MP07-08 THE PLATEAU EFFECT OF NUMBER OF PROSTATE CANCER RISK-ASSOCIATED SINGLE NUCLEOTIDE POLYMORPHISMS USED TO ASSESS GENETIC RISK

Abstract

You have accessJournal of UrologyProstate Cancer: Markers II1 Apr 2016MP07-08 THE PLATEAU EFFECT OF NUMBER OF PROSTATE CANCER RISK-ASSOCIATED SINGLE NUCLEOTIDE POLYMORPHISMS USED TO ASSESS GENETIC RISK Haitao Chen, Vignesh T. Packiam, Brian T. Helfand, Carly A. Conran, S. Lilly Zheng, William B. Isaacs, Charles B. Brendler, and Jianfeng Xu Haitao ChenHaitao Chen More articles by this author , Vignesh T. PackiamVignesh T. Packiam More articles by this author , Brian T. HelfandBrian T. Helfand More articles by this author , Carly A. ConranCarly A. Conran More articles by this author , S. Lilly ZhengS. Lilly Zheng More articles by this author , William B. IsaacsWilliam B. Isaacs More articles by this author , Charles B. BrendlerCharles B. Brendler More articles by this author , and Jianfeng XuJianfeng Xu More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2016.02.2211AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail INTRODUCTION AND OBJECTIVES Over 100 prostate cancer (PCa) risk-associated SNPs have been and continue to be identified through genome-wide association studies (GWAS). We further investigated the impact of increasing the number of PCa risk-associated SNPs used to predict inherited risk via Genetic Risk Scores (GRSs). METHODS Seventy-two PCa risk-associated SNPs identified from previous GWAS were available for calculating GRSs among 1,644 men in the placebo arm of the Reduction by Dutasteride of Prostate Cancer Events (REDUCE) cohort. SNPs were ranked from high to low by their impact on PCa risk based on their odds ratio and allele frequency. They were sequentially included to calculate 72 different GRSs, from the top SNP to top n SNPs, with n from 2 to 72. Performance of these GRSs in discriminating biopsy outcomes was assessed with three tests: area under the receiver operating characteristic curve (AUC), reclassification rate, and Net Reclassification Index (NRI). Reclassification rate is the percent of men reclassified as higher or lower PCa risk between two GRSs, using a GRS cutoff of 1.5. NRI measures the net positive effect of reclassification (PCa correctly reclassified higher and non-PCa correctly reclassified lower). RESULTS There was an initial fast rise in predictive ability of GRSs in discriminating biopsy outcomes, with AUC of 0.51 for the GRS using the top SNP vs AUC of 0.59 for GRS using the top 12 SNPs. The rise slowed afterwards, with AUC of 0.59, 0.60, 0.60, 0.60 and 0.61 for the GRSs using the top 24, 36, 48, 60, and 72 SNPs, respectively. In comparison, the AUC of family history was 0.52. The reclassification rate was high when additional SNPs were initially used to calculate GRSs, but was lower when less impactful SNPs were added (Table 1). PCa risk was reclassified in 11.7% of subjects between GRSs calculated using the top 12 vs 24 SNPs, compared to 3.83% between GRSs calculated using the top 60 vs 72 SNPs. Furthermore, the NRI was modest between different GRSs. The NRI was negative between GRSs of the top 60 vs 72 SNPs, suggesting that risk classification using all 72 SNPs was poorer than using the top 60 SNPs. CONCLUSIONS The 60 most significant PCa risk-associated SNPs identified to date are optimal for assessing genetic risk. Inclusion of additional less impactful SNPs may reduce predictive performance. © 2016FiguresReferencesRelatedDetails Volume 195Issue 4SApril 2016Page: e79 Advertisement Copyright & Permissions© 2016MetricsAuthor Information Haitao Chen More articles by this author Vignesh T. Packiam More articles by this author Brian T. Helfand More articles by this author Carly A. Conran More articles by this author S. Lilly Zheng More articles by this author William B. Isaacs More articles by this author Charles B. Brendler More articles by this author Jianfeng Xu More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...