the skull base.
The small size of the keyhole bone window limits the space for endoscopy, necessitating care to prevent excessive retraction of the frontal lobe that could lead to direct injury of the brain tissue or tearing of the olfactory nerve. When using the endoscope, it is crucial to pay attention to the optic nerves, arteries, and pituitary stalk, ensuring careful entry and exit to avoid iatrogenic injury. The area beneath the optic nerve often represents a blind spot, requiring meticulous observation with the endoscope; a 30-degree angled lens may be employed if necessary.
The endoscope should be used after tumor removal when intracranial pressure has been adequately controlled. Frontal lobe should be protected with brain cotton in case of scratch by endoscope. Entry and exit with the endoscope should be gentle, maintaining close contact with the anterior skull base. To minimize the visual blind spots typically found beneath the optic nerve, an alternative approach entering from the contralateral side can be adopted to enhance the vision.
A 42-year-old female patient presented with a history of right-sided blurred vision for over two months. Neurological examination revealed right temporal hemianopia. Magnetic resonance imaging (MRI) demonstrated right tuberculum sellae meningioma. To achieve optimal cosmetic outcomes and ensure complete resection of the tumor, we decided to perform endoscopic-assisted supra-orbital keyhole approach. The surgical procedures may be associated with several potential complications as follows. Injury of artery or optic nerve: if tumor was adherent to critical structures such as the carotid artery, optic nerves, or chiasm, forced dissection may risk damaging these structures, leading to massive bleeding, or further deterioration of visual acuity. Therefore, we opted for a left-sided approach which provided better visualization of the blind area below the optic nerve. Bleeding, seizure, infection and brain edema: these complications also occurred in standard open craniotomy surgery. Careful intraoperative hemostasis, mild frontal lobe retraction, and attention to draining venous protection were emphasized. Hypopituitarism and diabetes insipidus: injury of pituitary stalk or portal system may result in postoperative hypopituitarism and diabetes insipidus. Hormone replacement therapy may be required, and close monitoring of electrolyte disorder should be maintained. Wound healing and cosmetic problems: postoperative care should include vigilant observation of wound healing, as mild effusion, swelling, or CSF leak may occur, requiring early detection and management. Fortunately, the patient did not develop any of these complications and experienced excellent wound healing. Postoperative MRI confirmed GTR of the tumor. Follow-up evaluation of visual acuity revealed no change compared to preoperative status, while visual fields had returned to normal, and olfaction remained intact (Figs. , ).
Endoscopic assisted supraorbital keyhole approach is an effective minimally invasive transcranial approach for tuberculum sellae meningioma. Endoscopic visualization gives a direct view on tumor base and dural attachments. Be careful and not retract frontal tissue too much in case of frontal contusion. Gentle entry and exit of endoscopic instruments. Keep high stability and avoid any iatrogenic injury of optic nerve pituitary stalk and arteries. Protection of olfactory nerve during the operation. Cosmetic closure technique should be performed in supraorbital keyhole surgery. Contralateral approach is effective to minimize the blind area below optic nerve. Dural tail sign is far more extensive than MR image showed or observation under a microscope. Simpson Grade II resection is surgical goal for tuberculum sellae meningioma.
Below is the link to the electronic supplementary material. Supplementary file1 (MP4 19.6 MB)
|
Efficacy of Two Commercially Available Adsorbents to Reduce the Combined Toxic Effects of Dietary Aflatoxins, Fumonisins, and Zearalenone and Their Residues in the Tissues of Weaned Pigs | 4f3e9aad-cc3e-444f-b391-5ba906f8b8da | 10675588 | Microbiology[mh] | Mycotoxins are toxic compounds produced by certain fungi species that can grow on crops, food products, and in various environments [ , , ]. The most common mycotoxins are aflatoxins (AFs), ochratoxin A (OTA), deoxynivalenol (DON), fumonisins (FBs), and zearalenone (ZEN). When consumed or inhaled, these toxic compounds exert harmful effects on humans and animals. The most pronounced effects involve carcinogenicity, mutagenicity, and immunosuppression, depending on the type, dose, and exposure duration to these toxins [ , , ]. Carcinogenic effects are developed mostly by AFs and FBs , while immunosuppression is caused mainly by AFs, DON, and OTA, and infertility and endocrine disruption are attributed to ZEN [ , , ]. The severity of these toxic effects in production animals may vary according to several factors, including the individual’s age, health status, and susceptibility . Thus, control strategies for mycotoxins in food and feed are essential for health protection. This can be done by implementing proper food/feed storage and processing methods, as well as regular monitoring and testing of these toxicants in agricultural commodities and animal feed . Pigs are among the most sensitive animal species to mycotoxins, especially to AFs, FBs, and ZEN . The high sensitivity in pigs can be attributed to their single-chambered stomach, which facilitates toxin absorption in the gastrointestinal tract, coupled with their high feed consumption relative to body weight . Mycotoxin exposure can lead to a reduction in the growth rate, feed efficiency, and overall performance in pigs . In addition, AFs and DON also lead to immunosuppression disorder, while ZEN induces reproductive problems like infertility and abnormal estrous cycles [ , , ]. Due to their sensitivity to toxins and the potential economic impact on the swine industry, it is a common practice to closely monitor and manage mycotoxin contamination in their diet. This includes the use of mycotoxin binders or adsorbents in feed, applying quality control measures in feed production, and regular testing of feed ingredients for mycotoxin to reduce their related health risks and other issues in pigs [ , , , ]. Adsorbents for mycotoxins are substances or materials that can bind to mycotoxins and reduce their bioavailability, thus preventing their absorption in the gastrointestinal tract and reducing their potential harm when consumed . The common types of adsorbents for mycotoxins include bentonite clay, silicates, and zeolites . These adsorbents are commonly used in animal feed to mitigate the risk of mycotoxin contamination. However, the choice of adsorbent depends on the specific mycotoxin contamination issue and the intended application, such as in agriculture or animal husbandry. In this context, in vivo studies are needed to confirm the efficacy of adsorbents for mycotoxins. Minazel Plus ® is a product based on organically modified clinoptilolite (OMC) , whereas MycoRaid ® is a multicomponent mycotoxin detoxifying agent (MMDA) based on specially selected minerals, Bacillus sp., yeast cell wall, and herbal extract to remediate the effect of mycotoxins in animals . Therefore, the objectives of this study were: (1) to determine the efficacy of two commercial adsorbents, OMC and MMDA, to ameliorate the toxic effects of dietary AFs, FBs, and ZEN on piglets’ performance and serum chemistries and (2) to determine the efficacy of the adsorbents to reduce residual concentrations of mycotoxins metabolites in the liver and kidneys of piglets fed the combined mycotoxins.
2.1. Growth Performance The effects of dietary treatments on the growth performance of piglets fed mycotoxin-contaminated rations with or without commercial adsorbents for 42 days are shown in . Compared with the controls, the body-weight gain of animals receiving a mixture of AFs (sum of AFB 1 , AFB 2 , AFG 1, and AFG 2 ), FBs, and ZEN were lower ( p < 0.05), while pigs receiving these mycotoxins in combination with OMC or MMDA at 1.5 or 3.0 kg/ton had values similar ( p > 0.05) to the controls. However, feed consumption and feed gain did not differ ( p > 0.05) among treatments. 2.2. Serum Biochemistry The results of serum biochemistry are summarized in . No significant differences ( p > 0.05) were observed between the six treatments for total protein (TP), albumin (ALB), serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), or alkaline phosphatase (ALP). 2.3. Relative Organ Weights The individual body weights of pigs at the end of the trial and their respective organ weights were used to calculate the relative weights of the liver, kidneys, uterus, ovarium, and lungs. The results were expressed as g/kg body weight, as presented in . Livers from pigs fed with the mycotoxin mixture alone or with 1.5 kg/ton MP had higher ( p < 0.05) relative weights than controls, while the addition of 3.0 kg/ton OMC or MMDA (1.5 or 3.0 kg/ton) reduced the relative weight of the liver after 42 days of dietary exposure to AFs, FBs, and ZEN. The relative weight of the uterus from piglets in treatment B (BD + AFs + FBs + ZEN) was higher ( p < 0.05) than controls or animals from treatments D (BD + AFs + FBs + ZEN + 3.0 kg/ton OMC) and F (BD + AFs + FBs + ZEN + 3.0 kg/ton MMDA). No significant differences were observed ( p > 0.05) among the relative weights of kidneys, ovarium, or lungs. Size measurements were also performed for the vulvas from piglets after 42 days of intoxication, as given in . Compared with the controls, the width, height, and length of vulvas were higher ( p < 0.05) in animals from treatment B (mixed mycotoxins only), while the administration of OMC or MMDA reduced the size values, having a maximum effect with 1.5 kg/ton MMDA. 2.4. Histopathology No histopathological changes were observed in the evaluated uterus from any treatment. The histopathological findings in the liver, kidneys, lungs, and ovaries are presented in , , and , respectively. Control animals fed only with basal diet (BD) did not show any histopathological changes in the liver ( A), kidneys ( A), lungs ( A), and ovaries ( A). However, animals exposed to the mycotoxin mixture developed moderate liver dysplasia ( B) in 75% of cases, while those receiving the mycotoxin mixture and 1.5 kg/ton OMC showed mild hepatic dysplasia (25% of cases) ( C); those fed the mycotoxin mixture with 3.0 kg/ton OMC had no liver changes ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed mild hepatitis (50% of cases) without dysplasia ( D), while those exposed to the mycotoxin mixture and treated with 3.0 kg/ton MMDA showed moderate hepatitis (50% of cases) without dysplasia ( F). Regarding the kidneys, animals exposed to the mycotoxin mixture developed renal glomerular atrophy (25% of cases) ( B). However, pigs exposed to the mycotoxin mixture and receiving 1.5 kg/ton OMC exhibited no kidney changes ( C), while animals receiving the mycotoxin mixture and 3.0 kg/ton OMC exhibited only mild renal interstitial inflammation in 25% of cases ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed renal glomerular atrophy (25% of cases) ( D), while those exposed to the mycotoxin mixture and treated with 3.0 kg/ton MMDA showed mild renal interstitial inflammation (75% of cases) ( F). Lungs from animals exposed to the mycotoxin mixture developed moderate interstitial pneumonitis and mild pulmonary edema in 100 and 25% of cases, respectively ( B). Pigs fed the mycotoxin mixture plus 1.5 kg/ton OMC showed mild pulmonary edema (50% of cases) without interstitial inflammation ( C), while those receiving the mycotoxin mixture and 3.0 kg/ton OMC exhibited no lung changes ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed mild interstitial pneumonitis and pulmonary edema (75 and 25% of cases, respectively) ( D), while those exposed to the mycotoxin mixture and treated with 3.0 kg/ton MMDA showed pulmonary edema (25% of cases) but without inflammation ( F). As for the ovaries, animals receiving the mycotoxin mixture showed a reduced oocyte number in 25% of cases ( B). Pigs exposed to the mycotoxin mixture and receiving 1.5 kg/ton OMC also presented a reduced oocyte number (25% of cases) ( C), while those fed the mycotoxin mixture and 3.0 kg/ton OMC exhibited no ovarian changes ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed an increased oocyte number in 25% of cases ( D), and those receiving the mycotoxin mixture and treated with 3.0 kg/ton MMDA had no changes in the ovarium ( F). 2.5. Mycotoxin Residues in Liver and Kidneys The residual levels of AFs, FBs, ZEN, and their metabolites (α-ZEL and β-ZEL) in the liver and kidneys are presented in and , respectively. In , the addition of AFs, FBs, and ZEN to the BD (treatment B) leads to mean levels in the liver for AFM 1 , AFs, FBs, and ZEN of 0.92 ± 0.07, 3.99 ± 0.22, 2.59 ± 0.97, and 23.3 ± 3.7 µg/kg, respectively. ZEN metabolites (α-zeralenol, α-ZEL, and β-zeralenol, β-ZEL) were not detected in any liver sample. Lower levels ( p < 0.05) of residual FBs and ZEN were observed in the liver from treatments receiving the evaluated adsorbents. Regarding the residual AFs, no quantifiable levels were observed in the livers of pigs in treatments C-F. It suggests that the mycotoxin contamination was effectively reduced or bound in the presence of OMC or MMDA, as indicated by the variable levels of these toxins across treatments. The results in reveal that kidneys from pigs in treatment B, which involved combined AFs, FBs, and ZEN, exhibited mean concentrations of 2.63 ± 0.31, 9.13 ± 0.25, 4.32 ± 0.65, and 41.6 ± 5.7 µg/kg for AFM 1 , AFs, FBs, and ZEN, respectively. However, in treatment C (BD + AFs + FBs + ZEN + 1.5 kg/ton OMC), the levels were below the LOQ for AFM 1 , AFs, α-ZEL, and β-ZEL, while FBs and ZEN were observed at mean levels of 2.98 ± 1.11 and 34.3 ± 3.3 µg/kg, respectively. These results were similar in treatment D (BD + AFs + FBs + ZEN + 3.0 kg/ton OMC), with quantifiable levels of FBs (3.25 ± 1.56) and ZEN (30.5 ± 4.2). In the same order, treatment E (BD + AFs + FBs + ZEN + 1.5 kg/ton MMDA) provided mean levels for AFM 1 at 1.06 ± 0.20, AFs at 4.25 ± 0.53, FBs at 2.77 ± 1.02, and ZEN at 23.1 ± 2.9 µg/kg. In treatment F (BD + AFs + FBs + ZEN + 3.0 kg/ton MMDA), the mean values were lower for AFM 1 (0.80 ± 0.11), AFs (1.92 ± 0.24), FBs (2.63 ± 1.32), and ZEN (17.4 ± 2.1 d), if compared with the data for treatment B. The variations in mycotoxin levels in kidneys indicate that OMC or MMDA supplementation has the potential to reduce the residual levels of mycotoxins and associated health risks in piglets.
The effects of dietary treatments on the growth performance of piglets fed mycotoxin-contaminated rations with or without commercial adsorbents for 42 days are shown in . Compared with the controls, the body-weight gain of animals receiving a mixture of AFs (sum of AFB 1 , AFB 2 , AFG 1, and AFG 2 ), FBs, and ZEN were lower ( p < 0.05), while pigs receiving these mycotoxins in combination with OMC or MMDA at 1.5 or 3.0 kg/ton had values similar ( p > 0.05) to the controls. However, feed consumption and feed gain did not differ ( p > 0.05) among treatments.
The results of serum biochemistry are summarized in . No significant differences ( p > 0.05) were observed between the six treatments for total protein (TP), albumin (ALB), serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), or alkaline phosphatase (ALP).
The individual body weights of pigs at the end of the trial and their respective organ weights were used to calculate the relative weights of the liver, kidneys, uterus, ovarium, and lungs. The results were expressed as g/kg body weight, as presented in . Livers from pigs fed with the mycotoxin mixture alone or with 1.5 kg/ton MP had higher ( p < 0.05) relative weights than controls, while the addition of 3.0 kg/ton OMC or MMDA (1.5 or 3.0 kg/ton) reduced the relative weight of the liver after 42 days of dietary exposure to AFs, FBs, and ZEN. The relative weight of the uterus from piglets in treatment B (BD + AFs + FBs + ZEN) was higher ( p < 0.05) than controls or animals from treatments D (BD + AFs + FBs + ZEN + 3.0 kg/ton OMC) and F (BD + AFs + FBs + ZEN + 3.0 kg/ton MMDA). No significant differences were observed ( p > 0.05) among the relative weights of kidneys, ovarium, or lungs. Size measurements were also performed for the vulvas from piglets after 42 days of intoxication, as given in . Compared with the controls, the width, height, and length of vulvas were higher ( p < 0.05) in animals from treatment B (mixed mycotoxins only), while the administration of OMC or MMDA reduced the size values, having a maximum effect with 1.5 kg/ton MMDA.
No histopathological changes were observed in the evaluated uterus from any treatment. The histopathological findings in the liver, kidneys, lungs, and ovaries are presented in , , and , respectively. Control animals fed only with basal diet (BD) did not show any histopathological changes in the liver ( A), kidneys ( A), lungs ( A), and ovaries ( A). However, animals exposed to the mycotoxin mixture developed moderate liver dysplasia ( B) in 75% of cases, while those receiving the mycotoxin mixture and 1.5 kg/ton OMC showed mild hepatic dysplasia (25% of cases) ( C); those fed the mycotoxin mixture with 3.0 kg/ton OMC had no liver changes ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed mild hepatitis (50% of cases) without dysplasia ( D), while those exposed to the mycotoxin mixture and treated with 3.0 kg/ton MMDA showed moderate hepatitis (50% of cases) without dysplasia ( F). Regarding the kidneys, animals exposed to the mycotoxin mixture developed renal glomerular atrophy (25% of cases) ( B). However, pigs exposed to the mycotoxin mixture and receiving 1.5 kg/ton OMC exhibited no kidney changes ( C), while animals receiving the mycotoxin mixture and 3.0 kg/ton OMC exhibited only mild renal interstitial inflammation in 25% of cases ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed renal glomerular atrophy (25% of cases) ( D), while those exposed to the mycotoxin mixture and treated with 3.0 kg/ton MMDA showed mild renal interstitial inflammation (75% of cases) ( F). Lungs from animals exposed to the mycotoxin mixture developed moderate interstitial pneumonitis and mild pulmonary edema in 100 and 25% of cases, respectively ( B). Pigs fed the mycotoxin mixture plus 1.5 kg/ton OMC showed mild pulmonary edema (50% of cases) without interstitial inflammation ( C), while those receiving the mycotoxin mixture and 3.0 kg/ton OMC exhibited no lung changes ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed mild interstitial pneumonitis and pulmonary edema (75 and 25% of cases, respectively) ( D), while those exposed to the mycotoxin mixture and treated with 3.0 kg/ton MMDA showed pulmonary edema (25% of cases) but without inflammation ( F). As for the ovaries, animals receiving the mycotoxin mixture showed a reduced oocyte number in 25% of cases ( B). Pigs exposed to the mycotoxin mixture and receiving 1.5 kg/ton OMC also presented a reduced oocyte number (25% of cases) ( C), while those fed the mycotoxin mixture and 3.0 kg/ton OMC exhibited no ovarian changes ( E). Animals that received 1.5 kg/ton MMDA in addition to the mycotoxin mixture showed an increased oocyte number in 25% of cases ( D), and those receiving the mycotoxin mixture and treated with 3.0 kg/ton MMDA had no changes in the ovarium ( F).
The residual levels of AFs, FBs, ZEN, and their metabolites (α-ZEL and β-ZEL) in the liver and kidneys are presented in and , respectively. In , the addition of AFs, FBs, and ZEN to the BD (treatment B) leads to mean levels in the liver for AFM 1 , AFs, FBs, and ZEN of 0.92 ± 0.07, 3.99 ± 0.22, 2.59 ± 0.97, and 23.3 ± 3.7 µg/kg, respectively. ZEN metabolites (α-zeralenol, α-ZEL, and β-zeralenol, β-ZEL) were not detected in any liver sample. Lower levels ( p < 0.05) of residual FBs and ZEN were observed in the liver from treatments receiving the evaluated adsorbents. Regarding the residual AFs, no quantifiable levels were observed in the livers of pigs in treatments C-F. It suggests that the mycotoxin contamination was effectively reduced or bound in the presence of OMC or MMDA, as indicated by the variable levels of these toxins across treatments. The results in reveal that kidneys from pigs in treatment B, which involved combined AFs, FBs, and ZEN, exhibited mean concentrations of 2.63 ± 0.31, 9.13 ± 0.25, 4.32 ± 0.65, and 41.6 ± 5.7 µg/kg for AFM 1 , AFs, FBs, and ZEN, respectively. However, in treatment C (BD + AFs + FBs + ZEN + 1.5 kg/ton OMC), the levels were below the LOQ for AFM 1 , AFs, α-ZEL, and β-ZEL, while FBs and ZEN were observed at mean levels of 2.98 ± 1.11 and 34.3 ± 3.3 µg/kg, respectively. These results were similar in treatment D (BD + AFs + FBs + ZEN + 3.0 kg/ton OMC), with quantifiable levels of FBs (3.25 ± 1.56) and ZEN (30.5 ± 4.2). In the same order, treatment E (BD + AFs + FBs + ZEN + 1.5 kg/ton MMDA) provided mean levels for AFM 1 at 1.06 ± 0.20, AFs at 4.25 ± 0.53, FBs at 2.77 ± 1.02, and ZEN at 23.1 ± 2.9 µg/kg. In treatment F (BD + AFs + FBs + ZEN + 3.0 kg/ton MMDA), the mean values were lower for AFM 1 (0.80 ± 0.11), AFs (1.92 ± 0.24), FBs (2.63 ± 1.32), and ZEN (17.4 ± 2.1 d), if compared with the data for treatment B. The variations in mycotoxin levels in kidneys indicate that OMC or MMDA supplementation has the potential to reduce the residual levels of mycotoxins and associated health risks in piglets.
In this study, the absence of clinical signs in experimental piglets indicates low-to-moderate exposure of piglets to the evaluated mycotoxins and lack of stress due to the high comfort of animals. Moreover, these results indicate that the adsorbents evaluated at 1.5 or 3.0% of the diet did not negatively affect the health status of experimental animals. However, the combined mycotoxins tested negatively affected the body weights of female piglets after 42 days of intoxication ( ), although no effects were observed on feed consumption or feed gain. The inclusion of adsorbents OMC or MMDA at 1.5 or 3.0% decreased the negative effects of combined mycotoxin, thus increasing the body weights of piglets during 42 days of intoxication. One of the most significant economic effects of mycotoxicosis in pig production is the growth-rate reduction . In this study, a mixture of AFs, FBs, and ZEN was used, leading to combined toxic effects on the exposed animals. The effects of simultaneous exposure to multimycotoxins are complex and may be classified into synergic, additive, and antagonist categories . Previous studies have identified some possible synergistic and additive interactions of co-occurring mycotoxins, such as AFs and FBs associated with reduced body-weight gain , which agrees with the data reported here. Changes in protein synthesis, gene expression, and enzyme kinetics are considered the main mechanisms by which mycotoxins impair piglets’ performance . As for specific toxic effects, AFB 1 is well-known for both its carcinogenic and teratogenic properties . Hepatic microsomal cytochrome P450 (CYP450) catalyzes the formation of an unstable intermediate and highly reactive substrate, known as AFB 1 -8,9-epoxide, as a pivotal event in AFB 1 -induced toxicity mechanisms. This intermediate compound plays the main role in carcinogenic and other toxicities related to AFB 1 [ , , , ]. Additionally, AFB 1 is associated with immunotoxicity, oxidative stress, and epigenetic changes, such as DNA methylation and RNA alterations, among other effects that potentially contribute to hepatocellular carcinoma (HCC) . In utero, AFB 1 exposure impacts the offspring’s DNA methylation, highlighting the need for further research to understand the underlying epigenetic mechanisms . Amongst FBs, FB 1 has been shown to induce toxicity, like neurotoxic, teratogenic, hepatotoxic, and carcinogenic effects in animals . Exposure of pigs to FB 1 leads to the development of pig-specific clinical dysfunction, namely, pulmonary edema, which is associated with higher pulmonary capillary hydrostatic pressure . Exposure to FB 1 in pigs also results in damage to the hepatic, cardiovascular, gastrointestinal, and immune systems. Due to the nonsteroidal osteogenic structure, ZEN mimics natural hormones, leading to reproductive issues in animals by reducing estrogen activity and altering the associated metabolic pathways . For example, in pigs, short-term exposure in the first reproduction cycle led to an elevated return to estrus rates, abortions, and hyperestrogenism symptoms in newborn piglets . This exposure also resulted in ovarian follicle atresia, apoptotic-like changes in granule cells, and increased cell proliferation in the uterus and oviduct . In the present experiment, no significant differences were observed in the serum biochemical parameters. These findings are consistent with the previously published studies [ , , ]. Of note, these studies have reported that the abovementioned variables may not be satisfactory biomarkers that could indicate poisoning in pigs exposed to low levels of mycotoxins and/or for a short period of time. The mycotoxin mixture containing AFs, FBs, and ZEN increased the relative weight of the liver from pigs, also determining hepatocellular dysplasia in this organ. Additionally, these animals also presented interstitial pneumonitis, renal glomerular atrophy, and reduced oocyte number in the ovaries. However, the addition of 3.0 kg/ton OMC or MMDA (1.5 or 3.0 kg/ton) had a positive effect on the relative weight of the liver after 42 days of dietary exposure to AFs, FBs, and ZEN. A similar protective effect of OMC or MMDA was observed regarding the relative weight of the uterus. Moreover, OMC and MMDA treatments were able to reduce the incidence of hepatocellular dysplasia, renal glomerular atrophy, and interstitial pneumonitis. Curiously, the addition of 1.5 kg/ton MMDA increased the oocyte number in the ovaries. On the other hand, pigs supplemented with 3.0 kg/ton OMC or MMDA exhibited mild interstitial inflammatory infiltrate in the kidneys, and the MMDA-treated animals developed hepatitis. Residual levels of AFs (including AFM 1 ) were detected in piglets’ livers only in treatment B (BD + AFs + FBs + ZEN), while the concentrations of FBs and ZEN and their derivative metabolites (α- and β-ZEL) were found to be reduced in the remaining treatments C, D, E, and F, which were statistically lower than treatment B. These results indicate that the treatments with adsorbents were highly protective against the given mycotoxins in piglets’ livers. The protective effect of MMDA was previously demonstrated in an in vivo study with weaned pigs exposed to dietary ZEN and T-2 toxins, indicating a dose-dependent reduction in the residual levels of ZEN and T-2 in the liver, with the best inclusion of this adsorbent at 3.0 kg/ton. Similarly, the use of a purified clay mineral based on bentonite in pig diets during a 35-day trial also increased feed intake and weight gain . According to Raj et al. , the use of OMC in the feed of broilers exposed to AFB 1 and OTA also improved feed conversion and gain in the average body weight, which agrees with the data reported here. When kidneys were examined for the determination and quantification of residual levels of AFs, FBs, ZEN, and α- and β-ZEL, all treatments also had reduced levels of mycotoxins, compared with treatment B. Along with the evaluation of mycotoxins, AFs reduction by the adsorbents was found to be highly effective in treatments C and D, followed by treatments E and F. For FB 1 and ZEN, treatments E and F were more effective than the other treatments. These findings indicate that the applied adsorbents, OMC and MMDA, are highly effective against the AFs, also alleviating the effects of FBs and ZEN. Similar efficacy for MMDA was assessed in weaned pigs exposed to dietary ZEN and T-2, in which the residual levels of these toxins decreased significantly, compared with controls receiving only the mycotoxins . In line with the outcomes observed for OMC, a study found that the application of this adsorbent effectively reduced the residual levels of AFB 1 and OTA in broilers, also improving specific biochemical markers associated with liver health and performance metrics .
In this experiment, the effectiveness of two commercial adsorbents, OMC and MMDA, to reduce the combined adverse effects of dietary AFs, FBs, and ZEN and the residual concentrations of mycotoxin metabolites in the liver and kidney was evaluated in crossbred female pigs. After a 42-day dietary exposure, the mycotoxin cocktail decreased the body weight gain and increased the size of the vulva and the relative weights of the liver and uterus. The mycotoxin mixture also induced moderate histopathological changes in the liver, kidneys, lungs, and ovarium, although no effect was observed in the serum biochemistry parameters of the intoxicated animals. Both OMC and MMDA adsorbents ameliorated the toxic effects and significantly reduced the residual levels of mycotoxins in the liver and kidneys. Notably, OMC supplementation was able to reduce the initiation of liver carcinogenesis without causing hepatotoxic side effects. These findings underscore the effectiveness of OMC and MMDA for mitigating the adverse effects of dietary mycotoxins in piglets, with prospects for improving mycotoxin management strategies in pig-farming operations.
5.1. Animals, Diets, and Experimental Design The experimental work was evaluated and approved by the Animal Ethics Committee of the Institute of Animal Science and Pastures of Nova Odessa (protocol nº 326-2021). Twenty-four crossbred female piglets (21 days) were purchased from a commercial breeding center, allocated in individual cages, and allowed ad libitum access to feed and water. The health status of the animals was assessed by clinical examination upon arriving in the experimental facility, and at 7-day intervals during the entire experimental period, by a qualified veterinarian. After 14 days of adaptation period, the animals were randomly assigned into 6 experimental groups of 4 pigs each and were submitted during 42 days to the treatments summarized in . The basal diet (BD), based on a corn and soybean meal-type diet, was formulated to meet the nutritional requirements of growing pigs, as recommended by Grenier et al. . Mycotoxin’s culture materials containing Afs (sum of AFB 1 , AFB 2 , AFG 1 , and AFG 2 ) , FBs and ZEN, along with the commercial adsorbents (Minazel Plus ® , OMC, and MycoRaid ® , MMDA) were added to the BD and mixed in a horizontal/helicoidal mixer for 15 min to achieve the targeted concentration of the mycotoxins. The aflatoxins (AFB 1 , AFB 2 , AFG 1 , and AFG 2 ), fumonisins (FB 1 and FB 2 ), and ZEN concentrations were determined by an in-house validated liquid chromatography coupled with tandem mass spectrometry , as displayed in . In addition, all diets were screened by using the same analytical method and found to be free of, or with nondetectable levels, of ochratoxin A (limit of detection, LOD: 0.5 μg/kg) and deoxynivalenol (LOD: 6.1 μg/kg). The animals were weighed at baseline, and at 7-day intervals throughout the experiment. The piglets were also monitored daily for any sign of AFs, FBs, or ZEN toxicity. Feed consumption was measured weekly to calculate the feed conversion (FC). 5.2. Sample Collection, Biochemical and Histological Analyses Blood samples were collected at the beginning and at 14 d intervals throughout the experiment via jugular venipuncture in an evacuated blood-collection system in serum separator clot activator tubes Vacuette ® (Greiner Bio-one, Kremsmunster, Austria). Serum samples were split into two aliquots, one immediately used for serum biochemistry determinations and the other stored at –20 ℃ for further possible analysis of mycotoxin biomarkers. TP, ALB, AST, ALT, and ALP were measured using an automated biochemical analyzer. Results for AST, TP, and ALB were expressed as g/dL, while ALT and ALP data were displayed as international units (IU)/L. At the end of the trial, piglets were subjected to electrical stunning and euthanized by exsanguination. The liver, lungs, kidneys, uterus, and ovaries were separated for evaluation. After weighing, these samples were fixed in a 10 kg/ton buffered formalin. Vulvar measurements (height, width, and length) were performed immediately after euthanasia . Graded alcohol was used for the dehydration of the tissue samples, followed by cleaning with xylene, and then embedded in a liquid paraffin. A 5 µm section was stained with hematoxylin–eosin for a descriptive and semiquantitative histopathological analysis in each organ evaluated . 5.3. Determination of Mycotoxin Residues in Liver and Kidneys Duplicate samples of 1 g of ground tissues of the liver and kidney were extracted in acetonitrile: water: acetic acid (79:20:1), as described by Cao et al. and summarized in . AFB 1 , AFB 2 , AFM 1 , FB 1 , FB 2 , ZEN, α-zearalenol (α-ZOL), and β-zearalenol (β-ZOL) concentrations in the final extracts were determined using a Waters Acquity I-Class ultraperformance liquid chromatographic (UPLC) system (Waters, Milford, MA, USA) equipped with a BEH C18 column (2.1 × 50 mm, 1.7 μm) and coupled to a Xevo TQ-S mass spectrometer (Waters, Milford, MA, USA). The mass spectrometer (MS) was operated in multireaction monitoring (MRM) using electrospray ionization in positive and negative ion modes, with the main parameters as described in . Mycotoxin standard solutions and calibration curves were prepared using a work solution with mixed mycotoxins prepared in water: acetonitrile (50:50), containing AFM 1 , AFB 1 , AFB 2 , AFG 1 , AFG 2 , FB 1 , FB 2 , ZEN, α-ZEL, and β-ZEL at 100 ng/mL. This solution was used to prepare five matrix-matched calibration standards at the range levels expressed in . Additionally, isotopically labeled internal standards (IS) of [ 13 C 17 ]-AFB 1 (St. Louis, MO, USA), [ 13 C34]-FB 1 and [ 13 C 18 ]-ZEN (Biopure, Tulln, Austria) were also prepared in water: acetonitrile (50:50), and added to each sample prior extraction, to reach the concentration of 100 ng/mL for each IS. Five 5 μL of the extracts and standards were injected using gradient elution in a mobile phase made up of water (eluent A) and acetonitrile (eluent B), both containing 5 mM ammonium acetate and 0.1% acetic acid and kept at 0.6 mL/min, as described elsewhere . The total chromatography run for each sample was 10 min. Limits of detection (LOD) and quantification (LOQ) were determined considering signal-to-noise ratios of 1:3 and 1:10, respectively, and are displayed in . The analytical results were based on a standard calibration with added IS, which compensated for both recovery losses and matrix effects. 5.4. Statistical Analysis Data were submitted to analysis of variance (ANOVA) using the PROC GLM of the SAS for Windows program, version 9.4. For multiple comparisons between treatments, the Tukey test was performed. All statements of significance were based on the 0.05 level of probability.
The experimental work was evaluated and approved by the Animal Ethics Committee of the Institute of Animal Science and Pastures of Nova Odessa (protocol nº 326-2021). Twenty-four crossbred female piglets (21 days) were purchased from a commercial breeding center, allocated in individual cages, and allowed ad libitum access to feed and water. The health status of the animals was assessed by clinical examination upon arriving in the experimental facility, and at 7-day intervals during the entire experimental period, by a qualified veterinarian. After 14 days of adaptation period, the animals were randomly assigned into 6 experimental groups of 4 pigs each and were submitted during 42 days to the treatments summarized in . The basal diet (BD), based on a corn and soybean meal-type diet, was formulated to meet the nutritional requirements of growing pigs, as recommended by Grenier et al. . Mycotoxin’s culture materials containing Afs (sum of AFB 1 , AFB 2 , AFG 1 , and AFG 2 ) , FBs and ZEN, along with the commercial adsorbents (Minazel Plus ® , OMC, and MycoRaid ® , MMDA) were added to the BD and mixed in a horizontal/helicoidal mixer for 15 min to achieve the targeted concentration of the mycotoxins. The aflatoxins (AFB 1 , AFB 2 , AFG 1 , and AFG 2 ), fumonisins (FB 1 and FB 2 ), and ZEN concentrations were determined by an in-house validated liquid chromatography coupled with tandem mass spectrometry , as displayed in . In addition, all diets were screened by using the same analytical method and found to be free of, or with nondetectable levels, of ochratoxin A (limit of detection, LOD: 0.5 μg/kg) and deoxynivalenol (LOD: 6.1 μg/kg). The animals were weighed at baseline, and at 7-day intervals throughout the experiment. The piglets were also monitored daily for any sign of AFs, FBs, or ZEN toxicity. Feed consumption was measured weekly to calculate the feed conversion (FC).
Blood samples were collected at the beginning and at 14 d intervals throughout the experiment via jugular venipuncture in an evacuated blood-collection system in serum separator clot activator tubes Vacuette ® (Greiner Bio-one, Kremsmunster, Austria). Serum samples were split into two aliquots, one immediately used for serum biochemistry determinations and the other stored at –20 ℃ for further possible analysis of mycotoxin biomarkers. TP, ALB, AST, ALT, and ALP were measured using an automated biochemical analyzer. Results for AST, TP, and ALB were expressed as g/dL, while ALT and ALP data were displayed as international units (IU)/L. At the end of the trial, piglets were subjected to electrical stunning and euthanized by exsanguination. The liver, lungs, kidneys, uterus, and ovaries were separated for evaluation. After weighing, these samples were fixed in a 10 kg/ton buffered formalin. Vulvar measurements (height, width, and length) were performed immediately after euthanasia . Graded alcohol was used for the dehydration of the tissue samples, followed by cleaning with xylene, and then embedded in a liquid paraffin. A 5 µm section was stained with hematoxylin–eosin for a descriptive and semiquantitative histopathological analysis in each organ evaluated .
Duplicate samples of 1 g of ground tissues of the liver and kidney were extracted in acetonitrile: water: acetic acid (79:20:1), as described by Cao et al. and summarized in . AFB 1 , AFB 2 , AFM 1 , FB 1 , FB 2 , ZEN, α-zearalenol (α-ZOL), and β-zearalenol (β-ZOL) concentrations in the final extracts were determined using a Waters Acquity I-Class ultraperformance liquid chromatographic (UPLC) system (Waters, Milford, MA, USA) equipped with a BEH C18 column (2.1 × 50 mm, 1.7 μm) and coupled to a Xevo TQ-S mass spectrometer (Waters, Milford, MA, USA). The mass spectrometer (MS) was operated in multireaction monitoring (MRM) using electrospray ionization in positive and negative ion modes, with the main parameters as described in . Mycotoxin standard solutions and calibration curves were prepared using a work solution with mixed mycotoxins prepared in water: acetonitrile (50:50), containing AFM 1 , AFB 1 , AFB 2 , AFG 1 , AFG 2 , FB 1 , FB 2 , ZEN, α-ZEL, and β-ZEL at 100 ng/mL. This solution was used to prepare five matrix-matched calibration standards at the range levels expressed in . Additionally, isotopically labeled internal standards (IS) of [ 13 C 17 ]-AFB 1 (St. Louis, MO, USA), [ 13 C34]-FB 1 and [ 13 C 18 ]-ZEN (Biopure, Tulln, Austria) were also prepared in water: acetonitrile (50:50), and added to each sample prior extraction, to reach the concentration of 100 ng/mL for each IS. Five 5 μL of the extracts and standards were injected using gradient elution in a mobile phase made up of water (eluent A) and acetonitrile (eluent B), both containing 5 mM ammonium acetate and 0.1% acetic acid and kept at 0.6 mL/min, as described elsewhere . The total chromatography run for each sample was 10 min. Limits of detection (LOD) and quantification (LOQ) were determined considering signal-to-noise ratios of 1:3 and 1:10, respectively, and are displayed in . The analytical results were based on a standard calibration with added IS, which compensated for both recovery losses and matrix effects.
Data were submitted to analysis of variance (ANOVA) using the PROC GLM of the SAS for Windows program, version 9.4. For multiple comparisons between treatments, the Tukey test was performed. All statements of significance were based on the 0.05 level of probability.
|
Navigating Hope and Complexity: Turkish Parents’ Experiences with Savior Siblings | 7022643e-0a4a-496a-a172-9a1e02eaf539 | 11869153 | Surgical Procedures, Operative[mh] | Preimplantation genetic diagnosis (PGD) with human leukocyte antigen (HLA) typing constitutes a significant advancement in treating inherited hematological disorders, particularly thalassemia major. This technology enables the birth of healthy children who can serve as compatible stem cell donors for their affected siblings . Türkiye is a world leader in both PGD+HLA typing technology and hematopoietic stem cell transplantation (HSCT) from savior siblings born after PGD+HLA typing. In the multicenter study conducted by Kurekci et al. , transplantations from PGD/HLA-matched siblings achieved a 96% success rate among 52 patients between 2008 and 2014. This is the largest case series reported in the literature to date. The procedure, covered by Türkiye’s social security system since 2009, offers a promising option for patients without suitable donors. The study by Kurekci et al. confirmed Türkiye’s pioneering role in successfully implementing this technology for the treatment of various hematological disorders. In Türkiye, where thalassemia is prevalent and Islamic perspectives influence medical ethics, PGD and HLA matching have gained acceptance within the framework of medical necessity. While the existing literature focuses on medical outcomes and ethical analyses , there is limited research on families’ lived experiences while navigating this complex process. The present study addresses that gap by investigating the experiences of Turkish parents who successfully completed the savior sibling process, offering unique insights into how this advanced medical procedure is experienced within a predominantly Muslim society that combines traditional values with modern medical practices.
Research Design In this study, a descriptive phenomenological research model, as one of the qualitative research methods, was used . The research team consisted of four people. Two were faculty members in a medical school, one was a faculty member in a faculty of health sciences, and one was a qualitative research expert. Study Sample The study sample consisted of parents who had a savior sibling child to serve as a HSCT donor for a sick sibling. In phenomenological models, participants selected for the sample group must have experienced the phenomenon in all its aspects. Therefore, parents who were 18 years of age or older, who had a sick child who had undergone HSCT, and who also had a savior sibling child were included in this study. Using homogeneous purposive sampling, the study sample was selected so that all the sick children of the families who had received transplants from their savior siblings had thalassemia major. In other words, it was ensured that the members of the study sample had experienced the same processes from start to finish. This type of sampling tends to ensure that the final sample is adequately representative . Data saturation and homogeneity were considered in determining the number of participants for the sample ; it was anticipated that data saturation could be achieved with parents from 16 families. Participant Selection The study included 16 families, with the participation of 16 mothers and 14 fathers, who had successfully undergone the savior sibling process for a child with thalassemia major. Individual interviews were conducted with only mothers and fathers and separately, not together, to ensure that participants could give independent responses and maintain their privacy. Interviews were not conducted with savior siblings, sick children, or other family members. The average interview duration was 38 min (range: 30-45 min). All interviews were conducted online using secure video conferencing software between October 2022 and February 2023. Participant Demographics The mean age of the mothers was 36.4 years (range: 28-44) and that of the fathers was 39.2 years (range: 31-48). Their educational backgrounds varied: 37.5% had university degrees, 43.8% had high school education, and 18.7% had only primary school education. The families were also from diverse geographical regions of Türkiye, including the Marmara (31.25%), Central Anatolia (25%), Aegean (18.75%), Mediterranean (12.5%), and Eastern Anatolia (12.5%) regions. The mean age of the sick children was 8.3 years (range: 4-14) and that of the savior siblings was 3.2 years (range: 1-6). Data Collection Tools The study data were collected between October 2022 and February 2023 using a semi-structured interview form. The form consisted of five questions prepared in light of the relevant demographic information and the current literature on the subject: 1. How did you feel when your child was first diagnosed with the disease? 2. What were your initial thoughts when you were told that bone marrow transplantation (HSCT) was needed for your child’s treatment? 3. How did you decide to have a savior sibling as a bone marrow transplant (HSCT) donor? 4. How did you feel while making this decision? 5. What did you experience during this process? Additional questions were asked to deepen the interviews as needed, depending on the course of the conversation with other questions branching from the parent’s conversation. Data Collection Process Data were collected online using the individual in-depth interview technique. Before starting the study interviews, a pilot interview was conducted with two participants not included among the final participants of the study. The interview questions were modified based on those pilot interviews. Verbal consent was obtained from the participants before starting the interviews. All interviews were conducted individually with the mothers (n=16) and fathers (n=14) who participated. Mothers and fathers were not interviewed together to ensure that all participants could express themselves more comfortably. The average interview duration was 38 min (range: 30-45 min). Additional questions were asked based on the participants’ responses to gain deeper insights into their experiences. All interviews were recorded with the participant’s consent and transcribed verbatim. Transcripts were sent to participants for their approval. Data Analysis The data from the qualitative interviews were analyzed using the MAXQDA 20.0 statistical software package (VERBI GmbH, Berlin, Germany) and Colaizzi’s seven-step phenomenological analysis method, and thematic coding was performed . The analysis process included the following steps: reading the transcripts and taking notes; selecting significant statements; formulating meanings; grouping meanings into categories, themes, and sub-themes; integrating results into a comprehensive description of the phenomenon; formulating the fundamental structure of the phenomenon; returning to the participants for validation; and expert examination of themes and codes. Reliability of the Study The reliability of this study was established according to the criteria of credibility, dependability, transferability, and confirmability . Participant approval was obtained, the examined phenomenon was described in detail, the researchers discussed the process, direct quotations from participants’ statements were utilized, inter-coder consistency was ensured, and multiple data collection methods were used. Limitations of the Study Due to the qualitative design and limited sample size, definitive and generalizable results were not reached in this study. Retrospective sharing may have caused some details to be forgotten or misremembered by participants. The data of the study reveal the short-term experiences of the participants; investigating and defining the long-term experiences of the participants may be useful in future studies. Our study only included families who successfully had a savior sibling child and experienced a successful transplant from the savior sibling child to the sick child. Future studies may offer a broader perspective by including families whose processes were not successful. Ethical Aspects of the Study The study was carried out with the approval of the Clinical Research Ethics Committee of Afyonkarahisar Health Sciences University (dated: 05.08.2022 and numbered: 2022/9). The principle of confidentiality was applied, pseudonyms were used, and data were stored securely and will be destroyed after 3 years. The authors reported no competing interests.
In this study, a descriptive phenomenological research model, as one of the qualitative research methods, was used . The research team consisted of four people. Two were faculty members in a medical school, one was a faculty member in a faculty of health sciences, and one was a qualitative research expert.
The study sample consisted of parents who had a savior sibling child to serve as a HSCT donor for a sick sibling. In phenomenological models, participants selected for the sample group must have experienced the phenomenon in all its aspects. Therefore, parents who were 18 years of age or older, who had a sick child who had undergone HSCT, and who also had a savior sibling child were included in this study. Using homogeneous purposive sampling, the study sample was selected so that all the sick children of the families who had received transplants from their savior siblings had thalassemia major. In other words, it was ensured that the members of the study sample had experienced the same processes from start to finish. This type of sampling tends to ensure that the final sample is adequately representative . Data saturation and homogeneity were considered in determining the number of participants for the sample ; it was anticipated that data saturation could be achieved with parents from 16 families.
The study included 16 families, with the participation of 16 mothers and 14 fathers, who had successfully undergone the savior sibling process for a child with thalassemia major. Individual interviews were conducted with only mothers and fathers and separately, not together, to ensure that participants could give independent responses and maintain their privacy. Interviews were not conducted with savior siblings, sick children, or other family members. The average interview duration was 38 min (range: 30-45 min). All interviews were conducted online using secure video conferencing software between October 2022 and February 2023.
The mean age of the mothers was 36.4 years (range: 28-44) and that of the fathers was 39.2 years (range: 31-48). Their educational backgrounds varied: 37.5% had university degrees, 43.8% had high school education, and 18.7% had only primary school education. The families were also from diverse geographical regions of Türkiye, including the Marmara (31.25%), Central Anatolia (25%), Aegean (18.75%), Mediterranean (12.5%), and Eastern Anatolia (12.5%) regions. The mean age of the sick children was 8.3 years (range: 4-14) and that of the savior siblings was 3.2 years (range: 1-6).
The study data were collected between October 2022 and February 2023 using a semi-structured interview form. The form consisted of five questions prepared in light of the relevant demographic information and the current literature on the subject: 1. How did you feel when your child was first diagnosed with the disease? 2. What were your initial thoughts when you were told that bone marrow transplantation (HSCT) was needed for your child’s treatment? 3. How did you decide to have a savior sibling as a bone marrow transplant (HSCT) donor? 4. How did you feel while making this decision? 5. What did you experience during this process? Additional questions were asked to deepen the interviews as needed, depending on the course of the conversation with other questions branching from the parent’s conversation.
Data were collected online using the individual in-depth interview technique. Before starting the study interviews, a pilot interview was conducted with two participants not included among the final participants of the study. The interview questions were modified based on those pilot interviews. Verbal consent was obtained from the participants before starting the interviews. All interviews were conducted individually with the mothers (n=16) and fathers (n=14) who participated. Mothers and fathers were not interviewed together to ensure that all participants could express themselves more comfortably. The average interview duration was 38 min (range: 30-45 min). Additional questions were asked based on the participants’ responses to gain deeper insights into their experiences. All interviews were recorded with the participant’s consent and transcribed verbatim. Transcripts were sent to participants for their approval.
The data from the qualitative interviews were analyzed using the MAXQDA 20.0 statistical software package (VERBI GmbH, Berlin, Germany) and Colaizzi’s seven-step phenomenological analysis method, and thematic coding was performed . The analysis process included the following steps: reading the transcripts and taking notes; selecting significant statements; formulating meanings; grouping meanings into categories, themes, and sub-themes; integrating results into a comprehensive description of the phenomenon; formulating the fundamental structure of the phenomenon; returning to the participants for validation; and expert examination of themes and codes.
The reliability of this study was established according to the criteria of credibility, dependability, transferability, and confirmability . Participant approval was obtained, the examined phenomenon was described in detail, the researchers discussed the process, direct quotations from participants’ statements were utilized, inter-coder consistency was ensured, and multiple data collection methods were used.
Due to the qualitative design and limited sample size, definitive and generalizable results were not reached in this study. Retrospective sharing may have caused some details to be forgotten or misremembered by participants. The data of the study reveal the short-term experiences of the participants; investigating and defining the long-term experiences of the participants may be useful in future studies. Our study only included families who successfully had a savior sibling child and experienced a successful transplant from the savior sibling child to the sick child. Future studies may offer a broader perspective by including families whose processes were not successful.
The study was carried out with the approval of the Clinical Research Ethics Committee of Afyonkarahisar Health Sciences University (dated: 05.08.2022 and numbered: 2022/9). The principle of confidentiality was applied, pseudonyms were used, and data were stored securely and will be destroyed after 3 years. The authors reported no competing interests.
Demographics and Clinical Characteristics The study included parents from 16 families that had undergone successful savior sibling procedures for children with thalassemia major. The mean age of the mothers was 36.4±5.2 years and that of the fathers was 39.2±5.8 years. Most participating parents (81.3%) had at least high school education, with 37.5% having university degrees. Families were from diverse geographical regions of Türkiye, with the majority being from the Marmara (31.25%) and Central Anatolia (25%) regions. The mean age of the patients at diagnosis was 1.2±0.8 years and the mean time from diagnosis to HSCT was 4.1±1.9 years. Detailed demographic and clinical characteristics are presented in . Thematic Analysis Results Analysis of the interview data revealed six main themes reflecting parents’ experiences throughout the savior sibling process: disease stage, treatment, recovery process, social/family, support systems, and recommendations. A detailed breakdown of these themes, their categories, and associated codes is presented in . In qualitative research, the term “code” refers to frequently repeated expressions identified in interview data and textual content during the analysis process. Codes help researchers systematically organize and interpret data, facilitating the identification of underlying meanings and relationships. Within the theme of “disease stage,” four categories were identified: “learning the diagnosis,” “search for treatment methods,” “motivations for having a savior sibling child,” and “religious and cultural factors.” Parents reported initial fear and anxiety upon diagnosis, followed by active treatment research. One participant noted: “My biggest regret was when I learned about the disease... because I also have two children who are carriers, and we are both university graduates, my wife and I. Why didn’t we notice this before and take precautions?” (participant no. 1). The code of “community pressure on father to have savior sibling child” within the category of “religious and cultural factors” referred specifically to pressure to have a savior sibling child who could potentially be a donor for the sick child. Religious views about pregnancy termination were independent of fetal health status. All families were informed by their physicians about the risk of having another child with thalassemia major without PGD. Two parents reported no religious/cultural influence on their decisions, six mentioned the religious prohibition of pregnancy termination, and others were primarily focused on saving the sick child without considering religious implications. The theme of “treatment” contained four categories: “difficulties,” “having a baby by IVF,” “emotional state,” and “preparation for surgery.” Key challenges included transportation issues, economic burdens, and social stigma. The IVF process particularly affected mothers, as expressed by one participant: “All the responsibility was on me again, if something happened, so I spent most of the time lying down” (participant no. 3). The theme of “recovery process” comprised two categories: “emotional changes” and “caring for the sick child after bone marrow transplantation.” Parents reported positive emotional transitions following the successful transplantation, although concerns about disease recurrence persisted. The isolation period proved challenging, as one participant described: “He had a lot of pain... He stayed in the hospital for two months without going out, in a separate room where no one saw him; he only saw his father through the glass” (participant no. 8). The theme of “social/family” included three categories: “perspective of family members,” “having a savior sibling child,” and “meaning attributed to the savior sibling.” Families maintained normal relationships with the savior siblings, who were aware of their roles but too young for informed consent at the time of donation. In the theme of “support systems,” participants emphasized the importance of professional help and social support networks. Some experienced delayed psychological effects, with one mother sharing the following: “I guess because your world is that child, I always put myself in second place... I always felt like I shouldn’t be happy while my child is sick” (participant no. 11). The theme of “recommendations” highlighted the importance of trusting medical professionals, seeking psychological support, and maintaining determination throughout the process. Being determined/persistent and seeking psychological support were the most frequently repeated recommendations, each mentioned seven times. , presenting the code map, illustrates the interconnections between frequently mentioned codes. Codes are frequently repeated expressions identified in interviews and textual contents during the analysis process and they are expressed with numbers in parentheses in . For example, “Doctor’s Recommendation (27)” means that this expression was repeated 27 times by participants across all interviews conducted within the scope of the study, while “Saving the Child (23)” means that this expression was repeated 23 times by participants across all interviews. The code map particularly highlights the relationships between healing the sick child and factors such as doctors’ recommendations, the lack of alternatives, economic difficulties, travel requirements, burnout, extended hospital stays, and family support. The code map illustrates the interconnections between frequently mentioned codes with line thicknesses representing the frequency and strength of the relationships.
The study included parents from 16 families that had undergone successful savior sibling procedures for children with thalassemia major. The mean age of the mothers was 36.4±5.2 years and that of the fathers was 39.2±5.8 years. Most participating parents (81.3%) had at least high school education, with 37.5% having university degrees. Families were from diverse geographical regions of Türkiye, with the majority being from the Marmara (31.25%) and Central Anatolia (25%) regions. The mean age of the patients at diagnosis was 1.2±0.8 years and the mean time from diagnosis to HSCT was 4.1±1.9 years. Detailed demographic and clinical characteristics are presented in .
Analysis of the interview data revealed six main themes reflecting parents’ experiences throughout the savior sibling process: disease stage, treatment, recovery process, social/family, support systems, and recommendations. A detailed breakdown of these themes, their categories, and associated codes is presented in . In qualitative research, the term “code” refers to frequently repeated expressions identified in interview data and textual content during the analysis process. Codes help researchers systematically organize and interpret data, facilitating the identification of underlying meanings and relationships. Within the theme of “disease stage,” four categories were identified: “learning the diagnosis,” “search for treatment methods,” “motivations for having a savior sibling child,” and “religious and cultural factors.” Parents reported initial fear and anxiety upon diagnosis, followed by active treatment research. One participant noted: “My biggest regret was when I learned about the disease... because I also have two children who are carriers, and we are both university graduates, my wife and I. Why didn’t we notice this before and take precautions?” (participant no. 1). The code of “community pressure on father to have savior sibling child” within the category of “religious and cultural factors” referred specifically to pressure to have a savior sibling child who could potentially be a donor for the sick child. Religious views about pregnancy termination were independent of fetal health status. All families were informed by their physicians about the risk of having another child with thalassemia major without PGD. Two parents reported no religious/cultural influence on their decisions, six mentioned the religious prohibition of pregnancy termination, and others were primarily focused on saving the sick child without considering religious implications. The theme of “treatment” contained four categories: “difficulties,” “having a baby by IVF,” “emotional state,” and “preparation for surgery.” Key challenges included transportation issues, economic burdens, and social stigma. The IVF process particularly affected mothers, as expressed by one participant: “All the responsibility was on me again, if something happened, so I spent most of the time lying down” (participant no. 3). The theme of “recovery process” comprised two categories: “emotional changes” and “caring for the sick child after bone marrow transplantation.” Parents reported positive emotional transitions following the successful transplantation, although concerns about disease recurrence persisted. The isolation period proved challenging, as one participant described: “He had a lot of pain... He stayed in the hospital for two months without going out, in a separate room where no one saw him; he only saw his father through the glass” (participant no. 8). The theme of “social/family” included three categories: “perspective of family members,” “having a savior sibling child,” and “meaning attributed to the savior sibling.” Families maintained normal relationships with the savior siblings, who were aware of their roles but too young for informed consent at the time of donation. In the theme of “support systems,” participants emphasized the importance of professional help and social support networks. Some experienced delayed psychological effects, with one mother sharing the following: “I guess because your world is that child, I always put myself in second place... I always felt like I shouldn’t be happy while my child is sick” (participant no. 11). The theme of “recommendations” highlighted the importance of trusting medical professionals, seeking psychological support, and maintaining determination throughout the process. Being determined/persistent and seeking psychological support were the most frequently repeated recommendations, each mentioned seven times. , presenting the code map, illustrates the interconnections between frequently mentioned codes. Codes are frequently repeated expressions identified in interviews and textual contents during the analysis process and they are expressed with numbers in parentheses in . For example, “Doctor’s Recommendation (27)” means that this expression was repeated 27 times by participants across all interviews conducted within the scope of the study, while “Saving the Child (23)” means that this expression was repeated 23 times by participants across all interviews. The code map particularly highlights the relationships between healing the sick child and factors such as doctors’ recommendations, the lack of alternatives, economic difficulties, travel requirements, burnout, extended hospital stays, and family support. The code map illustrates the interconnections between frequently mentioned codes with line thicknesses representing the frequency and strength of the relationships.
Our study provides significant insights into the experiences of Turkish families who successfully completed the savior sibling process for the treatment of a child with thalassemia major. Analysis of their demographics revealed that the participating families represented diverse geographical regions and socioeconomic backgrounds of Türkiye. This heterogeneous distribution of participants across different regions and socioeconomic strata significantly enhanced the study’s representational validity and allowed for a more comprehensive perspective of the national landscape regarding savior sibling experiences. The inclusion of families from different geographical and socioeconomic contexts strengthens the generalizability of our findings and offers valuable insights into the challenges and experiences faced by families across different settings within the Turkish healthcare system. This demographic diversity particularly enriches our understanding of how varying resources, cultural contexts, and healthcare accessibility influence the savior sibling journey in different regions of Türkiye, thereby providing a more objective and holistic view of the national situation. The mean time from diagnosis to HSCT of 4.1±1.9 years reflects the complex nature of the process, including decision-making, PGD+HLA procedures, and preparation for the transplantation. The findings of this study provide crucial insights into the complex journeys of Turkish families through the savior sibling process for thalassemia major treatment, revealing several key implications for clinical practice and healthcare policy. The phenomenon of the savior sibling, which has medical, psychological, sociological, bioethical, and health policy aspects, reflects the complexity and multifaceted nature of this process. Medical Perspectives Our study sheds light on the medical aspects of the savior sibling process. The use of PGD and HLA matching technologies reflects significant advances in the field of medical genetics. These technologies are promising for diseases that previously had no chance of a cure. However, our study has also revealed some concerns about the use of these technologies. Most parents expressed concerns about the complexity and potential risks of the PGD+HLA matching process, but the willingness of the parents to take these risks shows how strong their desires were to save their children. Our study also emphasizes the importance of long-term follow-up of the health status of children born with the savior sibling method. There are still important gaps in the literature on this subject. For example, Kahraman et al. showed that the short-term health status of savior siblings was normal. However, the long-term effects are still unclear. Our findings confirm Türkiye’s leading position in PGD+HLA technology implementation. The successful outcomes of all transplantations in our cohort of 16 families align with previous reports of high success rates in transplantations for children with thalassemia major in Turkish pediatric hematopoietic transplantation centers. The mean age at donation for savior siblings (2.1±0.6 years) indicates careful timing of transplantation, balancing urgency with donor safety. This timing aligns with current recommendations for optimal transplant outcomes in thalassemia major . Psychosocial Perspectives The emergence of burnout as a significant theme emphasizes the need for systematic psychological support throughout the process. This finding corresponds with the recent literature on caregiver burden in cases of chronic pediatric conditions . The strong role of extended family support reflects the unique cultural context of Turkish society and suggests potential benefits of formally incorporating family support systems into treatment protocols. Sociological Perspectives Our study illuminates the social and cultural context of the savior sibling process in Turkish society. Similar to the findings reported by Gürtin , we observed that family structure, gender roles, and cultural norms shape this process. In particular, the importance of extended family support in Turkish society is noteworthy. Many parents stated that the support of the extended family was critical in this challenging process. This finding is in line with other studies conducted in collectivist cultures . However, this support may also sometimes be a double-edged sword. Some parents reported that the intrusive behaviors of the extended family could be a source of stress. The centrality of the maternal role is another important finding of our study. Mothers generally stated that they bore the main burden of the process. This reflects the traditional gender roles in Turkish society. However, in some families, fathers were also observed to play active roles, which may be an indicator of changing family dynamics in Turkish society. Unexpectedly, we observed that traditional values and modern medical practices can coexist harmoniously. This finding shows how traditional and modern values are blended in the modernization processes of Turkish society. Bioethical Perspectives Our study reveals the ethical dimensions of the savior sibling process. Issues such as child autonomy, intended pregnancy, and genetic selection raise important ethical debates. These findings overlap with the ethical debates addressed in the study conducted by Pennings et al. . However, an unexpected finding of our study is that the parents attached more importance than expected to the future autonomy of the child born as a savior sibling. Many parents stated that the child would be able to make his or her own decisions about whether or not to donate stem cells as the child grows up. This reflects the increasing awareness of children’s rights and autonomy in Turkish society. Our study also provides interesting findings on the role of religious beliefs in the savior sibling process. Unexpectedly, most of the parents stated that they did not seek approval in a religious context when deciding on this treatment method. This finding reflects the complex relationship between the acceptance of modern medical practices and religious beliefs in Turkish society. Healthcare Policy Perspectives Our study reveals the relative ease of access to savior sibling treatment in Türkiye. This reflects the strengths of the country’s healthcare system. However, factors such as economic difficulties and the geographical distribution of health services stand out as barriers faced by families. In particular, it was observed that families living in rural areas had difficulty in accessing treatment centers. This finding is similar to the results obtained by Liu et al. in the United States and it indicates that the geographical distribution of health services should be improved. Our study also revealed the inadequacy of psychosocial support services provided to families during the savior sibling process. Many parents stated that they wanted to receive professional psychological support during this process, but they had difficulty in accessing these services. Ensuring a more balanced distribution of savior sibling treatment centers across the country, developing comprehensive psychosocial support programs for families and facilitating access to these services, and establishing special financial support programs for families beginning the savior sibling process are our recommendations for these issues. Although premarital thalassemia screening was initiated in 41 provinces in Türkiye in 2003 and expanded nationwide in 2018 , the birth rate of infants affected by thalassemia major continues to be a significant concern. The first HSCT in Türkiye from a healthy HLA-matched sibling born after PGD+HLA matching was performed on November 29, 2005, at Akdeniz University Hospital in Antalya. The donor was a 9-month-old male sibling born on February 28, 2005, and the recipient was his 6-year-old sister with thalassemia major. In our study group, the youngest thalassemia major patient who received transplantation from a savior sibling was born on January 27, 2018. Despite mandatory nationwide premarital screening since 2018, new cases of thalassemia major continue to emerge. For instance, the lead author currently follows two 2-year-old thalassemia major patients with pathogenic mutations whose parents did not undergo premarital screening, and these patients are currently awaiting transplantation. This situation clearly demonstrates that the premarital thalassemia screening program has not yet achieved full effectiveness throughout the country. Future research may offer a broader perspective by including families whose processes have not been successful. Long-term studies should examine the health status, identity development, and self-perceptions of savior siblings alongside the medical and psychological impacts of PGD+HLA matching technologies. Investigation of socioeconomic factors, gender roles, and family dynamics within Turkish society would enhance our sociological understanding of the process. Further bioethical analysis should focus on balancing child autonomy with parental decisions and examining the ethical implications of genetic technologies.
Our study sheds light on the medical aspects of the savior sibling process. The use of PGD and HLA matching technologies reflects significant advances in the field of medical genetics. These technologies are promising for diseases that previously had no chance of a cure. However, our study has also revealed some concerns about the use of these technologies. Most parents expressed concerns about the complexity and potential risks of the PGD+HLA matching process, but the willingness of the parents to take these risks shows how strong their desires were to save their children. Our study also emphasizes the importance of long-term follow-up of the health status of children born with the savior sibling method. There are still important gaps in the literature on this subject. For example, Kahraman et al. showed that the short-term health status of savior siblings was normal. However, the long-term effects are still unclear. Our findings confirm Türkiye’s leading position in PGD+HLA technology implementation. The successful outcomes of all transplantations in our cohort of 16 families align with previous reports of high success rates in transplantations for children with thalassemia major in Turkish pediatric hematopoietic transplantation centers. The mean age at donation for savior siblings (2.1±0.6 years) indicates careful timing of transplantation, balancing urgency with donor safety. This timing aligns with current recommendations for optimal transplant outcomes in thalassemia major .
The emergence of burnout as a significant theme emphasizes the need for systematic psychological support throughout the process. This finding corresponds with the recent literature on caregiver burden in cases of chronic pediatric conditions . The strong role of extended family support reflects the unique cultural context of Turkish society and suggests potential benefits of formally incorporating family support systems into treatment protocols.
Our study illuminates the social and cultural context of the savior sibling process in Turkish society. Similar to the findings reported by Gürtin , we observed that family structure, gender roles, and cultural norms shape this process. In particular, the importance of extended family support in Turkish society is noteworthy. Many parents stated that the support of the extended family was critical in this challenging process. This finding is in line with other studies conducted in collectivist cultures . However, this support may also sometimes be a double-edged sword. Some parents reported that the intrusive behaviors of the extended family could be a source of stress. The centrality of the maternal role is another important finding of our study. Mothers generally stated that they bore the main burden of the process. This reflects the traditional gender roles in Turkish society. However, in some families, fathers were also observed to play active roles, which may be an indicator of changing family dynamics in Turkish society. Unexpectedly, we observed that traditional values and modern medical practices can coexist harmoniously. This finding shows how traditional and modern values are blended in the modernization processes of Turkish society.
Our study reveals the ethical dimensions of the savior sibling process. Issues such as child autonomy, intended pregnancy, and genetic selection raise important ethical debates. These findings overlap with the ethical debates addressed in the study conducted by Pennings et al. . However, an unexpected finding of our study is that the parents attached more importance than expected to the future autonomy of the child born as a savior sibling. Many parents stated that the child would be able to make his or her own decisions about whether or not to donate stem cells as the child grows up. This reflects the increasing awareness of children’s rights and autonomy in Turkish society. Our study also provides interesting findings on the role of religious beliefs in the savior sibling process. Unexpectedly, most of the parents stated that they did not seek approval in a religious context when deciding on this treatment method. This finding reflects the complex relationship between the acceptance of modern medical practices and religious beliefs in Turkish society.
Our study reveals the relative ease of access to savior sibling treatment in Türkiye. This reflects the strengths of the country’s healthcare system. However, factors such as economic difficulties and the geographical distribution of health services stand out as barriers faced by families. In particular, it was observed that families living in rural areas had difficulty in accessing treatment centers. This finding is similar to the results obtained by Liu et al. in the United States and it indicates that the geographical distribution of health services should be improved. Our study also revealed the inadequacy of psychosocial support services provided to families during the savior sibling process. Many parents stated that they wanted to receive professional psychological support during this process, but they had difficulty in accessing these services. Ensuring a more balanced distribution of savior sibling treatment centers across the country, developing comprehensive psychosocial support programs for families and facilitating access to these services, and establishing special financial support programs for families beginning the savior sibling process are our recommendations for these issues. Although premarital thalassemia screening was initiated in 41 provinces in Türkiye in 2003 and expanded nationwide in 2018 , the birth rate of infants affected by thalassemia major continues to be a significant concern. The first HSCT in Türkiye from a healthy HLA-matched sibling born after PGD+HLA matching was performed on November 29, 2005, at Akdeniz University Hospital in Antalya. The donor was a 9-month-old male sibling born on February 28, 2005, and the recipient was his 6-year-old sister with thalassemia major. In our study group, the youngest thalassemia major patient who received transplantation from a savior sibling was born on January 27, 2018. Despite mandatory nationwide premarital screening since 2018, new cases of thalassemia major continue to emerge. For instance, the lead author currently follows two 2-year-old thalassemia major patients with pathogenic mutations whose parents did not undergo premarital screening, and these patients are currently awaiting transplantation. This situation clearly demonstrates that the premarital thalassemia screening program has not yet achieved full effectiveness throughout the country. Future research may offer a broader perspective by including families whose processes have not been successful. Long-term studies should examine the health status, identity development, and self-perceptions of savior siblings alongside the medical and psychological impacts of PGD+HLA matching technologies. Investigation of socioeconomic factors, gender roles, and family dynamics within Turkish society would enhance our sociological understanding of the process. Further bioethical analysis should focus on balancing child autonomy with parental decisions and examining the ethical implications of genetic technologies.
This qualitative study of Turkish families’ experiences with the savior sibling process reveals the complex interplay among medical advancement, psychological resilience, and cultural adaptation. Our findings demonstrate that successful implementation of PGD+HLA matching requires both technical expertise and robust support systems. The study’s insights suggest the importance of establishing comprehensive support programs, enhancing geographical access to specialized centers, and implementing culturally sensitive care protocols. These findings have significant implications for HSCT in Türkiye and other similar cultural contexts, demonstrating that advanced genetic technologies can be successfully integrated into traditional societies when supported by appropriate medical and psychosocial infrastructures. On behalf of the Turkish Pediatric Bone Marrow Transplantation Study Group (TPBMT-SG): İbrahim Eker, Akif Yeşilipek, Alphan Küpesiz, Vedat Uygun, Gülsün Karasu, Funda Tayfun Küpesiz, Orhan Gürsel, Barış Kuşkonmaz, Serap Aksoylar, Fatma Visal Okur, Gülcihan Özek, Musa Karakükcü, Başak Adaklı Aksoy, Özlem Tüfekçi, Zühre Kaya, Barış Malbora, Ahmet Emin Kürekçi, Ali Bülent Antmen.
Ethics Committee Approval: Clinical Research Ethics Committee of Afyonkarahisar Health Sciences University (dated: 05.08.2022 and numbered: 2022/9). Informed Consent: Consent was obtained from all participants.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.