基于自噬基因的度洛西汀抗抑郁疗效预测模型的构建  

A predictive model for antidepressant effect of duloxetine based on autophagy-related genes

作　　者：李偲媛 魏宇梅 和申 曾端 李华芳[1] LI Si-yuan;WEI Yu-mei;HE Shen;ZENG Duan;LI Hua-fang(Shanghai Jiao Tong University,Shanghai Mental Health Center,Shanghai,200030,China)

机构地区：[1]上海交通大学医学院附属精神卫生中心精神科,200030

出　　处：《临床精神医学杂志》2024年第2期113-117,共5页Journal of Clinical Psychiatry

基　　金：上海市精神心理疾病临床医学研究中心(19MC1911100);上海市精神卫生中心院级重点课题(2022zd02);多维组学数据整合的外周生物标志物引导抑郁症精准药物治疗的临床研究(SHDC2020CR2053B);上海申康医院发展中心医企融合创新支撑技能培训专项(SHDC2022CRS032);上海市科委生物医药领域科技支撑项目(22S21902300)。

摘　　要：目的:通过生物信息学方法构建基于自噬基因的度洛西汀抗抑郁疗效预测模型。方法:在高通量基因表达数据库中下载GSE146446数据集,该芯片包括96例患者接受抗抑郁药物度洛西汀8周的治疗,组织样本为全血样本,以度洛西汀治疗8周后是否有效分组,筛选两组间的差异表达基因,与自噬基因集取交集。利用最小绝对值收敛和选择算法回归(LASSO)及Logistic回归构建疗效预测模型。结果:SPNS1、ITPR3基因的表达水平均为度洛西汀抗抑郁疗效的影响因素(P均<0.05)。LASSO-Logistic回归模型:Logit(P)=33.7846+(-2.8615×SPNS1表达水平)+(-1.7716×ITPR3表达水平),其中Logit(P)=ln[P/(1-P)]。结论:基于自噬相关基因(SPNS1、ITPR3)表达量的度洛西汀的抗抑郁疗效预测模型具有较好的区分度、校准度以及疗效预测效能,未来可能为抑郁症患者使用度洛西汀药物治疗提供更为科学可靠的证据。Objective:This paper aims to construct a prediction model of the efficacy of antidepressant treatment with duloxetine based on autophagy by bioinformatics methods.Method:GSE146446 dataset was downloaded from gene expression omnibus database,including 96 patients with major depressive disorder(MDD)treated with antidepressant duloxetine for 8 weeks,and the tissue is whole blood.After an 8-week treatment,patients were divided into responders and non-responders.The differentially expressed genes(DEGs)between the two groups were screened and intercrossed with the autophagy-related genes(ARGs)to find the key genes.The predictive model for antidepressant effect were established using least absolute shrinkage and selection operator(LASSO)and Logistic regression.Results:The expressions of two autophagy-related genes(SPNS1,ITPR3)were related to the efficacy of antidepressant treatment with duloxetine(all P<0.05).LASSO-Logistic regression prediction model:Logit(P)=33.7846+(-2.8615×the expression of SPNS1)+(-1.7716×the expression of ITPR3),and Logit(P)=ln[P/(1-P)].Conclusion:The predictive model for antidepressant effect of duloxetine based on two autophagy-related genes(SPNS1,ITPR3)can predict the treatment efficacy of duloxetine.The model has a good differentiation,calibration and efficacy,which may guide clinical medication of duloxetine in MDD in the future.

关 键 词：抑郁症 自噬 自噬相关基因 预测模型 最小绝对值收敛和选择算法回归-Logistic回归模型 

分 类 号：R749.4[医药卫生—神经病学与精神病学]