Objectives: A multi-institutional study of 280 radical prostatectomy specimens obtained from three independent academic hospitals was undertaken to validate a nomogram developed for the prediction of specimenconfined protstate cancer after prostatectomy.
Methods: Three preoperative factors - the Gleason score, prostatespecific antigen (PSA) and apical location of positive biopsies - that were identified with a previous logistic regression formula were collected. Links between margin status and preoperative criteria were confirmed by univariate methods. Subsequently, the predictive indexes of positive margins were calculated and compared to the actual margin status in terms of predictive characteristics.
Results: This control series, independent of the initial series that was used to identify the relevant preoperative factors, confirmed that positive apical biopsies(p<0.001), PSA (p<0.005) and the Gleason score (p<0.005) were strongly linked to the occurrence of positive margins. Different cutoff values for the predictive index were compared in a receiver operating characteristic curve. A value of 0.5, similar to the one described in the original series, gave an adequate compromise between sensitivity and specificity with respective values of 68 and 73% and a test accuracy of 72%. In practical terms, it was possible to predict 85% of negative margins, and to delineate two groups with different rates of positive margins (14.5 vs. 50%).
Conclusions: We demonstrated that PSA, the Gleason score and apical biopsy status are cumulative risk factors for positive margins. Risk of positive margins increases when it is not possible to obtain a wide excision of periprostatic fascia, as at the apex. This study substantiates the independent prognostic value of positive preoperative apical biopsies for predicting positive surgical margins.