肺部超声B线积分及新型炎症标志物对糖尿病肾病维持性血液透析患者心血管事件风险模型的构建及验证

Construction and validation of a cardiovascular event risk model for diabetic nephropathy-maintenance haemodialysis patients by lung ultrasound B-line score and novel inflammatory markers

目的构建并验证肺部超声B线积分(LUBS)及新型炎症标志物[单核细胞/高密度脂蛋白胆固醇比值(MHR)、系统免疫炎症指数(SII)、C反应蛋白/白蛋白比值(CAR)]预测糖尿病肾病维持性血液透析(DN-MHD)患者心血管事件的发生风险模型。方法选取北京市普仁医院2018年10月至2022年10月收治的122例DN-MHD患者,按照7∶3的比例分为训练集(86例)和验证集(36例)。将发生心血管事件的患者纳入发生组,分析DN-MHD患者心血管事件发生的影响因素,以此构建列线图预测模型并进行验证;受试者工作特征曲线评估列线图预测模型对DN-MHD患者心血管事件发生的预测效能;预测模型临床效益使用决策曲线分析。结果训练集33例患者发生心血管事件、验证集13例发生心血管事件。训练集与验证集临床一般资料比较,差异无统计学意义(P>0.05)。发生组与未发生组透析龄、透析中低血压、LUBS、MHR、SII、CAR比较,差异有统计学意义(P<0.05)。LUBS(OR=5.693)、MHR(OR=3.241)、SII(OR=4.877)、CAR(OR=4.052)是DN-MHD患者心血管事件发生的影响因素(P<0.05)。列线图预测模型区分度良好;校准曲线与理想曲线拟合度较好。训练集中列线图预测模型预测DN-MHD患者心血管事件发生的灵敏度为88.60%、特异度为90.20%,曲线下面积为0.883;验证集中列线图预测模型预测DN-MHD患者心血管事件发生的灵敏度为87.50%、特异度为89.10%,曲线下面积为0.874。列线图预测模型在阈值概率为0.00~0.23时可获得最大临床效益。结论基于LUBS、MHR、SII、CAR建立的列线图预测模型可较好地评估DN-MHD患者心血管事件的发生风险。

Objective To construct and validate a cardiovascular events risk model in diabetic nephropathy-maintenance haemodialysis(DN-MHD)patients by lung ultrasound B-line score(LUBS)and novel inflammatory markers(monocyte to high-density lipoprotein cholesterol ratio[MHR],systemic immune-inflammation index[SII],and C-reactive protein to albumin ratio[CAR]).Methods A total of 122 DN-MHD patients admitted to Beijing Puren Hospital from October 2018 to October 2022 were selected,and they were divided into training set(86 cases)and validation set(36 cases)according to the ratio of 7∶3.Patients with cardiovascular events were included in occurrence group,the influencing factors of cardiovascular events in patients with DN-MHD were analyzed,so as to construct a nomogram prediction model and verified;the predictive efficacy of the nomogram prediction model for the occurrence of cardiovascular events in DN-MHD patients was assessed by receiver operating characteristic curve;and predictive models for clinical benefit were analyzed by decision curves analysis.Results Cardiovascular events occurred in 33 patients in the training set and 13 patients in the validation set.There was no significant difference in general data between the training set and the validation set(P>0.05).There were significant differences in dialysis age,hypotension during dialysis,LUBS,MHR,SII,and CAR between two groups(P<0.05).LUBS(OR=5.693),MHR(OR=3.241),SII(OR=4.877),and CAR(OR=4.052)were the influencing factors of cardiovascular events in DN-MHD patients(P<0.05).The nomogram prediction model was well distinguished;the calibration curve fits well with the ideal curve.The sensitivity of the nomogram prediction model in the training set to predict the occurrence of cardiovascular events in DN-MHD patients was 88.60%,the specificity was 90.20%,and the area under the curve was 0.883;the sensitivity of the nomogram prediction model in the validation set to predict the occurrence of cardiovascular events in DN-MHD patients was 87.50%,the specificity was 89.10%and the area under the curve was 0.874.The maximum clinical benefit can be obtained when the threshold probability of the nomogram prediction model was 0.00-0.23.Conclusion The nomogram prediction model based on LUBS,MHR,SII,and CAR can better evaluate the risk of cardiovascular events in DN-MHD patients.