Data from 8,275 men in 27 RCTs were included. IPD could not be obtained for 183 participants in four other trials.
The authors reported that the typical duration of follow-up was almost 5 years. Data for cause of death were obtained for only 20 of the 27 trials.
There were 5,932 (72%) deaths overall. Heterogeneity in the treatment effect across all 27 trials was not statistically significant. The absolute difference in survival at 5 years was 1.8% (standard error, SE=1.3; log rank 2P=0.11). The difference was not statistically significant. The difference was also not statistically significant on omission of non-prostate cancer deaths from the analysis (2P=0.06).
There was no significant difference in overall mortality between metastatic and locally advanced disease, or between the age groups less than 65 years, 65 to 74 years, and 75 years and over, or according to whether AS was achieved by orchiectomy or drugs.
Based on 20 trials, 2,778 (80%) of the 3,475 deaths were attributed to prostate cancer. There was a non significant excess of non-prostate cancer deaths among men treated with MAB, but no association was found between this and age, stage, anti-androgen, or years of follow-up.
Trials of nilutamide (8 RCTs, 1,688 men; n adjusted) or flutamide (12 RCTs, 4,803 men) showed an absolute increase in 5-year survival of about 3% with MAB, whereas trials of cyproterone acetate (37 RCTs, 1,784 men; n adjusted) showed a 3% decrease. Some of the excess mortality among men treated with cyproterone acetate was accounted for by an excess of other deaths (i.e. not prostate cancer) in the cyproterone acetate trials, although non-prostate cancer deaths were not clearly significantly different between MAB and AS (2P=0.05).