Evaluating an integrated germline polygenic and clinical risk model (P-CARE) specifically for aggressive prostate cancer.

237 Background: Polygenic scores are strongly associated with age at diagnosis of prostate cancer (PCa) but have not been clearly shown to discriminate between indolent and aggressive PCa. P-CARE (Prostate CAncer integrated Risk Evaluation) is an integrated model that combines a 601-variant polygenic hazard score (PHS601), genetic ancestry, and family history. We hypothesize that among men diagnosed with PCa, those with higher P-CARE scores are more likely to develop metastatic disease. Methods: We analyzed genetic and phenotypic data from a diverse, national cohort of men diagnosed with PCa (Million Veteran Program, n=69,901, 6413 metastatic). We used Cox proportional hazards models to test P-CARE and PHS601 for association with age at onset of metastatic PCa (birth-met analysis). We also tested P-CARE for association with metastasis-free survival after diagnosis of localized PCa, with or without accounting for PSA, stage, and age at diagnosis (localized-met analysis). PCa was detected/treated too late if the patient had definitive treatment for localized disease but developed metastasis. Thus, onset of metastasis was time of first treatment or diagnosis of metastasis, whichever occurred first. Participants who never experienced metastasis were censored at time of first treatment or last follow-up/death.We repeated this analysis within men with African genetic ancestry (n=16,889). Confidence intervals for hazard ratios (HR) were estimated using 100-fold bootstrapping. For external validation of direction of effect, we tested P-CARE for association with age at diagnosis of clinically significant PCa in cohorts from the PRACTICAL Consortium (Cohort of Swedish Men [COSM, n=2163], ProtecT [n=1583]). Results: P-CARE was associated with birth-met and localized-met in the full MVP cohort and among men with African ancestry. In the full cohort, patients in the highest 20% of P-CARE (based on percentiles of men of European ancestry with PCa) compared with the lowest 20% had a HR (HR 80/20 ) of 1.58 [95% CI 1.48–1.68] and 1.32 [1.23–1.40] for birth-met and localized-met, respectively. The corresponding HR 80/20 values for PHS601 alone were 1.29 [1.20–1.37] and 1.16 [1.07–1.25]. When accounting for PSA, stage, and age at diagnosis, P-CARE HR 80/20 was 1.05 [0.99-1.11] for localized-met. Among men with African ancestry, HR 80/20 was 1.43 [1.26–1.62] and 1.17 [1.03–1.34] for birth-met and localized-met, respectively. Within the PRACTICAL cohorts, P-CARE was associated with development of clinically significant PCa (COSM: 1.33 [1.16–1.57], ProtecT: 1.67 [1.34–2.10]). Conclusions: Among men diagnosed with PCa, P-CARE and PHS601 are moderately specific for metastatic disease. We are currently studying P-CARE in ProGRESS (ClinicalTrials.gov ID NCT05926102), a nationwide randomized clinical trial to evaluate precision PCa screening in the VA healthcare system.