Correlation between histologic features and Decipher genomic risk in prostate cancer biopsies

The Decipher genomic classifier refines risk stratification in prostate cancer, but its incremental value over biopsy pathology remains uncertain. We evaluated the correlation between Decipher scores and clinicopathologic features in diagnostic prostate biopsies. We retrospectively reviewed 40 prostate biopsies with concurrent Decipher testing. Clinical variables (age, PSA, and clinical T stage) and pathologic features (ISUP Grade Group, percentage of Gleason pattern 4, cribriform morphology, tumor length, and perineural invasion) were analyzed. Decipher scores were assessed as continuous and categorical variables (low risk < 0.5; high risk ≥ 0.5). Biopsies were further classified as favorable or unfavorable histology, with unfavorable histology defined by high-risk features including large cribriform architecture (>0.25 mm). Of 40 cases, 13 (32.5%) were high genomic risk and 8 (20.0%) showed unfavorable histology. High-risk cases were significantly enriched for higher Grade Groups (92% vs. 26%), greater median Gleason pattern 4 (25% vs. 0%), and more frequent cribriform morphology (77% vs. 11%) (all p < 0.001). Decipher scores correlated moderately with Gleason pattern 4 percentage (R² = 0.345). Unfavorable histology demonstrated high specificity (96.3%) and positive predictive value (87.5%), but limited sensitivity, identifying 7 of 13 high-risk cases (53.8%). These findings suggest that while adverse histologic features partially predict high genomic risk, genomic testing may identify additional high-risk tumors not captured by morphology alone, supporting its selective use.