Background Earlier trials on the efficacy of poly (ADP-ribose) polymerase (PARP) inhibitors in platinum-sensitive relapsed ovarian cancer used the hazard ratio (HR) as an efficacy parameter. Objective The present meta-analysis was focused on improving the robustness and clinical interpretability of the efficacy evaluation of PARP inhibitors using the restricted mean survival time (RMST). Methods A search for relevant studies published up to July 31, 2020, was performed in electronic databases to identify eligible trials comparing PARP inhibitors with placebo. The difference in RMST was used as a PARP inhibitor efficacy parameter. Combined differences in RMST with 95% CIs across studies were calculated using a random-effects model. Results Four trials (6 articles) were assessed, including 1079 patients treated with PARP inhibitors and 598 with placebo. The combined RMST differences for up to 360 days (PARP inhibitors minus placebo: point estimate and 95% CI) among all patients and the patients of subgroups with BRCA mutations, homologous recombination-deficient (HRD) carcinoma, and BRCA wild-type carcinoma were 87 days (95% CI = 71, 102), 112 days (95% CI = 96, 129), 99 days (95% CI = 80, 119), and 69 days (95% CI = 47, 92), respectively. The combined RMST differences for up to 660 and 720 days were also larger among patients with BRCA mutations than among those with HRD carcinoma. Conclusion and Relevance Based on using the RMST difference as an alternative measure to the HR, this meta-analysis suggests that PARP inhibitors are the most effective for patients with BRCA mutations, followed by patients with HRD carcinoma.
Extrapolating Parametric Survival Models in Health Technology Assessment Using Model Averaging: A Simulation Study
