Epigenetic risk score improves prostate cancer risk assessment

Abstract

Background: Early detection of aggressive prostate cancer (PCa) remains crucial for effective treatment of patients. However, PCa screening remains controversial due to a high rate of overdiagnosis and overtreatment. To better reconcile both objectives, more effective methods for assessing disease severity at the time of diagnosis are needed.

Methods: The relationship between DNA-methylation and high-grade PCa was examined in a cohort of 102 prospectively enrolled men who received standard 12-core prostate biopsies. EpiScore, an algorithm that quantifies the relative DNA methylation intensities of GSTP1, RASSF1, and APC in prostate biopsy tissue, was evaluated as a method to compensate for biopsy under-sampling and improve risk stratification at the time of diagnosis.

Results: DNA-methylation intensities of GSTP1, RASSF1, and APC were higher in biopsy cores from men diagnosed with GS ≥ 7 cancer compared to men with diagnosed GS 6 disease. This was confirmed by EpiScore, which was significantly higher for subjects with high-grade biopsies and higher NCCN risk categories (both P < 0.001). In patients diagnosed with GS ≥ 7, increased levels of DNA-methylation were present, not only in the high-grade biopsy cores, but also in other cores with no or low-grade disease (P < 0.001). By combining EpiScore with traditional clinical risk factors into a logistic regression model, the prediction of high GS reached an AUC of 0.82 (95%CI: 0.73-0.91) with EpiScore, DRE, and atypical histological findings as most important contributors.

Conclusions: In men diagnosed with PCa, DNA-methylation profiling can detect under-sampled high-risk PCa in prostate biopsy specimens through a field effect. Predictive accuracy increased when EpiScore was combined with other clinical risk factors. These results suggest that EpiScore could aid in the detection of occult high-grade disease at the time of diagnosis, thereby improving the selection of candidates for Active Surveillance.

Keywords: Gleason grade; epigenetic; logistic regression model; prognosis; prostate cancer; risk score.