COMBI-I missed its endpoint and then reported a survival difference

Design

Randomised, placebo-controlled phase 3 trial; final overall survival analysis after at least five years

Population

532 patients with BRAF V600-mutant unresectable or metastatic melanoma (267 vs 265)

Primary outcome

Progression-free survival at 24 months (not met); overall survival reported here

Effect

Median overall survival 61.5 months vs 41.6 months, HR 0.760 (95% CI 0.598-0.966); grade 3 or worse treatment-related events 57.3% vs 36.7%

COMBI-I randomised 267 patients with BRAF V600-mutant unresectable or metastatic melanoma to spartalizumab plus dabrafenib and trametinib, and 265 to placebo plus dabrafenib and trametinib. It failed its primary endpoint of progression-free survival at 24 months. This final analysis reports overall survival after a median 76.9 months of follow-up, the trial having closed in August 2024.

Median overall survival was 61.5 months (95% CI 41.6 to not evaluable) with spartalizumab against 41.6 months (95% CI 30.6-56.9) without, a hazard ratio of 0.760 (95% CI 0.598-0.966). The toxicity difference is substantial: pyrexia in 65.9% against 46.2%, and grade 3 or worse treatment-related adverse events in 57.3% against 36.7%.

How to read a survival benefit that arrives after a missed primary endpoint is the whole question. It is not a positive trial. The hazard ratio's upper bound of 0.966 sits close to 1, the confidence interval on the experimental arm's median runs to 'not evaluable', and adding a checkpoint inhibitor to targeted therapy raised severe toxicity by more than twenty percentage points. For dermatologists this is context for a conversation with oncology colleagues and with patients who have read about the result, rather than a finding that changes referral or management.

• Confirm BRAF V600 status is documented before any discussion of targeted combination therapy
• Note pyrexia as the dominant treatment-related event when counselling a patient starting dabrafenib and trametinib
• A missed primary endpoint constrains how much weight a later survival analysis can carry
• Grade 3 or worse events in over half of the combination arm belongs in any summary of the benefit
• Melanoma systemic therapy decisions sit with oncology; this is background for the shared conversation

Why it matters

It tests whether a survival readout can rehabilitate a trial that failed its primary endpoint - and the toxicity makes that a real question.

Don't overread it

The trial did not meet its primary endpoint; this overall survival analysis is hypothesis-generating.

The statistics, in plain English

A hazard ratio of 0.760 with an interval reaching 0.966 means the data are compatible with a benefit as small as about 3%, and that estimate comes from a trial whose pre-specified primary question was answered negatively. Once a primary endpoint is missed, later analyses are no longer protected against the multiplicity of everything else that was measured, so they generate hypotheses rather than conclusions. The experimental arm's median overall survival of 61.5 months with an upper confidence bound of 'not evaluable' means too few events had accrued to pin the estimate down at the top end.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.