TRIDENT ML signature flags KRAS/STK11-mutant, EGFR/FGFR3/CDKN2A–WT non-squamous NSCLC for bigger tremelimumab add-on OS benefit

TRIDENT is a post hoc machine-learning analysis of the Phase 3 POSEIDON trial (NCT03164616; data cut-off March 12, 2021) evaluating whether adding tremelimumab (anti–CTLA-4) to durvalumab (anti–PD-L1) plus platinum-doublet chemotherapy better selects patients with metastatic non-squamous NSCLC for overall survival (OS) gain. Using clinical + genomic inputs, the top-ranked 50% “predicted-benefit” subgroup had improved OS with tremelimumab added (HR 0.56, 95% CI 0.33–0.97); key features associated with greater benefit included EGFR wild-type, FGFR3 wild-type, CDKN2A wild-type, plus KRAS and STK11 mutations (i.e., a profile often associated with immunotherapy resistance in PD-(L)1+chemo regimens). For context, the overall POSEIDON non-squamous subgroup benefit for tremelimumab+durvalumab+chemo vs chemo alone was more modest (HR ~0.70), and the closest SOC “CTLA-4 add-on” benchmark in this space (CheckMate 9LA: nivo/ipi+2-cycle chemo vs chemo) reported OS HR ~0.69—so TRIDENT is essentially proposing a biomarker-enriched slice where the CTLA-4 increment might be meaningfully larger than the class-average effect.