血小板微核糖核酸-26a-5p结合临床指标构建川崎病相关冠状动脉损伤的风险预测模型

Constructing a predictive model for coronary artery lesions in Kawasaki disease patients based on platelet-derived microRNA-26a-5p and clinical indicators

目的探讨血小板源性微核糖核酸(miRNA,简称miR)联合其他临床指标用于预测川崎病(KD)相关冠状动脉损伤(CAL)发生风险的潜在价值。方法收集2018年2—11月在上海市儿童医院(上海交通大学医学院附属儿童医院)心内科住院治疗的70例急性期KD患儿的临床资料和血小板源性miRNA表达水平,并根据心脏彩色多普勒超声(简称心脏彩超)检查获得的冠状动脉(简称冠脉)Z值将患儿分为CAL组(14例)和非冠脉损伤组(NCAL组,56例)。收集KD患儿在接受静脉注射免疫球蛋白(intravenous immunoglobulin,IVIG)治疗前急性期的基本信息、症状和体征、实验室检查结果、心脏彩超检查结果,并测定患儿外周血中血小板miRNA表达情况,以miR-126-3p为标准化基因,采用2-△△CT法计算相关血小板miRNA的相对表达量。将CAL与NCAL组间单因素分析中P<0.2的变量纳入多因素logistic回归分析,探究冠状动脉病变的风险因素,构建风险预测模型,并绘制ROC曲线评估模型预测效能。结果CAL与NCAL组间的性别构成、年龄、体重、发热持续时间和部分特征性临床表现占比的差异均无统计学意义(P值均>0.05),CAL组身高显著低于NCAL组(P=0.010)。CAL组血小板-淋巴细胞比值(PLR)有低于NCAL组的趋势,但差异无统计学意义(P>0.05,但<0.2)。CAL组ESR值和总蛋白值有低于NCAL组的趋势,但差异均无统计学意义(P值均>0.05,但均<0.2)。以miR-126-3p为标准化基因,CAL组血小板miR-26a-5p相对表达量为3.000±0.768,显著高于NCAL组的2.500±0.808(t=-2.120,P=0.038)。将单因素分析中P<0.2的指标,即ESR、总蛋白、PLR和miR-26a-5p,以及研究显示为KD患儿CAL发生的风险因素总胆红素(TBil)共同纳入多因素logistic回归分析,结果显示ESR、总蛋白、TBil和血小板miR-26a-5p是影响患儿CAL发生的风险因素(P值均<0.05)。联合上述3种临床指标及血小板miR-26a-5p构建CAL风险预测模型,其ROC的AUC为0.856(95%CI为0.703~1.000),以预测概率0.365作为截断值时,模型的约登指数最高为0.707,模型预测灵敏度和特异度分别为0.727和0.980。结论血小板miR-26a-5p、ESR、总蛋白和TBil是影响KD相关CAL发生的风险因素,联合miR-26a-5p与上述3种临床指标构建的CAL风险预测模型具有良好的预测效能。

Objective To explore the potential value of platelet-derived microRNAs(miRNA)in combination with clinical indicators in predicting the risk of coronary artery lesions(CAL)associated with Kawasaki Disease(KD).Methods Clinical data and platelet-derived miRNA expression levels were collected from 70 children with acute phase KD who were treated in the Department of Cardiology at Shanghai Children’s Hospital from February to November 2018.Based on the Z-score of coronary arteries obtained from color Doppler ultrasonography,the patients were divided into CAL group(n=14)and non-coronary artery lesion(NCAL)group(n=56).Basic information,symptoms and signs,laboratory indicators and ultrasound examination results were collected before intravenous immunoglobulin(IVIG)treatment.The platelet miRNA expression in peripheral blood was measured.Normalized by miR-126-3p,the relative expression level of related platelet miRNA was calculated by 2-ΔΔCT.Variables with P<0.2 in the univariate analysis between CAL and NCAL groups were included in the multivariate logistic regression analysis to explore the risk factors for CAL,construct a risk prediction model,and evaluate the prediction performance of the model by drawing an ROC curve.Results There was no statistically significant difference in gender distribution,age,weight,duration of fever,or proportion of partial characteristic clinical manifestations between the CAL and NCAL groups(all P>0.05).The body height of CAL group was significantly lower than that of NCAL group(P=0.010).The platelet-lymphocyte ratio(PLR),erythrocyte sedimentation rate(ESR)and total protein in the CAL group were lower than those in the NCAL group,but the differences were not statistically significant(0.05<P<0.2).Normalized by miR-126-3p,the level of miR-26a-5p in the CAL group was 3.000±0.768,which was significantly higher than that in the NCAL group(2.500±0.808,t=-2.120,P=0.038).The results of multivariate logistic regression analysis showed that ESR,total protein,miR-26a-5p,and total bilirubin(TBil)were risk factors for CAL in KD children(all P<0.05).A CAL risk prediction model was constructed,and the AUC of the ROC curve of this model was 0.856(95%CI:0.703-1.000).When the predictive probability was set at 0.365 as the cutoff value,the highest Youdens index of the model was 0.707,and the sensitivity and specificity were 0.727 and 0.980,respectively.Conclusion Platelet miR-26a-5p,ESR,total protein,and TBil are risk factors affecting the occurrence of CAL associated with KD.The CAL model constructed by combining miR-26a-5p with the aforementioned three clinical indicators exhibits good predictive performance.