Camrelizumab plus CAPOX with camrelizumab based maintenance versus CAPOX alone as initial treatment for gastric or gastro-oesophageal junction adenocarcinoma: randomised phase 3 trial.
Peng Z, Zhang Y, Xu H, Yang Y, Yang M, Zhang M, Cheng Y, Chen X, Pan Y, Wang F, Li Q, Xu N, Liu L, Gu K, Xiao J, Zhang R, Zhao Q, Chen J, Lin L, Chen Y, Deng Y, Cai S, Bai B, Wang Y, Shen L
- DOI
- 10.1136/bmj-2025-086115
- Record issued
- 2026-08-15
- Engine
- 7.39.0
- Exported
- 2026-09-22
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/657492f9-f73b-467f-a56a-b15d4540af64 is authoritative.
How this rating was calculated
Started at 5★ — no deductions. Nothing the checks ran surfaced a material problem.
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 well-conducted, transparently reported phase 3 randomized trial with a clear scientific premise, rigorous design, and comprehensive reporting of demographics, ethics, resources, statistics, and data availability. The only minor issues are copyedit-level (hyphenation, abbreviation consistency) and the absence of an explicit reporting guideline statement, which do not undermine the scientific integrity.
Both reviewers classified the study as interventional, and this was adopted. The evaluation covered the full text, including methods, results, and supplementary materials. Non-applicable sub-criteria (e.g., animal housing, cell line authentication) were excluded. The statistics verification covered only a subset of tests (2 recomputed consistently); other p-values were not machine-verified. The citation check found no retracted or non-existent references.
Numerical inconsistencies
None foundValues 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.
Checked — nothing surfaced.
Recomputed 2 tests: 2 consistent, 0 inconsistent; 2 via agent-written checks.
- CONSISTENTreported p = .020 · recomputed p = .033Reviewers 1, 2Check p-value for overall survival hazard ratio in PD-L1 positive population (camre+apa vs CAPOX).
“Overall survival was longer with camre+CAPOX followed by camre+apa than with CAPOX alone in the PD-L1 positive population (median 15.0 v 12.5 months; hazard ratio 0.80 (95% CI 0.65 to 0.98); one sided P=0.02)”
Taken as given: The hazard ratio is 0.80 with 95% CI 0.65 to 0.98.; The p-value is one-sided, so the two-sided p is doubled.; The CI is a 95% confidence interval.Method: Recomputed two-sided p from hazard ratio and 95% CI using normal approximation, then halved for one-sided.How we recomputed it: pCI(0.80, 0.65, 0.98, 1) - CONSISTENTreported p = .004 · recomputed p = .007Reviewers 1, 2Check p-value for overall survival hazard ratio in overall population (camre+apa vs CAPOX).
“Overall survival was longer with camre+CAPOX followed by camre+apa than with CAPOX alone in the overall population (median 13.5 v 12.1 months; hazard ratio 0.80 (0.68 to 0.94); one sided P=0.004)”
Taken as given: The hazard ratio is 0.80 with 95% CI 0.68 to 0.94.; The p-value is one-sided, so the two-sided p is doubled.; The CI is a 95% confidence interval.Method: Recomputed two-sided p from hazard ratio and 95% CI using normal approximation, then halved for one-sided.How we recomputed it: pCI(0.80, 0.68, 0.94, 1)
Overstated conclusions
1 finding · worst lowConclusions 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.
- Conclusions only partially backed by the presented evidenceAssessed
4 major claims checked against the paper's own evidence: 1 only partially supported (evidence backs part of the claim; gaps or caveats remain); the rest adequately supported.
- partialReviewers 1, 2The survival gain over CAPOX is driven mainly by camrelizumab.The data show that both camre-based regimens improve survival over CAPOX, but the claim that the gain is 'mainly' driven by camrelizumab is an interpretation, as the trial was not designed to isolate the contribution of camrelizumab alone.Evidence: Both camre+apa and camre groups showed improved overall survival versus CAPOX, but no direct comparison of camrelizumab alone versus CAPOX was performed.
“These findings indicate that the survival gain over CAPOX is driven mainly by camrelizumab, and that routine intensification of maintenance therapy with an antiangiogenic agent may not be justified in unselected patients.”
ConclusionFind in source - supportedReviewers 1, 2Camrelizumab plus CAPOX followed by camrelizumab based maintenance was associated with longer overall survival than CAPOX alone in HER2 negative, unresectable, locally advanced or metastatic gastric or gastro-oesophageal junction adenocarcinoma.The primary endpoint analysis shows a statistically significant improvement in overall survival for the camre+apa group versus CAPOX in both PD-L1 positive and overall populations.Evidence: Hazard ratio 0.80 (95% CI 0.65 to 0.98) in PD-L1 positive and 0.80 (0.68 to 0.94) in overall population, with one-sided p-values of 0.02 and 0.004, respectively.
“Overall survival was longer with camre+CAPOX followed by camre+apa than with CAPOX alone in the PD-L1 positive population (median 15.0 v 12.5 months; hazard ratio 0.80 (95% CI 0.65 to 0.98); one sided P=0.02) and in the overall population (median 13.5 v 12.1 months; hazard ratio 0.80 (0.68 to 0.94); one sided P=0.004).”
AbstractFind in source - supportedReviewer 1Adding apatinib to camrelizumab maintenance did not further prolong survival compared with camrelizumab alone.The exploratory comparison between camre+apa and camre showed no significant difference in overall survival, with hazard ratios close to 1.0 and wide confidence intervals.Evidence: Hazard ratio 1.08 (95% CI 0.83 to 1.40) in PD-L1 positive and 1.02 (0.82 to 1.26) in overall population, with one-sided nominal p-values of 0.29 and 0.45, respectively.
“No overall survival benefit was observed with camre+CAPOX followed by camre+apa versus camre+CAPOX followed by camre.”
AbstractFind in source - supportedReviewer 2Exploratory comparisons between the two camrelizumab based regimens showed no additional survival benefit, with higher rates of treatment related adverse events of grade ≥3 and treatment discontinuations when apatinib was added during maintenance.The exploratory analyses show no significant difference in overall survival between the two camre-based regimens (HR 1.08, p=0.29 in PD-L1 positive; HR 1.02, p=0.45 in overall), and safety data show higher grade ≥3 AEs and discontinuations with apatinib.Evidence: Results section and Table 4 show safety data; exploratory HRs are reported.
“Exploratory comparisons between the two camrelizumab based regimens showed no additional survival benefit, with higher rates of treatment related adverse events of grade ≥3 and treatment discontinuations when apatinib was added during maintenance.”
ConclusionFind in source
Efficacy claim is anchored to an adequate endpoint and a meaningful effect.
- ADEQUATESurrogate endpointThe primary endpoint is overall survival, a hard clinical outcome, not a surrogate.
“The primary endpoint was overall survival for camre+CAPOX followed by camre+apa versus CAPOX alone in the PD-L1 positive population (combined positive score >1) and the overall population who received at least one dose of study drug.”
- ADEQUATEEffect sizeThe effect size is a median overall survival improvement of 2.5 months in PD-L1 positive and 1.4 months in overall population, with hazard ratios of 0.80, which is clinically meaningful and consistent with other checkpoint inhibitor trials.
“Overall survival was longer with camre+CAPOX followed by camre+apa than with CAPOX alone in the PD-L1 positive population (median 15.0 v 12.5 months; hazard ratio 0.80 (95% CI 0.65 to 0.98); one sided P=0.02) and in the overall population (median 13.5 v 12.1 months; hazard ratio 0.80 (0.68 to 0.94); one sided P=0.004).”
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 research on gastric cancer, immunotherapy, and antiangiogenic agents, and acknowledges gaps in knowledge about maintenance strategies. The rationale for comparing the three regimens is clearly linked to the hypothesis that adding apatinib to camrelizumab maintenance may enhance benefit. Limitations of prior research are addressed by designing a phase 3 trial to provide definitive evidence.
“In a phase 1 study, camrelizumab (a humanised IgG4 monoclonal antibody with high affinity for programmed death 1) combined with apatinib showed encouraging antitumour activity and manageable toxicity in patients with previously treated gastric or gastro-oesophageal junction cancer and hepatocellular carcinoma.”
“These findings raised the hypothesis that introducing an antiangiogenic agent during maintenance might enhance or prolong the benefits of immune checkpoint inhibitor based initial treatment, but whether such intensification provides meaningful clinical benefit beyond immune checkpoint inhibitor maintenance alone, given the potential for additional toxicity, is unknown.”
“However, the optimal maintenance strategy and the potential role of treatment intensification during maintenance remain uncertain, and prospective phase 3 data directly comparing immune checkpoint inhibitors based maintenance approaches are lacking.”
“However, the optimal maintenance strategy and the potential role of treatment intensification during maintenance remain uncertain, and prospective phase 3 data directly comparing immune checkpoint inhibitors based maintenance approaches are lacking.”
The trial is a randomized, open-label, phase 3 study with a 2:2:1 randomization ratio, block randomization, and stratification by key prognostic factors. The sample size calculation is detailed, with power and alpha specified. Inclusion/exclusion criteria are clearly defined. The open-label design is acknowledged, and the primary endpoint (overall survival) is objective. The analysis population is defined as those who received at least one dose of study drug.
“Patients were randomly assigned through a centralised interactive web response system using block randomisation (block size of six), stratified by Eastern Cooperative Oncology Group performance status (0 v 1), presence of peritoneal metastasis (yes v no), and PD-L1 combined positive score (>1 v ≤1).”
“We estimated that 340 deaths would provide 88% power to detect a hazard ratio of 0.71 (median overall survival 16.2 v 11.5 months) in the PD-L1 positive population, and 576 deaths to provide 91% power to detect a hazard ratio of 0.76 (median overall survival 15.1 v 11.5 months) in the overall population.”
“This was an open label study: neither investigators nor patients were masked to treatment allocation.”
“Patients were randomly assigned through a centralised interactive web response system using block randomisation (block size of six), stratified by Eastern Cooperative Oncology Group performance status (0 v 1), presence of peritoneal metastasis (yes v no), and PD-L1 combined positive score (>1 v ≤1).”
“This was an open label study: neither investigators nor patients were masked to treatment allocation.”
The paper reports age, sex, ECOG performance status, primary tumor location, metastatic sites, and PD-L1 status in Table 1. Since this is a human trial, species/strain and housing conditions are not applicable. Sex is reported for both sexes, so sex_justified is not applicable. Demographics are adequately reported.
“Median (IQR) age (years) | 62 (55-67) | 62 (54-68) | 60 (54-68) | | Sex: | | Male | 250 (71.0) | 272 (77.9) | 128 (72.3) | | Female | 102 (29.0) | 77 (22.1) | 49 (27.7)”
“ECOG performance status: | | 0 | 105 (29.8) | 104 (29.8) | 51 (28.8) | | 1 | 247 (70.2) | 245 (70.2) | 126 (71.2)”
“Median (IQR) age (years) | 62 (55-67) | 62 (54-68) | 60 (54-68) | | Sex: | | Male | 250 (71.0) | 272 (77.9) | 128 (72.3) | | Female | 102 (29.0) | 77 (22.1) | 49 (27.7) |”
“ECOG performance status: | | 0 | 105 (29.8) | 104 (29.8) | 51 (28.8) | | 1 | 247 (70.2) | 245 (70.2) | 126 (71.2) |”
The paper states that the protocol was approved by the ethics committee at each participating centre and that written informed consent was obtained from all participants. It also mentions compliance with the Declaration of Helsinki and Good Clinical Practice guidelines. This meets the criteria for human research.
“The study protocol and all amendments were approved by the ethics committee at each participating centre (see supplementary table S1) and conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines.”
“Written informed consent was obtained from all participants before enrolment.”
“The study protocol and all amendments were approved by the ethics committee at each participating centre (see supplementary table S1) and conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines.”
“Written informed consent was obtained from all participants before enrolment.”
The drugs (camrelizumab, capecitabine, oxaliplatin, apatinib) are named with doses and regimens. The PD-L1 assay is described with the antibody clone (E1L3N) and central laboratory. Since this is a drug trial, antibodies, cell lines, and mycoplasma testing are not applicable. Software (SAS 9.4) is identified.
“A central laboratory (Shanghai Xiawei Biotechnology, Amoy Diagnostics, Shanghai, China; certified by the College of American Pathologists) evaluated PD-L1 combined positive score using an immunohistochemistry assay with the E1L3N antibody clone.”
“SAS software (version 9.4) was used for all statistical analyses.”
“A central laboratory (Shanghai Xiawei Biotechnology, Amoy Diagnostics, Shanghai, China; certified by the College of American Pathologists) evaluated PD-L1 combined positive score using an immunohistochemistry assay with the E1L3N antibody clone.”
“SAS software (version 9.4) was used for all statistical analyses.”
The paper names the statistical tests (log-rank, Cox proportional hazards, Clopper-Pearson, Newcombe) and provides exact p-values (e.g., P=0.02, P=0.004). Effect sizes are reported with 95% CIs. Software is identified. Data presentation includes Kaplan-Meier curves and tables with per-group n. The analysis is appropriate for a large clinical trial.
“We compared overall survival and progression-free survival with the log rank test, stratified by Eastern Cooperative Oncology Group performance status (0 v 1), presence of peritoneal metastasis (yes v no), and PD-L1 expression status (positive v negative; overall population only).”
“Overall survival was longer with camre+CAPOX followed by camre+apa than with CAPOX alone in the PD-L1 positive population (median 15.0 v 12.5 months; hazard ratio 0.80 (95% CI 0.65 to 0.98); one sided P=0.02)”
“Treatment v CAPOX (hazard ratio (95% CI), P value) | 0.80 (0.65 to 0.98), 0.02”
“We compared overall survival and progression-free survival with the log rank test, stratified by Eastern Cooperative Oncology Group performance status (0 v 1), presence of peritoneal metastasis (yes v no), and PD-L1 expression status (positive v negative; overall population only).”
“Treatment v CAPOX (hazard ratio (95% CI), P value) | 0.80 (0.65 to 0.98), 0.02”
The paper provides a data availability statement with a Mendeley Data link and states that the code is in the supplemental file. This meets the criteria for data and code sharing.
“The data underlying the findings in this paper are openly and publicly available and can be accessed here: https://data.mendeley.com/datasets/z22zyc2b2b/1”
“The code used to analyse the data in the paper can be found in the supplemental file.”
“The data underlying the findings in this paper are openly and publicly available and can be accessed here: https://data.mendeley.com/datasets/z22zyc2b2b/1 .”
“The code used to analyse the data in the paper can be found in the supplemental file.”
The trial is registered (NCT03813784). The paper includes a detailed discussion of limitations, and conclusions are proportional to the evidence. Funding and competing interests are disclosed. The reporting guideline is not explicitly mentioned, but the paper follows standard reporting for clinical trials.
“Trial registration ClinicalTrials.gov NCT03813784 (https://clinicaltrials.gov/ct2/show/NCT03813784)”
“Several limitations should be acknowledged. Firstly, this open label trial used multiple therapeutic agents, and tumour responses were not assessed by a blinded independent central review committee.”
“Trial registration ClinicalTrials.gov NCT03813784 (https://clinicaltrials.gov/ct2/show/NCT03813784)”
“Several limitations should be acknowledged. Firstly, this open label trial used multiple therapeutic agents, and tumour responses were not assessed by a blinded independent central review committee.”
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 26 references by DOI: 26 verified.
Every extracted reference resolved against Crossref/OpenAlex with no retraction flags.
1 data/code link checked; 1 live.
- dataMendeley DataLIVEHTTP 200https://data.mendeley.com/datasets/z22zyc2b2b/1Resolves to Mendeley Data (data repository).
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, grammar.
- MINORtypoAbstract, Results“one sided P=0.02”→ one-sided P=0.02Hyphenation of 'one sided' is inconsistent.
- MINORconsistencyTable 1 footnote“Camre+CAPOX followed by camre+apatinib.”→ Camre+CAPOX followed by camre+apa.Abbreviation expansion inconsistent with table header.
- MINORgrammarResults, Efficacy in PD-L1 positive patients“The corresponding overall survival rates at 12, 18, and 24 months were 56.8% (95% CI 50.1% to 63.0%), 41.9% (35.4% to 48.2%), and 33.8% (27.7% to 40.0%) with camre+CAPOX followed by camre+apa.”→ The corresponding overall survival rates at 12, 18, and 24 months were 56.8% (95% CI 50.1% to 63.0%), 41.9% (35.4% to 48.2%), and 33.8% (27.7% to 40.0%) with camre+CAPOX followed by camre+apa, respectively.Missing 'respectively' for clarity.
- MINORtypoAbstract, Results“454 of 592 (76.7%) deaths had occurred”→ Consider rephrasing to '454 of 592 (76.7%) patients had died' for clarity.Minor wording issue.
- MINORconsistencyTable 1, footnote“Camre+CAPOX followed by camre+apatinib.”→ Ensure consistent use of abbreviations throughout the manuscript.Abbreviation expansion is clear but could be standardized.
- MINORgrammarResults, Efficacy in PD-L1 positive patients“The corresponding overall survival rates at 12, 18, and 24 months were 56.8% (95% CI 50.1% to 63.0%), 41.9% (35.4% to 48.2%), and 33.8% (27.7% to 40.0%) with camre+CAPOX followed by camre+apa.”→ Consider breaking long sentences for readability.Long sentence but grammatically correct.
The published work is robust and well-reported; an informed reader should weigh the open-label design and the lack of a blinded independent review committee as the main limitations, but these are acknowledged. No erratum is warranted for the copyedit issues, though the authors may consider a minor correction for the hyphenation and abbreviation consistency. The data and code are available, and the trial is registered.
- 1.MEDIUMreportingAdd an explicit statement in the Methods or a separate section that the trial is reported in accordance with the CONSORT 2010 guideline.Both reviewers noted the absence of an explicit reporting guideline, which is a standard expectation for clinical trials and enhances transparency.
- 2.MEDIUMcopyeditFix the hyphenation of 'one sided' to 'one-sided' in the Abstract and throughout the manuscript.Consistency in terminology is important for professional presentation.
- 3.MEDIUMcopyeditStandardize the abbreviation 'camre+apa' vs 'camre+apatinib' in Table 1 footnotes and text.Inconsistent abbreviation expansion can confuse readers and should be unified.
- 4.LOWcopyeditAdd 'respectively' to the sentence listing overall survival rates at 12, 18, and 24 months in the Results section.Clarifies the correspondence between the rates and time points.
- 5.LOWcopyeditRephrase '454 of 592 (76.7%) deaths had occurred' to '454 of 592 (76.7%) patients had died' in the Abstract.Improves clarity and avoids awkward phrasing.
- 6.LOWdata codeConsider providing a more detailed description of the statistical analysis code in the supplemental file, including version control and dependencies.Enhances reproducibility and allows readers to verify the analyses.
- 7.LOWdata codeAdd a statement about the availability of the full protocol and statistical analysis plan in the supplementary materials.Increases transparency and allows independent verification of the pre-specified analyses.
- 8.LOWreportingIn the Discussion, explicitly mention the lack of a reporting guideline as a limitation and how it might affect interpretation.Acknowledging this gap demonstrates thoroughness and helps readers assess the reporting quality.
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.