Objectives: The presence of positive surgical margins (PSMs) after prostatectomy for prostate cancer has long been an indicator of poor survival outcomes. However, with the downstaging of cancer occurring in the prostate-specific antigen testing era, we sought to determine whether the risk associated with PSMs retains the same effect on prognosis as before the prostate-specific antigen testing era.
Methods: Of the 3460 patients in the Columbia University Urologic Oncology database, 2215 (64%) were identified who had undergone radical prostatectomy from 1991 to 2005 and had sufficient pathologic data to be analyzed and >or=1 year of follow-up. Three epochs were chosen: 1991-1995, 1996-2000, and 2001-2005.
Results: The median age, preoperative prostate-specific antigen, and Gleason score was 61.6 years, 6 ng/mL, and 7, respectively, and >50% of patients had pathologic Stage T2 disease. On multivariate analysis, PSMs were a risk factor for biochemical failure for each epoch (P < .01). The Wald's test indicated that the significance of PSMs had not changed over time (P = .8). The contribution of PSMs to the accuracy of predicting biochemical failure in a multivariate model was found only for the earliest epoch, because it improved the model by 0.15 (95% confidence interval 0.03-0.27). In the second epoch, it was 0.13 (95% confidence interval -0.01 to 0.27), and it was 0.13 for the third (95% confidence interval -0.06 to 0.32).
Conclusions: The results of this study suggest that the predictive contribution of PSMs to the accuracy of a multivariate model or nomogram used to predict the outcomes after prostatectomy has decreased during the past 15 years.