Preoperative parameters for predicting early prostate cancer recurrence after radical prostatectomy

Objectives. To determine whether easily measurable prostate biopsy features could complement Gleason score, prostate-specific antigen (PSA), or clinical stage in predicting recurrence-free survival after prostatectomy, Information relating preoperative parameters to recurrence-free survival is needed to counsel patients with newly diagnosed prostate cancer regarding expectations for post prostatectomy cancer control. Methods. The data of a cohort of 588 consecutive prostatectomy patients (median age 61 years, range 39 to 83) with ascertained preoperative data and up to 4 years of postprostatectomy follow-up were analyzed, Bivariate and multivariate Cox proportional hazards analysis evaluated preoperative factors (clinical stage, PSA, biopsy Gleason score, greatest percentage of a biopsy core involved by cancer [GPC], number of biopsy cores containing cancer, perineural invasion) to identify those relating significantly to recurrence-free survival. Functional forms of these factors were evaluated to optimize accuracy in predicting cancer control. Results. The baseline parameters significantly affecting PSA-free survival included PSA level (P <0.01), biopsy Gleason score (P = 0.04), and GPC (P <0.01). Although clinical stage and perineural invasion had a marginal association with PSA-free survival as univariate factors, this association was not independently significant in multivariable analysis. The multivariate Cox model using PSA, Gleason score, and GPC was highly predictive of PSA free-survival (chi-square = 48.2, P = 0.0001). A set of plots representing these data can be used to identify the risk of early postoperative PSA recurrence on the basis of specific preoperative PSA, Gleason score, and GPC values. Conclusions. These findings provide a highly significant model and a simple tool for assisting preoperative patient counseling regarding predicted cancer control after radical prostatectomy. (C) 2002, Elsevier Science Inc.