In Reply We thank Dr Baker for his Letter and agree that an optimal way to select men at highest risk for death, as well as for entry onto randomized clinical trials assessing the impact on survival of adding androgen deprivation therapy to novel agents shown to overcome castration resistance, is by combining prostate-specific antigen (PSA) nadir greater than 0.5 ng/mL with other important prostate cancer prognostic factors shown to have independent prognostic significance on multivariable analysis in which the end point is time to death. In our multivariable model that included age and known prostate cancer prognostic factors as covariates, which was appended in the Supplement (available online only), we reported adjusted hazard ratios (AHRs) and showed that increasing age (AHR, 1.08; 95% CI, 1.04-1.66; P < .001) in addition to PSA nadir greater than 0.5 ng/mL (AHR, 1.72; 95% CI, 1.17-2.52; P = .01) was significantly associated with the risk of death.¹ The established prostate cancer prognostic factor² Gleason score 7 and 8 to 10 as compared with 6 was not significant, with P values of .05 and .08, respectively, in the multivariable model; other established prognostic factors such as PSA level and clinical stage were also not significant (P = .24 and .50, respectively). Therefore, when selecting men at the highest risk for death it appears that PSA nadir greater than 0.5 ng/mL identified the most at-risk population and that established prognostic factors such as PSA level, clinical stage, and Gleason score do not add significantly to this risk assessment in men. Regarding age, it is expected that with increasing age survival would shorten but adding a drug that is aimed at killing prostate cancer is unlikely to change the impact of increasing age on the risk of death because the association of increasing age and increased risk of death is likely driven by competing risks.³ The reason we added additional covariates in our model was stated in the Methods section of our study: “because men were not stratified by comorbidity prior to randomization, we also included age and known PC [prostate cancer] prognostic factors—specifically PSA, Gleason score, and T stage—in the adjusted model because we only evaluated the subset of 157 men with no or minimal comorbidity, given that PSA failure has been shown to be associated with an increased risk of ACM [all-cause mortality] only in men with no or minimal comorbidity.⁴”^5(p654) As a result, our conclusion was that PSA nadir greater than 0.5 ng/mL “appears” to be a surrogate. This is the reason why we suggested the use of PSA greater than 0.5 ng/mL only as a patient selection factor to identify men for future randomized clinical trials who are at high risk for death if they are treated in accordance with current standards of practice, and not for immediate use as a surrogate end point for death without further validation.

在回复中，我们感谢贝克博士的来信，并同意选择死亡风险最高的男性的最佳方式，以及进入随机临床试验，评估在新药物中添加雄激素剥夺疗法对生存的影响，以克服去势抗性是通过将大于 0.5 ng/mL 的前列腺特异性抗原 (PSA) 最低点与其他重要的前列腺癌预后因素相结合，这些因素在多变量分析中显示具有独立的预后意义，其中终点是死亡时间。在我们的多变量模型中，包括年龄和已知的前列腺癌预后因素作为协变量，该模型附加在附录中（仅在线提供），我们报告了调整后的风险比（AHR）并显示年龄增加（AHR，1.08；95% CI， 1.04-1.66；P < .001) 除了 PSA 最低值大于 0.5 ng/mL (AHR, 1.72; 95% CI, 1.17-2.52; P  = .01) 与死亡风险显着相关。¹已确立的前列腺癌预后因素² Gleason 评分 7 和 8 至 10 与 6 相比不显着，在多变量模型中P值分别为 0.05 和 0.08；其他已确定的预后因素，如 PSA 水平和临床分期也不显着（P = .24 和 .50，分别）。因此，在选择死亡风险最高的男性时，似乎 PSA 最低值大于 0.5 ng/mL 确定了风险最高的人群，而 PSA 水平、临床分期和 Gleason 评分等既定预后因素并未显着增加男性的这种风险评估。关于年龄，预计随着年龄的增长，生存期会缩短，但添加一种旨在杀死前列腺癌的药物不太可能改变年龄增长对死亡风险的影响，因为年龄增长和死亡风险增加之间的关联是可能是由竞争风险驱动的。³我们在模型中添加额外协变量的原因在我们研究的方法部分中说明：“因为男性在随机化之前没有按合并症分层，我们还包括年龄和已知的 PC [前列腺癌] 预后因素——特别是 PSA、格里森评分, 和 T 阶段——在调整后的模型中，因为我们只评估了 157 名没有或最小合并症的男性子集，因为 PSA 失败已被证明与 ACM [全因死亡率] 风险增加仅在患有没有或最小的合并症。⁴英寸^{5 (p654)}因此，我们的结论是，大于 0.5 ng/mL 的 PSA 最低值“似乎”是一个替代指标。这就是为什么我们建议仅使用大于 0.5 ng/mL 的 PSA 作为患者选择因素来确定未来随机临床试验中如果按照当前实践标准接受治疗有高死亡风险的男性，并且未经进一步验证，不得立即用作死亡的替代终点。