EP795 Searching for the best maintenance therapy in platinum-sensitive recurrent ovarian cancer: bevacizumab or PARP-inhibitors? A network meta-analysis

Introduction/Background Patients (pts) with ovarian cancer experiencing a platinum-sensitive (PS) recurrence are generally re-exposed to platinum agents (PCT). The addition of bevacizumab (BEV) or PARP inhibitors (PARPi) as concomitant and/or maintenance therapy has shown to improve progression free survival (PFS). In the absence of direct comparisons coming from randomized trials (RCTs), we have performed a network meta-analysis to evaluate differences in terms of efficacy between BEV and PARPi in pts with PS recurrent ovarian cancer (rOC), according to BRCA status. Methodology We searched PubMed, Embase and Medline for RCTs involving pts with PS rOC treated with BEV (n=3, 1563 pts) or PARPi (n=5, 1839 pts). Only trials with PFS as primary endpoint were included. Analyses have been done pooling pts who had received PARPi in three groups, according to the available data on BRCA genes status: all comers (AC), BRCA mutated pts (BRCAm) and BRCA wild-type pts (BRCAwt). A frequentist approach has been used with R statistical software. To rank the effect size of treatments, surface under the cumulative ranking value (SUCRA) has been applied. Results In AC pts, PARPi improved PFS compared to BEV (hazard ratio [HR]=0.70, 95% CI 0.54–0.91). In BRCAm pts the gain in PFS for PARPi was even higher compared to BEV (HR=0.46, 95% CI 0.36–0.59). In BRCAwt pts the benefit of PARPi over BEV was not statistically significant (HR=0.87, 95% CI 0.63–1.20) but PARPi had the highest likelihood of being ranked as the best treatment in terms of efficacy according to SUCRA (90% and 60%, respectively for PARPi and BEV). Conclusion According to indirect comparisons, PARPi performed the best for the treatment of PS rOC, especially in BRCAm pts who had not previously received PARPi. BEV could be still an option in BRCAwt pts. Disclosure Fabio Puglisi: Roche, AstraZeneca (honoraria and research founding). No conflict of interest is to be declared for the remaining authors. The authors receive no financial support for this study.