Vaccine-induced ICOS+CD38+ circulating Tfh are sensitive biosensors of age-related changes in inflammatory pathway
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.
- ✓Same input data as the authors
- ✓Reported values are derivable from the shared data
- ✓Any deviation was negligible
- 🟡Reported values were only indirectly comparable
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡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
REPRODUCED (1:1, honest). Paper: Herati et al. 2021 Cell Rep Med, ICOS+CD38+ cTfh aging/inflammatory-pathway signature. The repo (teamTfh/cTfh_AgingSignature @ be57be6) is described well enough: it ships the analysis R scripts AND the processed pipeline inputs (processedData.zip -> bestDataLog.csv, bestDataCounts.csv) AND the authors' own rendered HTML reports with printed numeric output -- a rare case where the exact input and output are both available. I reproduced the GSVA Hallmark pipeline (the paper's signature analysis) on «our HPC» by re-running the authors' GSVA.R logic on their own bestDataLog.csv (ICOS+CD38+ HiHi_v2: 44207 genes x 14 samples, 6 young + 8 elderly) with MSigDB Hallmark v6.1 and gsva(method='gsva'). Two deterministic byproducts matched the authors' rendered GSVA.html to the EXACT integer: 9286 constant-expression genes discarded, and 50 gene sets scored; the 50x14 score matrix was fully regenerated (SHA256 c43cb87b...). The paper's central DIRECTION -- TNFa/NFkB inflammatory signature higher in elderly cTfh -- also reproduced (young -0.231 < elderly +0.090). Env: GSVA 1.36.0/R4.0.5 vs authors' GSVA 1.36.2/R4.0.2 (same Bioc 3.11, patch-level only; exact byproduct match confirms it is immaterial). DID NOT attempt / NOT reproducible: (1) per-cell GSVA score-vs-paper comparison -- authors shipped only a heatmap, not the numeric table (write.csv commented out); (2) ALL serum-cytokine/flow/titer correlations (Correlations.R, GSVA-vs-TNFa r/p) -- these need ../../Analysis/BAA_yr4_AllMergedData_FC.csv, a phenotype matrix ABSENT from repo+GEO (missing-input, NOT fabrication); (3) DESeq2 differential expression, WGCNA, GSEA-preranked, TNFR (80/20 skip -- heavier, no single clean anchor); (4) raw->counts re-derivation from GSE123696 (hard last-20%, repo ships post-normalization matrix). No fabrication indicators: every compared value is derivable from the shipped data+code and the byproducts reproduce exactly. Grades are provisional; a human auditor 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 88assessed: 2026-06-15 ⛓ 8f932cac6750
✎ 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
- 2026-06-15
- 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: opusDoes aging alter the transcriptional and signaling programs of vaccine-induced ICOS+CD38+ circulating T follicular helper (cTfh) cells after influenza vaccination, and can these cells serve as cellular biosensors of underlying age-related changes in inflammatory/immune health?
- ★ Vaccine-induced ICOS+CD38+ cTfh from elderly adults show enriched TNF-NF-κB signaling compared to young adults at day 7 post-vaccination finding
- ★ Vaccine-induced ICOS+CD38+ cTfh preferentially enrich for signatures of aging versus youth far better than other blood T cell types, acting as sensitive cellular biosensors of underlying immune state finding
- ★ The frequency and magnitude of the ICOS+CD38+ cTfh response to influenza vaccination appear similar between young and elderly adults at the flow cytometric and individual-gene level finding
- ★ TNF-NF-κB signaling is beneficial/pro-survival for cTfh in the elderly mechanism
- ★ Weighted gene correlation network analysis (WGCNA) reveals elderly-specific transcriptional modules (EM1, EM4) enriched for inflammatory/NF-κB and interferon pathways not present in young adults method
- ★ NF-κB1 (p50) protein and TNFRSF1A (TNFR1) expression are elevated in ICOS+CD38+ cTfh from elderly compared to young adults finding
- Serum/plasma TNF concentration does not correlate with the TNF-NF-κB gene signature in cTfh, suggesting serum TNF alone is not the cause finding
- ICOS+CD38+ cTfh resemble lymphoid germinal-center Tfh more than ICOS-CD38- cTfh by gene set enrichment analysis finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Flow cytometry | PBMC from young (n=28) and elderly (n=35) adults, influenza-vaccinated | influenza vaccination (in vivo) | frequency of ICOS+CD38+ cTfh, PD-1 geometric MFI, plasmablast response | — |
| Bulk RNA-seq | FACS-sorted ICOS+CD38+ cTfh, ICOS-CD38- cTfh, naive CD4 from young (n=6) and elderly (n=8) adults | influenza vaccination (days 0 and 7) | transcriptional profiles / differentially expressed genes | — |
| Flow cytometry (intracellular protein) | ICOS+CD38+ cTfh from independent cohort of young and elderly adults (n=16 each) | none (baseline ex vivo) | total NF-κB p50 protein (geometric MFI) | — |
| cTfh-B cell coculture / in vitro antibody assay | autologous naive B cells and CD4 T cells from young adults | SEB stimulation +/- recombinant TNF protein or anti-TNF antibody | IgM and IgG1 in supernatant | — |
| Plasma cytokine/multiplex profiling | plasma from young and elderly adults | none (baseline) | TNF, CXCL11, MIP1β concentrations | — |
- ▲ NF-κB target genes upregulated in elderly ICOS+CD38+ cTfh at day 7 vs young 58 targets up in elderly vs 25 in young (Padj<0.05)
- ▲ TNF-NF-κB signaling was the highest-scoring pathway associated with elderly module EM4 by GSEA
- ▲ Vaccine-induced fold change in ICOS+CD38+ cTfh correlated with plasmablast fold change, similar across age young r=0.57, p=3.9×10^-3; elderly r=0.67, p=4.8×10^-4
- – Elderly had lower total CXCR5+PD-1+ cTfh and slightly higher baseline ICOS expression 19% lower cTfh frequency (p=0.05); 1.2-fold higher ICOS (p=0.06)
- ▲ NF-κB1 (p50) protein elevated in elderly ICOS+CD38+ cTfh at baseline
- ▲ TNFRSF1A (TNFR1) increased in elderly cTfh after vaccination while TNFRSF1B (TNFR2) similar
- ▲ Plasma TNF, CXCL11, and MIP1β higher in elderly vs young at baseline
- – No correlation between plasma TNF concentration and TNF-NF-κB gene signature in cTfh
- correlation Pearson r = 0.57, p = 3.9 × 10^-3 (young: cTfh fold change vs plasmablast fold change)
- correlation Pearson r = 0.67, p = 4.8 × 10^-4 (elderly: cTfh fold change vs plasmablast fold change)
- count 58 NF-κB targets up in elderly vs 25 in young (Padj<0.05) (NF-κB target genes upregulated in ICOS+CD38+ cTfh day 7)
- fold_change 19% lower frequency (total CXCR5+PD-1+ cTfh in elderly vs young, p=0.05)
- fold_change 1.2-fold higher (ICOS expression at baseline in elderly, p=0.06)
- count 6 modules (young), 8 modules (elderly); EM1 and EM4 elderly-specific (WGCNA transcriptional modules at day 7)
- count ~1,500 NF-κB target gene list; 6 hub TFs (IRAK3, MYD88, TNFAIP3, STAT5A, REL, TNFSF11) (NF-κB target list and EM4 hub transcription factors)
- count n = 16 per group (independent cohort for NF-κB p50 protein measurement)
Statistical methods review
Model: opusA neutral, descriptive read of the statistical approach — what was done, and (for shared learning, not as criticism) what could also have been done.
The study profiled circulating Tfh subsets before and after influenza vaccination in young versus elderly adults, combining flow cytometry with bulk RNA-seq. Group differences at the single-feature level were assessed with correlation (Pearson) and group-comparison p values, while the transcriptional analyses relied on pathway- and network-level methods (GSEA against MSigDB HALLMARK, WGCNA module detection with Fisher's exact module-preservation testing, Ingenuity Pathway Analysis upstream regulators, and GSVA). Differential expression was reported with adjusted p values (Padj) and findings were illustrated with heatmaps, dot/violin plots, and enrichment plots.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Pearson correlation | cTfh fold change vs plasmablast fold change at day 7 (Figure 1E) | n=25 young, n=26 elderly | not stated |
| group comparison p value (test type not stated in text) | baseline cTfh frequency and ICOS expression, young vs elderly (Figures S1B, S1C; p=0.05, p=0.06) | n=28 young, n=35 elderly enrolled | not stated |
| Fisher's exact test | WGCNA module preservation/overlap between young and elderly modules (Figure 2D) | — | not stated |
| Gene set enrichment analysis (GSEA, MSigDB HALLMARK and curated gene sets) | pathway enrichment in ICOS+CD38+ cTfh, elderly vs young, and module-membership ranked genes (Figures 1, 2E, 2I, S1L-N, S2E) | n=6 young, n=8 elderly (RNA-seq) | na |
| Differential expression with adjusted p value (Padj) | NF-κB target genes upregulated in elderly vs young at day 7 (Figures 2H, S2K; Padj<0.05) | n=6 young, n=8 elderly | not stated |
| Weighted gene correlation network analysis (WGCNA); Ingenuity Pathway Analysis upstream regulators; GSVA | transcriptional module discovery, predicted upstream regulators, and TNF gene-set scoring vs plasma TNF (Figures 2B-G, S3A-B) | n=6 young, n=8 elderly | na |
-
Day-7 elderly-versus-young differences were emphasized through pathway- and network-level methods (GSEA, WGCNA) because individual-gene differences were minimal.↳ Could also: A complementary unsupervised approach such as principal component analysis or a formal differential-expression framework (e.g., DESeq2/limma-voom/edgeR) with shrinkage estimation could also be reported alongside the network analysis. — Reporting the per-gene model fit and dispersion estimates alongside enrichment gives readers a second, gene-level view of the same comparison and makes the link between individual genes and enriched pathways explicit.
-
Group comparisons of baseline cTfh features were summarized with p values (e.g., p=0.05, p=0.06) without the specific test named in this excerpt.↳ Could also: Stating the exact test (e.g., unpaired t-test or Mann-Whitney U) together with an effect size and a 95% confidence interval would also convey the same comparison. — Naming the test and adding an interval communicates both the magnitude and the precision of the difference, which is especially informative when p values sit near conventional thresholds.
-
Module overlap between young and elderly WGCNA networks was assessed with Fisher's exact test on gene membership.↳ Could also: WGCNA's built-in module-preservation Zsummary statistic (permutation-based) could also be used to evaluate whether modules are preserved across groups. — The preservation statistic incorporates network topology (connectivity and density), so it complements the gene-overlap view by indicating whether the internal structure of a module is retained.
-
Plasma cytokine and protein-level comparisons (TNF, p50, TNFR1/2) were made across multiple analytes and subsets.↳ Could also: Applying a single family-wise multiplicity adjustment (e.g., Benjamini-Hochberg) across the panel of analyte comparisons could also be reported. — A stated correction across the analyte family makes the control of false positives explicit when many related markers are tested together.
-
RNA-seq-based conclusions are drawn from n=6 young and n=8 elderly donors.↳ Could also: Reporting per-comparison effect sizes with confidence intervals, and noting the sample size as a stated consideration, could also accompany the enrichment results. — With modest n, intervals and effect sizes help readers gauge the precision of estimates and the strength of the pathway-level signals.
-
Spread/error bars in the figures are presented but the dispersion measure is not specified in this excerpt.↳ Could also: Explicitly labeling whether bars represent SD, SEM, or a 95% CI (SD or CI are often preferred for small n) could also be included. — Clearly identifying the dispersion measure lets readers correctly interpret variability versus precision, particularly with small per-group sample sizes.
Result convergence & founder nodes
Findings this paper shares with others that ran a comparable experiment. A node’s strength is how many independent papers report it (replication breadth) — not how often it is cited, so a heavily-replicated but under-cited founder still stands out.
-
NFKB1 (p50) protein is elevated at baseline in elderly ICOS+CD38+ cTfh compared to young adults.flow-cytometry human ctfh up 2021×1papers★ This paper is the founder (earliest)
-
Elderly adults have lower baseline CXCR5+PD-1+ cTfh frequency but slightly elevated ICOS expression compared to young adults.flow-cytometry human pbmc mixed 2021×1papers★ This paper is the founder (earliest)
-
Vaccine-induced ICOS+CD38+ cTfh fold expansion correlates positively with plasmablast response in both young and elderly adults after influenza vaccination.flow-cytometry human pbmc up 2021×1papers★ This paper is the founder (earliest)
-
Plasma TNF concentration does not correlate with the TNF-NF-κB gene expression signature in ICOS+CD38+ cTfh, suggesting cell-intrinsic rather than systemic TNF drives the pathway.other human plasma none 2021×1papers★ This paper is the founder (earliest)
-
Baseline plasma TNF, CXCL11, and CCL4 (MIP1β) concentrations are elevated in elderly compared to young adults.other human plasma up 2021×1papers★ This paper is the founder (earliest)
-
TNF-NF-κB signaling is the top-enriched transcriptional pathway in the elderly-associated ICOS+CD38+ cTfh gene module (EM4) by GSEA after influenza vaccination.RNA-seq human ctfh up 2021×1papers★ This paper is the founder (earliest)
-
NF-κB target genes are upregulated in elderly versus young ICOS+CD38+ cTfh at day 7 post-influenza vaccination (58 targets up in elderly vs 25 in young).RNA-seq human ctfh up 2021×1papers★ This paper is the founder (earliest)
-
TNFRSF1A (TNFR1) expression is upregulated in elderly ICOS+CD38+ cTfh after influenza vaccination while TNFRSF1B (TNFR2) is unchanged.RNA-seq human ctfh up 2021×1papers★ This paper is the founder (earliest)
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.
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.
- Repression of Divergent Noncoding Transcriptio... L1 No data access
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-34095875
Paper: Herati RS et al., "Vaccine-induced ICOS+CD38+ circulating Tfh are sensitive biosensors of age-related changes in inflammatory pathway", Cell Rep Med 2021. PMID 34095875 · PMCID PMC8149371 · DOI 10.1016/j.xcrm.2021.100262.
Code: https://github.com/teamTfh/cTfh_AgingSignature @ commit
be57be6f4ea998afaae240fdbf58421ee6be7048 (only commit; pushed 2021-03-14; GPL-3; R).
Repo ships the analysis R scripts plus the processed input data
(processedData.zip → bestDataLog.csv, bestDataCounts.csv) and the authors'
own rendered HTML reports (htmlReports.zip, one per script) that contain the
printed numeric output. This is the rare case where the authors shipped both the
exact pipeline input and their rendered output — the html reports serve as the
ground-truth values for an honest 1:1 re-run.
Data: GEO GSE123696 = the raw RNA-seq. Re-deriving the count matrix from raw
reads (alignment + PORT normalization) is the hard last-20% and is not attempted;
the repo ships the post-normalization matrices (bestDataLog.csv = log-expression,
44,207 genes × 83 samples) which are the direct inputs to every downstream script.
In scope (pipeline-derived, inputs shipped)
| Result | Pipeline | Script | Status |
|---|---|---|---|
| GSVA Hallmark pathway scores on ICOS+CD38+ (HiHi_v2) cTfh, day 7 | GSVA (Bioconductor) on bestDataLog.csv + MSigDB Hallmark v6.1 |
GSVA.R |
REPRODUCED |
| GSVA preprocessing byproducts (constant-gene filter count; n gene sets) | GSVA .filterFeatures |
GSVA.R |
REPRODUCED (exact) |
| TNFα-via-NFκB GSVA score, young vs elderly direction (paper's central inflammatory-pathway claim) | GSVA score contrast | GSVA.R |
REPRODUCED (direction) |
In scope but NOT attempted (80/20 — deliberately skipped)
- DESeq2 differential expression (
RNAseqDifferentialExpression.R): runnable frombestDataCounts.csv, but it is a large multi-contrast analysis (24 contrasts, full 83-sample model) and the html does not print a single clean headline DEG integer to anchor against without re-deriving the authors' thresholds. Skipped to stay focused; the GSVA anchors already give exact 1:1 matches. - WGCNA networks, PathwayAnalysis/GSEA (
*.GseaPreranked.*precomputed dirs not shipped), TNFR analysis: runnable in principle but lower-value / heavier.
Out of scope (required input NOT shipped → not reproducible)
- All correlation analyses in
Correlations.R/GSVA.Rthat regress GSVA scores against serum cytokines (TNFα pg/mL), flow-cytometry frequencies/MFI, and antibody titers. These read../../Analysis/BAA_yr4_AllMergedData_FC.csv(a phenotype matrix) which is NOT in the repo and not on GEO → the r/p values inCorrelations.htmlcannot be reproduced here. This is a missing-input limitation, distinct from a fabrication concern. - All wet-lab / flow-cytometry / ELISA measurements (manual, out of scope by rule).
Compute
All compute on «our HPC» («infra»), SLURM «job», env crmsubpathway-gsva136
(R 4.0.5, GSVA 1.36.0). «host» used only for orchestration + small results.
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.
Re-running the authors' GSVA.R on their own shipped bestDataLog.csv reproduced the deterministic byproducts exactly (9286 constant genes, 50 gene sets, 50×14 matrix) against the authors' rendered GSVA.html — strong evidence the shipped data is the genuine input and no fabrication. The central qualitative claim (TNFα/NFκB inflammatory signature higher in elderly ICOS+CD38+ cTfh) reproduces in direction (elderly +0.090 > young −0.231) but not in significance (p=0.117 on the only available test). The gap is on the data-availability side — the phenotype matrix that produced the paper's significant correlations was never deposited — not on the authors' computational logic, so this is a solid, honest reproduction with an explainable, data-driven limitation rather than a substantive 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.