Background: Preoperative treatment of prostate cancer (PCa) changes morphology of residual tumors so that the Gleason score is no longer valid.
Objective: To codify morphologic features of preoperatively treated PCa and identify potential classifiers predictive of outcome.
Design, setting, and participants: We performed a detailed morphologic evaluation of specimens obtained from 115 patients with high-risk PCa who had preoperative androgen ablation, alone or in combination with chemotherapy.
Measurements: Included hierarchical clustering analysis of morphologic characteristics, associations with other pathologic parameters, and univariate and multivariate analyses in search for associations with disease outcome.
Results and limitations: Based on hierarchical clustering analysis, we categorized pretreated prostate cancer in three morphologically distinct groups: group A, characterized by a predominance of cell clusters, cell cords, and isolated cells; group B tumors, by intact and fused small glands; and group C tumors by any degree of cribriform growth pattern or intraductal tumor spread. Univariate analysis identified associations between this grouping, pathologic tumor stage (p<0.01) and residual tumor volume (p<0.001). Presence of intraductal spread or cribriform pattern in biopsies was associated with group C tumors. The presence of cribriform or intraductal spread morphology and positive surgical margins were stronger predictors of biochemical relapse than pathologic stage on multivariate analysis. The number of specimens evaluated in this study was limited, and a prospective validation is warranted along with molecular studies to validate the proposed morphologic classifier.
Conclusions: If validated, this classification will introduce uniformity in the selection of tissue samples for biomarker studies, facilitate the comparison of trials among different institutions, and may provide a new prognostic tool for preoperatively treated PCa.
Copyright © 2009 European Association of Urology. Published by Elsevier B.V. All rights reserved.