Change in serum prostate-specific antigen as a marker of response to cytotoxic therapy for hormone-refractory prostate cancer.

PURPOSE: Prostate-specific antigen (PSA) has been used as a marker of advanced prostate cancer but remains controversial. To evaluate PSA as a predictor of survival, we analyzed data from sequential phase II trials of estramustine and etoposide. METHODS: A landmark analysis that used data from 62 men with PSA levels at baseline and 8 weeks was conducted. The best PSA measure (of six evaluated) was incorporated into a multiple regression model with performance status (PS); relative change in PSA level; and pretreatment PSA, alkaline phosphatase, and hemoglobin values. RESULTS: A decrease in PSA of 50% or greater at 8 weeks was associated with a significantly increased survival (P=.0005, two-sided log-rank test). Median survival from the landmark was 91 weeks in patients with a 50% or greater decrease at 8 weeks versus 38 weeks in those without this decrease. Modeling showed that PS, pretreatment hemoglobin level, and relative change in PSA level were significant prognostic factors, with a significant interaction between PS and pretreatment hemoglobin level. In the final model, a relative change in PSA level at 8 weeks of less than 50% had an adjusted relative risk of 2.20 (95% confidence interval, 1.21 to 4.00). A decrease in PSA level of 50% or greater at any time during therapy was associated with a response in measurable disease (P=.0369, two-sided Fisher's exact test). CONCLUSION: The PSA value after 8 weeks of this cytotoxic regimen does predict survival. A decrease in PSA level is associated with both survival and response in soft tissue lesions and should be incorporated into the response criteria and reporting of trials of cytotoxic agents in prostate cancer.