A real-world clinicogenomic study of androgen receptor mutations and co-mutations in prostate cancer: Clinical implications.

73 Background: While most prostate cancer (PCa) patients (pts) initially respond to androgen deprivation therapy, resistance commonly develops, resulting in castration-resistant PCa. In certain patients, androgen receptor ( AR ) mutations permit tumor cells to proliferate despite the absence of circulating androgens. This study leverages real-world clinicogenomic data to investigate the prevalence and clinical implications of AR mutations and co-mutations. Methods: We analyzed 3,753 pt records with PCa who underwent next-generation sequencing (NGS) from January 1, 2014 to October 1, 2024. NGS was ordered by >400 physicians from >180 community-based oncology clinics across the Sarah Cannon Research Institute network, and results were collated with pt medical records in the software platform, Genospace. Pts were stratified into AR -mutated ( AR m; 306 pts), AR -amplified ( AR amp; 231 pts), and AR wild-type ( AR WT; 3,216 pts) cohorts. Pts with multiple primary cancers were excluded. Results: AR mutations were detected in 306 pts (8%), with a higher detection rate in plasma- versus tissue-based NGS (15% vs 2%; p<0.01). AR mutations were more common in older pts, with 68% of AR m pts >70 years old compared to 53% of AR WT pts (p<0.01). The most common AR mutations included L702H (51%, 156 pts), T878A (38%, 115 pts), H875Y (23%, 70 pts) and W742C (8%, 25 pts), and over half of AR m pts reported multiple AR mutations (51%, 156 pts). Several genes were more frequently mutated in AR m versus AR WT pts, including TP53 (43% vs 32%), PTEN (10% vs 6%), BRCA2 (9% vs 4%), RB1 (5% vs 2%), and BRCA1 (4% vs 1%; p<0.01 for all). AR m pts reported a higher frequency of microsatellite instability (MSI) compared to AR WT pts (11% vs 2%; p<0.01). Within the AR m cohort, MSI was more frequently detected by tissue- versus plasma-based NGS (26% vs 7%; p<0.01). Specific AR mutations were enriched in MSI versus microsatellite stable (MSS) pts, including H875Y (50% vs 20%) and V716M (31% vs 2%; p<0.01 for both). Additionally, BRCA1/2 mutations were more common in AR m/MSI pts versus AR m/MSS pts ( BRCA1 : 15% vs 3%; BRCA2 : 35% vs 6%; p<0.01 for both). Pts in both the AR m (3.3 yrs) and AR amp (2.1 yrs) cohorts exhibited shorter overall survival from metastatic staging compared to the AR WT cohort (9.0 yrs; p<0.01 for both). Neither specific AR mutations nor MSI status correlated with survival outcomes. Conclusions: AR mutations are frequently detected by plasma-based NGS in the real-world setting, possibly due to test selection at disease progression and/or sampling of tumor heterogeneity. However, plasma-based NGS may miss other critical alterations, such as MSI status. The high co-occurrence of MSI in AR -mutated PCa suggests potential for combination therapies involving immunotherapy and novel AR degraders. These findings emphasize the need for tissue- and plasma-based genomic profiling to guide therapeutic decisions and optimize treatment strategies for advanced PCa.