PURPOSE In the PROpel study (ClinicalTrials.gov identifier: NCT03732820 ), olaparib plus abiraterone demonstrated statistically significant radiographic progression-free survival benefit versus placebo plus abiraterone in a first-line metastatic castration-resistant prostate cancer (mCRPC) population unselected by homologous recombination repair gene mutation (HRRm) status. From PROpel, we report exploratory biomarker analyses assessing HRRm and BRCA1 and/or BRCA2 (BRCAm) status using tumor tissue and plasma-derived circulating tumor (ct) DNA. METHODS Patients received (1:1) either olaparib (300 mg twice a day) or placebo in combination with abiraterone (1,000 mg once daily) plus prednisone/prednisolone (5 mg twice a day). Tumor tissue and ctDNA samples were analyzed using FoundationOne CDx and FoundationOne Liquid CDx tests, respectively. Biomarker results are presented by individual tests and as an aggregate analysis. RESULTS HRRm status was obtained for 778 of 796 (98%) patients randomly assigned. Aggregating tumor tissue and ctDNA results, 226 patients were identified as having HRRm (including 85 BRCAm). In matched tumor tissue and ctDNA samples (n = 491), and using the tumor tissue result as the reference, high concordance was observed for HRRm (overall percent agreement, 85.1%; negative predictive value, 93.7%) and BRCAm status (overall percent agreement, 93.9%; negative predictive value, 97.3%), with low ctDNA fraction being a limiting factor in rare cases where discordance was observed. Based on the observed agreement, there was a potential incidence of 1% unidentified BRCAm in aggregated non-BRCAm patients (n = 693). CONCLUSION Aggregating tumor tissue and ctDNA analyses maximized the number of patients (98%) with known biomarker status in PROpel while limiting the number of potential false negatives. This represents the most comprehensive means for identifying biomarker-positive and biomarker-negative subgroups and demonstrates the complementary utility of ctDNA testing in mCRPC, particularly when tissue testing is unable to provide a clinically useful result.
Armstrong et al. (Fri,) studied this question.