Combined DNA-methylation intensity and clinical risk score to stratify patients for high-grade disease.

Abstract

Listen

Translate article

51 Background: Prostate cancer (PCa) diagnostics remains challenging due to fear of over-diagnosis and overtreatment. Due to low accuracy of PSA too many men are biopsied that do not have a subsequent PCa diagnosis or that have indolent disease. Furthermore, persistent risk factors and fear of missed PCa leads to many unnecessary repeat biopsies. Most prostate tumors have epigenetic DNA-methylation aberrations, which display a field effect that can be observed in normal-appearing surrounding tissue, and that could help alleviate biopsy-sampling errors. Methods: A training cohort of methylation-positive men with a negative index biopsy followed by either a Gleason score (GS) ≥ 7 (n=43) or cancer-negative (n=226) repeat biopsy was evaluated. Using the initial negative biopsy, men were stratified for the likelihood of harboring high-grade PCa focusing on a methylation intensity algorithm involving GSTP1, RASSF1 and APC. This algorithm was validated in a cohort of 102 men, with either a PCa-free (n=20), GS6 (n=46), or GS≥7 (n=36) biopsies. Results: The methylation intensity-based algorithm was developed on PCa-negative index biopsies and optimized to predict the presence of GS≥7 cancer in a repeat biopsy. The methylation intensity was significantly higher in GS≥7 compared to PCa-free repeat biopsies (p<0.001). Men with GS6 PCa detected upon repeat biopsy exhibited intermediate intensities. When combined into one model with clinical risk factors (age, pathology, DRE, PSA), an area under the curve (AUC) of 0.762 was obtained, which was significantly higher than the AUC of PSA (0.574; p=0.004) or the AUC of the clinical risk as calculated by the PCPT risk calculator (0.618; p=0.029). In the validation set, an AUC of 0.818 was obtained, with higher intensities for men with GS≥7 disease compared to men with GS6 PCa (p=0.002). Conclusions: The risk score can identify clinically significant cancer in PCa-negative biopsies and is strongly correlated with the GS of PCa-positive biopsies. The risk score could better stratify men for the need for repeat biopsy and the risk of harboring occult clinically significant PCa. The same algorithm could be used to segregate likely under-graded men from active surveillance candidates.