Abstract 6143: Association of polygenic risk score and prostate tumor biomarkers with lethal prostate cancer

Abstract Prostate cancer (PCa) is a highly heritable cancer. While polygenic risk scores (PRS) stratify overall PCa risk among cancer-free men, they have not been reported to predict lethal outcomes in PCa patients. We aimed to examine to what extent PRS modifies associations between tumor biomarkers with cancer prognosis after diagnosis. We included men with primary, non-metastatic PCa in the Health Professionals Follow-up Study (HPFS) and Physicians’ Health Study (PHS). The PRS was constructed with 451 risk variants identified from prior PCa GWAS, measured on DNA from blood or buccal, standardized by adjusting for mean and standard deviation (SD) based on men without PCa with array-based genotyping data in HPFS/PHS. Prostate tumor biomarkers of molecular subtypes (TMPRSS2: ERG fusion, PTEN loss, and TP53 missense mutations from validated immunohistochemistry assays), insulin/lipid signaling pathways (FASN, IR, and IGF1-R), and cellular proliferation and tumor microenvironment (Ki-67, apoptosis rates, and angiogenesis) were measured on tumor tissue from radical prostatectomy or TURMP. Sample sizes varied by tumor biomarkers, ranging from 214 for angiogenesis markers to 598 for ERG protein expression. We first used either logistic or negative binomial regression to evaluate the association between the PRS and the protein expression (levels) of each individual biomarker, and then evaluated the addition of multiplicative interaction terms between PRS and biomarkers to logistic regression models for lethal PCa, defined as metastases or death from PCa. Lastly, we evaluated the predictive performance for lethal PCa by area under the curve (AUC). All models adjusted for age at diagnosis and Gleason score. Among all patients included in the analysis, the median age at diagnosis was 66 (interquartile range: 62-70) years, with 20% having a Gleason score≥8 and 10% having lethal PCa. PRS showed no association with lethal PCa (OR per SD increase PRS=0. 95, 95%CI 0. 85-1. 06) or most tumor biomarkers, except for the apoptosis index (per SD in PRS corresponding to 0. 22 units increase, 95%CI 0. 02-0. 42). Statistically significant interactions were found between PRS and p53 overexpression (p interaction = 1. 6x10-5), PTEN loss (p=3. 7x10-3), FASN (p=0. 03), and Ki-67 (p=1. 6x10-3), suggesting that compared to men with a PRS higher than the median, TP53 missense mutation, PTEN loss, higher FASN intensity, and higher Ki-67 indices were more strongly associated with lethal PCa in patients with lower PRS. Incorporating PRS with tumor biomarkers into predictive models enhanced the AUC beyond a model with tumor biomarker and age. The largest improvements were observed when combining the PRS with Ki-67 and apoptosis, showing absolute increases of 5% and 7% respectively. In conclusion, PRS integration with tumor biomarkers at diagnosis potentially improves the prediction of lethal PCa, thereby informing more targeted management strategies. Citation Format: Anqi Wang, Anna Plym, Konrad H. Stopsack, Lorelei A. Mucci, Kathryn L. Penney. Association of polygenic risk score and prostate tumor biomarkers with lethal prostate cancer 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 (6Suppl): Abstract nr 6143.