1型糖尿病患者QTc间期延长风险的预测模型的建立与验证

Development and validation of a predictive model for the risk of QTc interval prolongation in patients with type 1 diabetes mellitus

目的探讨1型糖尿病(T1DM)患者QTc间期延长的影响因素,建立预测模型并进行验证。方法本研究为横断面研究。选取2016年1月至2023年10月在甘肃省人民医院内分泌科住院的568例T1DM患者为研究对象。根据12导联常规心电图检查中QTc间期水平,将研究对象分为QTc间期正常组(423例)和QTc间期延长组(145例),并收集患者的一般资料和实验室检测指标,包括T1DM病程、尿白蛋白/肌酐比值(UACR)、糖化血红蛋白(HbA_(1c))、高密度脂蛋白胆固醇(HDL-C)、血肌酐(Scr)、左心室射血分数(LVEF),以及是否合并糖尿病周围神经病变(DPN)等。应用LASSO回归优化筛选变量,通过多因素logistic回归分析构建T1DM患者发生QTc间期延长风险的列线图模型。使用1000次增强Bootstrap法对模型进行内部验证,分别采用受试者工作特征(ROC)曲线、校准曲线、临床决策曲线(DCA)和临床影响曲线(CIC)综合评估模型的预测价值、校准度和临床实用性。结果应用LASSO回归分析筛选出7个预测变量,包括T1DM病程、UACR、HbA_(1c)、LVEF、DPN、HDL-C及Scr。多因素logistic回归分析进一步显示,T1DM病程≥10年(OR=4.951)、UACR>300 mg/g(OR=1.759)、HbA_(1c)≥7%(OR=7.988)、LVEF≤50%(OR=8.501)、DPN(OR=1.708)、HDL-C(OR=0.198)是T1DM患者发生QTc间期延长的影响因素(均P<0.05)。建立的预测模型内部验证结果显示,模型拟合度良好,ROC曲线下面积为0.822(95%CI 0.786~0.858),预测结果接近于实际。DCA显示,在0~0.8的阈值区间具有最大效益。CIC表明,预测模型可以在阈值概率范围内有效区分出QTc间期发生延长的高危患者。结论包含6个预测变量(T1DM病程、UACR、HbA_(1c)、LVEF、DPN、HDL-C)的列线图预测模型可用于预测T1DM患者发生QTc间期延长的风险,对早期甄别这类高风险人群具有一定的临床意义。

Objective To investigate the factors influencing the QTc interval prolongation in patients with type 1 diabetes mellitus(T1DM),and to develop and to validate a predictive model.Methods This was a cross-sectional study,and 568 patients with T1DM who were hospitalized in the Department of Endocrinology of Gansu Provincial People′s Hospital from January 2016 to October 2023 were selected as study subjects.Based on the length of QTc interval measurement in 12-lead conventional electrocardiography,the study subjects were divided into the normal QTc interval group(423 cases)and the prolonged QTc interval group(145 cases),and the general data and laboratory test indexes of the patients were collected,including the duration of T1DM,urinary albumin-creatinine ratio(UACR),glycated hemoglobin A_(1c)(HbA_(1c)),high-density lipoprotein cholesterol(HDL-C)and serum creatinine(Scr),left ventricular ejection fraction(LVEF),diabetic peripheral neuropathy(DPN).LASSO regression was applied to optimize the screening variables,and a column-line graphical model of the risk of QTc interval prolongation in T1DM patients was constructed by multifactor Logistic regression analysis.Internal validation of the model was performed using 1000 enhanced Bootstrap method,and the predictive value,calibration and clinical utility of the model were comprehensively assessed using receiver operating characteristic(ROC)curve,calibration curve,clinical decision curve(DCA)and clinical impact curve(CIC).Results LASSO regression analysis was applied to screen seven predictor variables,including T1DM duration,UACR,HbA_(1c),LVEF,DPN,HDL-C,and Scr.Multifactorial logistic regression analysis further showed that T1DM duration≥10 years(OR=4.951),UACR>300 mg/g(OR=1.759),HbA_(1c)≥7%(OR=7.988),LVEF≤50%(OR=8.501),DPN(OR=1.708),and HDL-C(OR=0.198)were the influencing factors for the occurrence of QTc interval prolongation in patients with T1DM(all P<0.05).The internal validation results of the established prediction model showed that the model fit was good,and the area under the ROC curve was 0.822(95%CI 0.786-0.858),which was close to the actual prediction results.DCA showed that there was a maximum benefit in the threshold interval of 0-0.8.CIC showed that the prediction model could effectively distinguish high-risk of the occurrence of prolongation of the QTc interval in the threshold probability range of patients.ConclusionA column chart prediction model including six predictive variables(T1DM course,UACR,HbA_(1c),LVEF,DPN and HDL-C)can be used to predict the risk of QTc interval extension in T1DM patients,which has clinical implications for early detection of such high-risk populations.