UCLA

Background

The randomized phase 3 COMBI-i trial did not meet its primary endpoint of improved progression-free survival (PFS) with spartalizumab plus dabrafenib and trametinib (sparta-DabTram) vs placebo plus dabrafenib and trametinib (placebo-DabTram) in the overall population of patients with unresectable/metastatic BRAF V600-mutant melanoma. This prespecified exploratory biomarker analysis was performed to identify subgroups that may derive greater treatment benefit from sparta-DabTram.

Methods

In COMBI-i (ClinicalTrials.gov, NCT02967692), 532 patients received spartalizumab 400 mg intravenously every 4 weeks plus dabrafenib 150 mg orally two times daily and trametinib 2 mg orally one time daily or placebo-DabTram. Baseline/on-treatment pharmacodynamic markers were assessed via flow cytometry-based immunophenotyping and plasma cytokine profiling. Baseline programmed death ligand 1 (PD-L1) status and T-cell phenotype were assessed via immunohistochemistry; BRAF V600 mutation type, tumor mutational burden (TMB), and circulating tumor DNA (ctDNA) via DNA sequencing; gene expression signatures via RNA sequencing; and CD4⁺/CD8⁺ T-cell ratio via immunophenotyping.

Results

Extensive biomarker analyses were possible in approximately 64% to 90% of the intention-to-treat population, depending on sample availability and assay. Subgroups based on PD-L1 status/TMB or T-cell inflammation did not show significant differences in PFS benefit with sparta-DabTram vs placebo-DabTram, although T-cell inflammation was prognostic across treatment arms. Subgroups defined by BRAF V600K mutation (HR 0.45 (95% CI 0.21 to 0.99)), detectable ctDNA shedding (HR 0.75 (95% CI 0.58 to 0.96)), or CD4⁺/CD8⁺ ratio above median (HR 0.58 (95% CI 0.40 to 0.84)) derived greater PFS benefit with sparta-DabTram vs placebo-DabTram. In a multivariate analysis, ctDNA emerged as strongly prognostic (p=0.007), while its predictive trend did not reach significance; in contrast, CD4⁺/CD8⁺ ratio was strongly predictive (interaction p=0.0131).

Conclusions

These results support the feasibility of large-scale comprehensive biomarker analyses in the context of a global phase 3 study. T-cell inflammation was prognostic but not predictive of sparta-DabTram benefit, as patients with high T-cell inflammation already benefit from targeted therapy alone. Baseline ctDNA shedding also emerged as a strong independent prognostic variable, with predictive trends consistent with established measures of disease burden such as lactate dehydrogenase levels. CD4⁺/CD8⁺ T-cell ratio was significantly predictive of PFS benefit with sparta-DabTram but requires further validation as a biomarker in melanoma. Taken together with previous observations, further study of checkpoint inhibitor plus targeted therapy combination in patients with higher disease burden may be warranted.

Trial registration number

NCT02967692.