Presentation Authors: Daniel Sjoberg, Emily Vertosick, New York, NY, Olle Melander, Anders Dahlin, Lund, Sweden, David Ulmert, Los Angeles, CA, Robert Klein, Weiqiang Li, New York, NY, Zsofia Kote-Jarai, Rosalind Eeles, London, United Kingdom, Andrew Vickers*, Hans Lilja, New York, NY
Introduction: Two recently published polygenic risk scores- Seibert and PRACTICAL- are known to predict a diagnosis of any grade prostate cancer. Given the questionable clinical relevance of this endpoint, and the possibility of confounding due to PSA screening, we assessed whether these scores can aid in the prediction of death from prostate cancer when measured in healthy men in a largely unscreened cohort with up to 20 years of follow-up.
Methods: The study is based on the Malmo Diet and Cancer cohort enrolling 11,506 unscreened men aged 45-73 years during 1991-1996, providing cryopreserved blood at enrollment and followed without PSA screening. We measured PSA in blood obtained at baseline and used GWAS-data to calculate the polygenic risk scores in the blood of 263 men who died of prostate cancer and 859 controls. 146 of the 147 PRACTICAL polygenic risk score SNPs and 46 of the 54 Seibert SNPs were utilized in the risk score calculations. We used logistic regression to assess the association between risk scores and cancer death. We repeated the analysis is men with PSA < 1, PSA â‰¥1, â‰¥2, â‰¥3, and â‰¥4 ng/mL. We also assessed whether the polygenic risk scores were independent predictors of prostate cancer death when accounting for PSA level.
Results: The PRACTICAL polygenic risk score was significantly associated with prostate cancer death univariately (AUC 0.629; p < 0.0001) and after inclusion of PSA (p < 0.0001). However, combining PRACTICAL and PSA did not increase discrimination of PSA alone (AUC 0.786) except in the subgroup of men with PSA < 1 ng/mL (AUC increased from 0.682 to 0.721). Risk reclassification in low PSA was of uncertain clinical relevance: 20-year risk of prostate cancer mortality of 0.2% among all men with PSA < 1 ng/mL increased two-fold to 0.4% for men with PSA < 1 ng/mL and the top quartile of PRACTICAL risk score and 3.5-fold to 0.7% for those in the top decile. The Seibert Hazard polygenic risk score was not significantly associated with prostate cancer death either overall (p = 0.6), or among men with elevated PSA (all p â‰¥ 0.2). In men with PSA < 1, the Siebert Hazard was associated with prostate cancer mortality (p = 0.003), but in the incorrect direction - higher Siebert Hazard scores contributed lower risk of prostate cancer death- suggesting a false positive related to multiple testing.
Conclusions: The PRACTICAL polygenic risk score is significantly associated with prostate cancer mortality but has low discrimination compared to PSA and does not add predictive value to PSA, but the possible value of the PRACTICAL polygenic risk score in the low PSA subgroup should be explored further. The Seibert polygenic risk score does not predict aggressive disease and has therefore has doubtful utility.
Source of Funding: This work was supported in part by the National Institutes of Health/National Cancer Institute (NIH/NCI) with a Cancer Center Support Grant to Memorial Sloan Kettering Cancer Center [P30 CA008748], a SPORE grant in Prostate Cancer to Dr. H. Scher [P50-CA92629], R01 CA175491 to Dr. Klein, the Sidney Kimmel Center for Prostate and Urologic Cancers, David H. Koch through the Prostate Cancer Foundation. This work was also supported in part by the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre Program in UK, the Swedish Cancer Society (CAN 2017/559), and the Swedish Research Council (VR-MH project no. 2016-02974).