Introduction

Current prognostic classification of prostate cancer (PCa) is based on Gleason scoring, which is subjective & lacks precision. We sought to identify discriminant methylated DNA markers with potential to enhance prognostic assessment.

Methods

For discovery, whole methylome sequencing was applied to DNA extracted from 54 frozen archival tissues [36 PCa cases (18 Gleason 3+3, 18 Gleason ≥7) and 18 normal-appearing prostate controls]. For tissue validation, top candidate markers were assayed by methylation-specific PCR on an independent set of 50 case and 35 control tissues. Tissues were obtained from radical prostatectomies done between 2003-2007; all were reviewed pathologically and micro-dissected prior to DNA extraction and bisulfite treatment. Marker levels were standardized to total human DNA. From the discovery set, marker distributions in cases and controls were assessed to identify discriminant candidates. Markers were correlated with progression (PSA >0.4 ng/ml or confirmed recurrence) using regression partitioning trees (rPart) in PCa cases. From the rPart model, subjects were categorized as low, medium, and high risk progression groups. The prognostic value of methylation markers relative to Gleason scoring was assessed with the likelihood ratio test of competing cox proportional hazards models.

Results

In the discovery set, 120 candidate markers exceeded filtering criteria (AUC>0.95, fold-change >10, p<0.005). Of these, the top 72 candidate markers were evaluated in the independent set. Numerous markers were identified that highly discriminated PCa from normal-appearing prostate controls, with some achieving AUCs > 0.99. Five prognostic markers (FAM78A, WNT3A, GAS6, LOC100129726, and MAXchr727) were selected by the rPart modeling. The risk grouping defined by methylated DNA markers added significant prognostic content in predicting progression-free survival relative to Gleason scoring (p<0.0001) whereas Gleason scoring had no added value relative to methylated DNA marker risk grouping (p=0.43) (figure)_x000D_

Conclusions

Based on these discovery and early validation data, we identify novel methylated DNA markers with potential to accurately predict progression in PCa. Further exploration is clearly warranted to corroborate and extend these intriguing findings.

Funding

Mayo Foundation