Cribriform carcinoma of the prostate and cribriform prostatic intraepithelial neoplasia: incidence and clinical implications.

Cribriform neoplasia of the prostate can be recognized easily. However, controversy persists regarding terminology, particularly with the intraductal spread of cribriform neoplasia; some consider this "intraductal carcinoma," whereas consensus meetings defined these lesions as high-grade cribriform prostatic intraepithelial neoplasia (HGCP). This study attempts to identify the incidence and clinical significance of HGCP and cribriform carcinoma (CC) by evaluating 114 radical prostatectomy specimens. Cases were divided into three histologic groups for statistical analysis: (1) pure acinar carcinoma: infiltrating acinar carcinoma without evidence of cribriform neoplasia; (2) CC: acinar carcinoma with CC; and (3) HGCP: acinar carcinoma with HGCP. High-grade cribriform prostatic intraepithelial neoplasia was defined as the presence of neoplastic cells spanning the entire lumen in a cribriform configuration in which a basal cell layer could be shown by immunohistochemistry. Similar areas in which no basal cell layer could be seen were diagnosed as CC. The incidence of cribriform neoplasia was 38% (43 of 114). The incidences of HGCP and CC were 13% (15 of 114) and 25% (28 of 114), respectively. Univariate analysis showed a strong association between HGCP and CC both and several preoperative and final pathology results, including digital rectal examination, pathology tumor stage, extraprostatic extension, surgical margin positivity, high Gleason sum (GS), and high tumor volume. Kaplan-Meier analysis showed HGCP to have a 61% cumulative prostate-specific antigen (PSA) failure rate in contrast with CC and pure acinar cancer, which had cumulative PSA failure rates of 15% and 13%, respectively (p = 0.0001, log-rank test). Multivariate Cox's proportional-hazards analysis found preoperative serum PSA, GS, tumor stage, and volume to be important predictors of PSA failure. In a second regression model that included serum PSA, GS, and pathology tumor stage, HGCP was an independent predictor of PSA failure. Both HGCP and CC are closely associated with several poor prognostic indicators, including advanced pathology tumor stage, a high GS, and serum PSA. Multivariate analysis showed HGCP as an independent prognostic indicator. The close association between high tumor volume and HGCP supports the theory that the development of HGCP is a late event in tumor progression, more compatible with the intraductal spread of tumor than dysplasia.