A validated strategy for side specific prediction of organ confined prostate cancer: A tool to select for nerve sparing radical prostatectomy

Purpose: Nerve sparing radical prostatectomy for prostate cancer should be restricted to patients who harbor tumors without capsular penetration. To o ur knowledge the selection criteria for nerve sparing radical prostatectomy are not clearly defined. We investigated a panel of preoperative tumor cha racteristics with respect to their ability to predict organ confined tumor growth for each lobe of the prostate to indicate unilateral or bilateral ne rve sparing radical prostatectomy. Materials and Methods: Nine preoperative tumor characteristics in 278 patie nts with clinically localized prostate cancer were included in retrospectiv e univariate and multivariate tree structured regression analysis. The asso ciation of clinical stage, serum prostate specific antigen (PSA), PSA densi ty, and results of transrectal ultrasound and systematic sextant biopsy, in cluding a quantitative assessment of cancer in the biopsies with organ conf ined tumor growth, was statistically evaluated. Except for serum PSA and PS A density preoperative characteristics were considered separately for each prostate lobe. Multivariate analysis results were validated prospectively i n 353 patients. Results: On univariate analysis the number of positive biopsies was the mos t useful single parameter with a positive predictive value of 83% in 274 lo bes and a negative predictive value of 55%, followed by mm. of tumor in the biopsy. Of all characteristics included in multivariate analysis only the number of biopsies with high grade cancer, the number of positive biopsies and serum PSA were independent for predicting organ confined cancer. When P SA was less than 10 ng./ml. and not more than 1 biopsy with high grade canc er was identified in a lobe, organ confined tumor growth was present in 86. 1% of cases. On prospective validation the same criteria led to an 88.5% in cidence of organ confined prostate cancer. Pooling the 2 most favorable gro ups led to 391 prostate lobes (70.8% of those investigated) with a positive predictive value of 82.1% (95% confidence interval 77.9% to 85.8%). Using the multivariate approach more prostate lobes were assigned to a favorable risk group than on univariate analysis. Clinical stage and simple Gleason g rade did not contribute independent information for predicting organ confin ed disease. Conclusions: Quantifying cancer and high grade cancer by systematic biopsy and serum PSA concentration are useful preoperative characteristics for pre dicting organ confined prostate cancer. Side specific analysis of these par ameters is a flexible and reliable tool for selecting patients for nerve sp aring radical prostatectomy.