Defining the Most Informative Intermediate Clinical Endpoints for Patients Treated with Salvage Radiotherapy for Prostate-specific Antigen Rise After Radical Prostatectomy

Abstract

Intermediate clinical endpoints (ICEs) might aid in trial design and potentially expedite study results. However, little is known about the most informative ICE for patients receiving salvage radiation therapy (sRT) after radical prostatectomy. To investigate the most informative ICE for patients receiving sRT, we used a multi-institutional database encompassing patients treated at eight tertiary centers. Overall, 1301 men with node-negative disease who had not received any form of androgen deprivation therapy were identified. Associations of biochemical (BCR) and clinical recurrence (CR) within 1, 3, 5, and 7yr after surgery with the risk of overall mortality were evaluated using multivariable Cox regression analyses fitted at the landmark points of 1, 3, 5, and 7yr after sRT. The discriminative ability of each model for predicting overall survival (OS) was assessed using Harrell’s c index. Median follow-up for survivors was 5.6yr (interquartile range 2.0–8.8). On multivariable analysis, progression to CR within 3yr from sRT (hazard ratio 4.19, 95% confidence interval 1.44–11.2; p= 0.008) was the most informative ICE for predicting OS (c index 0.78) compared to CR within 1, 5, and 7yr (c index 0.72, 0.75, and 0.71). In conclusion, progression to CR within 3yr after sRT, irrespective of the time of surgery, was the most informative ICE for prediction of OS. Our study is hypothesis-generating. If these results are confirmed in future prospective studies and surrogacy is met, this information could be applied for study design and could potentially expedite earlier release of results from ongoing randomized controlled trials. Patient summaryClinical recurrence of prostate cancer within 3yr after salvage radiation therapy, irrespective of the time of radical prostatectomy, represents the most informative intermediate clinical endpoint for the prediction of overall survival. This information could be applied in the design of future studies and could potentially expedite earlier release of results from ongoing randomized controlled trials.