Validation of a nomogram to predict disease progression following salvage radiotherapy after radical prostatectomy: results from the SEARCH database

Abstract

To externally validate the nomogram published by Stephenson et al. (termed the 'Stephenson nomogram') to predict disease progression after salvage radiotherapy (SRT) among patients with prostate cancer from the Shared Equal Access Regional Cancer Hospital (SEARCH) database. We analysed data from 102 men treated with SRT for prostate-specific antigen (PSA) failure after prostatectomy, of whom 30 (29%) developed disease progression after SRT during a median follow-up of 50 months. The predicted 6-year progression-free survival (PFS) was compared to the actuarial PFS using calibration plots. The accuracy of the nomogram to risk-stratify men for progression was assessed by the concordance index. The median PSA and PSA doubling time before SRT was 0.6 ng/mL and 10.3 months, respectively. The 6-year actuarial disease-free progression after SRT was 57% (95% confidence interval 42-69%). The overall concordance index of the Stephenson nomogram was 0.65. The nomogram predicted failure more accurately at the extremes of risk (lowest and highest) but in intermediate groups, the accuracy was less precise. Of the 11 variables used in the nomogram, only negative margins and high PSA level before SRT were significantly associated with increased disease progression. The Stephenson nomogram is an important tool to predict disease progression after SRT following radical prostatectomy. It adequately predicted progression in SEARCH with reasonable accuracy. Also, in SEARCH, disease progression was predicted by similar disease characteristics. However, the overall modest performance of the model in our validation cohort indicates there is still room for improvement in predictive models for disease progression after SRT.