Differential bone morphology and hypoxia activity in skeletal metastases of ER+ and ER- breast cancer.
The main results reproduced, with only marginal, non-material deviations.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓The central claim held under reproduction
- 🟡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
- 🟡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 without author contact, and reproduced largely 1:1 for the headline computational finding. This is a P16 third-party/own-code-equivalent reproduction: the paper's Code Availability says the analysis was 'based on and modified from' github.com/lethesea/Targeting-calcium-signaling-in-bone-micrometastases, whose repo ships the authors' own preprocessed GSE14020 object (breast-mets-gse14020.RData) and defines the scoring method (signature = colSums(clps[genes,]) re.scaled). We ran that verbatim on «our HPC» (R 4.2.3, SLURM «job») on the 16 bone-met samples. RESULT: the paper's central claim (Fig 3H) that HIF1A and a hypoxia signature are elevated in ER- vs ER+ bone metastases reproduces clearly and SIGNIFICANTLY (hypoxia sig p=0.0038, HIF1A p=0.020; robust to a median-split sensitivity check). The osteolysis-signature claim (Fig 2E) reproduces only in DIRECTION (ER->ER+) and is non-significant on the public bone-met data, with mixed per-gene signs -> partial. Overall status 'partial' because (a) Fig 2E is weak/NS and (b) the exact ER+/ER- per-sample assignment is undisclosed in the paper and on GEO, so we used an ESR1 surrogate (counts 6/10 forced; natural gap 7/9) -- the directional results are robust but the labels are provisional. NOT attempted (out of scope): all wet-lab/in vivo work (microCT, histomorphometry, qPCR, hypoxia-chamber), the mouse xenograft RNA-seq (GSE84114 / BioProject PRJNA1183437) with Xenome separation + ssGSEA + CIBERSORT (own-generated/restricted sequencing, non_pipeline for public-data scope), and the GSE110451 correlation (Fig 4A, secondary). No fabrication concern: every reproduced value is derivable from the shipped data+method.
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 68assessed: 2026-06-14 ⛓ 16ef86f37bd6
✎ 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
- 2026-06-14
- 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: opusDo ER+ and ER− breast cancers form bone metastases with different morphology, and is the more osteolytic phenotype of ER− tumors driven by elevated hypoxia/HIF signaling that can be therapeutically targeted?
- ★ ER− breast cancer forms more osteolytic bone metastatic lesions than ER+ breast cancer, while ER+ tumors are more osteoblastic/pre-osteolytic. finding
- ★ Bone metastases of ER− breast cancer are characterized by elevated hypoxia/HIF1A signaling in both tumor cells and bone stroma. mechanism
- ★ Hypoxia signaling stimulates cancer cells to secrete osteolytic inducers, driving osteoclast differentiation and bone destruction. mechanism
- ★ The HIF inhibitor 2-methoxyestradiol suppresses tumor growth and osteoclast differentiation in ER− MDA-MB-231 bone lesions in vivo. finding
- ★ ER+ cell lines depend on MSCs/osteogenic cells for growth, whereas ER− cell growth is suppressed by MSCs. finding
- Intra-iliac artery (IIA) injection models early-stage bone metastasis and enables comparison of ER+ and ER− subtypes. method
- Xenome in silico sorting separates human (cancer) and mouse (stromal) reads to profile tumor and host compartments separately. method
- HIF inhibitors offer a therapeutic rationale for treating skeletal complications of breast cancer bone metastases, especially ER− tumors. resource
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| Immunofluorescence (GFP/ALP and GFP/CTSK staining) | Mouse femur/tibia bone with MCF-7 (ER+) or MDA-MB-231 (ER−) micro-metastases | IIA injection of cancer cells; ± estradiol pellet (7 mg/mouse) | Osteogenic niche (ALP+ cells) and osteoclast (CTSK+) recruitment/contact | — |
| TRAP staining | Mouse bone lesions (MCF-7 day 35; MDA-MB-231 day 21) | IIA injection of cancer cells | Osteoclast activity / osteolysis | — |
| 2D co-culture proliferation assay (bioluminescence) | 4 ER+ and 5 ER− breast cancer cell lines with MSCs or U937 cells | Co-culture at various ratios | Bioluminescence signal (cancer cell growth) on Day 7, normalized to untreated | — |
| 3D co-culture assay | ER+ breast cancer cells with MSCs | Co-culture 7 days | MSC-conferred growth advantage normalized to control | — |
| Bulk RNA-seq with Xenome in silico sorting and ssGSEA/CIBERSORT | Size-matched MCF-7 vs MDA-MB-231 bone lesions (mouse stroma + human tumor); tumor-free bone ± E2 | Cancer cell type (ER+ vs ER−); ± E2 supplementation | Transcriptomic profiles, hallmark gene set enrichment, hypoxia signature, cell composition | — |
| RT-PCR / qPCR (mouse-specific and tumor primers) | Bone-resident cells and cancer cells from size-matched small/medium/large MCF-7 or MDA-MB-231 bone lesions | ± estradiol (7 mg/mouse pellet) | Expression of osteogenesis, osteolysis, and hypoxia-related genes (log2) | — |
| Clinical microarray dataset analysis (GSE14020) | Human breast cancer metastasis specimens (16 bone metastases; ER+ n=6, ER− n=10) | none | Transcriptional levels of osteolysis markers, HIF1A, hypoxia signature by ER status and metastatic site | — |
| In vivo HIF inhibitor treatment | Bone-tropic ER− MDA-MB-231 bone lesions in mice | 2-methoxyestradiol (HIF inhibitor) | Tumor growth and osteoclast differentiation in bone lesions | — |
- – All 4 ER+ cell lines showed growth dependence on MSCs, while all 5 ER− cell lines were growth-suppressed by MSCs in 2D co-culture ER+ 4/4 dependent; ER− 5/5 suppressed
- ▲ ER− MDA-MB-231 micrometastases showed CTSK+ osteoclasts early and strong TRAP staining with bone loss by day 21, while MCF-7 remained pre-osteolytic ~14 days longer despite comparable tumor burden MCF-7 day 35 vs MDA-MB-231 day 21
- ▲ Osteoclast-differentiation markers (Tnfsf11/RANKL, Tnfrsf11a/RANK, Csf1r, Itgb3, Ctsk, Src) increased in ER− MDA-MB-231 bone stroma vs ER+ MCF-7
- ▲ U937 precursors promoted growth of more ER− (4/5, 80%) than ER+ (2/4, 50%) cell lines 80% vs 50%
- ▲ Hypoxia hallmark gene set, Hif1a, and 12-gene hypoxia metagene (Vegfa, Slc2a1, Pgam1, Eno1, Ldha, Tpi1, P4ha1, Cdkn3, Tubb6) upregulated in MDA-MB-231 bone stroma vs MCF-7
- ▲ HIF1A and hypoxia metagene more enriched within MDA-MB-231 tumor cells; hypoxic response more pronounced in tumor cells than stroma
- ▲ In GSE14020, osteolysis markers and HIF1A/hypoxia signature globally more enriched in ER− bone metastases vs ER+
- ▲ MCF-7 induced more intensive osteogenesis in vivo (global enrichment of osteogenesis genes except Col1a1) vs MDA-MB-231
- count 4 of 4 ER+ cell lines growth-dependent on MSCs (2D MSC co-culture)
- count 5 of 5 ER− cell lines suppressed by MSCs (2D MSC co-culture)
- count U937 promoted 4 of 5 (80%) ER− vs 2 of 4 (50%) ER+ cell lines (2D U937 co-culture)
- count ER+ n=6, ER− n=10 bone metastasis specimens (16 bone met total, >60 mets) (GSE14020 clinical dataset)
- count Brain/Liver/Lung/Ovary metastases n=19/5/18/7; bone n=16 (HIF1A expression by site, GSE14020)
- count n=3 biologically independent samples (co-culture quantifications)
- count n=3 animals per group; three size-matched lesion pairs (in vivo RNA-seq/qPCR)
- other estradiol pellet 7 mg/mouse (sustained-release E2 supplementation)
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 study compared ER+ and ER− breast cancer bone metastases using in vivo mouse models (IIA injection), in vitro 2D/3D co-cultures, bulk RNA-seq with species-separated reads (Xenome), and a public clinical microarray dataset (GSE14020). Parametric tests (one-way ANOVA, two-tailed paired t-test, two-way ANOVA, two-tailed t-test) were applied to comparisons at the gene-expression and cell-growth levels. Pathway-level analysis used single-sample GSEA (ssGSEA), and dispersion was reported throughout as SD.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| One-way ANOVA | Cancer cell growth in 2D co-cultures with MSCs at varying ratios (Fig. 1B) | n=3 biologically independent samples per condition | not stated |
| Two-tailed paired t-test | qPCR of osteogenesis-related genes in bone-resident cells across size-matched lesion pairs (Fig. 1D) | n=3 animals (small/medium/large lesion pairs) | not stated |
| One-way ANOVA | Cancer cell growth in 2D co-cultures with U937 cells at varying ratios (Fig. 2C) | n=3 biologically independent samples per condition | not stated |
| Two-tailed paired t-test | qPCR of osteolysis-related genes in bone-resident cells across size-matched lesion pairs (Fig. 2D) | n=3 animals | not stated |
| Two-way ANOVA | Transcriptional levels of osteoclast markers in ER+ vs. ER− bone metastasis specimens from clinical dataset GSE14020 (Fig. 2E) | ER+ n=6 patients, ER− n=10 patients | not stated |
| Single-sample Gene Set Enrichment Analysis (ssGSEA) | Hallmark gene sets in stromal content of MDA-MB-231 vs. MCF-7 bone lesions (Fig. 3B) | n=3 animals per group | na |
| Two-tailed paired t-test | qPCR of hypoxia-related genes in bone-resident cells across size-matched lesion pairs (Fig. 3E) | n=3 animals | not stated |
| Two-tailed paired t-test | qPCR of hypoxia-related genes in cancer cells from size-matched bone lesion pairs (Fig. 3F) | n=3 animals | not stated |
| Two-tailed t-test | HIF1A expression and hypoxia signature in ER+ vs. ER− bone metastases in GSE14020 (Fig. 3H) | ER+ n=6, ER− n=10 | not stated |
-
Paired t-tests were used to compare gene expression between MCF-7 and MDA-MB-231 lesions across three size-matched pairs (n=3 animals), with pairing defined by lesion size (small/medium/large).↳ Could also: A nonparametric Wilcoxon signed-rank test could also be applied to the same paired data. — With only three pairs the paired t-test's normality assumption cannot be verified empirically; the Wilcoxon signed-rank test makes no distributional assumption and is a common alternative when n per group is very small.
-
Multiple individual genes within each heatmap (Figs. 1D, 2D, 3E, 3F) were each tested with separate paired t-tests without a stated multiplicity correction.↳ Could also: A single linear mixed model or repeated-measures ANOVA across all genes, followed by Benjamini-Hochberg FDR adjustment, could also be used. — Applying a correction across the family of gene-level tests within each panel would control the expected proportion of false discoveries; BH-FDR is a widely adopted approach for this setting and scales well to small numbers of comparisons.
-
SD was used as the sole dispersion measure throughout in vitro and in vivo experiments.↳ Could also: A 95% confidence interval (or SEM alongside n) could also be reported alongside or instead of SD. — With very small n (n=3), a CI explicitly communicates estimation uncertainty around the group mean, which can help readers judge how precisely the central tendency is measured; SD describes data spread but does not convey inferential precision.
-
A two-way ANOVA was applied to transcriptional levels in clinical specimens (GSE14020) comparing ER status and gene identity (Fig. 2E, n=6 ER+, n=10 ER−).↳ Could also: A linear model with a random effect for patient, or a permutation-based test, could also handle the small and unbalanced clinical sample sizes. — With only 6 ER+ patients the two-way ANOVA cell counts are sparse; permutation tests or rank-based alternatives (e.g., Kruskal-Wallis per gene with BH correction) make fewer parametric assumptions in this regime.
-
ssGSEA was used to score hallmark gene sets in the stromal compartment of each bone lesion (n=3 per group).↳ Could also: Bulk GSEA with gene-level ranking, or GSVA, could also be applied to the same RNA-seq count matrix. — ssGSEA scores each sample independently and the scores are then compared across groups; GSEA uses a ranked gene list derived from the differential expression between groups and provides a permutation-based p-value, which is an alternative way to assess enrichment significance and may be more powerful when group sizes are very small.
-
One-way ANOVA was used to compare growth across multiple MSC (or U937) ratio conditions within each cell line (Figs. 1B, 2C) without a stated post-hoc test identifying which ratios differed.↳ Could also: A one-way ANOVA followed by a Tukey HSD or Dunnett post-hoc test (comparing each ratio to the 0% control) could also be used. — Post-hoc testing after a significant omnibus ANOVA localizes which pairwise contrasts drive significance while controlling the family-wise error rate; without it, readers cannot determine which specific co-culture ratios differ from one another or from the control.
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.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-39572705
Paper: Das A, Barry MM, Ernst CA, et al. Differential bone morphology and
hypoxia activity in skeletal metastases of ER+ and ER− breast cancer.
Commun Biol 2024. PMID 39572705 · PMCID PMC11582807 · DOI 10.1038/s42003-024-07247-6
Code: https://github.com/lethesea/Targeting-calcium-signaling-in-bone-micrometastases
(default branch master, last push 2018-07-09). The paper's Code Availability
states: "Codes development is based on and modified from the R scripts published at
GitHub: …/Targeting-calcium-signaling-in-bone-micrometastases. Data analysis is
performed by R (version 4.2.1)." Per brief rule P16, applying these published R
scripts to the paper's own data is a fully valid reproduction.
Data: GEO GSE14020 (breast-cancer metastasis microarray, Affymetrix
HG-U133A/Plus2; 65 GSM across GPL96+GPL570). The repo ships the authors' own
preprocessed object breast-mets-gse14020.RData (allmets.clps = gene-collapsed
expression matrix; allmets.ann = per-sample metastasis-site annotation) — i.e. the
exact input the paper's modified code consumed.
Nature of the paper
A mixed wet-lab + bioinformatic study contrasting ER+ (MCF-7) vs ER− (MDA-MB-231) breast-cancer bone metastases. The bioinformatic component scores published gene signatures on public microarray data of human bone-metastasis specimens.
In scope (pipeline-derived, attempted) — all on public GSE14020
The authors' method (from Figure 1 and 5.Rmd) is: a signature score =
apply(allmets.clps[genes,], 2, sum) then re.scale() (linear rescale to [0,1]);
groups compared by ANOVA/t-test. We replicate that verbatim on the 16 bone-met
samples, stratified ER+ vs ER−.
- C1 — cohort: GSE14020 contains 16 bone-metastasis specimens; paper splits them ER+ (n=6) vs ER− (n=10) (Fig 2E, 3H text).
- C2 — HIF1A (Fig 3H): HIF1A expression higher in ER− than ER+ bone mets (two-tailed t-test).
- C3 — Hypoxia signature (Fig 3H): signature = Σ(12 top-rank hypoxia genes + HIF1A); 9 named in Fig 3C: VEGFA, SLC2A1, PGAM1, ENO1, LDHA, TPI1, P4HA1, CDKN3, TUBB6 (+HIF1A). Higher in ER− than ER+ bone mets.
- C4 — Osteolysis signature (Fig 2E): signature = Σ6 osteoclast-differentiation genes — TNFSF11(RANKL), TNFRSF11A(RANK), CSF1R(c-fms), ITGB3, CTSK, SRC. Z-score globally more enriched in ER− than ER+ bone mets (two-way ANOVA).
ER-status assignment (the documented hard-spot / "20%")
The paper says ER status was "identified" for the 16 bone samples but discloses no method, and GEO GSE14020 metadata contains no ER field (only metastasis site in the sample title). We therefore assign ER status by the standard surrogate ESR1 (estrogen-receptor gene) expression — ER+ = high ESR1 — and rank-split to the paper's reported 6/10. We report the ESR1 distribution and a median-split sensitivity analysis. The exact per-sample ER labels are thus provisional; the direction of each comparison is the robust, reproducible scientific claim. Flagged for the human auditor (possible-undisclosed-step note), not a fabrication claim.
Out of scope / not attempted (with reason)
- Mouse xenograft RNA-seq (GSE84114 / BioProject PRJNA1183437), Xenome host/tumor
separation, ssGSEA on mouse stroma, CIBERSORT bone-cell deconvolution — different
data modality + wet-lab-generated sequencing;
non_pipelinefor this RU's public-data scope (would require the restricted/own-generated RNA-seq). - All in vivo / histomorphometry / µCT / qPCR / hypoxia-chamber experiments (Fig 1A–C, 2A–D wet, 3A–G, 4C–F, 5–7) — wet-lab, no public-data pipeline.
- GSE110451 HIF1A↔osteolysis correlation (Fig 4A) — small pre-clinical RData also shipped; secondary, may add if core claims land. Not primary.
- The "remaining 3 of 12" hypoxia genes are not enumerated in the paper; we use the 9 named + HIF1A and note the incompleteness (no fabrication implied).
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 headline computational claim (Fig 3H: HIF1A and a hypoxia signature elevated in ER- vs ER+ bone metastases) reproduces clearly and significantly (p=0.0038, p=0.020) using the authors' own shipped GSE14020 object and scoring method, with no fabrication concern. The main deviations are on our/authors-shared side, not fabrication: the ER+/ER- assignment is undisclosed in the paper and absent from GEO, so labels rest on an ESR1 surrogate (natural gap suggests 7/9 vs the paper's 6/10), and the secondary Fig 2E osteolysis claim reproduces only in direction (NS, p=0.83). Severity is moderate — the central conclusion holds, but the underspecified ER method and the weak Fig 2E keep this yellow overall.
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.