First-line nivolumab plus chemotherapy versus chemotherapy alone for advanced gastric, gastro-oesophageal junction, and oesophageal adenocarcinoma (CheckMate 649): a randomised, open-label, phase 3 trial
Source: View publication →
In patients with previously untreated, non-HER2-positive advanced gastroesophageal adenocarcinoma, the addition of first-line nivolumab to chemotherapy significantly improved overall and progression-free survival compared to chemotherapy alone, particularly in those with a PD-L1 combined positive score (CPS) ≥ 5.
Key Findings
Study Design
Study Limitations
Clinical Significance
CheckMate 649 established a new first-line standard of care for HER2-negative advanced gastric, gastroesophageal junction, and esophageal adenocarcinomas, particularly for tumors with a PD-L1 CPS ≥ 5. By breaking the historic 1-year overall survival barrier in this setting, it led to global regulatory approvals and fundamentally shifted the treatment paradigm to incorporate upfront immune checkpoint inhibition.
Historical Context
Historically, advanced gastric and esophageal adenocarcinomas have carried a dismal prognosis, with standard first-line platinum and fluoropyrimidine-based chemotherapy consistently yielding a median overall survival of under one year. Prior to CheckMate 649, immune checkpoint inhibitors had shown efficacy in later-line settings (e.g., ATTRACTION-2, KEYNOTE-059), but initial attempts to move them into the first line (such as KEYNOTE-062) yielded mixed or negative results regarding superiority over chemotherapy alone. CheckMate 649 was a landmark breakthrough, providing definitive phase 3 evidence that the addition of a PD-1 inhibitor to standard chemotherapy improves survival in a biomarker-enriched population.
Guided Discussion
High-yield insights from every perspective
How does the mechanism of action of nivolumab complement standard cytotoxic chemotherapy in advanced gastroesophageal adenocarcinoma, and what does the Combined Positive Score (CPS) measure compared to the traditional Tumor Proportion Score (TPS)?
Key Response
Nivolumab is a PD-1 inhibitor that blocks the inhibitory signal on T-cells, restoring anti-tumor immunity. Chemotherapy can cause tumor cell death and antigen release, potentially synergizing with immunotherapy by turning a 'cold' tumor 'hot'. CPS evaluates PD-L1 expression on both tumor cells and tumor-infiltrating immune cells (lymphocytes, macrophages), which is a more accurate predictive biomarker for immunotherapy response in gastroesophageal cancers than TPS, which only counts PD-L1 expression on tumor cells.
A patient is newly diagnosed with metastatic HER2-negative gastric adenocarcinoma. Based on the CheckMate 649 trial, what specific biomarker must be evaluated prior to initiating first-line therapy, and how would a PD-L1 CPS of 3 versus a CPS of 6 practically alter your management discussion?
Key Response
PD-L1 CPS must be evaluated alongside mismatch repair (MMR) status. CheckMate 649 established Nivolumab plus chemotherapy as the standard of care with the clearest survival benefit for patients with a PD-L1 CPS of 5 or greater. For a CPS of 6, adding nivolumab is strongly indicated. For a CPS of 3, the absolute benefit of adding immunotherapy is less clear and highly debated, requiring a nuanced discussion with the patient about the potential lack of survival benefit versus the added risk of immune-related adverse events, despite broad FDA all-comer approvals.
The FDA approved nivolumab plus chemotherapy for all patients with advanced gastroesophageal adenocarcinoma regardless of PD-L1 expression, despite the primary endpoint in CheckMate 649 focusing on CPS of 5 or greater. How do you reconcile the all-comer regulatory approval with the subgroup hazard ratios, and how does this impact your prescribing practice for a patient with a PD-L1 CPS of 0?
Key Response
While the all-randomized population showed a statistically significant OS benefit, forest plot subgroup analyses suggest this was heavily driven by the CPS 5 or greater cohort. In patients with CPS less than 1 or CPS 1-4, the hazard ratios for OS cross 1.0 or show marginal benefit. Fellows must learn to weigh regulatory approval against clinical evidence, often choosing to spare CPS 0 patients the toxicity and financial burden of nivolumab, reserving it for those with higher likelihood of response.
CheckMate 649 cemented a new standard of care, but it lacked a biomarker-stratified step-down statistical hierarchy to formally isolate and test the CPS less than 5 subgroups. As an attending, how do you navigate the shared decision-making process regarding the added risk of permanent immune-related adverse events versus the theoretical potential for long-term tail-of-the-curve survival in patients with low PD-L1 expression?
Key Response
The attending must balance the lack of definitive, isolated survival benefit in low PD-L1 expressors against the rare but real risk of life-altering irAEs like permanent endocrinopathies or severe pneumonitis. Shared decision-making requires framing the 'tail of the curve' hope against the reality that the overwhelming majority of the survival benefit in this trial was concentrated in the CPS 5 or greater population, preventing overtreatment in a palliative setting.
Scholarly Review
Critical appraisal through the lens of expert reviewers and guideline development
The CheckMate 649 trial employed a hierarchical testing strategy that evaluated the CPS 5 or greater population first, followed by CPS 1 or greater, and finally all-randomized patients. From a methodological standpoint, what are the limitations of this nested hierarchical approach in determining the true efficacy of the intervention in the CPS less than 5 subgroup, and how could an alternative trial design have better addressed this?
Key Response
The hierarchical testing structure successfully controls type I error but limits the ability to formally test the isolated, mutually exclusive subgroup of CPS less than 5. Because the all-randomized population includes the highly responsive CPS 5 or greater group, it mathematically dilutes the effect and masks the potential lack of efficacy in the lower expressors. A study design stratifying entirely by discrete CPS cohorts and powering specifically to detect an interaction effect would have provided a clearer estimate of the true treatment effect across the biomarker continuum.
As a peer reviewer for this manuscript, how does the open-label design of CheckMate 649 potentially introduce bias into the progression-free survival (PFS) endpoint and the utilization of subsequent lines of therapy, and does this methodological choice threaten the internal validity of the overall survival (OS) benefit observed?
Key Response
Open-label trials can bias investigator-assessed PFS because knowledge of the assigned treatment arm might influence the timing of scans or the subjective interpretation of borderline disease progression. Furthermore, control arm patients aware they are not receiving immunotherapy might seek it out off-protocol or as subsequent therapy, potentially confounding OS. A critical reviewer must scrutinize the rates of subsequent immunotherapy cross-over in the control arm and rely more heavily on blinded independent central review (BICR) to ensure the OS and PFS benefits are robust.
While the FDA granted a broad approval, NCCN and ESMO guidelines diverge slightly on the strictness of the CPS threshold for offering nivolumab based on CheckMate 649. How does this trial's evidence inform the strength of recommendation for nivolumab in CPS 5 or greater versus CPS 1-4, and how should international guidelines harmonize these differing interpretations of the all-comer data?
Key Response
NCCN guidelines include nivolumab plus chemotherapy as a Category 1 recommendation for HER2-negative patients with CPS 5 or greater, but a lower Category 2B for CPS less than 5. ESMO is similarly restrictive, prioritizing the CPS 5 or greater group for strong recommendations. The committee must weigh the statistically positive all-comer trial outcome against exploratory subgroup data indicating the OS benefit is driven almost entirely by CPS 5 or greater. Guidelines should clearly delineate the level of evidence to promote value-based care and prevent the over-extrapolation of positive trial data into biologically unselected populations.
Clinical Landscape
Noteworthy Related Trials
ToGA Trial
Tested
Trastuzumab plus chemotherapy
Population
Patients with HER2-positive advanced gastric or GEJ cancer
Comparator
Chemotherapy alone
Endpoint
Overall survival (OS)
KEYNOTE-062 Trial
Tested
Pembrolizumab with or without chemotherapy
Population
Patients with untreated, advanced gastric or GEJ adenocarcinoma and PD-L1 CPS >= 1
Comparator
Chemotherapy plus placebo
Endpoint
Overall survival (OS) and progression-free survival (PFS)
ATTRACTION-4 Trial
Tested
Nivolumab plus chemotherapy
Population
Asian patients with previously untreated, HER2-negative, advanced gastric or GEJ cancer
Comparator
Placebo plus chemotherapy
Endpoint
Progression-free survival (PFS) and Overall survival (OS)
Tailored to your role
Want this tailored to you?
Add your specialty or training stage to get role-specific takeaways and more questions.
Personalize this analysis