Population-adjusted indirect treatment comparison (PAITC) of maintenance PARP inhibitor (PARPi) with or without bevacizumab versus bevacizumab in women with newly diagnosed ovarian cancer (OC).

Abstract

Listen

Translate article

6052 Background: In patients (pts) with newly diagnosed OC, bevacizumab (B), PARPi, and PARPi + B have shown benefit as maintenance treatment options after platinum chemotherapy response. Phase III trials have demonstrated longer median progression-free survival (PFS) with PARPi + B (PAOLA-1, olaparib [O]; NCT02477644) vs placebo (P) + B and with PARPi alone (PRIMA, niraparib [N]; NCT02655016) vs P. As there are no randomized head-to-head trials comparing PARPi + B vs PARPi, or PARPi vs B, we performed indirect treatment comparison across these regimens. Methods: Unanchored PAITC was performed with individual pt data (IPD) from a PAOLA-1 subset comprising pts with stage IV disease, stage III with residual disease after primary surgery, inoperable stage III disease, or any patient who received neoadjuvant chemotherapy. Propensity weights were used to match the baseline (BL) characteristics of the PRIMA population. PRIMA dataset was reconstructed using published PFS curves. Both datasets were pooled; treatment efficacy was assessed by weighted Cox regression and Kaplan–Meier methods. PAITC was performed in all pts (biomarker unselected) and the homologous recombination repair deficiency positive (HRD+; cut-off 42) subgroup. Results: 595/806 (266/387 HRD+) PAOLA-1 pts were included. After matching, the effective sample size (ESS) for PAOLA-1 was 532 (242 HRD+; weights 0.241–2.37). Weighted BL data were balanced across cohorts. Conclusions: In biomarker-unselected and HRD+ pts, PAITC suggests that adding O to B significantly improved PFS vs. N or B alone. In biomarker-unselected pts, PAITC results show no significant difference in PFS between N and B. In HRD+, improved efficacy with N appears to translate into improved PFS vs. B alone, although follow-up was <2 years (14 vs 22 months, respectively). Results are hypothesis generating and could guide randomized trial design. Clinical trial information: NCT02477644 and NCT02655016. [Table: see text]