Development of Nomogram to Non-steroidal Antiandrogen Sequential Alternation in Prostate Cancer for Predictive Model

Abstract

To clarify clinical predictors for a prostate-specific antigen decrease ≥50% in response to alternative non-steroidal antiandrogen therapy and to develop a nomogram to predict the prostate-specific antigen decrease ≥50% in response to alternative non-steroidal antiandrogen therapy in patients with advanced prostate cancer that relapsed after initial combined androgen blockade. We previously reported that combined androgen blockade with an alternative non-steroidal antiandrogen is effective for advanced prostate cancer that has relapsed after initial combined androgen blockade. We enrolled 161 patients from 14 medical institutions with histologically confirmed prostate cancer who had been treated with combination therapy and in whom cancer progressed after first-line combined androgen blockade therapy. A nomogram for the prostate-specific antigen decrease ≥50% from baseline prostate-specific antigen in response to alternative non-steroidal antiandrogen therapy was developed based on the final logistic regression model. Overall prostate-specific antigen decreased ≥50% in 75 of 161 patients (46.6%) in response to alternative non-steroidal antiandrogen therapy. Using five independent risk factors (initial serum level of prostate-specific antigen, hemoglobin, C-reactive protein, prostate-specific antigen nadir to second hormone therapy and Gleason sum), a nomogram was developed for the prediction of prostate-specific antigen decrease ≥50% in response to alternative non-steroidal antiandrogen therapy. The receiver operating characteristic curve showed that the accuracy of the predicted probability was 72.5% for the model. This predictive nomogram could predict the prostate-specific antigen decrease ≥50% in response to alternative non-steroidal antiandrogen therapy and might be of benefit to determine the sequential treatment strategy in patients with relapse after first combined androgen blockade.