Population-Specific Genetic Markers of Prostate Cancer Risk in Kazakh Men: Association Analysis of 102 SNPs and Risk Prediction Modeling

Abstract

Background/Objectives: GWASs have identified more than 250 prostate cancer (PCa) predisposition loci, predominantly in European and partly Asian cohorts. The Kazakh population is markedly under-represented in international genetic studies, limiting existing risk models. This study aimed to analyze the distribution of 102 PCa-associated single-nucleotide polymorphism (SNP) genotypes and alleles and to identify reliable population-specific associations with PCa risk in Kazakh men. Methods: This retrospective case–control study included 941 Kazakh men (476 with histologically confirmed PCa and 465 cancer-free controls). Genomic DNA extracted from peripheral blood was genotyped with TaqMan® OpenArray® technology on a QuantStudio 12K Flex system. Associations were assessed by Pearson’s χ2 test and logistic regression, with genotypic and allelic odds ratios (OR) and 95% confidence intervals (CI). Two-step multiple-testing correction (Bonferroni and Benjamini–Hochberg false-discovery rate, FDR) was applied. Predictive models were built using classification and regression trees (CART) and stepwise logistic regression. Results: Of 102 SNPs, 39 showed nominally significant genotypic differences; 12 remained significant after Bonferroni correction and 2 after FDR (14 in total). Several of the corrected loci were significant at both the genotypic and allelic level. Allelic ORs ranged from 0.37 (protective rs10187424 T allele) to 4.81 (rs1545985). A parsimonious seven-SNP autosomal logistic-regression model achieved an apparent AuROC of 0.84 (10-fold cross-validated 0.82); adding age as a covariate raised discrimination to 0.87. Ten of the fourteen significant loci remained significant after age adjustment, and six of these formed a core signal robust to both age imbalance and genotyping-quality concerns. Conclusions: This first large-scale SNP-association study in Kazakh men shows allele-frequency profiles resembling East Asian rather than European populations, confirming the need for population-specific genetic risk-assessment tools. The seven-SNP model showed high discriminatory power in the training set and requires external validation before clinical application.