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

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.

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.

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.

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.