Correlation of large cribriform carcinoma and "unfavorable histology" with other Gleason pattern 4 subtypes: A proof-of-principle study evaluating 485 radical prostatectomy specimens with proposal for the concept of "borderline histology".

Prostatic adenocarcinomas with large cribriform glands/intraductal carcinoma (LC/IDC), or the recently proposed unfavorable histology, are associated with adverse outcomes after radical prostatectomy. However, Gleason pattern 4 carcinomas without LC/IDC (or unfavorable histology) have minimal risk for aggressive clinical behavior after prostatectomy. As proof-of-principle study, we collected a cohort of 485 radical prostatectomy specimens to assess correlations between different subtypes of Gleason pattern 4 disease and the presence of adjacent high-risk prostatic adenocarcinoma, defined as LC/IDC or unfavorable histology. All prostatectomies were completely embedded, and all slides re-reviewed to record Gleason score/Grade Group, diameter of the largest cribriform gland (i.e. the longest cross-sectional distance), and all architectural patterns of carcinoma utilizing previously described Canary methodology. The presence and percent of LC/IDC (defined as >0.25 mm) was determined. We also evaluated correlation with the recently proposed "unfavorable histology" as a secondary endpoint. Complex Gleason pattern 4 subtypes, distinct from LC/IDC and unfavorable histology, were termed "borderline histology" and defined as the presence of any of the following patterns: small cribriform/glomeruloid architecture (≤0.25 mm), dominant population of poorly formed glands/small nests, simple glomerulations, and epithelial complexity associated with extravasated mucin (beyond typical mucinous fibroplasia pattern and not containing cribriform >0.25 mm). Comparisons between recorded variables and LC/IDC (or unfavorable histology) utilized the Wilcoxon test for continuous variables and chi-squared test or Fisher's test for categorical variables. Pearson or phi correlation coefficients were used to assess the association between two variables. "Borderline histology" was significantly correlated to LC/IDC (r = 0.55) and unfavorable histology (r = 0.607), both p < 0.001. Specifically, small cribriform/small glomeruloid architecture had the strongest correlation, compared to the other "borderline histology" subtypes (r = 0.646). We demonstrate that "borderline histology" has a strong association with the concomitant presence of high-risk prostate cancer by current histologic definitions (i.e. LC/IDC and unfavorable histology). This proof-of-principle study suggests that large cohort biopsy-RP correlation studies are needed, as the presence of these patterns on biopsy could potentially aid preoperative risk stratification for patients without other high-risk features at initial evaluation.