目的 通过对美国食品药品监督管理局（FDA）不良事件报告系统（FAERS）、日本药物不良事件报告系统（JADER）、加拿大药物警戒不良反应在线数据库（CVARD）进行数据挖掘，分析女性患者服用第3代芳香化酶抑制剂（AIs）与痴呆不良事件的关联性，为临床安全用药提供参考。方法 收集FAERS (2004年第一季度—2024年第四季度)、JADER (2004年4月—2024年12月)和CVARD (1991年1月—2024年12月)期间接收到的患者服用第3代AIs（来曲唑、阿那曲唑、依西美坦）后报告痴呆的病例。采用标准化流程进行数据清洗（包括去重、药物名称统一化、MedDRA术语标准化），确保数据质量。使用报告比值比法（ROR）、比例报告比值比法（PRR）和信息成分法（IC）检测药物的高危信号。对不良事件诱发时间数据进行拟合优度检验（比较韦伯、对数正态、伽玛、指数分布），并采用最优模型分析时间风险特征。结果 共纳入44例第3代AIs相关痴呆报告（FAERS数据库36例，JADER数据库4例，CVARD数据库4例；来曲唑4例，阿那曲唑19例，依西美坦21例）。信号检测显示：在FAERS数据库中阿那曲唑与阿尔茨海默病型痴呆显著相关（ROR=2.36,95% CI:1.50～3.69），其风险高于依西美坦与痴呆的关联（ROR=1.72,95% CI:1.07～2.77）。CVARD中依西美坦（ROR=4.85,95% CI:1.82～12.95）和JADER中来曲唑（ROR=2.72,95% CI:1.02～7.26）与痴呆存在信号。拟合优度检验表明韦伯分布（AICc=41.35,BIC=44.65,-2lgL=39.85）最能描述不良事件诱发时间数据。韦伯分布检验（β=1.86,95% CI:1.05～2.77）显示不良事件风险随用药时间延长而增加（磨损型故障曲线）。72.8%的病例在用药1～3年后发生痴呆。亚组分析提示60～69岁患者（ROR=7.48）及英国患者（ROR=18.88）风险更高。结论 基于多国药物警戒数据库的分析，发现第3代AIs与痴呆不良事件存在统计学关联信号，尤其在长期用药的老年人群中风险需关注。

Objective To conduct research through the US Food and Drug Administration Adverse Event Reporting System（FDA） FAERS, the Japanese Adverse Drug Event Report Data mining was conducted using JADER and the Canada Vigilance Adverse Reaction Online Database（CVARD） to analyze the association between the use of third-generation aromatase inhibitors（AIs） in female patients and adverse events of dementia. Provide a reference for the safe use of drugs in clinical practice. Methods Cases of reported dementia in patients who were admitted during FAERS（Q1 2004—Q4 2024）, JADER（April 2004—December 2024）, and CVARD（January 1991—December 2024） after being induced with third-generation aromatase inhibitors（letrozole, anastrozole, exemestane） were collected. Standardized processes were adopted for data cleaning（including deduplication, unification of drug names, and standardization of MedDRA terms） to ensure data quality. The high-risk signals of drugs were detected using the Reporting Odds Ratio（ROR）, Proportional Reporting Ratio（PRR）, and Information Component（IC） methods. The goodness of fit test was conducted on the data of the induction time of adverse events（comparing Weber, lognormal, gamma, and exponential distribution）, and the optimal model was used to analyze the temporal risk characteristics. Results A total of 44 reports of dementia related to third-generation AIs were included(36 cases in the FAERS database, 4 cases in the JADER database, and 4 cases in the CVARD database; There were 4 cases of letrozole, 19 cases of anastrozole and 21 cases of exemestane. Signal detection showed that in the FAERS database, anastrozole was significantly associated with Alzheimer’s dementia（ROR = 2.36, 95%CI: 1.5—3.69）, and its risk was higher than the association between exemetam and dementia（ROR = 1.72, 95%CI: 1.07—2.77）. In CVARD, exemestane（ROR = 4.85, 95%CI: 1.82—12.95） and in JADER, letrozole（ROR = 2.72, 95%CI: 1.02—7.26） had signals associated with dementia. The goodness of fit test indicated that the Weber distribution（AICc = 41.35, BIC = 44.65,-2LogL = 39.85） could best describe the data of the induction time of adverse events. The Weber distribution test（β = 1.86, 95%CI: 1.05—2.77） showed that the risk of adverse events increased with the prolongation of medication time（wear-type failure curve）. 72.8% of the cases developed dementia 1 to 3 years after medication. Subgroup analysis suggested that patients aged 60—69（ROR = 7.48） and those in the UK（ROR = 18.88） had a higher risk. Conclusion Based on the analysis of the multi-national pharmacovigilance database, it was found that there are statistical association signals between the third-generation AIs and adverse events of dementia. Especially in the elderly population with long-term medication, the risk needs to be paid attention to.

R979.1

国家自然科学基金青年科学基金项目(82405004); 郑州市妇幼保健院科研项目(YNKY202409); 中国医药教育协会“聚火优才”科研项目(CMEAPC2024030); 郑州市医疗卫生领域科技创新指导计划项目(202SYLZDJH158)