Objectives: To assess the value of prostate-specific antigen (PSA) kinetics in predicting survival and relate this to the baseline variables in men with metastatic hormone-refractory prostate cancer (HRPC).
Methods: The data from 417 men with HRPC were included in a logistic regression model that included hemoglobin, PSA, alkaline phosphatase, Soloway score, and performance status pain analgesic score at baseline. The posttreatment variables included the PSA level halving time after the start of treatment, PSA level at nadir, interval to nadir, PSA velocity (PSAV), PSA doubling time after reaching a nadir, patient age, and treatment. These variables were added to the baseline model, forming new logistic regression models that were tested for net reclassification improvement.
Results: The area under the receiver operating characteristics curve for the baseline model was 0.67. Of all variables related to PSA kinetics, the PSAV was the best predictor. The addition of PSAV to the baseline model increased the area under the receiver operating characteristics curve to 0.81. Only a moderate increase in the area under the receiver operating characteristics curve (0.83) was achieved by combining the baseline model in a multivariate model with PSAV, PSA doubling time, interval to nadir, and patient age at diagnosis of HRPC.
Conclusions: The PSAV alone gave a better prediction of survival value than all other PSA kinetics variables. By combining PSAV with the variables available at baseline, a better ground for treatment decision-making in men with HRPC can be achieved.