Fig. 3: Subgroup analysis of rPFS.

a,b, Subgroup analysis of rPFS by baseline characteristics (a) and by gene subgroups (b) (assessed by blinded independent central review; HRR-deficient intention-to-treat population). The overall HR for all patients, and by BRCA1/BRCA2 alteration status, was based on a Cox proportional hazards model stratified by the randomization stratification factors. For all other subgroups, the HR was based on an unstratified Cox model with treatment as the only covariate. Data are presented as HRs with two-sided 95% CIs. P values are two sided. The asterisk indicates the inclusion of one patient in each treatment arm who received prior orteronel. †Excludes four patients who did not have HRR gene alterations but were incorrectly randomized to the HRR-deficient population; including these patients resulted in an HR of 0.72 (95% CI, 0.49 to 1.07) for the non-BRCA alterations subgroup. ‡Post hoc exploratory analysis; as this analysis was underpowered, the data are hypothesis-generating and should be interpreted with caution. Gene clustering alteration dominance hierarchy is any BRCA1/BRCA2 alteration (BRCA cluster), then any PALB2 (PALB2 cluster), next any CDK12 (CDK12 cluster), then any ATM (ATM cluster), and finally, any of all other genes (with each patient counted only once). For the single-gene subgroups, only patients bearing alteration(s) in the designated HRR gene and none of the other HRR genes tested are shown, with a prevalence cutoff for display of ≥10 across arms. PS, performance status.