Unveiling genetic variant-level biomarkers for aggressive prostate cancer

Abstract

Prostate cancer (PCa) represents the second most frequently diagnosed malignancy among males in the United States and ranks fourth in terms of general cancer prevalence on a global scale. A critical assessment of existing literature indicates a notable deficiency in the identification of biomarkers that are uniquely associated with aggressive forms of PCa. The principal objective of this paper is to discover biomarkers at the genetic variant level by deploying statistical methodologies to determine associations between such variants and the aggressive and lethal form of the disease. Employing the multiple comparisons technique, we identified four variants that were statistically significant at the 5 % significance level. Furthermore, we utilized Over-representation analysis (ORA) to identify the biological pathways linked with these genetic variants. To validate our findings, we employed a decision tree algorithm on an independent dataset comparing the proposed biomarkers with random subsets of variants. Results have shown that the predictive accuracy of aggressive samples was 97 % for the proposed biomarkers, while this figure dropped to 67 % when randomly selected variants were considered.