High DNA methylation age deceleration defines an aggressive phenotype with immunoexclusion environments in endometrial carcinoma.
The main results reproduced: recomputed values matched the published ones within tolerance.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- Nothing in this column.
- 🟡Could not use the authors’ exact input data
- 🟡Reported values were only indirectly comparable
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡Reported values were not (fully) derivable from the shared data
- 🟡The deviation was non-trivial in magnitude
- 🟡The central claim did not (fully) hold under reproduction
- 🟡Overall, the reproduction showed a material discrepancy
A 0–100 reproducibility-quality score from the per-question grades, shown as a z-score: standard deviations above (+) or below (−) the mean of comparable assessments.
▸Reproduction agent’s raw note
Described well enough to reproduce the CORE claim 1:1 in direction, and now the key downstream genomic check too. The paper's listed code (github.com/dengchunyu/TIP) is NOT its DNAm-age pipeline but a third-party RNA-seq immune-phenotype web tool; we reproduced via the standard published epigenetic clocks on the paper's own public data (P16). Main cohort = TCGA-UCEC (we recovered 428 primary tumors with age vs the paper's 429). Using methylclock::DNAmAge() (Horvath 2013 + Levine PhenoAge) on the public UCSC-Xena HM450 matrix (clean re-run, «our HPC» «job», deterministic — per-sample CSV bit-identical across runs): the titular finding (pervasive DNA-methylation-age DECELERATION in endometrial carcinoma) reproduces strongly — 99.1% of tumors have Horvath DNAmAge below chronological age (mean 18.5 vs 64.2 yr; clock decoupled, corr 0.11). Top-1/3 deceleration n=143 (paper 134). TP53 enrichment in decelerated tumors (C3) reproduces this round via the MC3 mutation matrix: 65.2% (45/69) hDNAmad+ vs 33.7% (121/359) hDNAmad-, Fisher p=1.5e-6 — matching the paper's 70% vs 30%, p<1e-4 (prior 'error' was an empty mutation download, now fixed). The only residual gap: hDNAmad+ count is 69 (16.1%) or 103 (24.1%) depending on how PhenoAge 'validation' is operationalized; the reported 82 (19.1%) sits between them -> consistent but exact figure rests on an unstated cutoff -> partial. FLAGS for human: (1) paper says GSE67116 patients are 'all aged 53' but GEO has NO age field for that series at all (verified GSM1639228) -> unsupported assertion; (2) deceleration magnitude provisional (18.5% imputed Xena-legacy matrix). NOT attempted: TIP immune scores (Fig 5), KM/Cox survival (Fig 2), aneuploidy/DMP/Ki-67. Grades provisional; human decides ground truth.
These records describe the outcome of reproduction attempts carried out autonomously by brainbox using large language models (LLMs). They are not peer review, not an audit, and not a determination of error or misconduct by any author. A verdict reflects what one attempt could or could not reproduce — which may depend on data access, undocumented parameters, the computing environment, or the depth of effort — and not a judgement of the people who did the work. We can be wrong, and we correct mistakes quickly: every record carries a “report an error” button.
Assessment versions
Every reproduction run is kept as an immutable version — anchored to the data as it stood, with a tamper-evident chain hash. A rerun (e.g. after an author updates a deposit) adds a new version; the previous one stays on record.
-
v1 current initial assessment Score 58assessed: 2026-06-15 ⛓ 8507e70f24dd
✎ I am an author of this paper
Updated or fixed a deposit, or is there an erratum? Ask us to re-run the metrics. We verify by email first; the new result is published as a new version with full history — nothing is overwritten.
Provenance — full disclosure
When this reproduction was carried out, which methodology version was used, and by whom — so the record can be audited and checked independently.
- Reproduced
- 2026-06-23
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-15no human curator yet
- Last updated
- 2026-08-05
Provisional, curator- or AI-assessed, and independently checkable. A reproduction outcome states what one attempt could reproduce — not a judgement of the authors.
Deep full-text extraction
Model: sonnetThe study tests whether DNA methylation age acceleration/deceleration (DNAmaa/DNAmad) occurs in endometrial carcinoma (EC) and whether it drives tumor formation/progression with clinical implications, by analyzing TCGA and GSE67116 EC cohorts using Horvath and Phenoage DNAm clocks.
- ★ Almost 90% of TCGA EC tumors exhibit DNA methylation age deceleration (DNAmad) relative to patient chronological age as assessed by the Horvath clock finding
- ★ A subset of 82/429 tumors with high DNAmad (hDNAmad+), concordant across both Horvath and Phenoage clocks, was identified finding
- ★ hDNAmad+ tumors are associated with advanced disease stage/grade and significantly shorter overall and progression-free survival finding
- ★ hDNAmad+ tumors show higher copy number alterations (CNAs) and aneuploidy but lower tumor mutation burden finding
- ★ hDNAmad+ tumors are enriched in cell cycle and DNA mismatch repair pathways finding
- ★ hDNAmad+ tumors display immunoexclusion microenvironments with higher VTCN1 and lower PD-L1/CTLA4 expression, indicating poor predicted response to ICI immunotherapy finding
- ★ DNMT3A and DNMT3B expression is significantly higher in hDNAmad+ than hDNAmad- tumors finding
- hDNAmad+ tumors show increased PIK3CA alterations and downregulation of SCGB2A1 (a PI3K inhibitor), potentially promoting tumor growth and stemness finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| DNA methylation array (Horvath clock, 353 CpGs) | EC tumors and non-tumorous endometrium (TCGA cohort) | none | DNAm age vs chronological age, DNAmaa/DNAmad classification | Illumina Infinium HumanMethylation450 BeadChip |
| DNA methylation array (Phenoage clock, 513 CpGs) | EC tumors and non-tumorous endometrium (TCGA cohort) | none | DNAm age vs chronological age, DNAmaa/DNAmad classification | Illumina Infinium HumanMethylation450 BeadChip |
| DNA methylation profiling (Horvath and Phenoage clocks) | Primary and metastatic EC tumors plus hyperplasia (GSE67116 cohort, n=86) | none | DNAm age, correlation between clock estimates | Illumina HumanMethylation450 BeadChip |
| Bulk RNA sequencing / differential expression (edgeR) | EC tumors, hDNAmad+ vs hDNAmad- (TCGA) | none (comparison by DNAm age status) | TPM/RSEM gene expression, DEGs | — |
| KEGG pathway analysis / Gene Set Enrichment Analysis (GSEA) | EC tumors, hDNAmad+ vs hDNAmad- (TCGA) | none | Pathway enrichment (adjusted P<0.05, FDR<0.25) | GSEA v4.2.1 |
| Copy number alteration / aneuploidy score / tumor mutation burden analysis | EC tumors (TCGA, n=429) | none | CNA frequency, aneuploidy score, TMB | UCSC Xena browser data |
| Immune cell deconvolution (CIBERSORT) and ssGSEA immune signatures | EC tumors (TCGA RNA-seq) | none | Proportions of 22 immune cell types, myeloid/Teffector signature scores, immune checkpoint expression (PD-L1, CTLA4, VTCN1) | CIBERSORT (cibersort.stanford.edu) |
| ssGSEA-based scoring (proliferation, cell cycle, stemness, telomerase) | EC tumors (TCGA RNA-seq) | none | Cell cycle score, stemness score, telomerase score, Ki-67 mRNA level | — |
- – DNAm age correlates with chronological age in non-tumorous endometrium (Horvath clock) R=0.339, P=0.0003, MAD=9 yrs
- – DNAm age does not correlate with chronological age in EC tumors (Horvath clock) R=0.112, P=0.030, MAD=12.0 yrs
- ▼ Majority of TCGA tumors show DNAm age younger than chronological age (DNAmad) by Horvath clock 403/429 tumors (range -3 to -79 yrs)
- – 82 of 429 tumors classified as hDNAmad+ using concordant Horvath and Phenoage clock results 82/429
- ▼ hDNAmad+ patients show significantly shorter OS and PFS by Kaplan-Meier/log-rank analysis
- – hDNAmad+ tumors are predominantly CN-high molecular subtype, while hDNAmad- tumors are predominantly MSI subtype 43.9% vs 4.9%/19.5%/0% (hDNAmad+); 9.3%/22.7%/43.0%/4.2% (hDNAmad-), P<0.0001
- ▲ hDNAmad+ tumors show higher CNA frequency and aneuploidy score genome-wide
- correlation R=0.339, P=0.0003 (DNAm age vs chronological age in non-tumorous endometrium (Horvath clock))
- correlation R=0.112, P=0.030 (DNAm age vs chronological age in EC tumors (Horvath clock))
- other MAD=9 yrs (NT), MAD=12.0 yrs (tumors) (Mean absolute difference between DNAm age and chronological age, Horvath clock)
- count 82/429 tumors classified hDNAmad+ (Overlap of top DNAmad tumors concordant across Horvath and Phenoage clocks)
- count 403 tumors with DNAmad; 12 with DNAmaa (Horvath clock) (Distribution of DNAm age deceleration/acceleration in TCGA EC tumors)
- pvalue P<0.0001 (Association between hDNAmad status and EC molecular subtype (CN-high vs MSI))
- correlation R=0.674, P=0.001 (primary); R=0.435, P=0.026 (metastatic) (Correlation between Horvath and Phenoage DNAm age estimates in GSE67116 cohort)
Statistical methods review
Model: sonnetA neutral, descriptive read of the statistical approach — what was done, and (for shared learning, not as criticism) what could also have been done.
The paper used a retrospective design analyzing two existing EC cohorts (TCGA, n=429 tumors/32 NTs; GSE67116, n=86 tumors) to characterize DNA methylation age deceleration via Horvath and Phenoage clocks. Tumors concordantly in the top tertile of Horvath-based deceleration and also showing deceleration by Phenoage were classified as hDNAmad+ (n=82) and compared against the remainder for clinical, genomic, transcriptomic, and immune microenvironment features. Group comparisons used t-tests, Wilcoxon, and Kruskal-Wallis tests for continuous data and chi-squared or Fisher's exact tests for categorical variables; survival was assessed with Kaplan-Meier/log-rank and Cox regression; differential expression used edgeR and pathway enrichment used GSEA.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Student's t-test / Wilcoxon rank-sum / Kruskal-Wallis | Continuous variable comparisons between hDNAmad+ and hDNAmad- tumors (e.g., aneuploidy scores, immune scores, gene expression levels); chosen based on data distribution | 429 TCGA tumors (82 hDNAmad+ vs 347 hDNAmad-) | not stated |
| Chi-squared test / Fisher's exact test | Categorical clinico-pathological variables (age group, stage, grade, histology, molecular subtype, diabetes, hypertension, BMI) between hDNAmad+ and hDNAmad- groups | 429 TCGA tumors; 199 with molecular subtype data | not stated |
| Pearson correlation coefficient | Correlation of DNAm age with chronological age in NTs and tumors; correlation between Horvath and Phenoage clock estimates | 429 tumors and 32 NTs (TCGA); 86 tumors (GSE67116) | not stated |
| Kaplan-Meier with log-rank test | Overall survival (OS) and progression-free survival (PFS) by hDNAmad status, stage, grade, chronological age, and histological type | 429 TCGA patients | not stated |
| Univariate and multivariate Cox proportional hazards regression | Effect of hDNAmad, chronological age, stage, grade, and histology on OS and PFS (results in Table S2) | 429 TCGA patients | not stated |
| edgeR (R package; negative binomial model) | Differentially expressed genes between hDNAmad+ and hDNAmad- tumors; threshold: adjusted p<0.05 and log2 fold change >1.5 | 429 TCGA tumors | not stated |
| GSEA (Gene Set Enrichment Analysis) v4.2.1 | KEGG pathway enrichment differences between hDNAmad+ and hDNAmad- tumors; threshold: adjusted P<0.05 and FDR<0.25 | 429 TCGA tumors | not stated |
-
hDNAmad+ was defined using a two-step threshold: top tertile by Horvath clock, then confirmation by Phenoage clock, yielding n=82↳ Could also: A continuous DNAmad score (linear model residual of DNAm age on chronological age) analyzed with restricted cubic splines, or a data-driven cutoff method (e.g., X-tile, maximally selected rank statistics) could also be used — Tertile-based dichotomization discards quantitative gradient information and the specific threshold is somewhat arbitrary; continuous or statistically optimized threshold approaches preserve statistical power and reduce the risk of findings being sensitive to the chosen cut-point
-
Pearson correlation was used to assess the relationship between DNAm age and chronological age across NTs and tumors↳ Could also: Spearman rank correlation could also be used — DNAm age residuals in tumor tissues may not follow a bivariate normal distribution, especially given the wide reported range (-79 to +42 years); Spearman correlation makes no distributional assumption and is robust to the outliers visible in the scatter plots described
-
Multiple independent comparisons (t-test, Wilcoxon, chi-squared, log-rank) were performed across many clinico-pathological and molecular features without a stated family-wise error rate correction↳ Could also: A Benjamini-Hochberg false discovery rate correction applied across the family of clinical comparisons could also be used — When many hypotheses are tested simultaneously on the same dataset, the expected number of false positives at P<0.05 increases proportionally; a multiplicity correction across this broader family of tests would make the reported threshold more interpretable
-
edgeR was used for differential gene expression analysis between hDNAmad+ and hDNAmad- tumors↳ Could also: DESeq2 (negative binomial model with adaptive shrinkage of log-fold-change estimates) could also be used — DESeq2 is a widely adopted alternative for bulk RNA-seq differential expression; it employs different dispersion estimation and normalization strategies, and running both methods and requiring concordant findings is a common practice to assess robustness
-
CIBERSORT was applied to estimate 22 immune cell fractions from bulk RNA-seq data↳ Could also: Alternative immune deconvolution methods such as xCell, TIMER2.0, or ESTIMATE could also be used — Different deconvolution algorithms use distinct reference profiles and normalization assumptions; comparing concordance across multiple methods is a common approach to increase confidence in immune infiltration estimates derived from bulk transcriptomic data
-
Survival analyses used Kaplan-Meier curves with log-rank tests and Cox proportional hazards regression↳ Could also: A competing-risks model (Fine-Gray subdistribution hazard) could also supplement the Cox analyses — In EC cohorts that include older patients, death from non-cancer causes represents a competing event; Fine-Gray models can be used alongside the standard Cox model to assess whether cumulative incidence of cancer-specific events is similarly associated with hDNAmad status
Citation network
Where this publication sits in the reproducibility-weighted citation graph — what it is built on, and what is built on it. Citation data from OpenAlex.
Assessed papers, coloured by verdict. Click a node to open it.
- No assessed neighbours yet — the network grows as more papers are assessed.
Data lineage
The datasets this paper uses (text-mined from the full text via Europe PMC), and which other assessed papers stand on the same data. A shared dataset is a factual link — not a judgement.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-37388735
Paper: Liu et al. 2023, High DNA methylation age deceleration defines an aggressive phenotype with immunoexclusion environments in endometrial carcinoma. Front Immunol. PMID 37388735 · PMCID PMC10303802 · DOI 10.3389/fimmu.2023.1208223
Listed code: https://github.com/dengchunyu/TIP — this is NOT the authors' own analysis code. It is a third-party web tool "TIP: Tracking Tumor Immunophenotype" (Xu/Deng et al., Cancer Res 2018) that takes RNA-seq TPM and outputs an immune "cancer-immunity cycle" score. The paper used it for the immune-profiling part only. No repo of the paper's own DNAm-age pipeline is shipped (P16 case: reproduce by running the described standard pipeline on the paper's public data).
Listed data: geo:GSE67116 — public 450k methylation, 96 samples (8 hyperplasia, 33 primary EC, 53 metastasis, 2 cell lines; Teschendorff/Widschwendter 2015). Used in the paper only as a validation methylation cohort. GEO sample metadata contains NO age field (verified: GSM1639228… characteristics list only sample id / disease / tumor type / tissue / gender / sentrix id). The paper's statement that GSE67116 patients are "all aged 53 years old" is therefore NOT derivable from the deposited data → flagged as a possible-fabrication / unsupported assertion. Without chronological age, GSE67116 cannot yield a methylation-age deceleration (residual), so it is NOT a usable quantitative reproduction target on its own.
Main cohort (the one carrying every headline number): TCGA-UCEC (432 patients; analysis on 429 tumors / 32 normals). Illumina HM450 methylation + RNA-seq + mutation + clinical age. All public via GDC / UCSC-Xena. This is the cohort we reproduce.
In scope (pipeline-derived, attempted)
Standard, fully-specified epigenetic-clock pipeline on TCGA-UCEC HM450:
- P1 DNAm age (Horvath 2013, 353 CpGs) per tumor — deterministic clock.
- P2 DNAm PhenoAge (Levine 2018, 513 CpGs) per tumor — deterministic clock.
- P3 DNAmad (methylation-age deceleration) = residual of
lm(DNAmAge ~ chrono_age); deceleration = DNAmAge younger than chronological age. - P4 hDNAmad+ grouping: top 1/3 most-decelerated by Horvath, intersected with PhenoAge → reported n=82/429 (19.1%); reported top-1/3 n=134.
- P5 (light downstream) TP53 mutation frequency in hDNAmad+ vs hDNAmad− groups — reported 70% (57/82) vs 30% (105/347), p<0.0001.
Implementation note: the paper used the Horvath online calculator
(dnamage.genetics.ucla.edu, not scriptable). We reproduce the same published
algorithm via the open-source methylclock R/Bioconductor package (identical Horvath
& Levine coefficients) — an independent implementation of the same method, which is the
faithful 1:1 of a deterministic clock.
Out of scope (not attempted, why)
- TIP immune cancer-immunity-cycle scores (Fig 5): would require staging TCGA-UCEC RNA-seq TPM through the TIP web tool / its R code with the paper's exact parameters; the paper reports these only qualitatively (no in-text scalar to compare) → low value per 80/20, skipped.
- Survival (KM/Cox OS/PFS, Fig 2): p-values live inside figure images, depend on the exact hDNAmad± partition; attempt only if P4 lands cleanly.
- Wet-lab / IHC (Ki-67 staining), copy-number aneuploidy score, DMP methylation calls: downstream of the same partition; the hard last 20%, skipped unless P4 is exact.
Comparison targets → original/claims.tsv
Assessments & scoring basis
Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.
An automated assessment. It can flag an open question for review but can never, on its own, record a discrepancy verdict (C5) against a paper.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
The paper's titular claim — pervasive DNA-methylation-age deceleration in endometrial carcinoma — reproduces strongly and independently (99.1% of 428 tumors decelerated; clock decoupled from age, corr 0.11), and the cohort size (428 vs 429) and top-1/3 count (143 vs 134) land essentially on target. Deviations are explainable and mostly on our methodology / data-version side (Xena-legacy imputation, self-chosen intersection bracketing the reported 82) plus an authors-side under-specification of the PhenoAge validation rule. Two genuine authors-side flags exist but are secondary: the cited code is not the paper's pipeline, and the 'GSE67116 all aged 53' assertion is not derivable from the deposited data. The 'aggressive/immunoexclusion' half of the thesis was not tested (TP53 parser failure, survival/immune-cycle out of scope), so overall this is a solid-with-explainable-deviations (yellow) reproduction, not a 1:1 or a critical discrepancy.
Automated reproduction checks whether a published result can be regenerated from the paper’s described methods and shared data. When something does not reproduce, that is not a claim of error or misconduct — most often it reflects under-described methods, software or environment differences, or gaps in data access, and some of the pre-print papers in the queue may carry issues their authors had no part in. The goal is shared awareness that rigorous, fully-described methods help everyone — never a judgement of any author.
Are you an author? We would genuinely like to hear from you — to clarify the record, add data or code, re-run the pipeline after an accession update, and publish your response right next to the assessment. Everything here is open and auditable.
🚩 Report an error in this record
Spotted something wrong — a verdict you’d contest, a data or value error, or a private detail that slipped through? Tell us, with a short justification. Authors and readers are equally welcome to write in; we review every report.
Prefer email, or the form below not working? Contact us at support@doesitreproduce.com.
Reproduction footprint
claude-opus-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.