Deciphering epigenetic regulation of enhancers in high-risk prostate cancer

Abstract

Prostate cancer (PCa) is associated with widespread promoter hypermethylation. We hypothesized that aberrant DNA methylation also targets gene enhancers, modulating their activity and contributing to disease etiology. A patient discovery set (n = 37) was used for differential methylation analysis and biomarker identification using the Infinium methylation EPIC array, on high-risk (n = 13), low-risk (n = 11), and histologically benign (n = 13) tissues. Enhancers were primarily hypermethylated. However proportionally, hypomethylated enhancers were more prominent in high-risk (n = 385, 15%) than low-risk (n = 105, 10%) disease, primarily targeting genes involved in development and enriched for oncoprotein binding motifs, including FOXA1. The clinical significance of enhancer methylation was evaluated by identifying a 17 enhancer differentially methylated probe (DMP) signature using a Least Absolute Shrinkage and Selection Operator model in the discovery set. A large external dataset (n = 746) obtained from four publicly available prostate tissue methylation array studies was used to assess the enhancer signature through logistic regression models trained on a 2/3 training set and tested in a 1/3 test set. This delivered an area under the curve of 0.81 (95% bootstrapped CI 0.78–0.9) for selective detection of high-risk PCa, achieving a 0.71 sensitivity and 0.76 specificity. Array-wide aberrant DNA methylation at enhancers highlighted their epigenetic perturbance in high-risk disease. A clinically significant enhancer signature from this study could be used for detecting high-risk PCa.