甲状腺眼病中替妥木单抗相关耳毒性的发生率和特征：系统评价和荟萃分析

文献信息

中文摘要

背景： 替妥木单抗是首个获批用于甲状腺眼病的疗法，疗效显著，但越来越多地与耳科不良事件相关，促使美国食品药品监督管理局于2023年发出关于听力损害的警告。各研究报告的发生率差异很大，毒性的真实负担仍不确定。方法：我们对PubMed/MEDLINE、Embase和Cochrane CENTRAL从建库至2026年5月进行了系统评价和荟萃分析（PROSPERO CRD420261337579），共识别出22项符合条件的研究（5项随机对照试验和17项观察性研究），包括856例接受替妥木单抗治疗的患者。结果：任何耳科不良事件的合并发生率为27.3%（95% CI，20.8-34.3%），包括主观听力损失、耳鸣、耳闷胀感和自听过强。临床显著耳毒性发生于7.7%（95% CI，3.1-13.8%），而在接受连续听力学监测的患者中，客观听力学耳毒性的检出率为20.0%（95% CI，10.0-32.0%）。与安慰剂相比，替妥木单抗使耳科不良事件风险增加近四倍（风险比，3.6；95% CI，1.3-10.1）。在受影响患者中，39.0%（95% CI，18.0-62.0%）在末次随访时仍有持续性损害。确定方法是研究间异质性的主要来源，主动监测发现的事件率显著高于被动报告（42.0% vs. 20.0%）。既往听力损失、年龄较大和糖尿病与风险增加相关。尽管重叠调整后的发现一致，但设计、确定方法、定义和随访方面的异质性意味着这些估计值应被视为近似而非确定性结论。结论：这些结果支持基线听力学评估、治疗期间主动症状监测以及就持续性听力损害可能性进行知情咨询。注册：PROSPERO CRD420261337579。

英文摘要

Teprotumumab, the first approved therapy for thyroid eye disease, is highly effective, but has been increasingly associated with otologic adverse events, prompting a United States Food and Drug Administration warning regarding hearing impairment in 2023. Reported incidence varies widely across studies, and the true burden of toxicity remains uncertain. We conducted a systematic review and meta-analysis of PubMed/MEDLINE, Embase, and Cochrane CENTRAL from inception through May, 2026 (PROSPERO CRD420261337579), identifying 22 eligible studies (5 randomized controlled trials and 17 observational studies) comprising 856 teprotumumab-treated patients. The pooled incidence of any otologic adverse event was 27.3% (95% CI, 20.8-34.3%), including subjective hearing loss, tinnitus, aural fullness, and autophony. Clinically significant ototoxicity occurred in 7.7% (95% CI, 3.1-13.8%), while objective audiometric ototoxicity was detected in 20.0% (95% CI, 10.0-32.0%) of patients undergoing serial audiometric monitoring. Compared with placebo, teprotumumab increased the risk of otologic adverse events nearly four-fold (risk ratio, 3.6; 95% CI, 1.3-10.1). Among affected patients, 39.0% (95% CI, 18.0-62.0%) had persistent impairment at last follow-up. Ascertainment method was the major source of between-study heterogeneity, with active surveillance identifying substantially higher event rates than passive reporting (42.0% vs. 20.0%). Preexisting hearing loss, older age, and diabetes were associated with increased risk. Although overlap-adjusted findings were consistent, heterogeneity in design, ascertainment, definitions, and follow-up means these estimates should be regarded as approximate rather than definitive. These results support baseline audiometric assessment, active symptom surveillance during treatment, and informed counseling regarding the potential for persistent hearing impairment. REGISTRATION: PROSPERO CRD420261337579.