基于BCPNN法对来迪派韦/索磷布韦药物警戒信号的挖掘

Pharmacovigilance signal detection of ledipasvir/sofosbuvir based upon Bayesian confidence propagation neural network

目的:利用贝叶斯置信传播神经网络法(Bayesian confidence propagation neural network,BCPNN)挖掘来迪派韦/索磷布韦上市后的安全警戒信号,以期为临床合理用药提供参考。方法:检索2014年10月10日至2020年9月29日期间美国食品药品监督管理局不良事件报告系统数据库(FDA adverse event reporting system,FAERS)中以来迪派韦/索磷布韦为首要怀疑对象的药品不良事件(adverse drug events,ADEs),运用BCPNN法进行药物警戒信号挖掘,信号生成条件为信息因子(information component,IC)的95%CI下限>0且报告数a≥3。结果:共检索得到来迪派韦/索磷布相关ADEs 18492份,利用BCPNN法识别得到20个来迪派韦/索磷布韦药物警戒相关信号,其中有9个是药品说明书中未出现的可疑药物警戒信号,分别为:肾小球滤过率降低、脑病、耳鸣、便秘、血肌酐升高、流感样病例、异常做梦、口渴、色尿症。结论:挖掘来迪派韦/索磷布韦可疑药物安全警戒信号可为国内临床合理安全用药提供参考依据。

OBJECTIVE To employ Bayesian confidence propagation neural network(BCPNN)for mining the safety warning signals of ledipasvir/sofosbuvir after marketing so as to provide rationales for clinical rational use of drugs.METHODS The database of adverse event reporting system(FDA Adverse Event Reporting System,FAERS)was searched during the period from 2014/10/10 to 2020/09/29 since ledipasvir/sofosbuvir was a primary suspect for adverse drug events(ADEs).BCPNN method was utilized for pharmacovigilance signal mining.The signal generation conditions were set as 95%CI lower limit of information component IC>0 and the number of reports a≥3.RESULTS A total of 18492 ADEs related to ledipavir/sofosbuvir were retrieved.BCPNN method was employed for identifying 20 suspicious pharmacovigilance-related signals of ledipavir/sofosbuvir.Nine suspicious pharmacovigilance signals were not mentioned within package inserts.There were lowered glomerular filtration rate,encephalopathy,tinnitus,constipation,elevated blood creatinine,flu-like cases,abnormal nightmare,thirst and colored urine.CONCLUSION The mining of suspicious drug safety warning signals of ledipavir/sofosbuvir may provide rationales for reasonable and safe dosing of clinical drugs in China.