Correlation between radiographic progression-free survival (rPFS) and overall survival (OS): Results from PREVAIL.
Morris, MJ; Beer, TM; Loriot, Y; Higano, CS; Armstrong, AJ; Sternberg, CN; De Bono, JS; Tombal, BF; Parli, T; Bhattacharya, S; Krivoshik, AP ...
Published in: Journal of Clinical Oncology
182 Background: In PREVAIL, enzalutamide (ENZA) significantly improved OS (hazard ratio [HR] 0.71; P < 0.0001) and rPFS (HR 0.19; P < 0.0001) compared with placebo in chemotherapy-naïve metastatic castration-resistant prostate cancer (mCRPC). Using data from the final analysis at 439 rPFS events we showed a strong correlation between investigator and central review. Here we report sensitivity analyses (SAs) from the final analysis of investigator-assessed rPFS and the association between investigator-assessed rPFS and OS. Methods: The coprimary endpoint of rPFS was defined as time from randomization to the earliest objective evidence of centrally assessed radiographic progression, defined by PCWG2 guidelines for bone disease and RECIST v1.1 for soft-tissue disease, or death within 168 days of treatment discontinuation. Bone progression was captured using a validated bone scan data capture assay. Four SAs were performed on investigator-assessed rPFS to evaluate the impact of: (1) progression in bone, (2) clinical progression, (3) a required confirmatory scan for soft-tissue disease progression, and (4) all deaths, regardless of length of time after study drug discontinuation. Associations of investigator-assessed rPFS and OS were calculated using Pearson’s correlation coefficient, Spearman’s rho, and Spearman’s rho estimated through the Clayton copula. Results: Treatment effects remained significant with each SA, with HRs of (1) 0.21 (95% confidence interval [CI] 0.18, 0.27), (2) 0.21 (95% CI 0.17, 0.26), (3) 0.23 (95% CI 0.19, 0.30), and (4) 0.23 (95% CI 0.19, 0.30) (P < 0.0001 for each). Results of rPFS and OS associations are presented in the table below. Conclusions: SAs of rPFS in PREVAIL demonstrated a robust and consistent treatment benefit with ENZA. We observed a modest association between rPFS and OS. However, some unaccounted factors, such as post-protocol treatment, could have reduced the strength of the association of rPFS and OS in the placebo arm. Clinical trial information: NCT01212991. [Table: see text]
