Prediction of Survival of Metastatic Prostate Cancer Based on Early Serial Measurements of Prostate Specific Antigen and Alkaline Phosphatase

Abstract

We determined how serial measurements of prostate specific antigen and alkaline phosphatase can be used to predict survival early in the course of hormone treated metastatic prostate cancer. The study was based on a prospective randomized trial of 915 patients with metastatic prostate carcinoma designed to compare parenteral estrogen (polyestradiol phosphate) vs total androgen blockade. We included 697 men who survived at least 6 months and had complete serial measurements of prostate specific antigen and alkaline phosphatase. Six models were constructed based on prostate specific antigen and alkaline phosphatase at start, and after 6 months of treatment, alkaline phosphatase flare and relative prostate specific antigen velocity. We constructed time dependent receiver operating characteristic curves with corresponding area under the curve to predict death from prostate cancer within 3 years. The best variables to predict outcome were alkaline phosphatase at 6 months (AUC 0.79 for polyestradiol phosphate and 0.72 for total androgen blockade), alkaline phosphatase at baseline (AUC 0.70 for polyestradiol phosphate and total androgen blockade) and prostate specific antigen at 6 months (AUC 0.70 for polyestradiol phosphate and total androgen blockade). Prostate specific antigen and alkaline phosphatase levels 6 months after start of treatment give better prediction of survival than baseline levels. Alkaline phosphatase at start of treatment and alkaline phosphatase and prostate specific antigen after 6 months can be used to predict survival of hormone treated metastatic prostate cancer.