Design: randomised, double-blind, placebo-controlled, three-arm phase 3 trial at 224 centres in 22 countries (KEYLYNK-001)
Population: 1,367 patients with BRCA non-mutated stage III-IV epithelial ovarian, primary peritoneal or fallopian tube cancer; median age 61
Primary outcome: progression-free survival, tested first in PD-L1 CPS 10 or higher, then intention-to-treat
Effect: pembrolizumab-olaparib vs control, final analysis HR 0.66 (95% CI 0.53-0.83) in CPS 10 or higher and 0.71 (0.61-0.84) in ITT; pembrolizumab alone HR 0.95 (0.77-1.19; p=0.33); grade 3 or higher treatment-related adverse events 66% vs 51%

KEYLYNK-001 randomised 1,367 patients with stage III-IV epithelial ovarian, primary peritoneal or fallopian tube cancer without a BRCA mutation, across 224 centres in 22 countries, to three arms: pembrolizumab plus chemotherapy followed by pembrolizumab-olaparib maintenance; pembrolizumab plus chemotherapy followed by pembrolizumab alone; or chemotherapy with placebo maintenance.

The pembrolizumab-olaparib arm improved progression-free survival against control, with a final-analysis hazard ratio of 0.66 in the PD-L1 CPS 10 or higher population and 0.71 in the intention-to-treat population, sustained at nearly 50 months of median follow-up.

The three-arm design is what makes this informative, and the second comparison is the one to read carefully. Pembrolizumab without olaparib gave a hazard ratio of 0.95 against control — no benefit at all. Since both active arms contained pembrolizumab, the difference between them is olaparib. The reasonable inference is that a PARP inhibitor delivers benefit in BRCA non-mutated disease, and that adding a checkpoint inhibitor to achieve it is not supported by these data.

The cost of the combination was real: grade 3 or higher treatment-related adverse events in 66% against 51% on control, serious events in 24% against 9%, and four treatment-related deaths in the combination arm.

Read this as evidence for maintenance olaparib in BRCA non-mutated disease, not as evidence for pembrolizumab
Quote the toxicity honestly: two-thirds had grade 3 or higher treatment-related events, a quarter had serious ones
Weigh a progression-free survival gain against that toxicity in a patient who is currently well after chemotherapy
Note PD-L1 CPS stratification; the effect was larger in the CPS 10 or higher group but present in the whole population
Check homologous recombination deficiency status where testing is available, as it shapes the PARP inhibitor conversation

Don't overread it

This is a progression-free survival result. No overall survival benefit is reported, and the three-arm design shows pembrolizumab contributed nothing on its own.

The statistics, in plain English

The primary endpoint was progression-free survival, tested hierarchically in the PD-L1 CPS 10 or higher population first — a design that protects against false positives from multiple testing but means later comparisons depend on earlier ones passing. The pembrolizumab-alone comparison (HR 0.95; 95% CI 0.77-1.19; p=0.33) failed, which is why no formal intention-to-treat test was performed for that arm. Progression-free survival is not overall survival: a hazard ratio of 0.66 tells you scans stay clear longer, not that patients live longer, and ovarian cancer trials have repeatedly shown progression-free gains that do not carry through to survival.