合成代谢-雄激素类固醇使用、全因死亡率与主要心血管结局：一项系统综述与荟萃分析。
完整摘要 · 中文译文
背景/目的：非医疗用途的合成代谢雄激素类固醇（AAS）使用与不良心脏重构相关，但既往使用者中超额全因死亡率的幅度及心脏表型仍未被充分界定。方法：我们系统综述了报告AAS暴露个体停药后死亡率、主要心血管结局或心脏功能的研究。全因死亡率从采用随机效应模型、个体化确认暴露的对照队列中合并；既往使用者的整体纵向应变（GLS）和左心室射血分数以均数差合并。偏倚风险采用ROBINS-E评估，证据确定性采用GRADE评估。结果：共纳入15篇原始报告。在三项队列中，AAS暴露与较高的全因死亡率相关（风险比2.57，95% CI 1.98-3.33；I2 = 0%；95%预测区间0.47-13.87），敏感性分析结果一致。在单项大型匹配队列中，心血管结局的校正风险比为2.26至8.90，其中心肌病最高。既往使用者的GLS较未使用AAS的对照者更差（3项研究；均数差+0.90个百分点，95% CI +0.13至+1.66），而射血分数的合并估计为探索性且异质性极高。结论：在三项观察性队列中，AAS暴露与较高的全因死亡率相关，但残余混杂、随访差异及证据基础薄弱使得无法进行因果推断。来自单项队列的心血管关联需要独立的重复 并发症情况，以及关于既往使用者残余心肌功能障碍的证据，其确定性非常低。
完整摘要 · 英文原文
Background/Objectives: Non-medical anabolic-androgenic steroid (AAS) use has been associated with adverse cardiac remodelling, but the magnitude of excess all-cause mortality and the cardiac phenotype of former users remain incompletely defined. Methods: We systematically reviewed studies reporting mortality, major cardiovascular outcomes, or cardiac function after cessation in AAS-exposed individuals. All-cause mortality was pooled from comparative cohorts with individually ascertained exposure using random-effects models; global longitudinal strain (GLS) and left ventricular ejection fraction in former users were pooled as mean differences. Risk of bias was assessed with ROBINS-E and certainty with GRADE. Results: Fifteen primary reports were included. Across three cohorts, AAS exposure was associated with higher all-cause mortality (risk ratio 2.57, 95% CI 1.98-3.33; I2 = 0%; 95% prediction interval 0.47-13.87), with consistent sensitivity analyses. In a single large matched cohort, adjusted hazard ratios for cardiovascular outcomes ranged from 2.26 to 8.90, highest for cardiomyopathy. Former users had worse GLS than AAS-naïve controls (3 studies; mean difference +0.90 percentage points, 95% CI +0.13 to +1.66), whereas the pooled ejection-fraction estimate was exploratory and highly heterogeneous. Conclusions: AAS exposure was associated with higher all-cause mortality in three observational cohorts, but residual confounding, differing follow-up, and the small evidence base preclude causal inference. Cardiovascular associations from a single cohort require independent replication, and evidence on residual myocardial dysfunction in former users is of very low certainty.