Atezolizumab plus FOLFOX for Stage III Mismatch Repair-Deficient Colon Cancer.
Sinicrope FA, Ou FS, Arnold D, Peters WR, Behrens RJ, Lieu CH, Matin K, Cohen DJ, Potter SL, Nixon AB, Kottschade LA, Kathol E, Frankel WL, Shergill A, Hsu D, Reinacher-Schick A, Mehan P, Gold PJ, Khalil MF, Zemla T, Gatten C, O'Reilly EM, Meyerhardt JA
- DOI
- 10.1056/NEJMoa2507874
- 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/ae0bd40d-5531-4587-b880-5f1ec1290804 is authoritative.
How this rating was calculated
- StatisticsStatistic did not reproduce−0.5★
- IntegrityIntegrity concern−0.5★
- ReportingData & code availability partially met−0.25★
- 01Reported statistic does not recompute
Check the p-value for the primary DFS analysis from the reported HR and CI.
“hazard ratio (HR) for recurrence or death, 0.50; 95% CI, 0.35 to 0.73; P<0.0001”
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 phase 3 randomized trial with rigorous design, clear reporting of methods and results, and appropriate ethical approvals. The main weaknesses are the lack of an explicit data availability statement and minor reporting inconsistencies (e.g., CI bound discrepancy, blinding not explicitly stated).
Both reviewers agreed on all dimensions; no divergence. The study is an interventional trial; all non-applicable sub-criteria (e.g., animal housing, cell line authentication) were excluded. The statistics verification covered only 2 tests, one of which showed a minor p-value discrepancy; this does not affect the overall assessment.
Numerical inconsistencies
2 findings · worst highValues 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.
- Reported statistics do not recomputeRecomputed
- Internal contradictions in the reported numbersAssessed
Recomputed 2 tests: 1 consistent, 1 inconsistent; 2 via agent-written checks.
- INCONSISTENTreported p < .001 · recomputed p = <.001Reviewers 1, 2Check the p-value for the primary DFS analysis from the reported HR and CI.
“hazard ratio (HR) for recurrence or death, 0.50; 95% CI, 0.35 to 0.73; P<0.0001”
Taken as given: The HR is 0.50 and the 95% CI is 0.35 to 0.73.; The CI is two-sided at 95%.; The p-value is two-sided.Method: Compute the two-sided p-value from the HR and its 95% CI using the normal approximation for the log HR.How we recomputed it: pCI(0.50, 0.35, 0.73, 1) - CONSISTENTreported p < .001 · recomputed p = <.001Reviewers 1, 2Check the p-value for the stratified log-rank test for DFS.
“The stratified log-rank p-value was 0.0002.”
Taken as given: The log-rank test statistic is approximately normal with mean 0 and variance 1.; The reported p-value is two-sided.; The Z value is derived from the p-value.Method: Convert the two-sided p-value to a Z-score using the inverse normal CDF.How we recomputed it: pZ(3.71)
- lowinternal contradictionThe abstract reports the 3-year DFS upper CI as 89.9, while the Results section reports 89.8.
“86.3% (95% CI, 81.8 to 89.9) in the atezolizumab plus mFOLFOX6 arm and 76.2% (95% CI, 70.9 to 80.6) in the mFOLFOX6 arm [hazard ratio (HR) for recurrence or death, 0.50; 95% CI, 0.35 to 0.73; P<0.0001].”
AbstractFind in source
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
5 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 therapeutic benefit of adjuvant atezolizumab may depend on adequate chemotherapy exposure.The subgroup analysis by chemotherapy duration shows a benefit only in patients receiving >6 cycles, but this is exploratory and not definitive.Evidence: Pre-specified subgroup analysis: HR 0.41 (95% CI 0.27-0.64) for >6 cycles vs HR 0.97 (95% CI 0.44-2.11) for ≤6 cycles.
“These exploratory findings suggest, but do not prove, that the therapeutic benefit of adjuvant atezolizumab may depend on adequate chemotherapy exposure.”
Discussion ¶4Find in source - supportedReviewers 1, 2The addition of atezolizumab to mFOLFOX6 significantly improved disease-free survival in stage III dMMR colon cancer.The primary endpoint DFS was met with a stratified HR of 0.50 and a p-value of 0.0002, supporting the claim.Evidence: DFS at 3 years was 86.3% vs 76.2%, HR 0.50 (95% CI 0.35-0.73), p=0.0002.
“The addition of atezolizumab to mFOLFOX6 significantly improved disease-free survival in stage III dMMR colon cancer.”
ConclusionFind in source - supportedReviewers 1, 2The addition of atezolizumab to mFOLFOX6 was associated with a 50% reduction in the hazard of disease recurrence or death.The HR of 0.50 directly supports a 50% reduction in hazard.Evidence: Stratified HR 0.50 (95% CI 0.35-0.73).
“The addition of atezolizumab to mFOLFOX6 was associated with a ten percentage point absolute reduction and a 50% reduction in the hazard of disease recurrence or death compared to mFOLFOX6 alone.”
Discussion ¶1Find in source - supportedReviewers 1, 2No difference in overall survival has been observed between the study arms at this first analysis.The OS analysis showed no statistically significant difference (p=0.6787), supporting the claim.Evidence: OS HR 0.90 (95% CI 0.55-1.47), p=0.6787.
“No difference in overall survival has been observed between the study arms at this first analysis with a median follow-up of 40.9 months, and longer follow-up is required for mature estimates.”
Discussion ¶1Find in source - supportedReviewer 1The addition of atezolizumab to mFOLFOX6 resulted in a significant reduction in recurrence or death indicating it is an effective adjuvant treatment.The primary endpoint result supports this claim, though OS data are immature.Evidence: DFS HR 0.50, p=0.0002.
“In conclusion, the addition of atezolizumab to mFOLFOX6 resulted in a significant reduction in recurrence or death indicating it is an effective adjuvant treatment of patients with resected dMMR stage III colon cancer.”
DiscussionFind in source
Efficacy claim is anchored to an adequate endpoint and a meaningful effect.
- ADEQUATESurrogate endpointThe primary endpoint is disease-free survival (DFS), a hard clinical outcome (time to recurrence or death), not a surrogate biomarker. The trial is a phase 3 randomized controlled trial with a clinically meaningful endpoint.
“The primary endpoint was disease-free survival, defined as the time from randomization to disease recurrence or death from any cause.”
- ADEQUATEEffect sizeThe effect size is reported as a hazard ratio of 0.50 (95% CI, 0.35 to 0.73) for recurrence or death, with 3-year DFS rates of 86.3% vs 76.2%, representing a 10 percentage point absolute improvement. This is statistically significant and clinically meaningful in the adjuvant setting.
“the 3-year disease-free survival was 86.3% (95% CI, 81.8 to 89.9) in the atezolizumab plus mFOLFOX6 arm and 76.2% (95% CI, 70.9 to 80.6) in the mFOLFOX6 arm [hazard ratio (HR) for recurrence or death, 0.50; 95% CI, 0.35 to 0.73; P<0.0001].”
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
1 finding · worst mediumRequired detail the manuscript never states — study design, biological variables, ethics approval and consent, key resources, statistical reporting, data and code availability, and overall transparency.
- Data/code availability incompleteAssessed
The introduction cites prior evidence on dMMR/MSI-H colon cancer, the role of immune checkpoint inhibitors, and the rationale for combining atezolizumab with FOLFOX. It acknowledges the resistance of dMMR tumors to fluoropyrimidines and the benefit of oxaliplatin. The hypothesis follows logically from the cited evidence. Limitations of prior work are implicitly addressed by the design of a randomized phase 3 trial.
“In patients with metastatic dMMR colorectal cancer which encompasses those with microsatellite instability (MSI-H), the immune checkpoint inhibitor pembrolizumab was shown to be superior to fluoropyrimidine-based chemotherapy where it produced significantly longer progression-free survival with fewer adverse events as first-line therapy (KEYNOTE-177).”
“The phase 3 ATOMIC study was designed to investigate whether the addition of atezolizumab, an anti-PD-L1 antibody, to standard FOLFOX chemotherapy can improve disease-free survival in patients with curatively resected, dMMR stage III colon cancer.”
“In patients with metastatic dMMR colorectal cancer which encompasses those with microsatellite instability (MSI-H), the immune checkpoint inhibitor pembrolizumab was shown to be superior to fluoropyrimidine-based chemotherapy where it produced significantly longer progression-free survival with fewer adverse events as first-line therapy (KEYNOTE-177).”
“The phase 3 ATOMIC study was designed to investigate whether the addition of atezolizumab, an anti-PD-L1 antibody, to standard FOLFOX chemotherapy can improve disease-free survival in patients with curatively resected, dMMR stage III colon cancer.”
Randomization method (permuted block design with block size 4) and unit (patient) are reported. Blinding is not described, but the paper states the trial was open-label (implied by the design and the lack of blinding mention; however, the paper does not explicitly state open-label, but the design of an adjuvant trial with different treatment durations makes blinding infeasible; the paper does not provide a rationale for lack of blinding, but this is not a major issue for a phase 3 trial). Power analysis is reported with effect size, alpha, and power. Inclusion/exclusion criteria are pre-specified. Outlier handling is not explicitly described, but the analysis population (ITT and per-protocol) is defined. Controls are the mFOLFOX6 arm. Independent replication is not applicable for a single pivotal trial.
“Patients were randomized (1:1), via permuted block design with block size of 4, within 10 weeks after surgery to atezolizumab (840 mg intravenously every 2 weeks for 12 cycles) (6 months) plus modified (m) FOLFOX6 (folinic acid, fluorouracil, and oxaliplatin) followed by atezolizumab monotherapy for 13 cycles (12 months total) versus mFOLFOX6 [(5-fluorouracil 400 mg/m 2 bolus and 2400 mg/m 2 continuous infusion over 46 hours), oxaliplatin (85 mg/m 2 ), and leucovorin (400 mg/m 2 )] alone for 12 cycles (6 months total).”
“The trial was designed to randomize 700 patients, with 165 events providing an overall 90% power to detect a hazard ratio (HR) of 0.6 favoring the atezolizumab plus mFOLFOX6 arm at the one-sided significance level of 0.025.”
“Eligible patients were ≥ 12 years of age with completely resected (R0), histologically confirmed, stage III colon adenocarcinoma (any T, N1-2 M0) and dMMR status determined by local site-directed immunohistochemistry or at a site-selected reference laboratory; retrospective central confirmation was also performed.”
“Patients were randomized (1:1), via permuted block design with block size of 4, within 10 weeks after surgery”
“The trial was designed to randomize 700 patients, with 165 events providing an overall 90% power to detect a hazard ratio (HR) of 0.6 favoring the atezolizumab plus mFOLFOX6 arm at the one-sided significance level of 0.025.”
Sex is reported (55.1% female). Age is reported (median 64 years). Demographics include sex, age, and tumor characteristics. Health status is implied by ECOG performance status. Species/strain and housing conditions are not applicable for a human trial. Sex justification is not applicable because both sexes are enrolled.
“Median patient age was 64 years, 55.1% were female”
The trial was approved by the NCI central institutional review board (IRB). All patients provided written informed consent. Regulatory compliance is stated with adherence to Good Clinical Practice and the Declaration of Helsinki.
“The protocol was approved by the NCI central institutional review board (IRB).”
“All the patients provided written informed consent before enrollment.”
“The trial was conducted in accordance with the Good Clinical Practice guidelines of the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use and the principles of the Declaration of Helsinki.”
“The protocol was approved by the NCI central institutional review board (IRB).”
“All the patients provided written informed consent before enrollment.”
“The trial was conducted in accordance with the Good Clinical Practice guidelines of the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use and the principles of the Declaration of Helsinki.”
Atezolizumab is identified as an anti-PD-L1 antibody with dose and regimen. mFOLFOX6 components are specified with doses. Statistical software (SAS version 9.04) is identified. No antibodies, cell lines, or mycoplasma testing are applicable as this is a clinical trial without wet-lab assays.
“atezolizumab (840 mg intravenously every 2 weeks for 12 cycles) (6 months) plus modified (m) FOLFOX6 (folinic acid, fluorouracil, and oxaliplatin) followed by atezolizumab monotherapy for 13 cycles (12 months total)”
“Statistical analyses were performed using SAS software, version 9.04 (SAS Institute).”
“atezolizumab (840 mg intravenously every 2 weeks for 12 cycles) (6 months) plus modified (m) FOLFOX6 (folinic acid, fluorouracil, and oxaliplatin)”
“Statistical analyses were performed using SAS software, version 9.04 (SAS Institute).”
Tests are named (stratified log-rank, stratified Cox proportional-hazards). Assumptions are handled by design (Cox model). Exact p-values are reported (e.g., P<0.0001, p=0.0002). Effect sizes with CIs are reported. Software is identified. Data presentation includes Kaplan-Meier curves and forest plots. Mathematical plausibility is not applicable for large-N continuous outcomes.
“Disease-free and overall survival were compared between treatment groups using stratified log-rank tests; their hazard ratios were estimated using a stratified Cox proportional-hazards model without covariates adjustment.”
“The stratified log-rank p-value was 0.0002.”
“hazard ratio (HR) for recurrence or death, 0.50; 95% CI, 0.35 to 0.73; P<0.0001”
“stratified HR, 0.50; 95% confidence interval [CI] , 0.35 to 0.73”
No data availability statement is present. The paper mentions supplementary material but does not specify where data are deposited. For a clinical trial, a data availability statement is required. The trial is registered, but data access is not described.
“protocol available with the full text of this article at NEJM.org”
The trial is registered (NCT02912559). Methods are detailed. Limitations are discussed (e.g., overall survival immature, subsequent therapy confounds). Conclusions are proportional to the evidence. Funding and COI are disclosed.
“NCT02912559 (https://clinicaltrials.gov/ct2/show/NCT02912559)”
“No difference in overall survival has been observed between the study arms at this first analysis with a median follow-up of 40.9 months, and longer follow-up is required for mature estimates.”
“No difference in overall survival has been observed between the study arms at this first analysis with a median follow-up of 40.9 months, and longer follow-up is required for mature estimates.”
Registered (1 ID: ClinicalTrials.gov). Reporting guideline cited: CONSORT.
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 24 references by DOI: 24 verified.
Every extracted reference resolved against Crossref/OpenAlex with no retraction flags.
1 of 2 data/code links checked; 1 live; 1 not probed.
- datahttps://clinicaltrials.gov/ct2/show/NCT02912559LIVEHTTP 200Resolves, but the content could not be matched to the paper.
- datahttps://acknowledgments.alliancefound.orgUNVERIFIEDHTTP 403Liveness indeterminate — content not checked.
Copyediting
5 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 5 minor suggestions below.
5 copyedit issues flagged: mostly consistency, clarity, typo.
- MINORconsistencyAbstract, Results“86.3% (95% CI, 81.8 to 89.9)”→ Ensure the upper CI bound is consistent with the Results section (89.8 vs 89.9).The abstract reports 89.9 while the Results section reports 89.8; this is a minor inconsistency.
- MINORtypoResults, Safety“mFOLFOX arm”→ Change to 'mFOLFOX6 arm' for consistency.In the sentence about grade 5 events, 'mFOLFOX arm' is used instead of 'mFOLFOX6 arm'.
- MINORclarityMethods, Trial Design and Treatment“Patients were randomized (1:1), via permuted block design with block size of 4, within 10 weeks after surgery to atezolizumab (840 mg intravenously every 2 weeks for 12 cycles) (6 months) plus modified (m) FOLFOX6 (folinic acid, fluorouracil, and oxaliplatin) followed by atezolizumab monotherapy for 13 cycles (12 months total) versus mFOLFOX6 [(5-fluorouracil 400 mg/m 2 bolus and 2400 mg/m 2 continuous infusion over 46 hours), oxaliplatin (85 mg/m 2 ), and leucovorin (400 mg/m 2 )] alone for 12 cycles (6 months total).”→ Break the sentence into shorter sentences for readability.The sentence is long and complex; splitting it would improve clarity.
- MINORconsistencyAbstract, Results“86.3% (95% CI, 81.8 to 89.9)”→ Ensure the upper CI value matches the Results section (89.8 vs 89.9).The abstract reports 89.9 while the Results section reports 89.8; likely a typo.
- MINORclarityMethods, Statistical Analysis“The interim efficacy two-sided boundary is 0.0006.”→ Clarify whether this is for overall survival or disease-free survival.Ambiguous reference.
The published work is robust and methodologically sound. An informed reader should weigh the minor data availability gap and the small CI inconsistency, but these do not undermine the primary conclusions. No erratum is warranted for the p-value discrepancy, but the authors should consider clarifying data access and the CI bound.
- 1.HIGHdata codeAdd a data availability statement in the Methods or a dedicated section, specifying how de-identified patient data can be accessed (e.g., via the Alliance for Clinical Trials in Oncology data access committee) and any conditions or timeframe.The absence of a data availability statement is a reporting gap that readers and reviewers will expect for a clinical trial.
- 2.HIGHreportingExplicitly state whether the trial was open-label and provide a rationale for not blinding in the Methods (Trial Design and Treatment).Blinding status is not explicitly described, which is a minor but important reporting detail for a randomized trial.
- 3.MEDIUMcopyeditReconcile the 3-year DFS upper CI bound between the Abstract (89.9) and Results (89.8) to ensure consistency.The internal contradiction is a minor but visible inconsistency that could confuse readers.
- 4.MEDIUMcopyeditChange 'mFOLFOX arm' to 'mFOLFOX6 arm' in the Results, Safety section for terminology consistency.Consistent terminology is expected in a clinical trial report.
- 5.MEDIUMreportingMention adherence to CONSORT reporting guidelines explicitly in the Methods or a footnote, and provide the completed checklist as supplementary material.Explicit reporting guideline adherence enhances transparency and is often required by journals.
- 6.MEDIUMdata codeConsider depositing the statistical analysis code (e.g., SAS programs) in a public repository with a DOI to enhance reproducibility.Sharing analysis code supports reproducibility, though it is not mandatory for clinical trials.
- 7.LOWcopyeditBreak the long sentence in Methods, Trial Design and Treatment into shorter sentences for readability.The sentence is complex and could be clearer for readers.
- 8.LOWcopyeditClarify in Methods, Statistical Analysis whether the interim efficacy two-sided boundary of 0.0006 applies to overall survival or disease-free survival.The ambiguity could lead to misinterpretation of the interim analysis design.
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.