Androgen receptor activity in biopsy specimens at initial diagnosis of prostate cancer and correlation with outcomes and treatment response.

409 Background: Androgen receptor activity (AR-A) has been described after radical prostatectomy (RP) and metastatic castration-sensitive prostate cancer (mCSPC). In RP specimens low AR-A is associated with basal subtypes, decreased DNA repair and increased immune activity. In mCSPC, low AR-A is associated with poor overall survival (OS) and time to progression to CRPC. However, AR-A has not been well characterized in localized disease at initial diagnosis. Here we sought to assess AR-A signatures in biopsy samples from patients across the prostate cancer risk continuum and assess correlations between AR-A and outcomes. Methods: We analyzed 150,162 biopsy samples tested (2016-2024) with the Decipher prostate genomic classifier (Veracyte, Inc. San Diego, CA). Transcriptome-wide expression data and clinical factors were retrieved from the Decipher GRID (NCT02609269). Patients with low and high AR-A expression as defined by Spratt et al 2019 were compared using Chi-square tests. Clinical and pathologic outcomes for specific cohorts in the overall sample population were analyzed using Cox regression. Results: Overall, 11,752 (7.8%) patients had low AR-A expression. 9.8% of patients > 70 years of age had low AR-A compared to 7.6% of patients <70 (p<0.0001). Low AR-A was enriched in samples with poor prognostic clinical factors such as very high Decipher (p<0.0001), Grade Group (GG) 5 (p<0.0001) and very high NCCN risk (p<0.0001). The same differences were present when looking only at patients with a PSA <4 ng/mL. Like in RP samples, AR-A expression in biopsy samples positively correlated with intact DNA repair and negatively correlated with basal subtypes, PORTOS and immune infiltration scores (all p<0.001). Low AR-A was prognostic of poor clinical outcomes across 4 independent retrospective cohorts. In cohort 1, intermediate risk disease (n=647), low AR-A correlated with adverse pathology at time of RP (p <0.05). In cohort 2, intermediate risk disease treated with radiation therapy (RT) (n=121), low AR-A correlated with biochemical failure (p<0.05). In cohort 3, high risk disease (n=405), low AR-A correlated with decreased OS (p<0.05) in all patients and metastasis (p<0.05) after RT and androgen deprivation therapy (ADT). Finally, in cohort 4, high risk disease treated with RT+ADT (n=100), low AR-A correlated with metastasis (p<0.01). Conclusions: Overall, in a large cohort of biopsy specimens, low AR-A was associated with increased age, very high Decipher score, very high NCCN risk, and GG5 disease. Subpopulation analyses suggest that low AR-A portends a poor prognosis. Given that patients with low AR-A had decreased DNA repair activity and increased PORTOS scores, clinicians should consider post-operative RT and novel clinical trials with PARP inhibitors for these patients.