Clinical implications of prostate specific antigen (PSA) flare and response dynamics in TALAPRO-2: A post-hoc analysis.

168 Background: In the phase 3 TALAPRO-2 study, talazoparib (TALA) + enzalutamide (ENZA) significantly improved radiographic progression-free survival (rPFS) and overall survival (OS) vs placebo (PBO) + ENZA as first-line treatment (tx) in patients (pts) with metastatic castration-resistant prostate cancer (mCRPC) unselected (cohort 1) and selected for homologous recombination repair gene alterations (HRRm cohort). Here, we report post-hoc analyses of PSA kinetics from the final OS analysis cutoff date. Methods: Pts were randomized 1:1 to ENZA 160 mg + either TALA 0.5 mg or PBO once daily. The primary endpoint was rPFS; OS, a key secondary endpoint. We performed post hoc analyses of PSA flare (defined as a rise in PSA above baseline followed by a return to baseline or lower within 12 months), PSA response ≥50% (PSA50), PSA response ≥90% (PSA90), undetectable PSA (defined as PSA <0.2 ng/mL), and duration of PSA response. OS and rPFS were analyzed in PSA50 responders. Results: At data cutoff (Sep 3, 2024), PSA flare was observed in 2–3% of pts in both tx arms of the unselected and HRRm cohorts (Table). The median time for PSA to decline back to or below baseline after a rise above baseline was 57 days in both tx arms in the unselected cohort, and 60 and 117 days with TALA + ENZA and PBO + ENZA, respectively, in the HRRm cohort (Table). A greater proportion of pts taking TALA + ENZA achieved PSA50, PSA90, and undetectable PSA compared with PBO + ENZA in both cohorts (Table). In both cohorts, pts in the TALA + ENZA arm also had a longer median duration of PSA50 response, PSA90 response, and undetectable PSA vs PBO + ENZA (Table). In the unselected cohort, rPFS and OS directionally favored TALA + ENZA vs PBO + ENZA in PSA50 responders (rPFS HR: 0.72 [95% CI 0.57, 0.89], median 39.8 vs 27.6 months; OS HR: 0.90 [0.72, 1.13], median 53.3 vs 47.3 months). Similarly, in the HRRm cohort, rPFS and OS favored TALA + ENZA in PSA50 responders (rPFS HR: 0.51 [0.38, 0.70], median 36.1 vs 18.3 months; OS HR: 0.62 [0.44, 0.86], median not reached vs 38.2 months). Conclusions: In both cohorts, a greater proportion of pts in the TALA + ENZA arm achieved PSA responses with longer durations vs PBO + ENZA. Also, PSA flare was unusual (2–3%) and resolved in ~60 days (both cohorts) with TALA + ENZA vs ~60 days (unselected) and ~120 days (HRRm) with PBO + ENZA. In both cohorts, OS and rPFS numerically favored TALA + ENZA vs PBO + ENZA in PSA50 responders. Analyses of PSA responses for specific genes (e.g. BRCA2 ) are underway. Clinical trial information: NCT03395197 . Unselected Cohort HRRm Cohort TALA + ENZA (n=402) PBO + ENZA (n=403) TALA + ENZA (n=200) PBO + ENZA (n=199) PSA flare, n (%) 12 (3) 11 (3) 5 (2) 5 (3) Median time for PSA decline back to or below baseline, days 57 57 60 117 PSA50, n (%) 335 (83) 285 (71) 173 (86) 126 (63) Median duration, days 430 284 420 225 PSA90, n (%) 253 (63) 182 (45) 139 (70) 80 (40) Median duration, days 480 419 589 308 Undetectable PSA, n (%) 159 (40) 107 (27) 87 (44) 44 (22) Median duration, days 644 539 650 365