基于Lasso Cox回归模型的早期阿尔茨海默病转归预测研究  

Prediction of early Alzheimer′s disease progression based on Lasso Cox regression model

作　　者：韩红娟 秦瑶[1] 乔果果 陈杜荣[1] 崔靖 马艺菲 张荣 余红梅 HAN Hongjuan;QIN Yao;QIAO Guoguo;CHEN Durong;CUI Jing;MA Yifei;ZHANG Rong;YU Hongmei(Department of Health Statistics,School of Public Health,Shanxi Medical University,Taiyuan 030001,China;Department of Mathematics,School of Basic Medical Sciences,Shanxi Medical University,Taiyuan 030001,China;Teaching Experiment Center,School of Public Health,Shanxi Medical University,Taiyuan 030001,China;Shanxi Provincial Key Laboratory of Major Diseases Risk Assessment,Taiyuan 030001,China)

机构地区：[1]山西医科大学公共卫生学院卫生统计学教研室,太原030001 [2]山西医科大学基础医学院数学教研室,太原030001 [3]山西医科大学公共卫生学院教学实验中心,太原030001 [4]重大疾病风险评估山西省重点实验室,太原030001

出　　处：《中华疾病控制杂志》2023年第8期907-915,共9页Chinese Journal of Disease Control & Prevention

基　　金：国家自然科学基金(81973154);山西省青年自然科学基金(201901D211330,20210302123242);山西省研究生创新项目(2022Y367)。

摘　　要：目的探讨认知正常(normal cognition,NC)进展为轻度认知功能障碍(mild cognitive impairment,MCI)、MCI逆转为NC和MCI进展为阿尔茨海默病(Alzheimer′disease,AD)的相关因素并建立预测模型。方法利用阿尔茨海默病神经影像学倡议(Alzheimer′s Disease Neuroimaging Initiative,ADNI)数据,分别构建NC进展为MCI、MCI逆转为NC和MCI进展为AD的Lasso Cox回归模型。结果共纳入502例NC和990例MCI患者,其中NC进展为MCI者104例,MCI逆转为NC者90例,MCI进展为AD者370例。结果显示较低的雷氏听觉和语言学习测试-学习成绩(HR=0.8840,95%CI:0.7780~0.9900)、内嗅皮层萎缩(HR=0.9996,95%CI:0.9992~0.9999)、较高的功能活动量表得分(HR=1.5520,95%CI:1.2920~1.8130)和Tau蛋白(HR=1.0030,95%CI:1.0010~1.0060)均是NC进展为MCI的相关因素;较低的AD评定量表-13条目得分(HR=0.9250,95%CI:0.8570~0.9930)、功能活动量表得分(HR=0.8260,95%CI:0.6910~0.9610)和较低浓度的Tau蛋白(HR=0.9970,95%CI:0.9940~0.9990)均是MCI逆转为NC的相关因素;携带载脂蛋白E4(apolipoprotein E4,APOE4)基因(HR=1.3040,95%CI:1.0690~1.5400)、较长的受教育年限(HR=1.0550,95%CI:1.0170~1.0930)、较高的临床痴呆评定量表得分(HR=1.1920,95%CI:1.0600~1.3240)、AD评定量表-13条目(HR=1.0360,95%CI:1.0110~1.0610)、功能活动量表得分(HR=1.0650,95%CI:1.0390~1.0910)和磷酸化Tau蛋白(HR=1.0180,95%CI:1.0090~1.0270)、较低的雷氏听觉和语言学习测试-即时回忆(HR=0.9730,95%CI:0.9560~0.9900)和较低浓度的Aβ淀粉样蛋白(HR=0.9992,95%CI:0.9988~0.9996)均是MCI进展为AD的相关因素。结论AD的疾病转归受多个因素影响,不同阶段的风险预测有利于对重点人群的健康管理干预和预防性护理,减少其未来疾病进展的风险。Objective The study aims to explore predictors related to the progression from normal cognition(NC)to mild cognitive impairment(MCI),the reversion from MCI to NC,and the progression from MCI to Alzheimer′s disease(AD),and to establish corresponding prediction models.Methods Based on the Alzheimer′s Disease Neuroimaging Initiative(ADNI)database,three Lasso Cox regression models were developed to predict the transition from NC to MCI,the reversion from MCI to NC,and the progression from MCI to AD.Results The study included 502 NC and 990 MCI patients,of which 104 NC progressed to MCI,90 MCI reverted to NC,and 370 MCI progressed to AD.The covariate effects of the predictors showed that lower Rey′s auditory verbal learning test⁃learning(HR=0.8840,95%CI:0.7780-0.9900),atrophy of the entorhinal cortex(HR=0.9996,95%CI:0.9992-0.9999),higher functional activities questionary score(HR=1.5520,95%CI:1.2920-1.8130)and Tau protein(HR=1.0030,95%CI:1.0010-1.0060)were associated with the progression from NC to MCI.Lower scores of Alzheimer′s disease assessment scale⁃13(HR=0.9250,95%CI:0.8570-0.9930),functional activity scale scores(HR=0.8260,95%CI:0.6910-0.9610)and lower concentrations of Tau protein(HR=0.9970,95%CI:0.9940-0.9990)were all relevant factors for MCI reversal to NC.The apolipoprotein E4(APOE4)gene(HR=1.3040,95%CI:1.0690-1.5400),longer years of education(HR=1.0550,95%CI:1.0170-1.0930),higher clinical dementia rating sum of boxes score(HR=1.1920,95%CI:1.0600-1.3240),Alzheimer′s disease assessment scale⁃13 entries(HR=1.0360,95%CI:1.0110-1.0610),Rey′s auditory verbal learning test⁃immediate(HR=0.9730,95%CI:0.9560-0.9900),functional activities questionary score(HR=1.0650,95%CI:1.0390-1.0910),and phosphorylation Tau(HR=1.0180,95%CI:1.0090-1.0270),lower Rey′s auditory verbal learning test⁃immediate(HR=0.9730,95%CI:0.9560-0.9900)and lower concentrations of Aβamyloid(HR=0.9992,95%CI:0.9988-0.9996)were associated with the progression from MCI to AD.Conclusions The transformation among different stages

关 键 词：阿尔茨海默病 轻度认知功能障碍 认知正常 Lasso Cox回归模型 

分 类 号：R741[医药卫生—神经病学与精神病学] R181[医药卫生—临床医学]