Background: In the Phase IIIb OReO/ENGOT Ov-38 trial (NCT03106987), maintenance olaparib (O) rechallenge significantly improved progression-free survival (PFS) vs placebo in patients (pts) with platinum-sensitive relapsed ovarian cancer (PSROC) regardless of their BRCA status (Pujade-Lauraine, ESMO 2021). The impact of prognostic factors on this PFS benefit were unknown. Methods: Pts had PSROC, one prior line of PARP inhibitor (PARPi) maintenance and were in response to platinum-based chemotherapy (Cx). Pts were enrolled into two cohorts – BRCA mutated (BRCAm) and non-BRCAm – and randomized to receive maintenance O (300 mg bid) or placebo. Primary endpoint was PFS. Post-hoc individual patient data meta-analysis was used to combine both cohorts using fixed and random effect models, interaction, and stratified tests in the Cox model. Stepwise Cox multivariate model for PFS and logistic regression models were used for late relapse, defined as disease progression occurring >24 weeks after randomization. Randomization stratification criteria and cohort effect were included in all models. Results: This study included 220 pts from the BRCAm (n=112) and non-BRCAm (n=108) cohorts recruited between October 3, 2017 and February 10, 2021. No heterogeneity was detected between cohorts: Cochran’s Q test P=0.53 (fixed and random); interaction test P=0.63. The O effect was consistent, and no significant interactions were observed between subgroups. In the multivariate PFS analyses, the main independent factors for prognosis were CA-125 level, visceral disease (liver, lung, pleura, brain), and treatment arm. These factors were independent predictive factors for late relapse among the 181 evaluable pts with the opportunity of at least 6 months of follow-up (Table). Conclusions: In the OReO/ENGOT Ov-38 trial, CA-125 level and presence of visceral disease at baseline were the best predictors of patient outcome. Maintenance olaparib rechallenge was effective overall regardless of prognostic subgroup. Clinical trial information: NCT03106987.

*Cox Proportional-Hazards Model.