A Model That Predicts the Probability of Positive Imaging in Prostate Cancer Cases With Biochemical Failure After Initial Definitive Local Therapy

Purpose: Managing biochemical failure in patients following initial treatment of localized prostate cancer is a relatively common clinical problem. Imaging studies to document metastatic disease are frequently obtained but are often uninformative. In this study we identified clinical parameters that were predictive of positive imaging studies. Materials and Methods: From CaPSURE, a national disease registry, all patients with a detectable prostate specific antigen after definitive therapy with radical prostatectomy or radiation therapy and who had undergone at least 1 imaging study (bone scan, computerized tomography or magnetic resonance imaging of the abdomen and pelvis) were identified. Patient characteristics, trigger prostate specific antigen (prostate specific antigen before the imaging), prostate specific antigen doubling time and velocity prior to imaging for association with a positive imaging test were analyzed. The results were incorporated into a predictive model. Results: We identified 292 patients (66% radical prostatectomy and 34% radiation therapy) who had recurrence and had available imaging data, and 31 (11%) patients had a positive imaging study. On multivariate analysis age, imaging type, trigger prostate specific antigen and prostate specific antigen doubling time were significantly associated with imaging results. A multivariate model including age (younger than 60 vs 60 to 69 vs 70 years or older), primary imaging type (bone scan vs computerized tomography vs magnetic resonance imaging), trigger prostate specific antigen (5 or less vs more than 5 ng/ml) and prostate specific antigen doubling time (less than 10 vs 10 or more months) had a concordance index of 84% in predicting positive imaging. Conclusions: Age, imaging type, trigger prostate specific antigen and prostate specific antigen doubling time were significantly associated with imaging results. Imaging studies are unlikely to be useful when trigger prostate specific antigen is 5 or less ng/ml and prostate specific antigen doubling time is 10 or more months.