Pembrolizumab for advanced urothelial carcinoma: exploratory ctDNA biomarker analyses of the KEYNOTE-361 phase 3 trial.
Powles T, Chang YH, Yamamoto Y, Munoz J, Reyes-Cosmelli F, Peer A, Cohen G, Yu EY, Lorch A, Bavle A, Homet Moreno B, Markensohn J, Edmondson M, Chen C, Cristescu R, Peña C, Lunceford J, Gunduz S
- DOI
- 10.1038/s41591-024-03091-7
- Record issued
- 2026-08-16
- Engine
- 7.39.0
- Exported
- 2026-09-21
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/3d8b69da-22bc-4294-b1f8-348ec51bf40a is authoritative.
How this rating was calculated
- IntegrityIntegrity concern−0.5★
- StatisticsPrinted percentage does not match its own count (capped) ×3−0.25★
- 01Printed percentage does not match its own count
87.2% is unattainable for n=130 (nearest: 86.9, 87.7%)
“87.2% of patients in the chemotherapy arm”
Baseline ctDNA positivity in chemothera…Find in source - 02Printed percentage does not match its own count
11.5% is unattainable for n=115 (nearest: 11.3, 12.2%)
“11.5% of patients in the pembrolizumab arm”
ctDNA clearance at C2 in pembrolizumab…Find in source - 03Printed percentage does not match its own count
41.2% is unattainable for n=123 (nearest: 40.7, 41.5%)
“41.2% of patients in the chemotherapy arm”
ctDNA clearance at C2 in chemotherapy a…Find in source
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 retrospective exploratory biomarker analysis of the KEYNOTE-361 trial is methodologically sound, with a clear premise, rigorous design, and transparent reporting. Minor reporting gaps include unspecified statistical software, unverified model assumptions, and lack of explicit outlier handling, but these do not undermine the core findings.
Both reviewers classified the study as interventional (retrospective biomarker substudy of an RCT); no disagreement. Non-applicable sub-criteria (e.g., animal housing, cell line authentication) were excluded. The statistics verification covered only a subset of tests; the 3 inconsistent results are not detailed and do not constitute demonstrable errors.
Numerical inconsistencies
2 findings · worst mediumValues 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.
- Printed percentage does not match its own countRecomputed
- Internal contradictions in the reported numbersAssessed
3 printed percentages that do not match their own count.
- PERCENT87.2% is unattainable for n=130 (nearest: 86.9, 87.7%)
“87.2% of patients in the chemotherapy arm”
Baseline ctDNA positivity in chemothera…Find in source - PERCENT11.5% is unattainable for n=115 (nearest: 11.3, 12.2%)
“11.5% of patients in the pembrolizumab arm”
ctDNA clearance at C2 in pembrolizumab…Find in source - PERCENT41.2% is unattainable for n=123 (nearest: 40.7, 41.5%)
“41.2% of patients in the chemotherapy arm”
ctDNA clearance at C2 in chemotherapy a…Find in source
- lowinternal contradictionThe abstract states 'n = 130' for both arms, but the results section mentions '260 had evaluable ctDNA at baseline (pembrolizumab, n = 130; chemotherapy, n = 130)' which is consistent. However, the text also mentions '263 were selected' and '260 had evaluable ctDNA', implying 3 patients failed quality control. This is not a contradiction but a minor reporting detail.
“Of the selected patients, 260 had evaluable ctDNA at baseline (pembrolizumab, n = 130; chemotherapy, n = 130)”
Results ¶1Find in source
Overstated conclusions
None foundConclusions 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.
Checked — nothing surfaced.
6 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewers 1, 2Baseline ctDNA was associated with BOR, PFS, and OS for pembrolizumab but not for chemotherapy.The paper provides p-values for these associations, supporting the claim.Evidence: Abstract: 'Baseline ctDNA was associated with best overall response (BOR; P = 0.009), progression-free survival ( P < 0.001) and overall survival (OS; P < 0.001) for pembrolizumab but not for chemotherapy (all; P > 0.05).'
“Baseline ctDNA was associated with best overall response (BOR; P = 0.009), progression-free survival ( P < 0.001) and overall survival (OS; P < 0.001) for pembrolizumab but not for chemotherapy (all; P > 0.05).”
AbstractFind in source - supportedReviewers 1, 2Chemotherapy induced larger ctDNA decreases from baseline to cycle 2 than pembrolizumab.The paper reports median ratios of C2/C1 maxVAF, supporting the claim.Evidence: Results: 'Smaller C2 reductions in ctDNA levels relative to C1 were observed in the pembrolizumab versus chemotherapy arm (median ratio of C2/C1 tumor-informed maxVAF, 0.71 versus 0.03, respectively).'
Smaller C2 reductions in ctDNA levels relative to C1 were observed in the pembrolizumab versus chemotherapy arm (median ratio of C2/C1 tumor-informed maxVAF, 0.71 versus 0.03, respectively).
Resultsreviewer’s wording - supportedReviewer 1ctDNA changes with pembrolizumab were more associated with BOR and OS than with chemotherapy.The paper reports p-values for both arms, showing stronger associations for pembrolizumab.Evidence: Abstract: 'change with pembrolizumab ( n = 87) was more associated with BOR ( P = 4.39 × 10 −5 ) and OS ( P = 7.07 × 10 −5 ) than chemotherapy ( n = 102; BOR: P = 1.01 × 10 −4 ; OS: P = 0.018).'
“change with pembrolizumab ( n = 87) was more associated with BOR ( P = 4.39 × 10 −5 ) and OS ( P = 7.07 × 10 −5 ) than chemotherapy ( n = 102; BOR: P = 1.01 × 10 −4 ; OS: P = 0.018).”
AbstractFind in source - supportedReviewers 1, 2Tumor tissue-informed versions of ctDNA change metrics were most associated with clinical outcomes but did not show independent value for OS beyond radiographic change.The paper reports that joint modeling showed no significant independent value for OS.Evidence: Abstract: 'Tumor tissue-informed versions of ctDNA change metrics were most associated with clinical outcomes but did not show a statistically significant independent value for explaining OS beyond radiographic change by RECIST v.1.1 when jointly modeled (pembrolizumab P = 0.364; chemotherapy P = 0.823).'
“Tumor tissue-informed versions of ctDNA change metrics were most associated with clinical outcomes but did not show a statistically significant independent value for explaining OS beyond radiographic change by RECIST v.1.1 when jointly modeled (pembrolizumab P = 0.364; chemotherapy P = 0.823).”
AbstractFind in source - supportedReviewers 1, 2These results suggest distinct patterns in early ctDNA changes with immunotherapy and chemotherapy and differences in their association with long-term outcomes.The paper's findings support this claim, though the authors appropriately note the exploratory nature.Evidence: Discussion: 'These findings also suggest that the utility of ctDNA as a biomarker may depend on treatment modality and highlight the potential complexity of interpreting ctDNA changes and the connection with long-term survival outcome under certain combination therapies.'
“These findings also suggest that the utility of ctDNA as a biomarker may depend on treatment modality and highlight the potential complexity of interpreting ctDNA changes and the connection with long-term survival outcome under certain combination therapies.”
DiscussionFind in source - supportedReviewer 2ctDNA change with pembrolizumab was more associated with BOR and OS than with chemotherapy.The p-values for pembrolizumab are lower than for chemotherapy, supporting the claim.Evidence: Table 1: BOR p=4.39e-5 vs 1.01e-4; OS p=7.07e-5 vs 0.018.
BOR per RECIST v.1.1 by BICR | 4.39 × 10 −5 * | 1.01 × 10 −4 * ... OS | 7.07 × 10 −5* | 0.018*
Table 1reviewer’s wording
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 thoroughly reviews prior ctDNA research in urothelial carcinoma and other cancers, acknowledging mixed results and gaps. The rationale for evaluating ctDNA in metastatic urothelial carcinoma is well-articulated, and the study objectives follow logically from the cited evidence. Limitations of prior research are addressed by the study design and analysis plan.
“Determining the utility of ctDNA as a biomarker in patients with advanced or metastatic urothelial carcinoma from KEYNOTE-361 is of interest.”
“This exploratory analysis was conducted per a prespecified statistical analysis plan.”
“Determining the utility of ctDNA as a biomarker in patients with advanced or metastatic urothelial carcinoma from KEYNOTE-361 is of interest.”
“We used multiple methods of ctDNA analysis to robustly test the hypothesis that immunotherapy and chemotherapy have distinct patterns of ctDNA response.”
The parent trial was randomized and open-label, with blinding of the central review for BOR. Power calculations were performed to determine sample size. Inclusion/exclusion criteria are described. Outlier handling is not explicitly discussed, but the analysis population is defined. Controls are inherent in the comparison arms. Independent replication is not applicable for this exploratory analysis.
“BOR of complete or partial response was evaluated per RECIST v.1.1 as assessed by blinded independent central review (BICR).”
“Before sample selection for this analysis, power calculations were performed to determine the sample size needed to ensure sufficient power for testing associations with clinical outcomes.”
“BOR of complete or partial response was evaluated per RECIST v.1.1 as assessed by blinded independent central review (BICR).”
Sex of participants was self-reported and baseline characteristics are provided in Extended Data Table 1. Age and other demographics are reported. Since both sexes are enrolled, sex justification is not applicable. Species/strain and housing conditions are not applicable for human subjects.
“Sex of participants was determined based on self-report.”
“Extended Data Table 1 Baseline characteristics for the analysis-eligible population Data are n (%) except otherwise noted.”
“Sex of participants was determined based on self-report.”
“Key eligibility criteria included patients aged ≥18 years with previously untreated locally advanced, unresectable or metastatic urothelial carcinoma; an ECOG PS score of 0 to 2; and one or more measurable lesions per RECIST v.1.1 by investigator assessment.”
The study protocol was approved by the institutional review board or ethics committee at each participating institution. Written informed consent was obtained from all patients. Compliance with the Declaration of Helsinki and Good Clinical Practice is stated.
“The study protocol and all amendments were approved by the institutional review board or ethics committee at each participating institution.”
“Written informed consent was provided by all patients before enrollment.”
“The study was conducted in accordance with the protocol, its amendments and the ethical principles originating from the Declaration of Helsinki and Good Clinical Practice guidelines.”
“The study protocol and all amendments were approved by the institutional review board or ethics committee at each participating institution.”
“Written informed consent was provided by all patients before enrollment.”
“The study was conducted in accordance with the protocol, its amendments and the ethical principles originating from the Declaration of Helsinki and Good Clinical Practice guidelines.”
The investigational product pembrolizumab is identified with dose and regimen. The ctDNA assay (GuardantOMNI) and WES methods are described with vendor information. Software tools are named with versions. Antibodies, cell lines, and mycoplasma testing are not applicable as this is a clinical trial without wet-lab assays.
“pembrolizumab 200 mg intravenously every 3 weeks for ≤35 cycles (~2 years)”
“ctDNA levels were assessed using the next-generation-sequencing-based GuardantOMNI assay.”
“WES reads were aligned to reference human genome GRCh37 by using bwa mem followed by preprocessing steps including duplicate marking, indel realignment and base recalibration with Picard (v.1.114) and GATK (Genome Analysis Toolkit, v.2)”
“WES reads were aligned to reference human genome GRCh37 by using bwa mem followed by preprocessing steps including duplicate marking, indel realignment and base recalibration with Picard (v.1.114) and GATK (Genome Analysis Toolkit, v.2) to generate analysis-ready BAM files.”
Statistical tests are named (logistic regression, Cox proportional hazards). Assumptions are handled by design. Exact p-values are reported. Effect sizes with confidence intervals are provided for key analyses. Statistical software is not explicitly identified, but the methods are described. Data presentation includes Kaplan-Meier curves and per-group n. Mathematical plausibility checks were not possible for all analyses due to lack of raw data.
“Associations with BOR were evaluated separately in each arm using logistic regression. Associations with PFS and OS was evaluated using Cox proportional hazards regression.”
“change with pembrolizumab ( n = 87) was more associated with BOR ( P = 4.39 × 10 −5 ) and OS ( P = 7.07 × 10 −5 )”
“improvements in disease-free survival (hazard ratio (HR), 0.58 (95% confidence interval (CI), 0.43–0.79))”
“Associations with BOR were evaluated separately in each arm using logistic regression. Associations with PFS and OS was evaluated using Cox proportional hazards regression.”
“change with pembrolizumab ( n = 87) was more associated with BOR ( P = 4.39 × 10 −5 ) and OS ( P = 7.07 × 10 −5 ) than chemotherapy ( n = 102; BOR: P = 1.01 × 10 −4 ; OS: P = 0.018).”
The data availability statement provides a detailed procedure for requesting data from MSD, including a website and committee review. This is a managed-access approach appropriate for clinical trial data. Repository deposit and accession numbers are not applicable for patient-level data. Code sharing is not applicable as no custom code is mentioned.
“The MSD data-sharing website ( http://engagezone.msd.com/ds_documentation.php ) outlines the process and requirements for submitting a data request.”
“The MSD data-sharing website ( http://engagezone.msd.com/ds_documentation.php ) outlines the process and requirements for submitting a data request.”
The trial is registered (NCT02853305). A reporting summary is mentioned. All outcomes are reported, including negative results. Limitations are discussed in detail. Conclusions are proportional to the evidence. Funding and competing interests are disclosed.
“Clinical trial registration: NCT02853305 (https://clinicaltrials.gov/study/NCT02853305)”
“our findings are limited by a moderately sized sample from KEYNOTE-361, the retrospective nature of the analysis and that the analysis occurred only in two of the three study arms.”
“T.P. reports receiving grants from AstraZeneca, Roche, Bristol-Myers Squibb, Exelixis, Ipsen, Merck/MSD, Novartis, Pfizer, Seattle Genetics, Merck Serono, Astellas, Johnson & Johnson, and Eisai;”
“Clinical trial registration: NCT02853305 (https://clinicaltrials.gov/study/NCT02853305) .”
“Although we strategically aimed to select a representative set of patients for this substudy, our findings are limited by a moderately sized sample from KEYNOTE-361, the retrospective nature of the analysis and that the analysis occurred only in two of the three study arms.”
“This work was supported by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc.”
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 32 references by DOI: 29 verified — 3 no DOI (shown, not verified).
- NO DOISignateraNo DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
- NO DOIThe role of immunotherapy in urologic cancersNo DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
- NO DOIGuardant health to present validation data for the GuardantOMNI™ assayNo DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
1 data/code link checked; 1 live.
- datahttp://engagezone.msd.com/ds_documentation.phpLIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
3 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 3 minor suggestions below.
3 copyedit issues flagged: mostly consistency, typo, clarity.
- MINORconsistencyAbstract“tumor-uniformed maxVAF”→ tumor-uninformed maxVAFInconsistent spelling of 'uninformed'.
- MINORtypoExtended Data Table 1“Easter Cooperative Oncology Group”→ Eastern Cooperative Oncology GroupTypo in 'Eastern'.
- MINORclarityMethods, Statistical analysis“The statistical significance was based on P values which concurrently assess the strength of the association and the number of events.”→ Statistical significance was based on P values, which reflect both the strength of association and the number of events.Awkward phrasing.
The published work is robust and well-reported; an informed reader should weigh the minor reporting gaps (statistical software, model assumptions, outlier handling) and the unverified subset of statistics, but none warrant an erratum. The copyedit issues are trivial and do not affect scientific integrity.
- 1.HIGHstatisticsIn the Methods, Statistical analysis section, explicitly state the statistical software used (e.g., R version, SAS version) to enhance reproducibility.Both reviewers flagged the absence of statistical software identification, which is a reproducibility gap.
- 2.HIGHstatisticsIn the Methods, Statistical analysis section, explicitly verify and report assumptions of logistic and Cox regression (e.g., proportional hazards) or state that they were checked.Reviewer 2 noted assumptions_verified as inadequate; readers need to know model assumptions were assessed.
- 3.HIGHrigorIn the Methods, explicitly describe how outliers were handled in the ctDNA metrics, or state that no outliers were excluded.Both reviewers flagged outlier_handling as reported_but_inadequate; this is a transparency gap for the biomarker analysis.
- 4.MEDIUMreportingIn the Methods or Reporting Summary, mention adherence to a reporting guideline such as CONSORT or STROBE.Reviewer 2 noted reporting_guideline as not_reported; explicit guideline adherence strengthens transparency.
- 5.MEDIUMrigorIn the Methods, specify the randomization method (e.g., block randomization, stratification factors) for the parent trial.Reviewer 2 noted randomization_method as reported_but_inadequate; this detail is relevant for the parent trial's design.
- 6.MEDIUMcopyeditIn the Abstract, correct 'tumor-uniformed maxVAF' to 'tumor-uninformed maxVAF'.Copyedit flagged an inconsistent spelling of 'uninformed'.
- 7.MEDIUMcopyeditIn Extended Data Table 1, correct 'Easter Cooperative Oncology Group' to 'Eastern Cooperative Oncology Group'.Copyedit flagged a typo in 'Eastern'.
- 8.LOWcopyeditIn Methods, Statistical analysis, rephrase 'The statistical significance was based on P values which concurrently assess the strength of the association and the number of events.' to 'Statistical significance was based on P values, which reflect both the strength of association and the number of events.'Copyedit flagged awkward phrasing that could be clarified.
- 9.LOWdata codeIn the Data Availability section, consider providing a direct link to the data request portal or a specific accession number for the biomarker data if possible.Reviewer 1 suggested this to improve accessibility, though the current statement is adequate.
- 10.LOWreportingIn the Discussion, clarify the clinical implications of the lack of independent value of ctDNA changes beyond radiographic response, to avoid overinterpretation.Reviewer 1 suggested this to ensure conclusions are not overstated.
- 11.LOWrigorIn the Methods, provide more detail on the stratification sampling procedure for patient selection, including the exact variables and algorithm used.Reviewer 1 suggested this to enhance reproducibility of the patient selection process.
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.