Search is not available for this dataset
text
stringlengths 1
129k
| id
stringlengths 9
16
| article_id
stringlengths 6
10
| section_title
stringlengths 1
1.26k
| educational_score
float64 0.46
5.16
| domain
stringclasses 3
values | document_type
stringclasses 4
values | domain_scores
listlengths 3
3
| document_type_scores
listlengths 4
4
| language
stringclasses 42
values | language_score
float64 0
1
|
|---|---|---|---|---|---|---|---|---|---|---|
Our research demonstrates the feasibility of extracting intact, functional mitochondrial material using the Single Cellome™ System SS2000. This commercial system allowed to first visualize mitochondria in single cells using confocal microscopy and second to decipher mitochondrial properties and function. This is a huge improvement regarding other methods, where mitochondria need to be sampled from thousands to millions of cells or from bulk tissue.
|
39753747_p13
|
39753747
|
Conclusion
| 4.020912 |
biomedical
|
Study
|
[
0.9997001886367798,
0.00011956926755374297,
0.00018031620129477233
] |
[
0.9980295300483704,
0.0012359947431832552,
0.0006566056399606168,
0.00007780588930472732
] |
en
| 0.999995 |
By applying the system to an investigation of Aβ-deposit-derived changes in nucleic and mitochondrial gene expression in microglia, a clear distinction due to the relative positioning of the cells was found . In future, this system might be used for sampling other particle-shaped organelles and it might be considered that mitochondria that can be collected in such rapid, function-sparing, and defined manner might also serve as a valuable source for mitochondrial transfer experiments 42 . Keeping in mind that a recent exciting paper from the Picard group 43 shows distinct mitochondrial profiles in different brain areas, this method might allow digging even deeper in elucidating cell-specific mitochondrial function between neuronal populations, glia cells, and astrocytes. In addition, the system also could allow sampling selectively from cells containing pre-dominantly fragmented versus from such with fused mitochondria and exploring their differential gene expression.
|
39753747_p14
|
39753747
|
Conclusion
| 4.11575 |
biomedical
|
Study
|
[
0.9996676445007324,
0.0001383782655466348,
0.00019396643619984388
] |
[
0.9974754452705383,
0.0005248739616945386,
0.0019318602280691266,
0.00006780545663787052
] |
en
| 0.999997 |
Spontaneously immortalized mouse microglia (SIM-A9, BIOZOL Diagnostica Vertrieb GmbH, KER-END001) 44 were cultivated in 10 cm culture dishes (Sarstedt AG & Co. KG) in a CO 2 incubator (Forma™ STERICULT CO 2 Incubator, Thermo Fisher Scientific Inc.) at 37 °C, 95% humidity, and 5% CO 2 . The culture medium consisted of phenol red-free Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM:F-12), supplemented with 10% v/v heat-inactivated fetal bovine serum (Gibco®, Life Technologies), 5% v/v heat-inactivated donor horse serum (Gibco®, Life Technologies), 1% v/v Penicillin–Streptomycin (Sigma-Aldrich), and 1% v/v L-Glutamine (Sigma-Aldrich). Upon reaching 80% confluence, the cells were passaged. Therefore, the conditioned medium was aspirated, and the cells were washed with 5 ml of pre-heated phosphate-buffered saline (PBS, 37 °C). Subsequently, 2.5 ml of Trypsin/EDTA-solution (Sigma-Aldrich) were added for 5 min at 37 °C, 95% humidity, and 5% CO 2 . The trypsin reaction was halted by adding 7.5 ml of culture medium. Following resuspension, the cells were transferred to a 50 ml CELLSTAR®, BLUE SCREW CAP Tube (Greiner Bio-One International GmbH) and centrifuged in a Megafuge 1.0 R (Heraeus) for 6 min at 600xg. The cell supernatant was removed, and the cell pellet was suspended in 20 ml of culture medium. Finally, 10 ml of suspended cells were distributed evenly between two culture dishes. Work with the SIM-A9 cell culture was conducted within a sterile biosafety cabinet (MSC-Advantage™, Thermo Fisher Scientific Inc.). For experiments, aliquots from cell suspension following the termination of the trypsin reaction were diluted with PBS and cells counted using the Cell Scepter TM 3.0 (Merck KGaA). Twenty-thousand cells per well were seeded in black, glass bottom 96-well plates (Greiner Bio-One GmbH).
|
39753747_p15
|
39753747
|
SIM-A9 Cell Culture
| 4.255779 |
biomedical
|
Study
|
[
0.9992693066596985,
0.00047456618631258607,
0.00025612072204239666
] |
[
0.9974101185798645,
0.0019320532446727157,
0.0004996893112547696,
0.00015804429131094366
] |
en
| 0.999997 |
The 4’,6-diamidino-2-phenylindole stock solution (1 mg/mL) was diluted 1:2000 with 1×PBS and 50 μL were added per well. The plate was incubated in darkness for 10 min at 37 °C using the Thermomixer comfort (Eppendorf AG). Following incubation, the DAPI solution was aspirated, and the cells underwent two washing steps with 200 μL of 1×PBS each. Subsequently, staining with LumiTracker Mito Red CMXRos was carried out.
|
39753747_p16
|
39753747
|
Staining of Nuclei with DAPI
| 4.031339 |
biomedical
|
Study
|
[
0.9978219270706177,
0.001352806342765689,
0.00082523183664307
] |
[
0.6564010381698608,
0.341235488653183,
0.0013214168138802052,
0.0010421181796118617
] |
en
| 0.999997 |
The LumiTracker Mito Red CMXRos stock solution was diluted 1:40,000 in pre-warmed CO 2 -independent culture medium (Gibco®, Life Technologies). 50 μL of the prepared solution were added per well, followed by a 15 min incubation at 37 °C in the Single Cellome™ System 2000 (Yokogawa Electric Corporation; Tokyo, Japan). After incubation, organelle material extraction was performed.
|
39753747_p17
|
39753747
|
Staining of Mitochondria with LumiTracker Mito Red CMXRos
| 3.997692 |
biomedical
|
Study
|
[
0.9992430210113525,
0.0003712274774443358,
0.00038587491144426167
] |
[
0.9388065934181213,
0.060061320662498474,
0.000682795129250735,
0.00044927801354788244
] |
en
| 0.999997 |
For peptide aggregation, aliquots containing 40 µL of 350 µM of Aβ1-42 or scrambled Aβ1-42 in sterile 1xPBS were incubated at 37°C as described previously 26 . To induce aggregation, these aliquots were subjected to 50 pipette strokes using 20 µL filter-pipette tips. The pipetting regimen commenced immediately on day 1, followed by repetitions after 48 hours on day 3, and on day 6. After 168 hours (on day 7 from the start of incubation), the resulting deposits of Aβ1-42 or scrambled Aβ1-42 were utilized for cell culture experiments. Aggregate formation within the Aβ1-42 solution was confirmed by ThT assay and by assessing toxicity using Cell Titer Glo assay . Toxicity was probably not elicited by the larger fibrillary aggregates that could be visualized during extraction but by oligomeric forms 45 . Ten µL of aggregated peptide solution or of 1xPBS were added to the cells (in 90 µL culture medium volume). The added peptide preparation or solvent were carefully mixed with the cultivation medium by ten pipette strokes with 50 µL volume. The cells where subsequently cultivated for 24 h in an incubator (Forma™ STERICULT CO 2 Incubator, Thermo Fisher Scientific Inc.) at 37 °C, 95% humidity, and 5% CO 2 .
|
39753747_p18
|
39753747
|
Peptide Aggregation
| 4.188917 |
biomedical
|
Study
|
[
0.9993682503700256,
0.00045518495608121157,
0.000176538378582336
] |
[
0.9989191293716431,
0.0005584760801866651,
0.00041483205859549344,
0.00010753374954219908
] |
en
| 0.999997 |
ThT-solution was freshly diluted to 10 µM with PBS . Thirty µl of peptide-containing solution were mixed with 55 µl of ThT solution in black 96 well plates and fluorescence measured at 37°C with Ex/Em=440 nm/484 nm (Fluostar Omega, BMG Labtech).
|
39753747_p19
|
39753747
|
ThT assay
| 4.002603 |
biomedical
|
Study
|
[
0.999526858329773,
0.00025517118046991527,
0.00021798427042085677
] |
[
0.9919769167900085,
0.007451784331351519,
0.0003955608990509063,
0.00017573777586221695
] |
en
| 0.999995 |
Using the built-in custom ruler tool integrated with the pixel-to-metric unit translator of the Single Cellome™ System 2000, the distance between two deposits was measured by delineating a line from the edge of one deposit to the edge of the other. The midpoint of this line was identified as the farthest point, while the starting points of the line were defined as the nearest point. Subsequently, cells were categorized into three groups: those far from the deposits, situated more than 10 µm away from the midpoint between two deposits; those near the plaques, within 10 µm of the edge of the deposit; and those in between. Only cells categorized as far or near to a deposit were selected for organelle harvesting.
|
39753747_p20
|
39753747
|
Measuring Cell-Peptide-Deposit Distance
| 4.079803 |
biomedical
|
Study
|
[
0.9994996786117554,
0.00024242633662652224,
0.0002578209387138486
] |
[
0.9991735816001892,
0.0005660228198394179,
0.00020341368508525193,
0.00005691853948519565
] |
en
| 0.999998 |
Mitochondrial and nuclear materials were extracted employing the Single Cellome™ System 2000. The Single Cellome™ System 2000 was programmed to manual entry of the cell and automatic discharge of collected material into a collection plate. The collection loader was cooled to 4 °C, while the cell incubation loader was heated to 37 °C. The microscope channels were configured for fluorescence visualization, with 561 nm Exc/ 617 nm Em (red fl.) or 405 nm Exc/ 447 nm Em (blue fl.). The objective lens was adjusted to a magnification of “40x Dry.” Upon selection of the respective cell, the tip was inserted into the cell with controlled velocity (10 μm/s) and pressure (1 Pa). The material was extracted by changing the pressure to −10 Pa for 5 s and automatically discharged into the collection plate (MicroAmp™ optical 96-well reaction plate, Applied Biosystems™), pre-filled with 6 μL of sterile 1×PBS for use in RT-qPCR and qPCR or 6 μL of sterile measurement buffer (250 mM sucrose (Sigma-Aldrich), 15 mM KCl (Carl Roth GmbH + Co. KG), 5 mM MgCl 2 (Carl Roth GmbH + Co. KG), 30 mM K 2 HPO 4 (Carl Roth GmbH + Co. KG), 50 mM succinate (Sigma-Aldrich), pH 7.4) for the Oxygen Consumption Assay. For each sample, material from three cells from the same well was extracted and pooled. The total extraction process took approximately 4 minutes, with several distinct steps. The identification of a suitable cell required around 30 seconds, followed by 2 to 3 min for internal machine positioning and loading of the extraction tips. The subsequent alignment of the tip relative to the cell took an additional 20–30 s. Finally, the actual extraction of the cellular material was completed in approximately 10 s.
|
39753747_p21
|
39753747
|
Organelle Material Extraction Using Single Cellome™ System 2000
| 4.298485 |
biomedical
|
Study
|
[
0.9990938901901245,
0.0006452101515606046,
0.00026094214990735054
] |
[
0.9917725920677185,
0.007102964911609888,
0.0008579952991567552,
0.00026643028832040727
] |
en
| 0.999998 |
To ascertain that the extracted material did not vary in general amount of to be measured molecules (DNA, RNA, and protein), we assessed the respective data by using the absorption at 260 nm for nucleic acids (Nanodrop) and the ProteOrange kit for measuring small amounts of protein. All sample means did not differ statistical from each other or from the control .
|
39753747_p22
|
39753747
|
Organelle Material Extraction Using Single Cellome™ System 2000
| 3.635123 |
biomedical
|
Study
|
[
0.9991565942764282,
0.00023260412854142487,
0.0006107473745942116
] |
[
0.998553454875946,
0.00108378566801548,
0.00028778924024663866,
0.00007496320904465392
] |
en
| 0.999997 |
The extracted material was transferred to a black 384-well plate with transparent bottom (Greiner Bio-One GmbH), and supplemented with 4 μL of nuclease-free water. Fluorescence signals were measured using the FLUOstar® Optima microplate reader (BMG Labtech GmbH), with 485 nm Exc/ 520 nm Em (DAPI blue fl.) or with 540 nm Exc/ 580 nm Em (LumiTracker Mito Red CMXRos, red fl.).
|
39753747_p23
|
39753747
|
Fluorescence Analysis of Extracted Organelle Samples
| 4.070223 |
biomedical
|
Study
|
[
0.9995489716529846,
0.00023594641243107617,
0.0002151211374439299
] |
[
0.9934050440788269,
0.006039176136255264,
0.00040220239316113293,
0.00015366432489827275
] |
en
| 0.999997 |
Six μL of samples were combined with the Cell lysis mix , excluding DNAseI, resulting in a total volume of 40 μL. 32 μL were transferred into a new vial and the lysis protocol proceeded according to the manufacturer’s instructions. For qPCR reactions, the remaining 8 μL of sample lysis mix underwent the manufacturer’s protocol, substituting DNAseI with 10 μM RNase A (Carl Roth GmbH + Co. KG) and incubation for 10 min at 37 °C. RT-qPCR was carried out with Luna ® Cell Ready One-Step RT-qPCR Kit (New England Biolabs) and qPCR with primaQUANT SYBRGreen qPCR Blue with ROX (Steinbrenner Laborsysteme GmbH) adhering to the manufacturer’s guidelines. The sample volume was adjusted to 20 μL per sample and a primer concentration of 0.4 μM (Qiagen or produced by EurofinsMWG). The reactions were carried out on a StepOnePlus™Real-Time PCR System (Applied Biosystems™). For primer sequences see Table 1 . Table1 Primer sequences Primer Sequence 5ʹ- > 3ʹ Reference mt-Atp8_for GGCACCTTCACCAAAATCAC 46 mt-Atp8_rev TTGTTGGGGTAATGAATGAGG mt-Co1_for CCTAGATGACACATGAGCAAAAG (forward primer modified) 47 mt-Co1_rev AGCGTCGTGGTATTCCTGAAA mt-Nd1_for ACGCTTCCGTTACGATCAAC 46 mt-Nd1_rev ACTCCCGCTGTAAAAATTGG mt-DNA_for CTAGAAACCCCGAAACCAAA 48 mt-DNA_rev CCAGCTATCACCAAGCTCGT mGapdh QuantiTect Primer Assay (Qiagen) Tfam_for CCAAAAAGACCTCGTTCAGC 49 Tfam_rev ATGTCTCCGGATCGTTTCAC Pnpt1_for AATCGGGCACTCAGCTATTTG PrimerBank ID 12835817a1 50 Pnpt1_rev CAGGTCTACAGTCACCGCTC PGC1α_for CGGAAATCATATCCAACCAG 51 PGC1α_rev TGAGGACCGCTAGCAAGTTTG Cox2_for GGGTGTGAAGGGAAATAAGG 52 Cox2_rev TGTGATTTAAGTCCACTCCATG Itgax_for CCATGCTGGCTGTAGATGACC 53 Itgax_rev GTCATCCTGGCAGATGTGGTC Inpp5d_for GAGCTACTTTCCAGAGCCG 54 Inpp5d_rev CACAATTCCGGAACAGCACG
|
39753747_p24
|
39753747
|
Reverse Transcription quantitative Polymerase Chain Reaction (RT-qPCR) and qPCR
| 4.287495 |
biomedical
|
Study
|
[
0.9993137121200562,
0.00038427067920565605,
0.0003020101285073906
] |
[
0.9938086867332458,
0.00539743946865201,
0.0006300804670900106,
0.00016381009481847286
] |
en
| 0.999997 |
The Oxygen Consumption Rate Assay Kit was performed in accordance with the manufacturer’s guidelines with the reagent volumes adjusted to 40 μL in a black 384-well plate with transparent bottom (Greiner Bio-One GmbH), using 6 μL of organelle material in measurement buffer. Antimycin A (1 μM; Cayman Chemicals) was employed for inhibition, while glucose oxidase was utilized as a positive control for oxygen consumption. Fluorescence signal was measured by the FLUOstar® Optima microplate reader (BMG LABTECH) with an excitation wavelength of 380 ± 12 nm, and an emission wavelength of 630 ± 12 nm. The linear phase from the kinetic measurement was used for calculation of oxygen consumption.
|
39753747_p25
|
39753747
|
Oxygen Consumption Assay
| 4.144167 |
biomedical
|
Study
|
[
0.9995542168617249,
0.00026206692564301193,
0.00018371416081208736
] |
[
0.9971691966056824,
0.002267979783937335,
0.00045060590491630137,
0.00011221485328860581
] |
en
| 0.999996 |
Proteins were subjected to 10% SDS PAA gels. As control, 40 µg (lysate high, Lh) and 14 µg (lysate low, Ll) of protein derived from lysed SIM-A9 cells was used and 14 µg of protein from extracted material (nuclear or mitochondrial). Proteins were transferred to nitrocellulose membrane and the membrane was blocked with I-block solution (0.2% in PBS) (Thermo Fisher Scientific) including 0.05% Tween 20 (AppliChem). Primary antibody incubation took place overnight at 4 °C with anti-Calnexin or anti-Gapdh in combination with HRP-labeled secondary antibodies . Signals were detected after incubation with SuperSignal West Femto chemiluminescent substrate using a CCD-camera imaging system (Stella Camera, Raytest, Straubenhardt).
|
39753747_p26
|
39753747
|
Western blotting
| 4.110462 |
biomedical
|
Study
|
[
0.999488115310669,
0.00023456172493752092,
0.0002772661973722279
] |
[
0.9971069693565369,
0.002480599330738187,
0.00031792177469469607,
0.00009449903154745698
] |
en
| 0.999998 |
Data were obtained from at least two independent experiments as indicated in the figure legends. All statistical analyses were conducted using GraphPad Prism 6 or 8 software (GraphPad Software). All data are presented as mean ± standard deviation. Statistical significance of differences between two groups was determined using two-tailed Student’s t-tests. One-way analysis of variance (ANOVA) was used for three or more groups and a post-hoc pairwise comparison as indicated. Outlier analysis was performed with GraphPadPrism (ROUT 1%). A p -value < 0.05 was considered statistically significant.
|
39753747_p27
|
39753747
|
Statistics and Reproducibility
| 3.543136 |
biomedical
|
Study
|
[
0.999442994594574,
0.000205004500458017,
0.0003519404854159802
] |
[
0.9737608432769775,
0.023954160511493683,
0.0019903688225895166,
0.00029465131228789687
] |
en
| 0.999997 |
Schematics were created by using Biorender.
|
39753747_p28
|
39753747
|
Images
| 1.40632 |
biomedical
|
Other
|
[
0.9415069818496704,
0.0035091976169496775,
0.054983776062726974
] |
[
0.06406097859144211,
0.926696240901947,
0.0064081428572535515,
0.002834669314324856
] |
en
| 0.999999 |
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
|
39753747_p29
|
39753747
|
Reporting summary
| 0.843296 |
biomedical
|
Other
|
[
0.5440846681594849,
0.0033791586756706238,
0.4525361955165863
] |
[
0.011882489547133446,
0.9854373931884766,
0.0019243218703195453,
0.0007558096549473703
] |
en
| 0.999998 |
Supplementary Information Description of Additional Supplementary File Supplementary data 1 Reporting Summary
|
39753747_p30
|
39753747
|
Supplementary information
| 1.603749 |
biomedical
|
Other
|
[
0.9694399237632751,
0.001160050043836236,
0.029400015249848366
] |
[
0.14051735401153564,
0.8525478839874268,
0.00561775965616107,
0.001317030400969088
] |
en
| 0.999998 |
As of 2024, endometrial cancer (EC) ranks as the sixth most common malignancy in women globally, contributing significantly to cancer-related morbidity and mortality . Despite recent advancements in the early detection of EC and its treatment modalities, a significant number of patients with EC continue to suffer from poor prognoses, particularly those with advanced-stage and high-grade disease . This persistent challenge underscores the limitations of current prognostic tools and therapies, thereby necessitating a more nuanced understanding of the molecular and immunological aspects of EC.
|
39751650_p0
|
39751650
|
Introduction
| 3.967898 |
biomedical
|
Review
|
[
0.9979860782623291,
0.0010628938907757401,
0.0009510900708846748
] |
[
0.0803922563791275,
0.004396265372633934,
0.9146897792816162,
0.0005217301659286022
] |
en
| 0.999998 |
The Cancer Genome Atlas (TCGA) has been instrumental in identifying four key genomic subgroups in EC: POLE (ultramutated), microsatellite instability (MSI) (hypermutated), copy-number low (CN low) (endometrioid), and copy-number high (CN high) (serous-like) . These findings have significantly advanced our approach to personalized EC treatments and prognoses. Notably, tumors with POLE mutations have shown better outcomes, even in high-grade cases, whereas the CN high subtype, particularly with TP53 mutations, often indicates a poorer prognosis . This diversity in outcomes among the EC subgroups underscores the necessity for additional markers to refine the risk stratification and personalize treatments . The proactive molecular risk classifier for endometrial cancer (ProMisE), based on TCGA's genomic subgroups, represents a significant step forward in this context . By integrating POLE exonuclease domain mutations and immunohistochemistry for mismatch repair proteins and TP53, the ProMisE categorizes EC into four distinct groups: POLE-EDM, MMR-D, p53wt, and p53abn . It extends TCGA's foundational work by providing comprehensive prognostic information, enhancing patient management within the International Federation of Gynecology and Obstetrics 2023 staging system .
|
39751650_p1
|
39751650
|
Introduction
| 4.499127 |
biomedical
|
Study
|
[
0.9984373450279236,
0.0008535799570381641,
0.0007089930004440248
] |
[
0.5644580125808716,
0.0028213495388627052,
0.43195390701293945,
0.0007667781319469213
] |
en
| 0.999996 |
The distribution of tumor-infiltrating lymphocytes (TILs) was reported to be associated with the prognosis of patients with colorectal cancer in 2006 . The concept of immunophenotypes (i.e., inflamed, excluded, and desert) based on the spatial distribution of CD8 + TILs, which recognized as vital players in modulating tumor progression and influencing responses to immunotherapies, was introduced in 2016 . These immunophenotypes are known to affect the effectiveness of treatments, particularly treatment with immune checkpoint inhibitors (ICIs) in some solid tumors . In EC, the amount of CD8 + TILs has been reported to be associated with patients’ prognosis ; however, the association between the distribution of CD8 + TILs and clinical outcomes has not been investigated. Although ICIs are increasingly being applied clinically in the treatment of EC, research on the tumor immune microenvironment of EC is lacking.
|
39751650_p2
|
39751650
|
Introduction
| 4.081398 |
biomedical
|
Study
|
[
0.9996044039726257,
0.0001963321556104347,
0.00019923763466067612
] |
[
0.9955867528915405,
0.00033768449793569744,
0.003991348203271627,
0.00008423181861871853
] |
en
| 0.999998 |
We sought to bridge this knowledge gap by exploring the distribution and prognostic relevance of CD8 + TIL-based immunophenotypes in EC, stratified by ProMisE molecular classification. We hypothesized that each molecular subtype may be associated with a distinct immunophenotype, which could have significant implications for patient prognoses and treatment responses. Our goal is to integrate genomic and immunological data to provide a more holistic view of the EC landscape, potentially paving the way for novel, targeted immunotherapeutic strategies for EC.
|
39751650_p3
|
39751650
|
Introduction
| 4.102594 |
biomedical
|
Study
|
[
0.9995381832122803,
0.00032492782338522375,
0.00013688878971152008
] |
[
0.9976060390472412,
0.00042508961632847786,
0.0018657143227756023,
0.00010313771053915843
] |
en
| 0.999998 |
A prospective analysis was conducted to examine the relationship between immunophenotype and molecular classification, as well as survival outcomes. The cases of 60 patients with EC treated during the period from January 2019 to December 2022 at Nagoya University Hospital (Nagoya, Japan) were prospectively enrolled. Following a definitive pathological diagnosis, these patients underwent surgical intervention. Ethical approval of this study was granted by the Institutional Review Board of Nagoya University . Both fresh frozen and formalin-fixed paraffin-embedded (FFPE) tumor tissues were collected for analysis.
|
39751650_p4
|
39751650
|
Prospective cohort
| 4.049722 |
biomedical
|
Study
|
[
0.9979051351547241,
0.0018359562382102013,
0.00025899242609739304
] |
[
0.9990512728691101,
0.0006028121570125222,
0.00017956169904209673,
0.0001662789873080328
] |
en
| 0.999996 |
We retrospectively analyzed the cases of a separate cohort of patients with high-grade EC ( n = 85, including endometrioid grade 3, serous, and clear cell histology) treated between January 2002 and December 2017 at Nagoya University Hospital. FFPE tumor tissues were collected for analysis.
|
39751650_p5
|
39751650
|
Retrospective cohort
| 3.806277 |
biomedical
|
Study
|
[
0.9987239241600037,
0.0009688363061286509,
0.00030719381175003946
] |
[
0.9992363452911377,
0.00044926241389475763,
0.00020167999900877476,
0.00011274792632320896
] |
en
| 0.999998 |
All 145 of the patients underwent a simple or semiradical hysterectomy with a bilateral salpingo-oophorectomy, pelvic, and/or para-aortic lymphadenectomy. The surgical specimens were staged according to the FIGO 2008 staging system. Postoperative treatment followed the Japan Society of Gynecologic Oncology guidelines . Risk-based adjuvant therapy was administered; low-risk patients received no additional treatment, and the intermediate- to high-risk patients received radiotherapy or platinum-based chemotherapy.
|
39751650_p6
|
39751650
|
Diagnosis and treatment
| 3.956275 |
biomedical
|
Study
|
[
0.8908725380897522,
0.10838084667921066,
0.0007465503294952214
] |
[
0.9453375935554504,
0.04200030863285065,
0.0027433340437710285,
0.009918774478137493
] |
en
| 0.999997 |
Hotspot mutations in exons 9, 13, and 14 of the POLE gene were identified using Sanger sequencing. Genomic DNA extraction was conducted in accord with the manufacturers' protocols for both fresh frozen (NucleoSpin® DNA Rapidlyse, Macherey–Nagel, Düren, Germany) and FFPE tissues (QIAamp DNA FFPE Advanced UNG Kit, Qiagen, Hilden, Germany). The samples were enriched using the Blend Taq Plus (Toyobo, Osaka, Japan). The polymerase chain reaction (PCR) amplification and conditions were as described . A portion of the PCR products was run on a 2% agarose gel in 1× TAE buffer to verify the presence of a single band approx. 200–300 base pairs in size. The rest of the PCR products was then purified using the QIAquick Gel Extraction Kit (Qiagen), following the manufacturer's instructions.
|
39751650_p7
|
39751650
|
Sanger sequencing for POLE exonuclease domain
| 4.138988 |
biomedical
|
Study
|
[
0.9995056390762329,
0.0003453933459240943,
0.00014888541772961617
] |
[
0.998766303062439,
0.0007842041668482125,
0.00034011510433629155,
0.0001093155806302093
] |
en
| 0.999998 |
The DNA concentrations were measured with a NanoDrop One spectrophotometer (Thermo Fisher Scientific, Waltham, USA). The subsequent DNA sequencing was outsourced to Eurofins Genomics (Tokyo). For the sequence analysis, we used SnapGene Viewer ver. 6.1.2 (GSL Biotech, San Diego, CA) to examine the waveform patterns. Mutation identification was conducted using the Nucleotide BLAST tool (Basic Local Alignment Search Tool) from the U.S. National Center for Biotechnology Information (NCBI). In this study, we defined ' POLE pathogenic variants' as the nine single-nucleotide substitutions on exons 9, 13, and 14: c.857 C > G (P286R) (exon9), c.884 T > G (M295R) (exon9), c.890 C > T (S297F) (exon9), c.1231 G > T/C (V411L) (exon13), c.1270 C > A (L424I) (exon13), c.1307 C > G (P436R) (exon13), c.1331 T > A (M444K) (exon13), c.1366 G > C (A456P) (exon14), and c.1376 C > T (S459F) (exon14) .
|
39751650_p8
|
39751650
|
Sanger sequencing for POLE exonuclease domain
| 4.126885 |
biomedical
|
Study
|
[
0.9994733929634094,
0.0003486259374767542,
0.00017797209147829562
] |
[
0.9991540908813477,
0.0005654809065163136,
0.00017889575974550098,
0.00010148641013074666
] |
en
| 0.999998 |
Immunohistochemistry (IHC) was conducted on 4-µm sections of FFPE tumor tissues, which included sections from both the central tumor (CT) and the invasive margin (IM). The primary antibodies used for the IHC included anti-human CD8 (clone C8/144b, Dako, Glostrup, Denmark; 1:100), PMS2 (clone A16-4, Biocare Medical, Walnut Creek, CA; 1:100), MSH6 (clone BC/44, Biocare Medical, 1:100), and p53 (clone DO-7, Dako; 1:100). The sections were deparaffinized and rehydrated, subjected to antigen retrieval in 10 mM sodium citrate (pH 6.0) or 1× Immunoactive (pH 9.0, Matsunami, Osaka, Japan) for 20 min at 95 °C in a microwave, and treated with 0.3% hydrogen peroxide in methanol for 20 min. Blocking was performed using the Histofine SAB-PO kit (Nichirei, Tokyo), followed by overnight incubation at 4 °C with the diluted primary antibodies. After the primary antibody incubation, the sections were incubated with biotin-labeled secondary antibody, peroxidase-labeled streptavidin, and developed using 3,3'-diaminobenzidine (DAB) substrate-chromogen for specific time durations. Then, the sections were counterstained with hematoxylin, dehydrated, and mounted.
|
39751650_p9
|
39751650
|
Immunohistochemistry analysis
| 4.186735 |
biomedical
|
Study
|
[
0.9993128776550293,
0.0005014144699089229,
0.00018570144311524928
] |
[
0.9971089959144592,
0.0021073927637189627,
0.0006273849285207689,
0.00015631067799404263
] |
en
| 0.999996 |
Our application of immunohistochemistry for PMS2 and MSH6 was based on reports suggesting their effectiveness in screening for mismatch repair deficiency (MMRd) . MMRd was identified by the complete absence of nuclear staining for either protein with internal positive controls including unaltered nuclear staining in adjacent normal endometrium, stromal cells, and inflammatory cells. Representative examples of loss of MSH6 or PMS2 expression are shown in Supplementary Fig. S1 A–D. Abnormal p53 staining (p53abn) was characterized as either a strong, diffuse nuclear staining pattern in > 80% of carcinoma cells, or a complete lack of staining ("null pattern"), using adjacent non-tumor cells as an internal control. Wild-type tumor cells exhibited weak and heterogeneous staining patterns . Representative examples of normal and abnormal p53 immunohistochemical staining patterns are shown in Supplementary Fig. S1 E-G.
|
39751650_p10
|
39751650
|
Immunohistochemistry analysis
| 4.143143 |
biomedical
|
Study
|
[
0.9994834661483765,
0.000323768996167928,
0.00019275287922937423
] |
[
0.9993621706962585,
0.00024039750860538334,
0.0003251005837228149,
0.00007235258090076968
] |
en
| 0.999998 |
Multiplex immunofluorescent (IF) staining was performed by the TSA method using Opal IHC kit (PerkinElmer, Waltham, MA) according to the manufacture’s instructions. Anti-pan cytokeratin (clone AE1/AE3, Abcam, Cambridge, UK; 1:600) and anti-human CD8 (clone C8/144b, Dako, Glostrup, Denmark; 1:100) were used as primary antibodies. The antigen retrieval process was carried out as described above, and blocking was performed using the opal kit reagent, followed by incubation at room temperature for 30 min for anti-pan cytokeratin and overnight incubation at 4 °C for anti-CD8. After the primary antibody incubation, the sections were incubated with peroxidase-labeled secondary antibody, followed by incubation with opal 520 or opal 620 reagent. Then, the sections were stained with 4’, 6-diamidino-s-phenylindole (DAPI) (Dojindo, Kumamoto, Japan) and mounted. Multiplexed fluorescent labeled images were captured with a BZ-X800 (Keyence, Osaka, Japan).
|
39751650_p11
|
39751650
|
Immunohistochemistry analysis
| 4.165201 |
biomedical
|
Study
|
[
0.9994592070579529,
0.0003426572075113654,
0.0001981433160835877
] |
[
0.993770182132721,
0.00540211470797658,
0.0006346910377033055,
0.00019295699894428253
] |
en
| 0.999995 |
The assessment of TILs in this study followed the guidelines established by the International Immuno-Oncology Biomarker Working Group . We did not differentiate between intratumoral TILs and stromal TILs during this evaluation. After capturing stained slide images with a VS120-S5 (Evident, Tokyo, Japan), CD8 + TILs were quantified automatically in both the CT and IM using QuPath ver. 0.3.0 . This quantification was performed over five distinct areas, each being a square with 0.25 mm on each side . The average number of CD8 + TILs per square millimeter was calculated for these regions.
|
39751650_p12
|
39751650
|
Immunophenotyping
| 4.072474 |
biomedical
|
Study
|
[
0.9995438456535339,
0.00028488095267675817,
0.00017118656251113862
] |
[
0.9993178844451904,
0.0003515901626087725,
0.0002654635172802955,
0.00006516242865473032
] |
en
| 0.999997 |
The distinction between the three immunophenotypes (inflamed, excluded, and desert) was based on previous research , but as there is no established definition for the density of CD8 + TILs in EC, we adopted 1000 cells/mm 2 in this study. Tumors with a CD8 + TIL density ≥ 1000 cells/mm 2 in both the CT and IM regions were classified as 'inflamed' phenotype. The tumors with a CD8 + TIL density < 1000 cells/mm 2 in the CT but > 1000 cells/mm 2 at the IM were designated as the 'excluded' phenotype. Conversely, tumors were classified as the 'desert' phenotype when the density of CD8 + TILs was < 1000 cells/mm 2 in both the CT and IM areas.
|
39751650_p13
|
39751650
|
Immunophenotyping
| 4.076801 |
biomedical
|
Study
|
[
0.9994933605194092,
0.0002540592395234853,
0.0002525841409806162
] |
[
0.9993784427642822,
0.0003230453876312822,
0.00025022815680131316,
0.00004831719343201257
] |
en
| 0.999997 |
We conducted the molecular classification of tumors using an adapted approach from the ProMisE methodology, aligned with the steps outlined in the World Health Organization (WHO) classification . This approach involved a sequential assessment of specific molecular markers. Initially, all tumor samples underwent Sanger sequencing to identify mutations in the POLE exonuclease domain, specifically targeting exons 9, 13, and 14 as noted above. Tumors harboring pathogenic variants in these regions were classified as 'POLEmut.' Next, the tumors exhibiting a complete absence of nuclear staining for PMS2 or MSH6 protein by IHC were categorized as 'MMRd.' Then, the tumors demonstrating abnormal p53 expression patterns by IHC were classified as 'p53abn.' Finally, tumors that did not exhibit any of the aforementioned molecular characteristics were classified as 'NSMP,' indicating a no specific molecular profile.
|
39751650_p14
|
39751650
|
The ProMisE molecular classification
| 4.12149 |
biomedical
|
Study
|
[
0.9994450211524963,
0.00036587018985301256,
0.00018913066014647484
] |
[
0.998828113079071,
0.00039316152106039226,
0.0006967636290937662,
0.00008196639828383923
] |
en
| 0.999997 |
We defined progression-free survival (PFS) as the duration from the initiation of a patient's treatment to the point of observed disease progression. Overall survival (OS) was determined as the period from the commencement of treatment to either the death of a patient due to any reason or the patient's last confirmed survival status, with data collected up until March 2023 in retrospective cohort or until November 2024 in prospective cohort. We used the Kaplan–Meier method to estimate the 145 patients' PFS and OS rates. To assess the impact of immunophenotypes on the prognosis of EC patients within each molecular classification, survival curves were compared across the three immunophenotypes (inflamed, excluded, and desert) within each ProMisE subtype.
|
39751650_p15
|
39751650
|
Survival analysis
| 4.15277 |
biomedical
|
Study
|
[
0.9974029660224915,
0.002387515502050519,
0.00020949541067238897
] |
[
0.9981060028076172,
0.0011311962734907866,
0.0005429196171462536,
0.0002198413567384705
] |
en
| 0.999998 |
Given the limited sample size of our cohort, the abundance of CD8 + TILs from the Cancer Genome Atlas (TCGA) database was estimated using the CIBERSORT algorithm in R . Patients in the TCGA database were stratified into high and low CD8 + T-cell groups within the four genomic subgroups of EC (POLE, MSI, CN low, and CN high) using a receiver operating characteristic (ROC) curve. An optimal cutoff value of 0.0536085 was adopted for predicting 1-year OS.
|
39751650_p16
|
39751650
|
Survival analysis
| 4.096845 |
biomedical
|
Study
|
[
0.9994767308235168,
0.00028470699908211827,
0.00023847856209613383
] |
[
0.9995156526565552,
0.00022209039889276028,
0.00020560504344757646,
0.000056611424952279776
] |
en
| 0.999996 |
The RNA-sequencing analysis was carried out on samples from the 40 of the 60 patients in prospective cohort from whom RNA samples were available. The extraction of RNA from fresh frozen tumor tissue was done using the NucleoSpin RNA Plus kit (Macherey–Nagel), following the manufacturer's guidelines. We measured the total RNA concentration with the NanoDrop One spectrophotometer. RNA-sequencing was then performed by Novogene Japan (Tokyo). The obtained raw FASTQ data were uploaded to Galaxy, an open-source web-based platform tailored for data-intensive biomedical research. Quality control of the data was executed using FastQC and Trimmomatic. The clean, paired-end data were then processed for gene expression quantification using the Kallisto quant tool, referencing the GENCODE GRC38.p13 transcript (genecode. v41.transcript).
|
39751650_p17
|
39751650
|
RNA-sequencing analysis
| 4.11084 |
biomedical
|
Study
|
[
0.9995148181915283,
0.00029734038980677724,
0.00018796029326040298
] |
[
0.9993603825569153,
0.0003217098710592836,
0.00024273166491184384,
0.00007529588037868962
] |
en
| 0.999996 |
Post-processing, the data were aggregated using the tximport package (ver. 1.18.0) in R software (ver. 4.0.3) and RStudio. For the subsequent analyses, scaled transcripts per million (TPM) counts were used. The TPM counts were processed with the use of the web portal for integrated differential expression and pathway analysis (iDEP) . We also used iDEP 2.01 for a principal component analysis (PCA). A gene set enrichment analysis (GSEA) was conducted employing the GSEA software (ver. 4.3.2), allowing for the identification of significantly altered pathways and gene sets in the dataset.
|
39751650_p18
|
39751650
|
RNA-sequencing analysis
| 4.07827 |
biomedical
|
Study
|
[
0.9994956254959106,
0.0001608775492059067,
0.0003434491518419236
] |
[
0.9989172220230103,
0.0007220542174763978,
0.00030816736398264766,
0.000052615039749071
] |
en
| 0.999997 |
We used GraphPad Prism software, ver. 9.2.0 (GraphPad Software, San Diego, CA) for the statistical analyses. To compare the relationships between different groups, we applied two distinct statistical tests depending on the data structure and distribution: the Wilcoxon matched-pairs signed-rank test was used for paired data comparisons, and the nonparametric Mann–Whitney U-test was applied for unpaired data sets. To compare distributions between the observed and expected data, we used the χ 2 -test.
|
39751650_p19
|
39751650
|
Statistical analysis
| 3.644018 |
biomedical
|
Study
|
[
0.9992092251777649,
0.0001844880753196776,
0.0006062726606614888
] |
[
0.9805481433868408,
0.018478624522686005,
0.0008070127223618329,
0.0001662077265791595
] |
en
| 0.999999 |
The Kaplan–Meier method was used for the survival analyses, i.e., the PFS and OS rates. This allowed us to plot survival curves and estimate survival probabilities over time. The differences in survival rates between groups were evaluated by the log-rank test. Throughout the analyses, a p value threshold < 0.05 was set for determining statistical significance.
|
39751650_p20
|
39751650
|
Statistical analysis
| 3.908514 |
biomedical
|
Study
|
[
0.9995005130767822,
0.0002795895270537585,
0.00021986923820804805
] |
[
0.9976478219032288,
0.0017413586610928178,
0.0005177836283110082,
0.00009301928366767243
] |
en
| 0.999997 |
We conducted a prospective study over a 4-year period, enrolling consecutive 60 EC patients representing a full spectrum of histology. The clinicopathological characteristics of these patients are provided in Table 1 . The cohort had a median follow-up of 38.4 months (range 0.6–66.0 months) and a median age of 58 years (range 32–84 years). Most of the patients (61.7%) presented with endometrioid G1/2 histology, and 66.7% were diagnosed at early stages . The ProMisE classification revealed that 11.7% of these patients fell into the POLEmut subtype, 25.0% into MMRd, 48.3% into NSMP, and 15.0% into p53abn (Table 2 ). The detailed POLE pathogenic variants are presented in Supplementary Table S1 . The distribution of these categories aligns with another investigation of Japanese patients with EC . Notably, higher ages were observed in the p53abn subtype, and the NSMP subtype predominantly consisted of endometrioid G1 and G2 histology. The serous histology tumors were exclusively categorized as p53abn. Table 1 Prospective cohort; EC patients in 2019–2022 Total patients, n 60 Follow–up period, mos.; median (range) 38.4 (0.6–66.0) Age, yrs; median (range) 58.0 (32–84) Histology Endometrioid G1/2 37 (61.7) Endometrioid G3 20 (33.3) Serous 3 (5.0) Stage I 35 (58.3) II 5 (8.3) III 18 (30.0) IV 2 (3.3) Risk of recurrence Low 13 (21.7) Intermediate 18 (30.0) High 29 (48.3) The data are numbers and percentages EC endometrial cancer; FIGO international federation of gynecology and obstetrics Table 2 ProMisE molecular classification in prospective cohort POLEmut ( n = 7) MMRd ( n = 15) NSMP ( n = 29) p53abn ( n = 9) Proportion, % 11.7 25.0 48.3 15.0 Age, yrs; median 57.0 58.0 58.0 66.0 (range) (48–75) (50–65) (32–84) (52–75) Histology Endometrioid G1/2 5 6 23 3 Endometrioid G3 2 9 6 3 Serous 0 0 0 3 Stage I 6 7 19 3 II 0 1 4 0 III 1 6 6 5 IV 0 1 0 1 Risk of recurrence Low 2 2 8 1 Intermediate 4 4 10 0 High 1 9 11 8 ProMisE proactive molecular risk classifier for endometrial cancer; POLEmut polymerase-epsilon mutation; MMRd mismatch repair deficiency; NSMP no specific molecular profile; p53abn p53 abnormality; FIGO international federation of gynecology and obstetrics
|
39751650_p21
|
39751650
|
The clinicopathological characteristics and ProMisE classification of prospective cohort
| 4.209706 |
biomedical
|
Study
|
[
0.998971700668335,
0.0008013054030016065,
0.00022699905093759298
] |
[
0.9992750287055969,
0.00027298330678604543,
0.0003370408376213163,
0.0001149654999608174
] |
en
| 0.999996 |
Figure 1 provides representative histological images of the three immunophenotypes; inflamed , excluded , and desert . The classification was based on the distribution patterns of CD8 + TILs. The inflamed phenotype showed abundant CD8 + TILs in both the CT and the IM , whereas the excluded phenotype had a higher concentration in the IM than the CT . The desert phenotype was characterized by sparse CD8 + TILs in both areas . To further evaluate the spatial relationship between tumor cells and CD8 + TILs, we performed multiplex IF staining for pan cytokeratin (tumor cell marker) and CD8. In the inflamed phenotype, CD8 + TILs were abundant in both the CT and the IM , and CD8 + TILs well infiltrated into the tumor cells in both regions. In the excluded phenotype, there were many CD8 + TILs in the IM , but few in the CT . In the desert phenotype, there were few CD8 + TILs in both the CT and the IM . Fig. 1 Representative images of three immunophenotypes based on the distribution patterns of CD8 + tumor-infiltrating lymphocytes (TILs) in endometrial cancer. The immunohistochemistry images of CD8 + TILs in the inflamed phenotype ( A , D , G ), excluded phenotype ( B , E , H ), and desert phenotype ( C , F , I ) are shown. The middle rows ( D , E , F ) show CD8 + TILs in the CT, and the bottom rows ( G , H , I ) show those in the IM. Abbreviations: CT, central tumor; IM, invasive margin
|
39751650_p22
|
39751650
|
Three distinct immunophenotypes based on the distribution of CD8+ TILs
| 4.126618 |
biomedical
|
Study
|
[
0.9994951486587524,
0.00032526697032153606,
0.0001795618300093338
] |
[
0.9991843104362488,
0.000254719314398244,
0.00048170232912525535,
0.00007929361163405702
] |
en
| 0.999996 |
The CT and IM regions on the hematoxylin and eosin stain are shown in Fig. 2 A. CD8 + TILs were automatically quantified five distinct areas in both the CT and IM, as illustrated in Fig. 2 B. The average density of CD8 + TILs (positive cells per square millimeter) in the CT (blue bar) and IM (red bar) for 60 EC patients is shown in Fig. 2 C. Among all 60 patients, 17 (28.3%) were classified as the inflamed phenotype, 22 (36.7%) as the excluded phenotype, and 21 (35%) as the desert phenotype. The median density of CD8 + TILs in the CT was significantly lower than that in the IM in the inflamed phenotype , in the excluded phenotype , and in the desert phenotype . Fig. 2 The relationship between the ProMisE classification and the immunophenotypes in prospective cohort. The approximate areas of the CT and IM in the hematoxylin and eosin stain are shown ( A ). CD8 + tumor-infiltrating lymphocytes (TILs) were automatically counted at five areas in the CT ( surrounded by blue lines ) and IM ( surrounded by red lines ) ( B ). The average density of CD8 + TILs (positive cells per square millimeter) in the CT ( blue bar ) and IM ( red bar ) for 60 EC patients is shown ( C ). Comparison of the density of the CD8 + TILs in the CT and IM in the inflamed phenotype ( D ), excluded phenotype ( E ), and desert phenotype ( F ). Comparison of the density of the CD8 + TILs by the four ProMisE subtypes in the CT ( G ) and IM ( H ). Abbreviations: CT, central tumor; IM, invasive margin; TILs, tumor-infiltrating lymphocytes; POLEmut, polymerase-epsilon mutation; MMRd, mismatch repair deficiency; NSMP, no specific molecular profile; p53abn, p53 abnormality
|
39751650_p23
|
39751650
|
The relationship between the ProMisE classification and the immunophenotypes in prospective EC cohort
| 4.124865 |
biomedical
|
Study
|
[
0.9992486834526062,
0.0005005151615478098,
0.0002507969329599291
] |
[
0.9994946718215942,
0.00022912415442988276,
0.00020762236090376973,
0.00006850773934274912
] |
en
| 0.999998 |
Figure 2 G depicts the density of CD8 + TILs in the CT in each ProMisE category. The median density of CD8 + TILs in the CT was not significantly different between the POLEmut and MMRd subtypes or between the NSMP and p53abn subtypes . This value was significantly higher in the POLEmut subtype compared to the NSMP and p53abn subtypes and was significantly higher in the MMRd subtype versus the NSMP and p53abn subtypes .
|
39751650_p24
|
39751650
|
The relationship between the ProMisE classification and the immunophenotypes in prospective EC cohort
| 4.033228 |
biomedical
|
Study
|
[
0.9993988275527954,
0.0003528881352394819,
0.00024829135509207845
] |
[
0.999464213848114,
0.0002556093968451023,
0.00021796934015583247,
0.00006228005076991394
] |
en
| 0.999998 |
Figure 2 H illustrates the density of CD8 + TILs in the IM in each of the ProMisE categories. The median density of CD8 + TILs in the IM was not significantly different between the POLEmut and MMRd subtypes but was significantly higher in p53abn subtype than in NSMP subtype unlike the densities in the CT. In terms of the median density of CD8 + TILs in the CT and IM, the POLEmut and MMRd subtypes showed the inflamed phenotype with more CD8 + TILs in both areas; the NSMP subtype showed the desert phenotype with fewer CD8 + TILs in both areas; and the p53abn subtype showed the excluded phenotype with fewer CD8 + TILs in the CT and more in the IM. However, looking at individual cases, some of the POLEmut and MMRd subtypes showed the non-inflamed phenotypes, and some of the NSMP subtype showed the inflamed phenotype.
|
39751650_p25
|
39751650
|
The relationship between the ProMisE classification and the immunophenotypes in prospective EC cohort
| 4.095935 |
biomedical
|
Study
|
[
0.9993681311607361,
0.00034024828346446157,
0.0002915870863944292
] |
[
0.999497652053833,
0.0001994389749597758,
0.000248976779403165,
0.000053895044402452186
] |
en
| 0.999995 |
To further investigate the relationship between the ProMisE classification and immunophenotype, we added that a retrospective cohort consists of 85 EC patients treated prior to 2017 to our analysis. These patients' clinicopathological characteristics are summarized in Supplementary Table S2 . The median follow-up period was 75.8 months (range 0.7–173.3 months). Most of the retrospective cohort (70.6%) presented with endometrioid G3 histology; the other patients presented serous, clear, and mixed histology. The ProMisE classification revealed that 7.1% of the retrospective cohort were the POLEmut subtype, 22.3% had the MMRd subtype, 42.4% showed NSMP subtype, and 28.2% were classified as p53abn subtype (Supplementary Table S3 ). The detailed POLE pathogenic variants are presented in Supplementary Table S4 .
|
39751650_p26
|
39751650
|
The analysis of the relationship between the ProMisE classification and the immunophenotypes in prospective and retrospective EC cohort
| 4.119604 |
biomedical
|
Study
|
[
0.9992393255233765,
0.0005242613842710853,
0.00023649699869565666
] |
[
0.9995205402374268,
0.0001596342772245407,
0.00024261687940452248,
0.00007726362673565745
] |
en
| 0.999998 |
Table 3 summarizes the distribution of the three immunophenotypes in each ProMisE category in prospective and retrospective cohort. Notably, the inflamed phenotype was most frequently observed in the POLEmut and MMRd subtypes, while the desert phenotype was predominant in the NSMP subtype; however, other immunophenotypes were also observed. No cases of the inflamed phenotype were observed in the p53abn subtype. The distribution of the three immunophenotypes in each cohort is shown in Supplementary Table S5 (prospective cohort) and S6 (retrospective cohort). Table 3 Relationship between the ProMisE classifications and immunophenotypes in prospective and retrospective cohorts POLEmut ( n = 13) MMRd ( n = 34) NSMP ( n = 65) p53abn ( n = 33) p value Inflamed 6 22 9 0 p < 0.0001 Excluded 6 9 13 23 Desert 1 3 43 10 ProMisE proactive molecular risk classifier for endometrial cancer; POLEmut polymerase-epsilon mutation; MMRd mismatch repair deficiency; NSMP no specific molecular profile; p53abn p53 abnormality; FIGO international federation of gynecology and obstetrics
|
39751650_p27
|
39751650
|
The analysis of the relationship between the ProMisE classification and the immunophenotypes in prospective and retrospective EC cohort
| 4.100888 |
biomedical
|
Study
|
[
0.9994199275970459,
0.00035063986433669925,
0.00022950720449443907
] |
[
0.9994656443595886,
0.00020089096506126225,
0.0002762535586953163,
0.00005722581045120023
] |
en
| 0.999996 |
The survival analysis revealed distinct prognostic differences across the ProMisE classifications and immunophenotypes. The POLEmut and MMRd subtypes exhibited favorable PFS and OS rates, whereas the p53abn and NSMP subtypes were associated with poorer outcomes . Moreover, the patients with the excluded or desert phenotypes demonstrated significantly worse survival rates compared to those with the inflamed phenotype , highlighting the prognostic relevance of this immunophenotypic categorization. Within each ProMisE subtype, the prognostic trends of the immunophenotypes were generally maintained, with the inflamed phenotype associated with better outcomes compared to the excluded and desert phenotypes. However, due to the small number of patients in each subgroup, these trends did not reach statistical significance. Survival analysis by immunophenotype in each ProMisE subtype is depicted in Supplementary Figure S3 . Fig. 3 Survival analysis of 145 patients stratified by ProMisE classification, and the immunophenotypes. The progression-free survival and overall survival rates according to four ProMisE categories comprising POLEmut ( light blue line ), MMRd ( yellow-green line) , NSMP ( orange line ), and p53abn ( red line ) subtypes are shown ( A , B ). The progression-free survival and overall survival rates according to three immunophenotypes comprising the inflamed ( red line ), excluded ( violet line ), and desert (blue line ) phenotypes are shown ( C , D ). Abbreviations: ProMisE, proactive molecular risk cassifier for endometrial cancer; POLEmut, polymerase-epsilon mutation; MMRd, mismatch repair deficiency; NSMP, no specific molecular profile; p53abn, p53 abnormality
|
39751650_p28
|
39751650
|
Survival analysis of the 145 EC patients stratified by the ProMisE classification and their immunophenotypes
| 4.121298 |
biomedical
|
Study
|
[
0.9991262555122375,
0.0006168709951452911,
0.00025691380142234266
] |
[
0.9991664886474609,
0.00024973266408778727,
0.0004992548492737114,
0.00008452178008155897
] |
en
| 0.999998 |
We utilized the TCGA database to verify whether the CD8 + T-cell fraction, estimated using the CIBERSORT algorithm, serves as a prognostic factor across the four molecular subtypes of EC in a larger cohort. Since the TCGA database does not provide direct immunophenotype data, we categorized cases based on the estimated CD8 + T-cell fraction. Cases with high CD8 + T-cell fractions were classified as CD8-high (analogous to the inflamed type), while those with low CD8 + T-cell fractions were grouped as CD8-low (representing non-inflamed types, including excluded and desert phenotypes). There were few cases in the CD8-low group in POLE subtype, and no differences in prognosis were observed between the two groups . Although it was not significant, there was a trend for the prognosis of the CD8-low group to be worse than that of the CD8-high group in MSI and CN low subtypes. In the CN high subtype, the prognosis was significantly worse in the CD8-low group than in the CD8-high group .
|
39751650_p29
|
39751650
|
Survival analysis of the differences in the abundance of CD8+ T-cells in the four genomic subgroups of EC in the TCGA database
| 4.096725 |
biomedical
|
Study
|
[
0.999406099319458,
0.0003480463637970388,
0.00024594893329776824
] |
[
0.9994926452636719,
0.00018439862469676882,
0.00026394418091513216,
0.00005904473800910637
] |
en
| 0.999997 |
Finally, to investigate factors that produce the different distribution patterns of CD8 + TILs between the non-inflamed (excluded and desert) and inflamed phenotypes, we performed an RNA-sequencing (RNA-seq) analysis. The principal component analysis (PCA) was performed with RNA-seq data of 40 EC samples from prospective cohort. The first two principal components (PCs) are plotted and colored according to the ProMisE classification or immunophenotype . The GSEA of the MMRd subtype with the non-inflamed phenotypes compared to that with the inflamed phenotype revealed that the expressions of MYC target gene sets were more enriched in the non-inflamed phenotypes versus the inflamed phenotype . Figure 4 D shows the enrichment plots for the top two datasets that were enriched in the GSEA hallmark analysis. The heat map of the top 50 marker genes for each phenotype in the MMRd subtype with the non-inflamed phenotypes and that with the inflamed phenotype is provided as Fig. 4 E. The top two genes that were upregulated in the non-inflamed phenotype were CD99 and NLGN1 . The top gene that was upregulated in the inflamed phenotype was BRINP1. Fig. 4 The results of the RNA-sequencing analysis in Cohort 2. Principal component analysis plots for the data of 40 samples in Cohort 2 are shown. The first two principal components are plotted and colored according to the ProMisE classification ( A ) or immunophenotype ( B ). The results of a gene set enrichment analysis (GSEA) of the MMRd subtype with the non-inflamed (excluded and desert) phenotypes versus that with the inflamed phenotype comparison are illustrated, as are the results of the GSEA hallmark analysis showing significantly enriched gene sets (FDR < 25% and a nominal p value < 5%). A positive normalized enrichment score indicates enrichment in the MMRd subtype with the non-inflamed phenotypes ( C ). Enrichment plots for the top two datasets enriched in the GSEA hallmark analysis, showing the profile of the running enrichment score and the positions of gene set members on the rank-ordered list ( D ). The heat map of the top 50 marker genes for each phenotype with the non-inflamed phenotypes ( left column ) versus that with the inflamed phenotype ( right column ). Expression values are represented as colors and range from red (high expression), pink (moderate), and light blue (low) to dark blue (lowest expression) ( E ). Abbreviations: PC, principal component; POLEmut, polymerase-epsilon mutation; MMRd, mismatch repair deficiency; NSMP, no specific molecular profile; p53abn, p53 abnormality
|
39751650_p30
|
39751650
|
Comparison of gene expression between the non-inflamed (excluded and desert) and inflamed phenotypes
| 4.200213 |
biomedical
|
Study
|
[
0.9994043111801147,
0.00034038297599181533,
0.0002553451049607247
] |
[
0.9994358420372009,
0.00019763009913731366,
0.0002987385669257492,
0.00006781948468415067
] |
en
| 0.999999 |
Our GSEA of the NSMP subtype with the non-inflamed phenotypes compared to that with the inflamed phenotype revealed that the expressions of type-1 interferon response gene sets were more enriched in the non-inflamed phenotypes versus the inflamed phenotype . The enrichment plots for the top two datasets enriched in the GSEA hallmark analysis are given in Fig. 5 B. The heat map of the top 50 marker genes for each phenotype in the NSMP subtype with the non-inflamed phenotypes and that with the inflamed phenotype is shown in Fig. 5 C. The top gene that was upregulated in the non-inflamed phenotype was OVOL2 . The top gene that was upregulated in the inflamed phenotype was HDC. Fig. 5 Gene set enrichment analysis (GSEA) results of the NSMP subtype with the non-inflamed (excluded or desert) phenotypes versus that with the inflamed phenotype in Cohort 2. The results of the GSEA hallmark analysis showing significantly enriched gene sets (FDR < 25% and a nominal p value < 5%). A positive normalized enrichment score indicates enrichment in the NSMP subtype with the non-inflamed phenotypes, and a negative score indicates enrichment in that with the inflamed phenotype ( A ). Enrichment plots for the top two datasets enriched in GSEA hallmark analysis, showing the profile of the running enrichment score and the positions of gene set members on the rank-ordered list ( B ). Heat map of the top 50 marker genes for each phenotype with the non-inflamed phenotypes ( left column ) versus that with the inflamed phenotype ( right column ). Expression values are represented as colors and range from red (high expression), pink (moderate), and light blue (low) to dark blue (lowest expression) ( C ). Abbreviation: NSMP, no specific molecular profile
|
39751650_p31
|
39751650
|
Comparison of gene expression between the non-inflamed (excluded and desert) and inflamed phenotypes
| 4.176056 |
biomedical
|
Study
|
[
0.9993776679039001,
0.00035522188409231603,
0.0002671488036867231
] |
[
0.9994921684265137,
0.00020160048734396696,
0.00024506915360689163,
0.00006118028977653012
] |
en
| 0.999999 |
Our investigation into the distribution and prognostic significance of CD8 + TILs in EC according to a molecular classification revealed crucial findings. Most notably, we identified three distinct immunophenotypes—inflamed, excluded, and desert—based on CD8 + TILs in EC patients. These immunophenotypes provide a more granular understanding of the immune landscape in EC, reflecting the diverse immunological responses triggered by tumor development. Our results demonstrated that inflamed phenotypes were associated with better prognosis, while excluded and desert phenotypes correlated with poorer outcomes. Importantly, the integration of immunophenotypes with the ProMisE molecular classification underscores the complexity of EC and highlights the need for personalized therapeutic strategies that consider both molecular and immunological characteristics.
|
39751650_p32
|
39751650
|
Discussion
| 4.191726 |
biomedical
|
Study
|
[
0.9994639754295349,
0.00034727973979897797,
0.0001886919344542548
] |
[
0.9979532957077026,
0.00025169638684019446,
0.0016959793865680695,
0.00009907047206070274
] |
en
| 0.999998 |
Interestingly, we observed that the EC patients with the excluded or desert phenotypes had poorer prognoses than those exhibiting inflamed phenotypes. This observation aligns with the increasing body of evidence suggesting that a robust anti-tumor immune response, represented by a high level of CD8 + TILs, is associated with better outcomes in various cancers, including EC . Additionally, TCGA-based analysis using the CIBERSORT algorithm supported this observation, revealing that cases with high CD8 + T-cell fractions (analogous to inflamed phenotypes) demonstrated better prognoses. These results emphasize the value of incorporating immunophenotypic evaluation, alongside molecular classification, into prognostic assessments for EC.
|
39751650_p33
|
39751650
|
Discussion
| 4.135058 |
biomedical
|
Study
|
[
0.9995218515396118,
0.0002909668837673962,
0.00018723240646068007
] |
[
0.9990623593330383,
0.00021336817007977515,
0.0006480502197518945,
0.00007624901627423242
] |
en
| 0.999996 |
When evaluated by the median density of CD8 + TILs in the CT and IM, the POLEmut and MMRd subtypes showed the inflamed phenotype; the NSMP subtype showed the desert phenotype; and the p53abn subtype showed the excluded phenotype. This insight suggests a potential correlation between the genomic background and CD8 + T-cell anti-tumor response in EC. It has been reported that POLEmut and MMRd subtypes have high tumor mutation burdens which contribute to the expression of neoantigens and they cause a strong anti-tumor response by CD8 + T-cell . The limited CD8 + T-cell response in the NSMP subtype suggests that this subtype is less likely to elicit a CD8 + T-cell response due to its low immunogenicity and may contribute to its resistance to immunotherapy strategies. Our observation that the p53abn subtype exhibited the excluded phenotype suggests that this subtype, despite eliciting an immune response, is resistant to immunotherapy because it has mechanisms that exclude CD8 + TILs from the central tumor. However, looking at individual cases, some of the POLEmut and MMRd subtypes showed the non-inflamed phenotypes, and some of the NSMP subtype showed the inflamed phenotype. It is thought that the formation of different immunophenotypes is due to some factors other than the number of genetic mutations or differences in immunogenicity.
|
39751650_p34
|
39751650
|
Discussion
| 4.254195 |
biomedical
|
Study
|
[
0.9994481205940247,
0.0003304209094494581,
0.00022152249584905803
] |
[
0.9990831613540649,
0.0002508223697077483,
0.0005831972812302411,
0.00008277082815766335
] |
en
| 0.999999 |
Our RNA-seq analysis provided valuable insights into the molecular mechanisms underlying the formation of inflamed and non-inflamed phenotypes within the ProMisE subtypes. In the MMRd subtype, non-inflamed phenotypes exhibited upregulation of CD99 and NLGN1, both of which have been implicated in immune suppression and tumor progression. There have been reports that high expression of CD99 in tumors is involved in the infiltration of immunosuppressive macrophages, in addition to the malignant transformation of tumor cells themselves , and some study reported that high expression of NLGN1 was a poor prognostic factor in colorectal cancer, prostate cancer, and pancreatic cancer . In contrast, inflamed phenotypes showed upregulation of BRINP1, a gene associated with immune cell differentiation and PD-L1 regulation in tumor cells, suggesting its role in fostering robust anti-tumor immune responses . In the NSMP subtype, the non-inflamed phenotypes were characterized by elevated expression of OVOL2, a potential tumor suppressor gene , However, the role of OVOL2 in tumor immunity remains unclear. In contrast, inflamed phenotypes showed increased expression of HDC, an enzyme involved in histamine production, which has recently been associated with immune modulation in the tumor microenvironment . There is scope for further exploration of the impact of these genes on tumor immunity.
|
39751650_p35
|
39751650
|
Discussion
| 4.367057 |
biomedical
|
Study
|
[
0.9994245767593384,
0.00035709995427168906,
0.00021835890947841108
] |
[
0.9981860518455505,
0.0003030765801668167,
0.0013970118016004562,
0.0001138294319389388
] |
en
| 0.999997 |
Our RNA-seq analysis also provided another layer of insight, suggesting that MYC target genes or type-1 interferon response genes might play a role in determining these immunophenotypes. The MYC oncogene is a well-documented driver of tumorigenesis in many cancers, and the type-1 interferon pathway is an important pathway in the antiviral response; however, their roles in the formation of the tumor immunosuppressive microenvironment remain unclear. Our findings suggest that the MYC signaling pathway or type-1 interferon pathway may, directly or indirectly, influence the distribution of CD8 + TILs and form a different immunophenotype, representing a novel avenue for future research.
|
39751650_p36
|
39751650
|
Discussion
| 4.188513 |
biomedical
|
Study
|
[
0.9995742440223694,
0.00021953867690172046,
0.00020623925956897438
] |
[
0.9993038177490234,
0.0002284653455717489,
0.00040334570803679526,
0.00006434023089241236
] |
en
| 0.999998 |
There are some limitations to our research. First, we only assessed CD8 + TILs in the evaluation of the immune microenvironment of endometrial cancer, To clarify the immune microenvironment, it is necessary to assess the distribution patterns of immune cells other than CD8 + T-cells and the interaction of each immune cell and also necessary to assess the expression of immune evasion molecules such as PD-L1 expression on tumor cells, which is a future task. Next, there is a lack of functional analysis of the MYC target genes and type-1 IFN genes in terms of their effects on immune phenotypes and EC patients’ clinical outcomes, which is another topic for future research. Furthermore, because the number of tumors in which RNA-seq was performed was limited, it was not possible to find the factors that cause the differences between the excluded and desert phenotypes. As it is predicted that it will be difficult to find the differences in these two non-inflamed phenotypes using RNA-seq of only one region in the CT, we believe that it will be necessary to compare the gene expression in the CT and the IM in future research.
|
39751650_p37
|
39751650
|
Discussion
| 4.048442 |
biomedical
|
Study
|
[
0.9995443224906921,
0.00021513162937480956,
0.00024058193957898766
] |
[
0.999208390712738,
0.00022032081324141473,
0.0005148059572093189,
0.00005648122532875277
] |
en
| 0.999995 |
In conclusion, our results suggest that evaluating not only the molecular classification but also the immunophenotype may lead to more accurate patients’ prognosis prediction in EC. Future studies exploring the role of the MYC signaling pathway or type-1 interferon pathway in shaping the immune landscape will undoubtedly provide further insights into the complex biology of EC. Elucidating the mechanisms that underlie the formation of the three immunophenotypes could lead to the discovery of novel immunotherapy targets.
|
39751650_p38
|
39751650
|
Discussion
| 4.035459 |
biomedical
|
Study
|
[
0.9995903372764587,
0.00024894715170376003,
0.00016080641944427043
] |
[
0.9967760443687439,
0.0003536942240316421,
0.002783561358228326,
0.00008670848910696805
] |
en
| 0.999995 |
Below is the link to the electronic supplementary material. Supplementary file 1 Supplementary file 2 (PDF 41 KB)
|
39751650_p39
|
39751650
|
Supplementary Information
| 1.044788 |
other
|
Other
|
[
0.2562398314476013,
0.0028378127608448267,
0.7409223318099976
] |
[
0.009614438749849796,
0.9888092875480652,
0.0010562385432422161,
0.0005200589657761157
] |
en
| 0.999996 |
Malignant neoplasm of the breast was the fifth leading cause of death among women in Germany in 2020, with 18,500 deaths, according to the last available statistics of the German Federal Statistical Office . Socioeconomic inequalities in mortality due to breast cancer (BC) have been documented in Germany both at the individual and district level , where low-income women or women living in areas with higher levels of deprivation entail a higher mortality risk.
|
PMC11699213_p0
|
PMC11699213
|
Introduction
| 2.541626 |
biomedical
|
Other
|
[
0.981641411781311,
0.0009609742555767298,
0.017397593706846237
] |
[
0.3543992340564728,
0.6337065696716309,
0.010915271006524563,
0.0009789153700694442
] |
en
| 0.999998 |
The European Commission encouraged Member States to implement organised screening programmes (OSP) in 2003, with invitations being sent out on a biannual basis to women aged between 50 and 69 . Since then, numerous European studies have reported a decrease in breast cancer mortality rates and a reduction in inequalities in access to screening services .
|
PMC11699213_p1
|
PMC11699213
|
Introduction
| 2.110677 |
biomedical
|
Other
|
[
0.9540270566940308,
0.003922629170119762,
0.042050402611494064
] |
[
0.02295364998281002,
0.956952691078186,
0.019348541274666786,
0.000745164230465889
] |
en
| 0.999996 |
Germany initiated the implementation of the OSP in 2005 and achieved full implementation by 2009 . The participation rate following an invitation has fluctuated between 43% and 55% over the past two decades, failing to reach the 70% benchmark recommended by the European Commission . Additionally, in 2020, 10.38% of the targeted women reported that they had never attended BCS in their lifetime .
|
PMC11699213_p2
|
PMC11699213
|
Introduction
| 1.866198 |
biomedical
|
Study
|
[
0.8613728880882263,
0.008029746823012829,
0.13059736788272858
] |
[
0.6009232401847839,
0.3960455358028412,
0.0018203724175691605,
0.001210808171890676
] |
en
| 0.999998 |
Several studies have investigated the sociodemographic characteristics of women who are at higher risk of not participating in breast cancer screening (BCS) programmes. In the most recent international systematic review, Mottram et al. observed that migrant women, women with lower socioeconomic status, without home ownership, and those who experienced false positives had the lowest attendance rates . In a scoping review of the German context, Pedrós Barnils et al. identified native women, women with lower incomes, women living in rural areas, and those not cohabiting with their partners as those with the lowest lifetime BCS attendance rates. However, the author also highlighted considerable heterogeneity in methods and, therefore, results .
|
PMC11699213_p3
|
PMC11699213
|
Introduction
| 3.87073 |
biomedical
|
Review
|
[
0.9924246072769165,
0.002244804287329316,
0.005330507643520832
] |
[
0.038380395621061325,
0.0008221991010941565,
0.9605700969696045,
0.00022724605514667928
] |
en
| 0.999996 |
Usually, inequalities in attendance are documented based on independent social dimensions. To explore correlations between social dimensions and BCS attendance, most studies incorporate variables deemed relevant (i.e. based on specific assumptions) into statistical models and then, in multivariate analyses, estimate the independent effect of each social dimension with the effects of other covariates held constant. However, as no individual can be defined by a single social dimension alone , it is unlikely that examining the independent effect of each social dimension will provide a comprehensive understanding of the inequalities in accessing cancer screenings.
|
PMC11699213_p4
|
PMC11699213
|
Introduction
| 3.640365 |
biomedical
|
Study
|
[
0.9949877262115479,
0.0003706666175276041,
0.004641667474061251
] |
[
0.9497860074043274,
0.008065477013587952,
0.0420067124068737,
0.00014173243835102767
] |
en
| 0.999999 |
Instead, individuals sit at the intersection of different social dimensions, and this needs to be considered when assessing who is at higher risk of not attending BCS. Methodologically and conceptually, the way the risk for not attending BCS of a person with a migration background and low educational attainment can be seen differently: either as the sum of (presumably) independent discrimination dimensions or as accounting for the discrimination of being a migrant from a lower social class simultaneously . It is, therefore, essential to employ a framework that allows to detect the inherent complexity of inequalities when attempting to understand the underlying factors influencing BCS attendance. The most appropriate approach is to adopt the framework of intersectionality . Intersectionality theory, as first proposed by law scholar Kimberlé Crenshaw in 1989, posits that the experiences of discrimination (e.g. classism, racism) based on disadvantaged social positions (e.g. low social class, migration background) overlap and derive into unique experiences of discrimination .
|
PMC11699213_p5
|
PMC11699213
|
Introduction
| 2.886972 |
other
|
Other
|
[
0.16629062592983246,
0.0012510694796219468,
0.8324582576751709
] |
[
0.348322331905365,
0.6138604283332825,
0.03697798773646355,
0.0008391969022341073
] |
en
| 0.999997 |
Over the past two decades, in the field of population health, quantitative intersectionality has given rise to new methodological approaches. The most commonly used methods for describing intersectional inequalities within a population range from simple cross-classification descriptions or regressions to methods that account for discriminatory accuracy (e.g. analysis of individual heterogeneity and discriminatory accuracy (AIHDA) and multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA)) or are data-driven (e.g. decision trees) .
|
PMC11699213_p6
|
PMC11699213
|
Introduction
| 3.664 |
biomedical
|
Review
|
[
0.9571534991264343,
0.0009019523859024048,
0.04194458946585655
] |
[
0.192295640707016,
0.07898081094026566,
0.7282319068908691,
0.000491628423333168
] |
en
| 0.999998 |
To build cross-classification regression and AIHDA or MAIHDA, the (potentially) relevant social dimensions are usually selected on the basis of the available evidence, and these dimensions are combined to identify intersectional subgroups. This is a deductive approach. In contrast, decision trees and analogous heuristic procedures employ an inductive methodology to identify which variables are most predictive of an outcome assuming non-linear relationships between the variables . This enables a data-driven determination of the social dimensions that will constitute intersectional subgroups, often previously unnoticed . Decision trees have been applied as statistical exploratory tools for classification in population health .
|
PMC11699213_p7
|
PMC11699213
|
Introduction
| 3.655246 |
biomedical
|
Study
|
[
0.9304715991020203,
0.0005193323013372719,
0.06900910288095474
] |
[
0.6514842510223389,
0.30743634700775146,
0.04065079614520073,
0.00042859293171204627
] |
en
| 0.999997 |
To the author's knowledge, no explicit comparisons between these approaches to identify intersectional inequalities in breast cancer screening have been conducted; besides, no study has employed an intersectional approach for reporting inequalities in breast cancer screening in Germany. Consequently, the present study aims to identify intersectional groups of women aged 50–69 who are at higher risk of never attending BCS in Germany comparing two analytical strategies: a) evidence-informed regression and b) decision tree-based regression.
|
PMC11699213_p8
|
PMC11699213
|
Introduction
| 3.846583 |
biomedical
|
Study
|
[
0.9965218305587769,
0.0004181053664069623,
0.0030600696336477995
] |
[
0.9994733929634094,
0.0002525287272874266,
0.0002331001014681533,
0.00004107772474526428
] |
en
| 0.999998 |
For this analysis, we employed cross-sectional data from the European Health Interview Survey (EHIS) third wave conducted in Germany in 2019. The survey sample size was 23,001 respondents, corresponding to 21.6% of the invited participants . EHIS is conducted every 5 years and focuses on individuals aged 15 and above residing in private households .
|
PMC11699213_p9
|
PMC11699213
|
European Health Interview Survey
| 2.12578 |
biomedical
|
Study
|
[
0.9640954732894897,
0.0012380687985569239,
0.03466654196381569
] |
[
0.9927207231521606,
0.00674560246989131,
0.000337662931997329,
0.00019613005861174315
] |
en
| 0.999996 |
The primary outcome of this study was self-reported breast cancer screening attendance via mammography at least once in a lifetime for women aged 50–69 in Germany. Responses were dichotomised, excluding those who indicated “unknown” or left the question unanswered to prevent uncertainty about whether the respondent reported never attending BCS (no = 0, yes = 1).
|
PMC11699213_p10
|
PMC11699213
|
Primary outcome
| 3.365308 |
biomedical
|
Study
|
[
0.9973124265670776,
0.001588690560311079,
0.0010988481808453798
] |
[
0.9991344809532166,
0.000600750558078289,
0.00016425331705249846,
0.00010053873847937211
] |
en
| 0.999996 |
The explanatory variables to predict BCS derive from the PROGRESS-Plus characteristics: place of residence, race, ethnicity, culture and language, occupation, sex, education, socioeconomic status, social capital and plus (i.e. other potentially discriminatory factors) . These variables have been widely used to disentangle social inequities in health .
|
PMC11699213_p11
|
PMC11699213
|
Explanatory variables
| 2.162663 |
biomedical
|
Study
|
[
0.978552520275116,
0.0007704416057094932,
0.020676983520388603
] |
[
0.5896098017692566,
0.3990786075592041,
0.010526750236749649,
0.0007848451496101916
] |
en
| 0.999997 |
Place of Residence was determined through the degree of urbanisation of the municipality and the specific region ( Bundesland ). The first variable was composed of three categories: cities (densely populated areas), towns and suburbs (intermediate-density areas), and villages (thinly populated areas). The second variable indicated the federal states ( Bundesländer ) in Germany.
|
PMC11699213_p12
|
PMC11699213
|
Explanatory variables
| 1.416028 |
other
|
Other
|
[
0.17245592176914215,
0.0011443640105426311,
0.8263997435569763
] |
[
0.3177298307418823,
0.6801803112030029,
0.0009546726942062378,
0.0011351787252351642
] |
en
| 0.999995 |
Race, ethnicity, culture, and language were indicated by proxy variables: since the EHIS did not assess either of these explicitly. We selected the respondent's country of origin and nationality and then classified them as either born in Germany, in Europe or outside Europe. Although short at measuring complexities of identity, these variables have shown utility as ethnicity proxies in European countries where no information on race or ethnicity is gathered .
|
PMC11699213_p13
|
PMC11699213
|
Explanatory variables
| 1.616578 |
other
|
Study
|
[
0.3094514310359955,
0.001284637488424778,
0.6892638802528381
] |
[
0.803642749786377,
0.19449862837791443,
0.0010495439637452364,
0.000809083750937134
] |
en
| 0.999995 |
Occupation was operationalised based on the respondents’ current working situation: in paid employment, unemployed, retired, unable to work, (unpaid) household work and others.
|
PMC11699213_p14
|
PMC11699213
|
Explanatory variables
| 1.209596 |
other
|
Other
|
[
0.07796751707792282,
0.001427055336534977,
0.920605480670929
] |
[
0.2584998905658722,
0.7377960681915283,
0.0019136543851345778,
0.0017903796397149563
] |
en
| 0.999998 |
Sex (to identify as a female) was a prerequisite for participant inclusion in the analysis. Gender and religion were not captured by the EHIS.
|
PMC11699213_p15
|
PMC11699213
|
Explanatory variables
| 1.713121 |
biomedical
|
Study
|
[
0.9354532957077026,
0.002613047370687127,
0.061933714896440506
] |
[
0.8470478653907776,
0.150987446308136,
0.0009748990414664149,
0.0009897294221445918
] |
en
| 0.999995 |
Education was measured following the ISCED-2011 classification . Since only 6 participants had primary education or less, the first three categories were combined into “less than upper secondary education”.
|
PMC11699213_p16
|
PMC11699213
|
Explanatory variables
| 1.659035 |
biomedical
|
Study
|
[
0.6866068840026855,
0.0017223514150828123,
0.3116707503795624
] |
[
0.8764363527297974,
0.1218227967619896,
0.0010707674082368612,
0.0006700129015371203
] |
en
| 0.999997 |
Socioeconomic status was operationalised through household income and was divided into five quintile groups: the 20% with the lowest income were coded 1, and the 20% with the highest income were coded 5 .
|
PMC11699213_p17
|
PMC11699213
|
Explanatory variables
| 1.746747 |
biomedical
|
Study
|
[
0.8668129444122314,
0.0020873500034213066,
0.13109982013702393
] |
[
0.8807166814804077,
0.11758095026016235,
0.0009646161342971027,
0.0007377054425887764
] |
en
| 0.999998 |
Social capital was considered through six variables: social network dimensions (none, 1–2, 3–5, 6 or more), perceived social support (a lot, some, uncertain, little, or no concern) and ease in available help (very easy, easy, possible, difficult, or very difficult). Further, three proxy variables were also included: marital status (single, married, legally separated/divorced or widowed), type of household (alone, with a partner, with a partner and children, with children, or other) indicating the availability of family support, and partner cohabitation (yes or no).
|
PMC11699213_p18
|
PMC11699213
|
Explanatory variables
| 1.794843 |
other
|
Study
|
[
0.3610469698905945,
0.0015686115948483348,
0.6373844742774963
] |
[
0.9790216684341431,
0.01997048780322075,
0.0006877469713799655,
0.00032010237919166684
] |
en
| 0.999996 |
For the Plus dimension, the Global Activity Limitations Indicator (GALI), a self-report of the extent of limitation experienced in the last six months was considered, with possible answers: severely limited, mildly limited, or not limited . Age (50–69 years old) was required to be included in the analysis and was treated as a confounder in the regression analyses.
|
PMC11699213_p19
|
PMC11699213
|
Explanatory variables
| 2.631799 |
biomedical
|
Study
|
[
0.9937084913253784,
0.0013324974570423365,
0.004958925303071737
] |
[
0.9971612691879272,
0.002479543210938573,
0.00022690618061460555,
0.0001323452452197671
] |
en
| 0.999997 |
Descriptive analytics, including frequencies and percentages, were calculated for all variables. A complete case analysis was conducted, i.e. cases with missing data were excluded listwise. The total sample was restricted to women aged 50–69 . Among these women, those who did not respond on whether they underwent mammography (n = 15), their place of residence (n = 384), the degree of urbanisation of their place of residence (n = 213), the household's income (n = 122), their level of education (n = 14), their social network dimensions (n = 11), their perceived social support (n = 46), the available help (n = 81), the type of household (n = 56), their marital status (n = 13), their partnership cohabitation status (n = 30), their working situation (n = 10), their country of origin (n = 11), their citizenship (n = 7), their GALI (n = 7) were excluded. Hence, the final total sample size of the study was 4761 participants.
|
PMC11699213_p20
|
PMC11699213
|
Analytic approach
| 3.675934 |
biomedical
|
Study
|
[
0.9979518055915833,
0.0006399523699656129,
0.0014082894194871187
] |
[
0.9995922446250916,
0.00027089272043667734,
0.00009054371912498027,
0.00004633212665794417
] |
en
| 0.999997 |
Sampling weights were not used in analyses, as the sampling weights provided in the German EHIS data were derived from variables included in the analyses (education, urbanisation and age), which could lead to multicollinearity and biased standard error estimation. We report sensitivity analyses applying the sampling weights in both analytical strategies in Appendix A and show the correlations between sampling weights and variables in the analyses in Appendix B . The central aim of this article was to compare the estimation of women at higher risk of never attending BCS using two different analytical strategies: (a) evidence-informed regression and (b) decision tree-based regression.
|
PMC11699213_p21
|
PMC11699213
|
Analytic approach
| 3.313433 |
biomedical
|
Study
|
[
0.996536374092102,
0.0006029200158081949,
0.002860784064978361
] |
[
0.9993801116943359,
0.0004055850731674582,
0.00015836914826650172,
0.00005591490844381042
] |
en
| 0.999995 |
The evidence-informed analytical strategy builds a full cross-classification matrix based on social dimensions identified as relevant in the literature. A recent scoping review pinpointed migration background, socioeconomic position (based on income), degree of urbanisation, and partner cohabitation as significant dimensions for BCS attendance prediction .
|
PMC11699213_p22
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 2.444877 |
biomedical
|
Review
|
[
0.8803834915161133,
0.0026834867894649506,
0.11693309247493744
] |
[
0.4176454544067383,
0.0329788513481617,
0.5478602647781372,
0.0015154173597693443
] |
en
| 0.999995 |
For this analysis, country of origin was dichotomised as born inside or outside Germany, income was dichotomised into low (categories 1 and 2) and high (categories 3, 4 and 5), degree of urbanisation was dichotomised in people living in cities (urban) and people living in towns, suburbs or rural areas (rural), and partner cohabitation was already a dichotomous variable (yes/no). The cross-classification of all social positions led to 16 intersectional groups: 2 (country of origin) ∗ 2 (income) ∗ 2 (degree of urbanisation) ∗ 2 (partner cohabitation) ( Table 1 ). Table 1 Evidence-informed intersectional groups on lifetime BCS attendance. Table 1 Country of origin Income Degree of urbanisation Partner cohabitation Intersectional group name Germany High Urban Yes HGUY No HGUN Rural Yes HGRY No HGRN Low Urban Yes LGUY No LGUN Rural Yes LGRY No LGRN Other than Germany High Urban Yes HOUY No HOUN Rural Yes HORY No HORN Low Urban Yes LOUY No LOUN Rural Yes LORY No LORN
|
PMC11699213_p23
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 2.376462 |
biomedical
|
Study
|
[
0.7482686042785645,
0.0015646711690351367,
0.25016677379608154
] |
[
0.9963394403457642,
0.003302958095446229,
0.00024020169803407043,
0.00011750867997761816
] |
en
| 0.999998 |
For the purpose of comparison, univariate models were initially constructed for each of the four individual predictors and age. Next, a multivariable model that included all main effects was estimated.
|
PMC11699213_p24
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 3.017295 |
biomedical
|
Study
|
[
0.9958648681640625,
0.0005494357901625335,
0.0035857628099620342
] |
[
0.9982624650001526,
0.0011984329903498292,
0.00044992376933805645,
0.0000892204261617735
] |
en
| 0.999996 |
Following this, a multivariate logistic regression with the full cross-classification matrix as the main predictor was performed to estimate the odds ratio (OR) of never attending BCS adjusted by age. Discriminatory accuracy (DA) was estimated through the area under the receiver operating characteristics curve (AUC) with a 95% confidence interval (CI), indicating how well each model discriminates between women attending and women never attending BCS. DA is considered absent or very small when 0.5 ≤AUC≤ 0.6, moderate when 0.6< AUC ≤0.7, large when 0.7< AUC ≤0.8 and very large AUC>0.8 . These statistical procedures were carried out using Stata version 17.0.
|
PMC11699213_p25
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 4.059789 |
biomedical
|
Study
|
[
0.9992583394050598,
0.0003967055235989392,
0.0003450067015364766
] |
[
0.9993851184844971,
0.00026927769067697227,
0.000291690812446177,
0.000053960666264174506
] |
en
| 0.999994 |
The second analytical strategy consisted of two steps. First, building an explorative decision tree with the total sample size to identify homogeneous subgroups of women at higher risk of never attending BCS in Germany. Second, performing a multivariate logistic regression using the outcome of the decision tree adjusted by age to estimate the OR of never attending BCS.
|
PMC11699213_p26
|
PMC11699213
|
Analytical strategy b: decision tree-based regressions
| 3.894325 |
biomedical
|
Study
|
[
0.9990373849868774,
0.0004279125714674592,
0.0005347570986486971
] |
[
0.9992870688438416,
0.0004384583153296262,
0.00022233287745621055,
0.00005221804167376831
] |
en
| 0.999996 |
There is no consensus on which decision tree better operates on binary outcomes. In this study, we trained three different algorithms: Classification and Regression Tree (CART), Conditional Inference Tree (CIT) and C5.0. The CART algorithm makes splitting decisions based on the lowest gini impurity (or entropy) coefficient among all potential splits (i.e. every category or step of every variable) . CART does not provide statistical significance measures and potentially overestimates the influence of variables with many categories. CIT addresses these limitations by utilising a formal statistical hypothesis in growing decision trees and mitigating variable selection bias by splitting the selection process into two steps . C5.0 uses the entropy coefficient of the imputed variables to generate splits plus adaptative boosting and winnowing .
|
PMC11699213_p27
|
PMC11699213
|
Analytical strategy b: decision tree-based regressions
| 4.020599 |
biomedical
|
Study
|
[
0.9910023808479309,
0.00031062751077115536,
0.008687086403369904
] |
[
0.9993066787719727,
0.00039490373455919325,
0.0002672579721547663,
0.000031186515116132796
] |
en
| 0.999998 |
All three decision tree algorithms (CART, CIT, C5.0) were built using the entire dataset and the same subdivision of the data when performing cross-validation. Cost weights were applied to distribute the sums of weights equally for cases and non-cases, given the (relative) rareness of the outcome in the dataset (10.38% prevalence). Parameters were hypertuned and optimised by two performance measures: sensitivity (i.e. enhancing detection of positive cases) and the Area Under the Precision-Recall Curve (i.e. improving overall precision-recall performance for unbalanced datasets) . Decision trees were grown using the tune function from the “mlr3tuning” optimisation R packages in R version 4.4.0. This package integrates essential packages for building CART “rpart” , CIT “partykit” , and C5.0 “C50” .
|
PMC11699213_p28
|
PMC11699213
|
Analytical strategy b: decision tree-based regressions
| 4.057607 |
biomedical
|
Study
|
[
0.999180257320404,
0.00023783621145412326,
0.000581839238293469
] |
[
0.9991123080253601,
0.0005429346929304302,
0.000304856599541381,
0.00003985732109867968
] |
en
| 0.999997 |
After inductively identifying the best-performing decision tree, the final nodes were deductively used as predictors for a multivariate logistic regression adjusted by age, where the ORs and DA of the model were estimated. This statistical procedure was performed using the Stata version 17.0. Estimations, performance and interpretability of both analytical strategies were compared and discussed.
|
PMC11699213_p29
|
PMC11699213
|
Analytical strategy b: decision tree-based regressions
| 3.816376 |
biomedical
|
Study
|
[
0.9992081522941589,
0.00022524093219544739,
0.000566640286706388
] |
[
0.9991944432258606,
0.0005584555910900235,
0.00019608999718911946,
0.00005106988101033494
] |
en
| 0.999997 |
Summary descriptive statistics of the sample can be found in Table 2 . The total sample size is 4761. Of those, 4267 attended BCS at least once in their lifetime, and 494 did not. Relative frequencies for never attending BCS among the different PROGRESS-Plus characteristics were assessed. As expected, women aged 65–69 had the lowest prevalence (6.55%), and women aged 50–54, had the highest prevalence (18.58%). For SES, women in the lowest quintile attended BCS the least (14.07%), and those in the highest quintile attended BCS the most (9.79%). Almost contradicting, women with the highest education attainment, doctoral or equivalent, attended BCS the least (14.75%) and women with bachelor or equivalent educational attainment the most (9.10%). Based on the country of origin, women born in Germany (10.56%) had the lowest attendance rates compared to women born elsewhere. However, women of another European nationality attended the least (12.05%) and women of German nationality the most (10.31%). Regarding the place of residence, women living in cities (11.16%), women living in Berlin (13.69%) and Saarland (13.39%) had the lowest BCS attendance rate. Table 2 Descriptive PROGRESS-Plus characteristics on BCS attendance among targeted women in Germany. Relative frequencies per column and variable are displayed. Table 2 Attended BCS Never attended BCS (N = 494) Total Age 50-54 916 (21.5%) 209 (42.3%) 1125 (23.6%) 55-59 1149 (26.9%) 116 (23.5%) 1265 (26.6%) 60-64 1104 (25.9%) 92 (18.6%) 1196 (25.1%) 65-69 1098 (25.7%) 77 (15.6%) 1175 (24.7%) Income 1Q 458 (10.7%) 75 (15.2%) 533 (11.2%) 2Q 687 (16.1%) 76 (15.4%) 763 (16.0%) 3Q 795 (18.6%) 89 (18.0%) 884 (18.6%) 4Q 1010 (23.7%) 111 (22.5%) 1121 (23.5%) 5Q 1317 (30.9%) 143 (28.9%) 1460 (30.7%) Educational group Lower secondary or lower 219 (5.1%) 28 (5.7%) 247 (5.2%) Upper secondary 1515 (35.5%) 170 (34.4%) 1685 (35.4%) Post-secondary 609 (14.3%) 78 (15.8%) 687 (14.4%) Bachelor or equivalent 1120 (26.2%) 112 (22.7%) 1232 (25.9%) Master or higher a 804 (18.8%) 106 (21.4%) 900 (19.2%) Country of origin Germany 3948 (92.5%) 466 (94.3%) 4414 (92.7%) Outside of Germany a 319 (7.5%) (28) b (5.6%) 347 (7.3%) Citizenship German or other a 4267 (100%) 494 (100%) 4761 (100%) Degree of urbanisation City 1712 (40.1%) 215 (43.5%) 1927 (40.5%) Town or suburb 1803 (42.3%) 210 (42.5%) 2013 (42.3%) Rural area 752 (17.6%) 69 (14.0%) 821 (17.2%) Region Baden-Württemberg 459 (10.8%) (43) b (8.7%) 502 (10.5%) Bavaria 522 (12.2%) 71 (14.4%) 593 (12.5%) Berlin/Brandenburg a 503 (11.8%) 83 (14.7%) 576 (12.1%) Hesse 260 (6.1%) (31) b (6.3%) 291 (6.1%) Lower Saxony/Bremen a 368 (8.8%) (35) b (7.1%) 403 (8.5%) North Rhine-Westphalia/Rhineland-Palatinate a 991 (23.2%) 102 (20.6%) 1093 (22.9%) Saarland 427 (10.0%) 66 (13.4%) 493 (10.4%) Saxony/Saxony-Anhalt/Thuringia a 377 (8.8%) (37) b (7.4%) 414 (8.7%) Schleswig-Holstein/Hamburg/Mecklenburg-Vorpommern a 360 (8.4%) (36) b (7.2%) 397 (8.3%) Quality of social network 1–2 or less a 557 (13.1%) 82 (16.6%) 639 (13.5%) 3-5 2086 (48.9%) 252 (51.0%) 2338 (49.1%) >6 1624 (38.1%) 160 (32.4%) 1784 (37.5%) Perceived social support A lot 930 (21.8%) 116 (23.5%) 1046 (22.0%) Some 2529 (59.3%) 260 (52.6%) 2789 (58.6%) Uncertain 523 (12.3%) 77 (15.6%) 600 (12.6%) Little or none a 275 (6.7%) (41) b (8.3%) 326 (6.8%) Available help Very easy 1414 (33.1%) 165 (33.4%) 1579 (33.2%) Easy 1702 (39.9%) 176 (35.6%) 1878 (39.4%) Possible 726 (17.0%) 93 (18.8%) 819 (17.2%) Difficult 291 (6.8%) (34) b (6.9%) 325 (6.8%) Very difficult 134 (3.1%) (26) b (5.3%) 160 (3.4%) Marital status Single 481 (11.3%) 95 (19.2%) 576 (12.1%) Married 2706 (63.4%) 267 (54.0%) 2973 (62.4%) Widowed 449 (10.5%) (37) b (7.5%) 486 (10.2%) Divorced 631 (14.8%) 95 (19.2%) 726 (15.2%) Type of household Alone 1219 (28.6%) 168 (34.0%) 1387 (29.1%) With children 145 (3.4%) (21) (4.3%) 166 (3.5%) With a partner 2100 (49.2%) 172 (34.8%) 2272 (47.7%) With a partner and children 435 (10.2%) 86 (17.4%) 521 (10.9%) Other 368 (8.6%) (47) b (9.5%) 415 (8.7%) Working situation In paid employment 2531 (59.3%) 332 (67.2%) 2863 (60.1%) Unemployed/Others a 135 (3.1%) (20) b (4.0%) 155 (3.3%) Retired 1258 (29.5%) 90 (18.2%) 1348 (28.3%) Household work (unpaid) 181 (4.2%) (25) b (5.1%) 206 (4.3%) Unable 162 (3.8%) (27) b (5.5%) 189 (4.0%) Partner cohabitation Yes 2767 (64.8%) 280 (56.7%) 3047 (64.0%) No 1500 (35.2%) 214 (43.3%) 1714 (36.0%) Experienced limitation Severely limited 320 (7.5%) (44) (8.9%) 364 (7.6%) Mildly limited 1339 (31.4%) 137 (27.7%) 1476 (31.0%) Not limited 2608 (61.1%) 313 (63.4%) 2921 (61.4%) a Multiple categories were displayed collapsed when cell sizes <20 observations to avoid re-identifiability according to EHIS anonymisation rules. b Cells containing between 20 and 49 observations are individually flagged according to EHIS anonymisation rules.
|
PMC11699213_p30
|
PMC11699213
|
Descriptive statistics of the sample
| 4.022606 |
biomedical
|
Study
|
[
0.9976499676704407,
0.0009239542414434254,
0.0014261096948757768
] |
[
0.9995941519737244,
0.00020176800899207592,
0.0001532530295662582,
0.00005087511453893967
] |
en
| 0.999998 |
When considering social capital, the highest prevalence of never attending BCS was among those with no social network (13.35%), those with little perceived social support (13.36%), and those who find it very difficult to get help from neighbours (16.25%).
|
PMC11699213_p31
|
PMC11699213
|
Descriptive statistics of the sample
| 1.847996 |
biomedical
|
Study
|
[
0.8445001840591431,
0.0037311671767383814,
0.15176858007907867
] |
[
0.9617258906364441,
0.03727707639336586,
0.0005963981384411454,
0.0004006493545603007
] |
en
| 0.999998 |
Single women (16.49%) showed the highest rates of never attending BCS among all marital statuses. Women living with a partner and children (16.51%), women unable to work (13.76%), unemployed women (13.68%), and women not cohabiting with a partner (12.49%) displayed the highest prevalences. Lastly, severely limited women (12.09%) had the lowest attendance rates among their PROGRESS-Plus dimension.
|
PMC11699213_p32
|
PMC11699213
|
Descriptive statistics of the sample
| 1.916037 |
biomedical
|
Study
|
[
0.8346059918403625,
0.0027878584805876017,
0.1626061648130417
] |
[
0.9833764433860779,
0.016077745705842972,
0.0003311636100988835,
0.000214670566492714
] |
en
| 0.999998 |
Based on a recent scoping review , four PROGRESS-Plus variables were relevant for predicting lifetime BCS attendance: migration background, income, urbanisation degree and partnership cohabitation .
|
PMC11699213_p33
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 1.860916 |
biomedical
|
Review
|
[
0.7655478715896606,
0.004816614091396332,
0.22963544726371765
] |
[
0.2238718718290329,
0.07781852781772614,
0.6961659789085388,
0.0021436724346131086
] |
en
| 0.999995 |
Univariate logistic regression analyses separately estimated the effects of these four variables ( Table 3 ). Only cohabitation significantly predicted BCS attendance, with women living alone having higher odds of never attending. Age also had a significant relationship with BCS attendance. Table 3 Univariate logistic regression on never attending BCS in Germany. Table 3 Sociodemographic variables OR 95% CI R 2 model AUC model Income High 1 Low 1.20 (0.98–1.47) 0.0010 0.5187 Country of origin Germany 1 Not Germany 0.74 (0.50–1.11) 0.0007 0.5090 Degree of urbanisation Urban 1 Rural 0.87 (0.72–1.05) 0.0007 0.5170 Partner cohabitation Yes 1 No 1.41∗∗∗ (1.17–1.70) 0.0039 0.5408 Age 50–54 1 55–59 0.44∗∗∗ (0.35–0.56) 60–64 0.37∗∗∗ (0.28–0.47) 65–69 0.31∗∗∗ (0.23–0.40) 0.0317 0.6225
|
PMC11699213_p34
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 4.072528 |
biomedical
|
Study
|
[
0.99875807762146,
0.0005880504031665623,
0.0006538303568959236
] |
[
0.9994789958000183,
0.00020839153148699552,
0.0002540854038670659,
0.00005854920163983479
] |
en
| 0.999995 |
Multivariate logistic regression was performed to capture the effects of each predictor when adjusting for covariates and age ( Table 4 ). Here, the only relationship that showed a statistically significant relationship with BCS attendance was partner cohabitation, with 1.45 higher odds (p < 0.001) for women not cohabitating with their partners. Table 4 Multivariate logistic regression on never attending BCS in Germany (main effects model). Table 4 Sociodemographic variables OR 95% CI Income High 1 Low 1.21 (0.98–1.49) Country of origin Germany 1 Not Germany 0.68 (0.46–1.02) Degree of urbanisation Urban 1 Rural 0.91 (0.75–1.11) Partner cohabitation Yes 1 No 1.45∗∗∗∗ (1.19–1.76) Age 50–54 1 55–59 0.43 ∗∗∗ (0.34–0.56) 60–64 0.36 ∗∗∗ (0.28–0.47) 65–69 0.29 ∗∗∗ (0.22–0.38) R 2 0.0394 AUC-ROC 0.6539 A complete case analysis only based on the variables would have resulted in 300 more participants, but the results do not change meaningfully – see Appendix C .
|
PMC11699213_p35
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 4.107833 |
biomedical
|
Study
|
[
0.9989368319511414,
0.0004893385921604931,
0.0005738186882808805
] |
[
0.999546229839325,
0.00018014578381553292,
0.00022041340707801282,
0.00005320612035575323
] |
en
| 0.999997 |
Sixteen intersectional groups were created based on the combination of the four variables identified in the literature. Fig. 1 depicts the size and prevalence of each group. Fig. 1 Prevalence and size across the sixteen evidence-informed intersectional groups. a Cells containing between 20 and 49 observations are individually flagged according to EHIS anonymisation rules. Fig. 1
|
PMC11699213_p36
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 1.979732 |
biomedical
|
Study
|
[
0.9674137234687805,
0.001758516184054315,
0.03082774206995964
] |
[
0.973055899143219,
0.025725645944476128,
0.0007089631399139762,
0.0005094856023788452
] |
en
| 0.999998 |
Following this, an unweighted logistic regression was performed . As a reference group, we chose the one expected to have the highest attendance rate - based on the multivariate regression and Pedrós Barnils et al. - high-income women born outside Germany, living in urban areas with a partner (HOUY). Table 5 Full cross-classified multivariate logistic regression with evidence-informed intersectional groups. Table 5 OR 95% CI Intersectional groups HOUY 1 HGUY 2.49 (0.88–7.04) HGUN 2.97∗ (1.04–8.45) HGRY 1.92 (0.69–5.40) HGRN 2.84 (0.99–8.14) LGUY 1.96 (0.62–6.18) LGUN 3.71∗ (1.27–10.89) LGRY 2.81 (0.98–8.08) LGRN 3.24∗ (1.11–9.47) HOUN 3.11 (0.85–11.39) HORY 0.74 (0.16–3.46) HORN 3.15 (0.64–15.48) LOUY 0.66 (0.07–6.26) LOUN 4.00 (0.91–17.49) LORY 0.52 (0.05–4.82) LORN 9.48∗∗ (2.24–40.10) Age 50–54 1 55–59 0.43∗∗∗ (0.33–0.54) 60–64 0.35∗∗∗ (0.27–0.45) 65–69 0.29∗∗∗ (0.22–0.38) R 2 0.0445 AUC-ROC 0.6618 ∗p-value <0.05; ∗∗p-value <0.01; ∗∗∗ p-value <0.001. HGUY - high-income, born in Germany, urban, cohabitation. HGUN - high-income, born in Germany, urban, no cohabitation. HGRY - high-income, born in Germany, rural, cohabitation. HGRN - high-income, born in Germany, rural, no cohabitation. LGUY - low-income, born in Germany, urban, with cohabitation. LGUN - low-income, born in Germany, urban, no cohabitation. LGRY - low-income, born in Germany, rural, cohabitation. LGRN - low-income, born in Germany, rural, no cohabitation. HOUY - high-income, born outside Germany, urban, cohabitation. HOUN - high-income, born outside Germany, urban, no cohabitation. HORY - high-income, born outside Germany, rural, cohabitation. HORN - high-income, born outside Germany, rural, no cohabitation. LOUY - low-income, born outside Germany, urban, cohabitation. LOUN - low-income, born outside Germany, urban, no cohabitation. LORY - low-income, born outside Germany, rural, cohabitation. LORN - low-income, born outside Germany, rural, no cohabitation. Fig. 2 Odds Ratio (OR) with evidence-informed intersectional groups on never attending BCS in Germany. Fig. 2
|
PMC11699213_p37
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 4.097666 |
biomedical
|
Study
|
[
0.9931782484054565,
0.0003724902926478535,
0.006449262145906687
] |
[
0.9994305968284607,
0.000340945553034544,
0.00019566668197512627,
0.00003277223731856793
] |
en
| 0.999998 |
Four intersectional groups were significantly associated with never attending BCS. Low income women not born in Germany and living in rural areas with no partner (LORN) showed the highest odds (OR = 9.48, p = 0.002). The confidence intervals for all these estimations were rather wide, increasing the uncertainty of the predicted estimations. The DA of the full cross-classification model was moderated and 0.0079 points higher than the main effects model. That indicates that the regression with intersectional groups discriminates slightly better between women attending or never attending BCS than the main effects model.
|
PMC11699213_p38
|
PMC11699213
|
Analytical strategy a: evidence-informed regression
| 3.160758 |
biomedical
|
Study
|
[
0.9887905120849609,
0.0009302208200097084,
0.01027924008667469
] |
[
0.999163031578064,
0.0005487273447215557,
0.0002273454301757738,
0.00006089474481996149
] |
en
| 0.999997 |
Out of the three algorithms, CART showed the highest sensitivity and balanced accuracy performance. For more information on the hypertuned models, see Appendix D . The inner performance (i.e. evaluated on trained data) of CART was: 72.47% sensitivity, 51.35% specificity, 61.91% balanced accuracy, 14.71% positive predictive value and 94.15% negative predictive value. The moderate sensitivity suggests reasonable confidence in CART detecting women not attending BCS. However, the low specificity suggests small confidence in CART to identify negative cases (i.e. women attending BCS). The small positive predictive value indicates that many cases classified as positive (i.e. not attending BCS) are false positives. Nevertheless, the high negative predictive value indicates very few false negatives and, therefore, very high confidence that those cases classified as negative are negative (i.e. not assuming that a woman is attending BCS when she is not). Fig. 3 and Table 6 show the final decision tree and the emerged intersectional groups. Fig. 3 CART decision tree on never attending BCS in Germany. Fig. 3 Table 6 Intersectional groups on never attending BCS in Germany based on CART. Table 6 Group Intersectional groups Rank a Size, Prevalence H Women living with a partner, retired or doing unpaid household work 1 N = 882 Pr = 0.0454 E Widowed women living alone, with children, with a partner and children or other arrangements, residing in Baden-Württemberg, Berlin, Hesse, Mecklenburg-Vorpommern, Lower Saxony, North Rhine-Westphalia, Rhineland-Palatinate, Saxony, Saxony-Anhalt, and Schleswig-Holstein or Thuringia 2 N = 316 Pr = 0.0506 C Single, married or divorced women living in other living arrangements, with some or no perceived social support 3 N = 211 Pr = 0.0616 G Women living with a partner, who are either employed, unemployed, unable to work, or in other categories, and residing in Baden-Württemberg, Brandenburg, Hesse, Mecklenburg-Vorpommern, Lower Saxony, North Rhine-Westphalia, Rhineland-Palatinate, Saxony, Saxony-Anhalt, or Schleswig-Holstein 4 N = 918 Pr = 0.0730 B Single, married or divorced women living alone, with children, with a partner and children, with some or no perceived social support 5 N = 953 Pr = 0.1301 F Women living with a partner who are either employed, unemployed, unable to work, or in other working categories and residing in Bavaria, Berlin, Bremen, Hamburg, Saarland or Thuringia 6 N = 472 Pr = 0.1377 D Widowed women living alone, with children, with a partner and children or other arrangements, residing in Bavaria, Brandenburg, Bremen, Hamburg, or Saarland 7 N = 136 Pr = 0.1471 A Single, married or divorced women living alone, with children, with a partner and children or other arrangements, with little, uncertain or a lot of perceived social support 8 N = 873 Pr = 0.1707
|
PMC11699213_p39
|
PMC11699213
|
Analytical strategy b: decision tree-based regression
| 4.125537 |
biomedical
|
Study
|
[
0.9971966743469238,
0.00045815607882104814,
0.002345192711800337
] |
[
0.9995241165161133,
0.00021487977937795222,
0.0002258570893900469,
0.000035153308999724686
] |
en
| 0.999996 |
CART identified household type, marital status, working situation, region and perceived social support as relevant variables. The first splitting point, the root node, is the household type, where women living with a partner are split from all other household types. Women living with a partner are further split into working situations. Here, women retired or doing unpaid household work form a final node , and women employed, unemployed, unable to work, or others further split based on their region .
|
PMC11699213_p40
|
PMC11699213
|
Analytical strategy b: decision tree-based regression
| 1.643796 |
other
|
Study
|
[
0.12354346364736557,
0.0006838637054897845,
0.8757727146148682
] |
[
0.6818530559539795,
0.31597834825515747,
0.001298646442592144,
0.0008699563331902027
] |
en
| 0.999997 |
Women living alone, with children, with a partner and children or in other arrangements are further split based on marital status. Here, widowed women are separated from single, married or divorced women. Widowed women were lastly split based on their region . On the other hand, single, married or divorced women further split based on their perceived social support. Those with some or no perceived social support are separated from those with little, uncertain or a lot of perceived social support, who form a final node . The first group split one last time based again on their type of household: living alone, with children, or with a partner and children , and other arrangements .
|
PMC11699213_p41
|
PMC11699213
|
Analytical strategy b: decision tree-based regression
| 1.173654 |
other
|
Other
|
[
0.012052912265062332,
0.0005274161230772734,
0.9874196648597717
] |
[
0.022759027779102325,
0.9760245680809021,
0.0005937468959018588,
0.0006226907134987414
] |
en
| 0.999996 |
Subsets and Splits
Clinical Cases Sample
Retrieves 100 samples of clinical cases, providing a basic overview of this specific document type.
High-Score Clinical Cases
The query retrieves a limited set of clinical case documents with a high educational score, providing a basic filtered view of the dataset.