Abstract 4801: Clinical polygenic model to stratify risk of any, metastatic, and fatal prostate cancer in a randomized controlled trial of precision screening

Abstract Introduction: Lifetime risk of prostate cancer (PCa) is strongly associated with three factors: ancestry, family history, and genetics. We previously developed a polygenic hazard score (PHS) for PCa comprised of 290 common genetic variants (PHS290). When accounting for family history and ancestry (self-reported or genetically determined), PHS290 remained a powerful independent predictor of age at PCa diagnosis, metastatic PCa diagnosis, and PCa death. Additional variants associated with PCa susceptibility or benign elevation of prostate-specific antigen (PSA) have since been published. Here, we evaluate their impact on lifetime risk predictions for metastatic and fatal PCa and develop an updated polygenic score. This score will inform the design of the Prostate Cancer, Genetic Risk, and Equitable Screening Study (ProGRESS, ClinicalTrials.gov ID NCT05926102), a nationwide randomized controlled trial evaluating precision PCa screening in the VA healthcare system. Methods: We utilized a machine learning approach to update the PHS. Over 450 variants associated with PCa, aggressive PCa, benign prostatic hyperplasia, or benign PSA elevation were evaluated for inclusion. First, we identified pairs of variants with highly correlated genotype (defined as R2 > 0.95) and utilized univariable Cox proportional hazards models (using age at diagnosis of PCa) to exclude one variant from each pair. We then evaluated all remaining candidate variants for inclusion in a new PHS using a LASSO-regularized Cox proportional hazards model. We combined the new PHS with clinical predictors (ancestry and family history) to create a comprehensive, quantitative polygenic score in a diverse, population-based cohort of n=590,750 (Million Veteran Program). Both self-reported and genetic ancestry (based on 2,309 ancestry informative markers) were assessed. We estimated hazard ratios (HRs) to compare men with high versus low polygenic score and evaluated absolute predictions of any, metastatic, and fatal PCa. Results: The updated polygenic score included 356 variants, as well as family history and genetic ancestry. When comparing the highest to the lowest quintile of the population for polygenic score, the HRs for any PCa, metastatic PCa, and fatal PCa were 5.6, 5.8, and 5.2 respectively. These HRs are all greater than those achieved with PHS290, which yielded 5.2, 4.9, and 4.4 respectively. Risk stratification improved within each ancestry group, including within the >100,000 participants of African ancestry—a group known to have elevated risk of fatal PCa. Absolute risk prediction was accurate for both metastatic and fatal PCa. Conclusion: The polygenic score accurately predicts risk of metastatic and fatal PCa in multiple ancestry groups. A clinical polygenic model has potential utility for precision PCa screening and will now be evaluated prospectively in the ProGRESS trial. Citation Format: Anna M. Dornisch, Roshan Karunamuni, Kara N. Maxwell, Julie Lynch, Isla P. Garraway, Adam Kibel, Charles A. Brunette, Morgan E. Danowski, Kyung M. Lee, Scott L. DuVall, J M. Gaziano, Brent S. Rose, Richard L. Hauger, Jason L. Vassy, Tyler M. Seibert. Clinical polygenic model to stratify risk of any, metastatic, and fatal prostate cancer in a randomized controlled trial of precision screening [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4801.