Correlation of a Commercial Genomic Risk Classifier with Histological Patterns in Prostate Cancer

RNA expression based molecular testing has the potential to improve clinical decision making as an adjunct to histopathological interpretation of prostate cancer biopsies. The GPS (Oncotype Dx Genomic Prostate Score®) assay has been proposed as a predictor of more severe pathology at prostatectomy but its true clinical value is uncertain. We hypothesized that some of the predictive usefulness of this assay relates to its correlation with histopathological features which are apparent but not typically reported on prostate biopsies.

In this retrospective, single center cohort we determined an RNA based GPS of prostate biopsies. We retrospectively reviewed the histopathological features of biopsy cores with the score and assessed tumor length, Gleason pattern 4 amount and type, and stromal reaction type. Associations between the GPS and histopathological features were assessed by linear mixed models.

From May 2013 to August 2015 a GPS was determined in 319 biopsies in a total of 296 patients. Of the types of Gleason pattern 4 the expansile cribriform, simple cribriform, poorly formed and fused patterns were associated with a higher GPS. The expansile cribriform pattern had the strongest association. The glomerulation pattern was associated with a lower GPS and an increasing stromal reaction also positively correlated with the GPS. A model incorporating these pathological features accounted for 36.9% of the variation in the score.

The stromal reaction and the type of Gleason pattern 4 are histopathological features which are not typically reported for prostate biopsies but they correlate with the GPS. These data suggest that more detailed analysis of prostate histopathology might substitute for some of the information gained from this molecular diagnostic assay.