Pembrolizumab and chemotherapy in high-risk, early-stage, ER(+)/HER2(-) breast cancer: a randomized phase 3 trial.
Cardoso F, O'Shaughnessy J, Liu Z, McArthur H, Schmid P, Cortes J, Harbeck N, Telli ML, Cescon DW, Fasching PA, Shao Z, Loirat D, Park YH, Fernandez MG, Rubovszky G, Spring L, Im SA, Hui R, Takano T, André F, Yasojima H, Ding Y, Jia L, Karantza V, Tryfonidis K, Bardia A
- DOI
- 10.1038/s41591-024-03415-7
- Record issued
- 2026-08-15
- Engine
- 7.39.0
- Exported
- 2026-09-20
Prepared by Alpha1. This document is confidential: it is intended for the recipient it was shared with and must not be redistributed. The live record at alpha1science.com/verify/8d38bda8-a710-48b3-aa54-a0023afd616d is authoritative.
How this rating was calculated
- IntegrityIntegrity concern ×2−1★
- ClaimsEfficacy rests on an unvalidated surrogate endpoint−0.5★
- 01Efficacy rests on an unvalidated surrogate endpoint
The primary efficacy claim is based on pathological complete response (pCR), a surrogate endpoint. The paper does not demonstrate target engagement at the tested dose (e.g., PK/PD data) nor cite validated evidence linking pCR to clinical outcomes in this specific ER+/HER2- population, beyond a general meta-analysis reference. The other primary endpoint, event-free survival, is not yet mature.
“Dual primary endpoints were pathological complete response and event-free survival in the intention-to-treat population. ... Event-free survival was not mature in this analysis.”
This Kaimen Rigor review uses Kaimen Rigor reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.
This is a rigorously designed and reported phase 3 randomized controlled trial. The methodology is strong across all dimensions, with clear randomization, blinding, power analysis, and statistical methods. Minor reporting gaps include lack of explicit CONSORT mention and regulatory compliance statement, but these do not undermine the overall integrity.
Both reviewers classified the study as interventional, and no divergence was noted. The evaluation covered all eight dimensions; non-applicable sub-criteria (e.g., animal housing, cell line authentication) were excluded. The statistics verification covered only 2 tests, so the statistical analysis is not fully independently verified.
Numerical inconsistencies
1 finding · worst lowValues that contradict each other or are impossible for the stated sample: recomputed p-values and test statistics, GRIM/GRIMMER checks on summary numbers, percentages against their own counts, totals against their parts, and estimates against their own confidence intervals.
- Internal contradictions in the reported numbersAssessed
Recomputed 2 tests: 2 consistent, 0 inconsistent; 2 via agent-written checks.
- CONSISTENTreported p < .001 · recomputed p = <.001Reviewer 1pCR rate comparison between pembrolizumab and placebo arms using Miettinen-Nurminen method (approximated by chi-square test).
“A pCR (ypT0/Tis ypN0) was observed in 154 of 635 patients (24.3%) in the pembrolizumab–chemotherapy arm and 100 of 643 patients (15.6%) in the placebo‒chemotherapy arm. The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table ).”
Taken as given: The 154 and 100 are the event counts in the pembrolizumab and placebo arms, respectively.; The 635 and 643 are the total numbers in each arm.; The chi-square test is a reasonable approximation to the Miettinen-Nurminen method for this comparison.Method: Two-tailed Pearson chi-square test on the 2x2 table of pCR counts.How we recomputed it: pChi2x2(154, 481, 100, 543) - CONSISTENTreported p < .001 · recomputed p = <.001Reviewer 2pCR rate comparison between pembrolizumab and placebo arms using stratified Miettinen-Nurminen method.
“A pCR (ypT0/Tis ypN0) was observed in 154 of 635 patients (24.3%) in the pembrolizumab–chemotherapy arm and 100 of 643 patients (15.6%) in the placebo‒chemotherapy arm. The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table ).”
Taken as given: The 154 and 100 are the event counts in the pembrolizumab and placebo arms, respectively.; The denominators are 635 and 643, respectively.; The p-value is two-sided.; The test is a chi-square approximation to the Miettinen-Nurminen method.Method: Two-tailed Pearson chi-square test on the 2x2 table (154/635 vs 100/643).How we recomputed it: pChi2x2(154, 481, 100, 543)
- lowinternal contradictionThe safety population (n=634 and n=642) is smaller than the ITT population (n=635 and n=643), which is expected as some patients may not have received treatment, but the paper does not explicitly state the reason for the one-patient difference in each arm.
“During the neoadjuvant phase, treatment-related adverse events (AEs) of any grade occurred in 624 of 634 patients (98.4%) in the pembrolizumab‒chemotherapy arm and 633 of 642 patients (98.6%) in the placebo‒chemotherapy arm”
ResultsFind in source - lowinternal contradictionThe number of patients who received the first neoadjuvant treatment (1,275) is less than the total randomized (1,278), which is expected due to dropouts, but the paper does not explicitly explain the discrepancy.
“At the first interim analysis (data cutoff, 25 May 2023; median duration of follow-up, 33.2 (range = 9.7–51.8) months), 1,275 patients had received the first neoadjuvant treatment”
ResultsFind in source
Overstated conclusions
1 finding · worst highConclusions that reach past what the paper's own results support — including a significance claim that no longer holds when the statistic is recomputed, and efficacy resting on an unvalidated surrogate endpoint.
- Efficacy rests on an unvalidated surrogate endpointAssessed
6 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewers 1, 2The addition of pembrolizumab to neoadjuvant chemotherapy significantly improved the pathological complete response rate in patients with high-risk, early-stage ER+/HER2- breast cancer.The claim is directly supported by the primary endpoint result: pCR rate 24.3% vs 15.6%, difference 8.5 percentage points, P=0.00005.Evidence: Results, Efficacy: pCR rates and treatment difference with CI and p-value.
“In summary, the addition of pembrolizumab to neoadjuvant chemotherapy significantly improved the pathological complete response rate in patients with high-risk, early-stage ER + /HER2 − breast cancer.”
AbstractFind in source - supportedReviewers 1, 2Safety was consistent with the known profiles of each study treatment.The safety data show expected adverse events with higher immune-related AEs in the pembrolizumab arm, consistent with known profiles.Evidence: Results, Safety: AE rates and types.
“Safety was consistent with the known profiles of each study treatment.”
AbstractFind in source - supportedReviewers 1, 2pCR benefit was observed across all prespecified subgroups.The subgroup analysis shows consistent benefit across subgroups, though some CIs are wide.Evidence: Results, Efficacy: subgroup analysis (Fig. 2).
“The between-group difference in pCR favored pembrolizumab‒chemotherapy across all prespecified subgroups, albeit with a differing magnitude of benefit and wide 95% CIs in some subgroups.”
Discussion ¶1Find in source - supportedReviewer 1The addition of pembrolizumab shifted more patients to lower residual cancer burden categories.The RCB analysis shows a shift to lower categories (RCB-0/1: 35.0% vs 23.6%).Evidence: Results, Efficacy: RCB analysis (Fig. 3).
“The addition of pembrolizumab to neoadjuvant chemotherapy also shifted more patients to lower RCB categories (RCB-0 and RCB-1), demonstrating the ability of the combination to reduce tumor tissue remaining after surgery among those without pCR.”
Discussion ¶2Find in source - supportedReviewers 1, 2Event-free survival was not mature at the first interim analysis.The paper explicitly states EFS was not mature and continues to be evaluated.Evidence: Results, Efficacy: 'EFS was not mature at the first interim analysis'.
“EFS was not mature at the first interim analysis; this endpoint continues to be evaluated by prespecified subsequent interim analyses and a final analysis.”
ResultsFind in source - supportedReviewer 2The addition of pembrolizumab shifted more patients to lower RCB categories.The RCB analysis shows a shift to lower categories (RCB-0/1: 35.0% vs 23.6%), supporting the claim.Evidence: Figure 3 and text describe RCB results.
“Analysis of the exploratory endpoint of residual cancer burden (RCB) showed that the addition of pembrolizumab to neoadjuvant chemotherapy shifted more patients to lower RCB categories (RCB-0 or RCB-1, 35.0% versus 23.6%; RCB-2, 40.8% versus 45.3%; RCB-3, 20.5% versus 28.9%; Fig. ).”
ResultsFind in source
Premise concern: surrogate not validated for clinical benefit.
- INADEQUATESurrogate endpointThe primary efficacy claim is based on pathological complete response (pCR), a surrogate endpoint. The paper does not demonstrate target engagement at the tested dose (e.g., PK/PD data) nor cite validated evidence linking pCR to clinical outcomes in this specific ER+/HER2- population, beyond a general meta-analysis reference. The other primary endpoint, event-free survival, is not yet mature.
“Dual primary endpoints were pathological complete response and event-free survival in the intention-to-treat population. ... Event-free survival was not mature in this analysis.”
- ADEQUATEEffect sizeThe absolute improvement in pCR is 8.5 percentage points (from 15.6% to 24.3%), which is statistically significant (P=0.00005) and the trial was powered to detect a 15 percentage point difference. The effect is presented with confidence intervals and subgroup analyses, and the paper discusses clinical meaningfulness in the context of prior studies and regulatory guidance.
“the pathological complete response rate was 24.3% ... in the pembrolizumab−chemotherapy arm and 15.6% ... in the placebo−chemotherapy arm (estimated treatment difference, 8.5 percentage points; 95% CI, 4.2–12.8; P = 0.00005).”
Data authenticity concerns
None foundAn adversarial read for patterns associated with data that may not be genuine: results that look too clean, implausibly large effects, duplicated data or images, and methods that do not match the results reported.
Checked — nothing surfaced.
Reporting gaps
None foundRequired detail the manuscript never states — study design, biological variables, ethics approval and consent, key resources, statistical reporting, data and code availability, and overall transparency.
Checked — nothing surfaced.
The introduction cites prior studies (I-SPY2, KEYNOTE-522) and a meta-analysis correlating pCR with EFS/OS, providing a strong rationale for the trial. It also acknowledges the heterogeneity of ER+ disease and the need for better treatments in high-risk subgroups. Limitations of prior research are addressed by designing a phase 3 trial with dual primary endpoints.
Randomization method (central interactive voice-response system) and unit (patient) are described. Blinding is double-blind with placebo control. Power analysis is provided (>99% power for pCR, 84% for EFS). Inclusion/exclusion criteria are detailed. Outlier handling is addressed via ITT and as-treated populations. Controls are the placebo arm. Independent replication is not applicable for a single pivotal trial.
“Patients were randomly assigned (in a 1:1 ratio) to the pembrolizumab–chemotherapy arm or the placebo–chemotherapy arm using a central interactive voice-response system with an integrated web-response system.”
“With an enrollment of approximately 1,240 patients, the trial had >99% power to detect a true difference of 15 percentage points for the comparison of the rate of pCR (stage ypT0/Tis ypN0) between the treatment arms at a one-sided α level of 0.005.”
“Patients were randomly assigned (in a 1:1 ratio) to the pembrolizumab–chemotherapy arm or the placebo–chemotherapy arm using a central interactive voice-response system with an integrated web-response system.”
“With an enrollment of approximately 1,240 patients, the trial had >99% power to detect a true difference of 15 percentage points for the comparison of the rate of pCR (stage ypT0/Tis ypN0) between the treatment arms at a one-sided α level of 0.005.”
Sex is reported (six men included), age is reported (median and range), and demographics are detailed in Table 1. Sex justification is not applicable as both sexes are enrolled. Species/strain/housing are not applicable for human trial. Demographics include age, sex, region, PD-L1 status, ECOG, tumor stage, grade, ER positivity, and menopausal status.
“A total of six men with ER + and HER2 − breast cancer were included in the study (three in each treatment arm).”
“Median (range) | 49 (24–82) | 49 (19–78)”
“Demographics and baseline disease characteristics were balanced between treatment arms (Table ).”
“A total of six men with ER + and HER2 − breast cancer were included in the study (three in each treatment arm).”
The protocol was approved by an ethics body at each study site, and patients provided written informed consent. Regulatory compliance is implied by adherence to ICH-GCP and the Declaration of Helsinki, though not explicitly named. The ethics statement is adequate as it names the approving body (ethics body at each site).
“The protocol was approved by an ethics body at each study site (see for a list of study sites).”
“Patients provided written informed consent”
“The protocol was approved by an ethics body at each study site (see for a list of study sites).”
“Patients provided written informed consent”
The investigational product (pembrolizumab) is identified with dose and regimen. The PD-L1 IHC 22C3 pharmDx assay is identified with vendor. Statistical software (SAS v9.4) is identified. Antibodies, cell lines, mycoplasma, and organisms are not applicable as this is a clinical trial without wet-lab assays.
“patients received four cycles of intravenous pembrolizumab (200 mg) or placebo once Q3W plus paclitaxel QW (80 mg m − 2 ; first neoadjuvant treatment)”
“PD-L1 expression in new or recent core needle biopsy samples was assessed at a central laboratory using PD-L1 IHC 22C3 pharmDx (Agilent Technologies).”
“Statistical analyses were conducted using SAS v9.4 (SAS Institute).”
“patients received four cycles of intravenous pembrolizumab (200 mg) or placebo once Q3W plus paclitaxel QW (80 mg m − 2 ; first neoadjuvant treatment)”
“PD-L1 expression in new or recent core needle biopsy samples was assessed at a central laboratory using PD-L1 IHC 22C3 pharmDx (Agilent Technologies).”
“Statistical analyses were conducted using SAS v9.4 (SAS Institute).”
The primary analysis uses the stratified Miettinen–Nurminen method, which is named. Exact p-values are reported (P = 0.00005). Effect sizes with 95% CIs are reported. Software is identified. Data presentation includes per-group n and CIs. Assumptions are handled by design (stratified analysis). Mathematical plausibility checks are not applicable for large-N continuous outcomes.
“We used the stratified Miettinen–Nurminen method , with weights proportional to the stratum sample size, to compare between-arm differences in percentages of patients with a pCR.”
“The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table ).”
“The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table ).”
“We used the stratified Miettinen–Nurminen method , with weights proportional to the stratum sample size, to compare between-arm differences in percentages of patients with a pCR.”
“The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table ).”
“The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table ).”
The data availability statement provides a concrete route for data access via the MSD data-sharing website and a review committee. Repository deposit and accession numbers are not applicable for patient-level data. Code sharing is not applicable as no bespoke code is mentioned.
“Feasible requests will be reviewed by a committee of MSD subject matter experts to assess the scientific validity of the request and the qualifications of the requestors.”
The trial is registered (NCT03725059). Methods are detailed. Limitations are discussed (e.g., EFS not mature, subgroup analyses underpowered). Conclusions are proportional, noting the need for replication. Funding and COI are disclosed.
“ClinicalTrials.gov identifier: NCT03725059”
“However, these results should be interpreted with caution as these subgroups are underpowered, and the only objective of subgroup analyses is to explore convergent validity.”
“To establish confidence, replication in other trials is absolutely essential.”
“However, these results should be interpreted with caution as these subgroups are underpowered, and the only objective of subgroup analyses is to explore convergent validity.”
“Funding for this research was provided by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA.”
Registered (1 ID: ClinicalTrials.gov). No reporting guideline cited.
Broken references and links
None foundReferences checked against Crossref, OpenAlex and Retraction Watch for retractions and resolvability, plus declared data and code links probed for whether they resolve to content matching the paper.
Checked — nothing surfaced.
Checked 27 references by DOI: 23 verified — 4 no DOI (shown, not verified).
- NO DOIPathological complete response in neoadjuvant treatment of high-risk early-stage breast cancer: use as an endpoint to support accelerated approval guidance for industryNo DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
- NO DOIThe role of pathological complete response as an endpoint in neoadjuvant breast cancer studiesNo DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
- NO DOINCCN clinical practice guidelines in oncology (NCCN guidelines). Breast cancer, version 4.2023No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
- NO DOICommon Terminology Criteria for Adverse Events (CTCAE), version 4.0No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
3 data/code links checked; 3 live.
- datahttps://clinicaltrials.gov/study/NCT03725059LIVEHTTP 200Resolves, but the content could not be matched to the paper.
- datahttps://clinicaltrials.gov/ct2/show/NCT03725059LIVEHTTP 200Resolves, but the content could not be matched to the paper.
- datahttp://engagezone.msd.com/ds_documentation.phpLIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
6 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 6 minor suggestions below.
6 copyedit issues flagged: mostly typo, consistency, clarity.
- MINORtypoAuthor list“O’Shaughnessy Joyce Liu Zhenzhen”→ Ensure author names are correctly formatted with proper punctuation.Potential missing comma or spacing.
- MINORconsistencyAbstract“P = 0.00005”→ Consider reporting p-value as <0.001 for readability.Exact p-value is acceptable but may be formatted differently.
- MINORclarityMethods, Trial design and treatment“Patients from Eastern Europe were further stratified by tumor PD-L1 status (CPS ≥ 1 versus < 1).”→ Clarify that this stratification applies only to Eastern Europe.The sentence is clear but could be rephrased for better flow.
- MINORtypoAbstract“ER + /HER2 − breast cancer”→ Ensure consistent spacing around plus/minus signs.Minor formatting inconsistency.
- MINORconsistencyResults, Safety“treatment-related acute non- Q wave myocardial infarction”→ Remove space after hyphen: 'non-Q wave'.Typographical spacing issue.
- MINORclarityMethods, Statistical analysis“The graphical method discussed in ref. was applied”→ Provide the reference number explicitly.Reference placeholder not resolved.
The published work is robust and well-reported. An informed reader should weigh the minor reporting gaps (e.g., explicit CONSORT and regulatory compliance statements) and the limited independent statistical verification, but these do not warrant a correction. The trial's conclusions are appropriately cautious and supported by the data.
- 1.MEDIUMreportingIn the Methods or Reporting Summary, explicitly state compliance with the CONSORT reporting guideline.Both reviewers flagged the absence of an explicit CONSORT mention, which is a standard expectation for RCT reporting.
- 2.MEDIUMethicsIn the Methods, explicitly state that the trial was conducted in accordance with the Declaration of Helsinki and ICH-GCP guidelines.One reviewer noted that regulatory compliance is only implied, and an explicit statement strengthens the ethics reporting.
- 3.MEDIUMreportingIn the Methods, provide more detail on the randomization sequence generation (e.g., block size) to enhance reproducibility.The randomization method is described, but additional details on sequence generation would improve transparency.
- 4.MEDIUMstatisticsIn the Statistical analysis section, clarify the handling of missing data for EFS and other secondary endpoints.The paper does not explicitly describe missing data imputation or handling, which is important for survival analyses.
- 5.MEDIUMdata codeIn the Data Availability section, clarify the timeframe for data availability after product approval.One reviewer suggested specifying when data will be available, which is a common requirement for data-sharing statements.
- 6.MEDIUMreportingIn the Methods, specify the exact version of the AJCC staging criteria used for pCR assessment.Providing the staging version ensures consistency and reproducibility of the pCR definition.
- 7.MEDIUMstatisticsIn the Results, report the number of patients with missing pCR data to fully address missing-data handling.Transparency about missing data is essential for interpreting the primary endpoint.
- 8.MEDIUMreportingIn the Discussion, explicitly acknowledge the lack of independent replication of the pCR finding within this trial.While the paper notes the need for replication, explicitly stating that this is a single trial without independent replication would be clearer.
- 9.MEDIUMreportingIn the Methods, describe the process for central confirmation of ER+/HER2- status to ensure consistency across sites.Central confirmation of biomarker status is critical for trial validity and should be described.
- 10.MEDIUMstatisticsIn the Results, consider presenting p-values for secondary endpoints to fully report all outcomes.Reporting p-values for secondary endpoints would enhance completeness, though it is not mandatory.
- 11.LOWcopyeditFix the author list formatting: 'O’Shaughnessy Joyce Liu Zhenzhen' should have proper punctuation and spacing.Correct author formatting is essential for indexing and professional presentation.
- 12.LOWcopyeditIn the Abstract, consider reporting the p-value as '<0.001' for readability, or keep exact value consistently.Consistency in p-value reporting improves readability.
- 13.LOWcopyeditIn the Methods, clarify that the stratification by PD-L1 status applies only to Eastern Europe.The sentence could be rephrased for better flow and clarity.
- 14.LOWcopyeditEnsure consistent spacing around plus/minus signs in 'ER + /HER2 − breast cancer' throughout the manuscript.Minor formatting consistency improves professional appearance.
- 15.LOWcopyeditIn the Results, Safety section, remove the space after the hyphen: 'non-Q wave' should be 'non-Q wave'.Typographical spacing issue that should be corrected.
- 16.LOWcopyeditIn the Methods, Statistical analysis, resolve the reference placeholder: 'The graphical method discussed in ref. was applied' should cite the specific reference.Unresolved reference placeholders are a copyedit issue that must be fixed before publication.
The star rating is the report’s one-glance summary. Every paper starts at 5★ and loses stars for the concrete problems the review finds — so a rating is never a vague average, it’s a running total you can read line by line under “How this rating was calculated.”
- Reporting — 8 dimensionseach dimension that fully fails−½★
- each dimension partially met−¼★
- Statistics · Integrity · Claimseach serious problem−1★
- each medium problem−½★
- Citationseach retracted or unverifiable reference−¼★
- Copyeditonly when the manuscript needs a full edit−½★
The rating never drops below 1★, and a demonstrable critical failure (an impossible statistic, a proven ethics violation) caps it at 1★ on its own — so the stars can never look healthy when the verdict is CRITICAL.
The rating draws on a panel of agents. Three independent Kaimen Rigor reviewers grade the eight dimensions below across several independent passes (the shown verdict is their majority vote — steadier than any single run), isolate the paper’s major claims and check its own evidence backs them, and flag integrity concerns. Alongside them, a citation agent resolves every reference against Crossref, OpenAlex, and Retraction Watch; a statistics agent recomputes reported tests; and rule-based checks verify that declared data/code links actually resolve. Full text is required — an abstract-only submission is not analyzed.
Graded against NIH, MDAR, ARRIVE 2.0, CONSORT, EQUATOR, and RRID guidelines. A dimension that doesn’t apply to the study type is skipped, never penalized.
This Kaimen Rigor review is model-assisted and is not a substitute for formal expert review. It complements human evaluation by surfacing potential methodological concerns — verify each finding against the source.