Intermediate end point for prostate cancer-specific mortality following salvage hormonal therapy for prostate-specific antigen failure: D’Amico AV, Moul JW, Carroll PR, Cote K, Sun L, Lubeck D, Renshaw AA, Loffredo M, Chen MH, Department of Radiation Oncology, Brigham and Women’s Hospital and Dana-Farber Cancer Institute, Boston, MA. J Natl Cancer Inst 2004;96:509–15

Abstract

Whether the prostate-specific antigen (PSA) response to salvage hormonal therapy can act as an intermediate end point for prostate cancer-specific mortality (PCSM) remains unclear. Therefore, we evaluated whether PSA response, defined as the absolute value of the ratio of the rate of PSA change after salvage hormonal therapy to the rate of PSA change before salvage therapy, is associated with the time to PCSM following salvage hormonal therapy. A single-institution and two pooled multi-institution databases containing baseline, treatment, and follow-up information on men who received salvage hormonal therapy for PSA failure following surgery or radiation therapy from January 1, 1988, to January 1, 2002, formed the study (n = 199) and validation cohorts (n = 1255), respectively. The ability of PSA response and its constituents (i.e., pre-salvage hormonal therapy PSA slope and post-salvage hormonal therapy PSA slope) to predict time to PCSM following salvage hormonal therapy was assessed using Cox regression analysis. For illustrative purposes, PSA response was analyzed as a dichotomous variable with a breakpoint for the ratio of PSA response of 1. All statistical tests were two-sided. PSA response was statistically significantly associated with time to PCSM following salvage hormonal therapy in both the study (P(Cox) = .0014) and validation (P(Cox) < .001) cohorts; however, its constituents were not (pre-salvage hormonal therapy PSA slope: P(Cox-study) = .97, P(Cox-validation) = .57; post-salvage hormonal therapy PSA slope: P(Cox-study) = .27, P(Cox-validation) = .31). Patients with a PSA response that was less than or equal to 1 had a statistically significantly shorter time to PCSM than patients with a PSA response of greater than 1 in both the study (hazard ratio [HR] = 3.6, 95% confidence interval [CI] = 1.3 to 10.3; P(Cox) = .01) and validation (HR = 12.8, 95% CI = 6.2 to 26.3; P(Cox) < .001) cohorts. The PSA response to salvage hormonal therapy can serve as an intermediate end point for PCSM in patients with a rising PSA level following surgery or radiation therapy.