作者：蒋荣莉，殷安康，徐礼科

单位：

1.225001江苏省扬州市，扬州大学附属医院全科医学科 2.225001江苏省扬州市，扬州大学附属医院内分泌科 通信作者：徐礼科，E-mail：lcy3a@163.com

单位（英文）：

1. Department of General Medicine, Affiliated Hospital of Yangzhou University, Yangzhou 225001, China 2. Department of Endocrinology, Affiliated Hospital of Yangzhou University, Yangzhou 225001, China Corresponding author: XU Like, E-mail: lcy3a@163.com

关键词：

糖尿病,2型; 无症状低血糖; 动态血糖监测; 心率变异性; 胰岛素; 预测;

关键词（英文）：

Diabetes mellitus, type 2; Asymptomatic hypoglycemia; Continuous glucose monitoring; Heart ratevariability; Insulin; Forecasting

中图分类号：

DOI：

10.12114/j.issn.1008-5971.2022.00.135

基金项目：

江苏省医院协会2018年度医院管理创新研究立项课题（JSYGY-2-2018-59）

摘要：

目的 分析心率变异性（HRV）参数对接受胰岛素治疗的2型糖尿病（T2DM）患者发生夜间无症状低血糖的预测价值。方法 选取2020年3月至2021年11月于扬州大学附属医院接受胰岛素治疗的T2DM患者207例。剔除夜间有症状低血糖患者19例，根据夜间低血糖发生情况将剩余患者分为夜间血糖正常组（137例）和夜间无症状低血糖组（51例）。比较两组患者基线资料、生化指标、HRV参数[低频功率归一化值（LFnu）、高频功率归一化值（HFnu）、低频功率（LF）/高频功率（HF）、正常R-R间期的标准差（SDNN）、SDNN中每5 min R-R间期平均值的标准差（SDANN）、相邻R-R间期之差的均方根（RMSSD）、相邻R-R间期差值>50 ms的个数占总R-R间期个数的百分比（PNN50）]。采用多因素Logistic回归分析探讨接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的影响因素。采用ROC曲线分析HRV参数对接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的预测价值。结果 夜间无症状低血糖组患者空腹血糖、糖化血红蛋白低于夜间血糖正常组（P<0.05）。夜间无症状低血糖组患者HFnu、SDANN、RMSSD、PNN50低于夜间血糖正常组，LF/HF高于夜间血糖正常组（P<0.05）。多因素Logistic回归分析结果显示，HFnu、LF/HF、SDANN、RMSSD、PNN50是接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的影响因素（P<0.05）。ROC曲线分析结果显示，HFnu、LF/HF、SDANN、RMSSD、PNN50预测接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的AUC分别为0.768[95%CI(0.680,0.856)]、0.776[95%CI(0.693,0.859)]、0.769[95%CI(0.693,0.845)]、0.767[95%CI(0.698,0.836)]、0.770[95%CI(0.695,0.844)]，最佳截断值分别为38.38 Hz、1.91、52.31 ms、22.23 ms、6.14 ms，灵敏度分别为81.80%、74.50%、74.50%、70.80%、73.70%；特异度分别为66.70%、72.30%、72.50%、72.50%、74.50%。HFnu、LF/HF、SDANN、RMSSD、PNN50联合预测接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的AUC为0.936[95%CI(0.892,0.980)]，灵敏度为70.60%，特异度为98.50%。HRV参数联合预测接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的AUC分别大于HFnu、LF/HF、SDANN、RMSSD、PNN50单独预测接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的AUC，差异有统计学意义（Z值分别为3.324、3.341、3.688、4.035、3.737,P值均<0.05）。结论 HRV参数中的HFnu、LF/HF、SDANN、RMSSD、PNN50对接受胰岛素治疗的T2DM患者发生夜间无症状低血糖具有一定预测价值，而其联合预测的价值更高，其有望成为接受胰岛素治疗的T2DM患者发生夜间无症状低血糖的新型诊断标志物。

英文摘要：

【Abstract】　Objective To analyze the predictive value of heart rate variability (HRV) parameters for nocturnalasymptomatic hypoglycemia in insulin-treated type 2 diabetes mellitus (T2DM) patients. Methods A total of 207 T2DM patientswho were treated with insulin in Affiliated Hospital of Yangzhou University from March 2020 to November 2021 were selected.Nineteen patients with symptomatic hypoglycemia at night were excluded. According to the occurrence of nocturnal asymptomatichypoglycemia, the remaining patients were divided into nocturnal normal blood sugar group (n=137) and nocturnal asymptomatichypoglycemia group (n=51) . The baseline data, biochemical indexes, HRV [low frequency power normalized unit (LFnu) ,high frequency power normalized unit (HFnu) , low frequency power (LF) /high frequnency power (HF) , standard deviation ofnormal R-R interval (SDNN) , standard deviation of 5-min average of normal R-R intervals (SDANN) , root mean square of thedifference of successive R-R intervals (RMSSD) , and percentage of successive R-R intervals differing from more than 50 ms(PNN50) ] were compared between the two groups. Multivariate Logistic regression analysis was used to explore the influencingfactors of nocturnal asymptomatic hypoglycemia in T2DM patients treated with insulin. ROC curve was used to explore thepredictive value of HRV parameters for nocturnal asymptomatic hypoglycemia in insulin-treated T2DM patients. ResultsThe fasting blood glucose and glycosylated hemoglobin of the patients in the nocturnal asymptomatic hypoglycemia group werelower than those in the nocturnal blood sugar normal group (P < 0.05) . The HFnu, SDANN, RMSSD and PNN50 of the nocturnalasymptomatic hypoglycemia group were lower than those of the nocturnal normal blood sugar group, and the LF/HF was higher thanthat of the nocturnal blood sugar normal group (P < 0.05) . The results of multivariate Logistic regression analysis showed that HFnu,LF/HF, SDANN, RMSSD and PNN50 were the influencing factors of nocturnal asymptomatic hypoglycemia in patients with T2DMtreated with insulin (P < 0.05) . The results of ROC curve analysis showed that the AUC of HFnu, LF/HF, SDANN, RMSSD, PNN50in predicting nocturnal asymptomatic hypoglycemia in T2DM patients treated with insulin was 0.768 [95%CI (0.680, 0.856) ] ,0.776 [95%CI (0.693, 0.859) ] , 0.769 [95%CI (0.693, 0.845) ] , 0.767 [95%CI (0.698, 0.836) ] , 0.770 [95%CI (0.695, 0.844) ] .The best cut-off values were 38.38 Hz, 1.91, 52.31 ms, 22.23 ms, 6.14 ms, respectively. The sensitivity was 81.80%, 74.50%,74.50%, 70.80%, 73.70% and the specificity was 66.70%, 72.30%, 72.50%, 72.50%, 74.50%, respectively. The AUC ofcombination of HFnu, LF/HF, SDANN, RMSSD, PNN50 in predicting nocturnal asymptomatic hypoglycemia in T2DM patientstreated with insulin was 0.936 [95%CI (0.892, 0.980) ] , the sensitivity was 70.60% and the specificity was 98.50%. The AUCof combination of HRV parameters in predicting nocturnal asymptomatic hypoglycemia in insulin-treated T2DM patients wasbigger than the AUC of HFnu, LF/HF, SDANN, RMSSD, and PNN50 alone in predicting nocturnal asymptomatic hypoglycemiain insulin-treated T2DM patients, respectively (Z values were 3.324, 3.341, 3.688, 4.035, 3.737, all P ＜ 0.05) .ConclusionHFnu, LF/HF, SDANN, RMSSD, PNN50 in HRV parameters have certain predictive value for nocturnal asymptomatichypoglycemia in T2DM patients treated with insulin, their combination has higher predictive value, and they are expected to benovel markers for the diagnosis of nocturnal asymptomatic hypoglycemia in T2DM patients treated with insulin.

参考文献：