Reply to J. Zhu et al

Abstract

We thank Zhu and Zheng for their interest and comments on our study. As indicated in our publication, we included heterogeneous stages of prostate cancer among and within studies. However, results of these trials were comparable with regard to time to development of castration-resistant disease, time off treatment in the intermittent strategy, quality of life, and survival. Furthermore, we believe that presence of metastasis versus rising PSA reflects more a difference between overt and occult metastasis (extent of disease), thereby allowing us to fit one recommendation to all. As indicated by Zhu and Zheng, most trials showed increased nonprostate cancer–related deaths with continuous androgen deprivation therapy (CAD) and increased prostate cancer–related deaths with intermittent androgen deprivation therapy (IAD). Although we agree that pre-existing comorbidity can influence toxicity of an intervention, none of the studies were powered to evaluate subgroups, and the randomized nature of the included studies should have distributed any comorbidity between the groups. Also, the baseline level of PSA in the study by Salonen et al, although numerically different between the groups, has a wide standard deviation around the mean, warranting cautious interpretation (116.0 173.4 ng/mL in the IAD group and 186.3 454.4 ng/mL in the CAD group). We do not agree that we should interpret the results of the two large trials by Crook et al and Hussein et al as having a trend to favor one strategy versus the other: the hazard ratio for overall survival with IAD compared with CAD was 1.03 for the first study and 1.10 for the second, with 95% CIs encompassing unity. Nine randomized controlled trials, three of them of good quality and four of fair quality according to standard prespecified criteria, have shown no consistent survival advantage of one strategy versus the other, with IAD being more favorable in terms of some quality of life scores, cost and convenience. We disagree that further resources should be spent to address the same question in further randomized controlled trials.