Three linked nomograms for predicting biochemical failure in prostate cancer treated with radiotherapy plus androgen deprivation therapy.

Abstract

Nomograms were established to predict biochemical recurrence (BCR) after radiotherapy (RT) with a low weight of the characteristic variables of RT and androgen deprivation therapy (ADT). Our aim is to provide a new stratified tool for predicting BCR at 4 and 7 years in patients treated using RT with radical intent. A retrospective, nonrandomized analysis was performed on 5044 prostate cancer (PCa) patients with median age 70 years, who received RT-with or without ADT-between November 1992 and May 2007. Median follow-up was 5.5 years. BCR was defined as a rise in serum prostate-specific antigen (PSA) of 2 ng/ml over the post-treatment PSA nadir. Univariate association between predictor variables and BCR was assessed by the log-rank test, and three linked nomograms were created for multivariate prognosis of BCR-free survival. Each nomogram corresponds to a category of the Gleason score-either 6,7, or 8-10-and all of them were created from a single proportional hazards regression model stratified also by months of ADT (0, 1-6, 7-12, 13-24, 25-36, 36-60). The performance of this model was analyzed by calibration, discrimination, and clinical utility. Initial PSA, clinical stage, and RT dose were significant variables (p < 0.01). The model showed a good calibration. The concordance probability was 0.779, improving those obtained with other nomograms (0.587, 0.571, 0.554) in the database. Survival curves showed best clinical utility in a comparison with National Comprehensive Cancer Network (NCCN) risk groups. For each Gleason score category, the nomogram provides information on the benefit of adding ADT to a specific RT dose.