Maximum androgen blockade in advanced prostate cancer: A meta-analysis of published randomized controlled trials using nonsteroidal antiandrogens

Objectives. To assess the survival benefit of maximum androgen blockade (MAB) using nonsteroidal antiandrogens (NSAAs) through meta-analysis of published randomized controlled trials (RCTs). Methods. All RCTs comparing treatment with NSAA plus either luteinizing hormone-releasing hormone (LHRH) or orchiectomy versus treatment with LHRH or orchiectomy alone were included if the necessary statistical summaries were present in the publication. Estimates and standard errors of log hazard ratio for overall survival and progression-free survival were derived from published studies using two methods: (1) reconstructing an annual life table from graphical presentations of survival distributions and fitting discrete proportional hazard models, and (2) reconstructing the log hazard ratio from reported P values and numbers of deaths. An alternative set of log hazard ratios was derived from figures presented in a summary report by the Prostate Cancer Trialists' Collaborative Group (PCTCG). Comparative meta-analyses were performed using the random effects approach of DerSimonian and Laird. Additionally, published studies were used in a random-effects-based meta-analysis of objective tumor response. Results. Nine studies provided enough information to perform a meta-analysis for survival using one of the two methods. Estimates of relative risks (RR) comparing treatment with NSAA plus either LHRH or orchiectomy versus treatment with LHRH or orchiectomy alone with respect to overall survival were 0.78 (95% confidence intervals [Cls] 0.67 to 0.90) using method 1, and 0.84 (95% CI 0.76 to 0.93) using method 2. Sensitivity analyses based on PCTCG data showed that a favorable survival result for MAB was associated with NSAAs but not with steroidal antiandrogens and depended on randomization blinding and overall trial quality. Additionally, random-effects-based meta-analysis of published studies showed a significant increase in time-to-progression (RR = 0.74; 95% CI 0.63 to 0.86) and an increase in objective tumor responses for MAB using NSAAs compared with castration alone (odds ratio = 0.65; 95% CI 0.51 to 0.81; P = 0.00022). Conclusions. Inconsistent results have been published about the benefit of MAB in advanced prostate cancer. This meta-analysis supports a beneficial effect for MAB using NSAAs compared with castration alone, and sensitivity analyses suggest that the design of future trials should carefully address issues of patient characterization, randomization blinding, and other study quality issues. Copyright 1997 by Elsevier Science Inc.