睡眠特征在区分轻度认知障碍和阿尔茨海默病中的临床价值

Clinical value of sleep characteristics in distinguishing mild cognitive impairment from Alzheimer′s disease

目的观察轻度认知障碍(MCI)和阿尔茨海默病(AD)人群的睡眠特征,评价其在区分MCI和AD人群中的价值。方法选取2020年6月至2022年3月在杭州市第七人民医院确诊的MCI组患者(50例)和AD组患者(50例)进行整夜多导睡眠监测(PSG),比较两组间总睡眠时间(TST)、睡眠结构占比(N1%、N2%、N3%、REM%)、睡眠效率(SE)、睡眠起始潜伏期(SL)、REM睡眠潜伏期(REML)、入睡后清醒时间(WASO)、睡眠纺锤波密度(ρ_(spindle))、K复合波密度(ρKC)指标差异,将组间差异有统计学意义的睡眠特征指标纳入logistic逐步回归,建立联合预测模型,通过受试者工作特征(ROC)曲线比较预测组合中单一指标与联合预测模型区分MCI和AD人群的价值。结果MCI与AD组的TST、N3、SE、ρ_(spindle)比较,差异均有统计学意义(t=3.315、2.798、3.682、6.488,均P<0.05)。Logistic逐步回归分析结果显示,N3、SE、ρ_(spindle)被纳入建模组合,联合预测模型为Logit(Pre)=-19.972-0.269N3-0.141SE-3.303ρ_(spindle)。N3、SE、ρ_(spindle)以及联合预测模型区分MCI和AD人群的灵敏度分别为64%、32%、96%、90%,特异度分别为58%、98%、50%、76%,ROC曲线下面积(AUC)分别为0.639、0.684、0.810和0.901,联合预测模型AUC显著优于单一指标(P<0.05)。结论在AD疑似人群中,基于N3、SE、ρ_(spindle)的联合预测模型对区分MCI和AD具有一定的预测价值,结合预测模型进行综合判断,有助于提高诊断的科学性和准确性。

Objective To observe the sleep characteristics of mild cognitive impairment(MCI)patients and Alzheimer disease(AD)patients,and to evaluate their values in distinguishing MCI and AD.Methods 50 patients with MCI and 50 patients with AD diagnosed in Hangzhou Seventh People′s Hospital from June 2020 to March 2022 were recruited in this study.All-night polysomnography(PSG)was performed for each patient.The total sleep time(TST),proportion of sleep structure(N1%,N2%,N3%,REM%),sleep efficiency(SE),sleep latency(SL),rapid eye movement sleep latency(REML),wake time after sleep onset(WASO),sleep spindle density(ρ_(spindle))and sleep k-complex density(ρKC)were compared between the two groups.The indexes of sleep characteristics with statistical significance between the two groups were included to perform logistic stepwise regression.The single-factorial and multi-factorial prediction models were established.Receiver operating characteristic(ROC)curve was used to compare the value of single-factorial model and multi-factorial model in distinguishing MCI and AD patients.Results There were statistically significant difference in TST,N3,SE,ρ_(spindle) between MCI and AD groups(t=3.315,2.798,3.682,6.488,all P<0.05).Logistic stepwise regression analysis showed that N3,SE,ρ_(spindle) was included in the modeling portfolio,and the joint prediction model was Logit(pre)=-19.972-0.269N3-0.141SE-3.303ρ_(spindle).The sensitivity of N3,SE,ρ_(spindle) and multi-factorial model for distinguishing MCI and AD was 64%,32%,96%,90%,and the specificity was 58%,98%,50%,76%,respectively.The area under the ROC curve(AUC)of the multi-factorial model and multi-factorial model were 0.639,0.684,0.810 and 0.901,respectively.The AUC of multi-factorial model was significantly better than that of single-factorial models(P<0.05).Conclusions In AD suspected population,the multi-factorial prediction model based on N3,SE andρ_(spindle) has a certain predictive value in distinguishing MCI and AD.Comprehensive judgment combined with the prediction model is helpful to improve the scientificity and accuracy of diagnosis.