Association of BRCA alteration (alt) type with real-world (RW) outcomes to PARP inhibitors (PARPi) in patients (pts) with metastatic castrate-resistant prostate cancer (mCRPC)

5527 Background: Inactivating alts in BRCA1/2 result in homologous recombination deficiency and are predictive of PARPi response in mCRPC. BRCA reversion mutations, which restore the protein function, are frequently observed in acquired resistance to PARPi. In tumors harboring homozygous gene deletions ( BRCAdel) reversions cannot develop; thus, we hypothesize that BRCAdel pts may have prolonged benefit from PARPi compared to pts harboring other BRCA alterations. Methods: Pts were included from the Flatiron Health (FH)-Foundation Medicine (FMI) de-identified clinico-genomic database (CGDB). Inclusion criteria were diagnosis of mCRPC, treatment in the FH network and an FMI comprehensive genomic profiling result between 1/1/2011 - 9/30/2019. Time to therapy discontinuation (TTD) and overall survival (OS) from start of PARPi were estimated with Kaplan-Meier analysis and unadjusted/adjusted (age at PARPi initiation, line number, practice type) hazard ratios (HR/aHR) from Cox proportional hazards models adjusted for survival bias. Results: Out of 829 mCRPC cases, BRCA1/2 alts were detected in 15 (1.8%) and 71 (8.6%) respectively, with 2 cases included in both groups. 26% of BRCAalts were BRCAdel, 67% were coding mutations, and 7% were genomic rearrangements. 25 (28%) BRCAalt pts were treated with PARPi, 11/25 in the 1st or 2nd line setting including 43% of BRCAdel and 44% of other BRCAalt cohorts. Median age at PARPi initiation was 70 yrs and 88% were treated in community practices. TTD was significantly longer in the BRCAdel (n = 7) cohort vs. other BRCAalt cohort (n = 18) (22.7 vs. 9.2 months; HR: 0.16 [0.03-0.74]; aHR: 0.13 [0.02 – 0.92]) while a statistically nonsignificant difference in median OS was observed (31.5 vs. 11.9 months; HR: 0.20 [0.02-1.58]; aHR: 0.24 [0.02-3.15]). In comparison, no statistically significant difference in TTD was observed for BRCAdel (n= 7) vs. other BRCAalts (n=19) pts treated with 1st line hormonal therapies (abiraterone or enzalutamide) (3.4 vs. 5.7 months; HR: 1.16 [0.45-2.98]; aHR: 0.72 [0.25-2.10]). Follow up analysis with more pts and somatic/germline status and zygosity of BRCAalts will be presented. Conclusions: These data suggest a differential benefit from PARPi therapy across BRCAalt subgroups. This observation may in part be explained by the inability to develop reversion mutations to restore BRCA function in tumors with BRCAdel. Further studies are warranted to fully assess the association of BRCAalt type with outcomes to PARPi-based treatments.