Back to 2011 Program

Introduction
Patients with localized prostate cancer undergoing radical prostatectomy (RP), external radiation therapy (XRT), or brachytherapy (BT) are often concerned about erectile function (EF) following treatment. We developed models for individualized sexual outcome expectations following prostate cancer treatment at academic centers and sought to validate them in a community-based cohort.
Methods
The PROSTQA cohort was utilized to create models predicting the likelihood of EF at 2 years following therapy for localized prostate cancer with RP, XRT, or BT (N=1027), based on pre-treatment patient, disease, treatment and HRQOL characteristics.
CaPSURE participants (N=1931), were used for validation by AUC from fitting univariable logistic regression of reported 2-year EF on model-predicted probability, and calibration by examining model-predicted probability vs. observed EF at 2 years.
Results
The PROSTQA models performed well in predicting EF at 2-years following treatment with AUC's of 0.76, 0.81, and 0.89 for men undergoing RP, XRT, and BT, respectively. Calibration showed that predicted rates of EF based on the PROSTQA-derived models corresponded to the observed outcome in the CaPSURE cohort across a broad range of predicted probabilities. (Table 1)
Conclusion
Validation in a community-based cohort of predictive models for recovery of EF following treatment of localized prostate cancer with RP, XRT, and BT at academic centers based on pretreatment EF and various patient and treatment characteristics suggest that these models are generalizable.