基于C反应蛋白表达预测川崎病并发冠状动脉损伤患儿风险的列线图模型构建

Construction of A Nomogram Model for Predicting the Risk of Kawasaki Dis⁃ease Complicated with Coronary Artery Injury in children Based on the Ex⁃pression of C-Reactive Protein

目的探讨C反应蛋白(C-reactive protein,CRP)早期诊断川崎病患儿并发冠状动脉损伤(coronary artery injury,CAL)的临床价值,并构建个体化预测川崎病并发CAL的列线图模型。方法收集2018年1月至2021年11月在保定市儿童医院接受治疗的川崎病患儿263例,将纳入患者按照7∶3分为建模组(184例)及验证组(79例)。采用受试者工作特征曲线(receiver operating characteristic curve,ROC)评价血清CRP对建模组川崎病患儿并发CAL的诊断价值。采用多因素Logistic回归分析建模组影响川崎病并发CAL的危险因素;采用R软件构建预测川崎病并发CAL的列线图模型,并使用ROC及校准曲线验证列线图模型的区分度和一致性。结果与非CAL组患儿相比,建模组中CAL组患儿的血清CRP浓度较高,差异有统计学意义(P<0.05)。ROC分析显示,建模组中CRP诊断川崎病并发CAL的曲线下面积(area under the curve,AUC)为0.842,其敏感度为76.90%,特异度为89.40%。建模组中单因素分析显示,年龄≤3岁、持续发热时间、心电图检查异常、心肌肌钙蛋白I(cardiac troponinl,cTnI)、白细胞计数、CRP均是川崎病并发CAL的影响因素(P<0.05)。建模组中Logistic分析显示,心电图检查异常(OR=3.743,95%CI=1.649~8.497)、cTnI>0.5μg/L(OR=3.636,95%CI=1.488~8.888)、白细胞计数≥10×10^(9)/L(OR=2.332,95%CI=1.071~5.078)、CRP>46 mg/L(OR=4.732,95%CI=1.698~13.185)是影响川崎病并发CAL的危险因素(P<0.05)。构建的列线图预测模型显示,建模组ROC-AUC为0.846;验证组ROC-AUC为0.917,预测模型具有较好的一致性;Hosmer-Lemeshow拟合优度检验显示,建模组χ^(2)=6.043,P=0.535;验证组χ^(2)=6.440,P=0.598。结论CRP是预测川崎病并发CAL的可靠指标,基于CRP构建的列线图模型可有效指导川崎病患儿并发CAL的个体化预测。

Objectives To explore the clinical value of C-reactive protein(CRP)in early diagnosis of Kawasaki dis⁃ease complicated with coronary artery injury(CAL)in children,and to construct a nomogram model for individualized prediction of Kawasaki disease complicated with CAL in children.Methods From January 2018 to November 2021,263 children with Kawasaki disease who were treated in Baoding Children′s Hospital were gathered,the included patients were divided into a modeling group(184 cases)and a validation group(79 cases)according to 7∶3.The receiver operating characteristic curve(ROC)was used to evaluate the diagnostic value of serum CRP in children with Kawasaki disease complicated with CAL in modeling group.Multivariate Logistic regression analysis was performed to screen the risk factors affecting Kawasaki disease complicated with CAL in modeling group.R software was performed to construct a nomogram model for predicting Kawasaki disease complicated with CAL,and the ROC and calibration curve were performed to verify the discrimination and consistency of the nomogram model.Results Compared with non-CAL group,the serum concentration of CRP in CAL group was higher in modeling group(P<0.05).The ROC showed that the area under the curve(AUC)of CRP for the diagnosis of Kawasaki disease complicated with CAL was 0.842 in modeling group,the sensitivity was 76.90%,and the specificity was 89.40%.In modeling group,univariate analysis showed that age≤3 years old,the duration of fever,abnormal electrocardiogram(ECG),cardiac troponinl I(cTnI),white blood cell count(WBC),and CRP were all influencing factors of Kawasaki disease complicated with CAL(P<0.05).In modeling group,Logistic analysis showed that abnormal ECG(OR=3.743,95%CI=1.649-8.497),cTnI>0.5μg/L(OR=3.636,95%CI=1.488-8.888),WBC≥10×10^(9)/L(OR=2.332,95%CI=1.071-5.078),CRP>46 mg/L(OR=4.732,95%CI=1.698-13.185)were the risk factors for Kawasaki disease complicated with CAL(P<0.05).The constructed nomogram prediction model showed that,the area under the curve(AUC)of ROC in modeling group was 0.846;the ROC-AUC of validation group was 0.917,the prediction model had good consistency.Hosmer-Lemeshow goodness of fit test showed that modeling groupχ^(2)=6.043,P=0.535;validation groupχ^(2)=6.440,P=0.598.Conclusions CRP is a reliable indicator for predicting Kawasaki disease complicated with CAL in children.The nomogram model based on CRP can effectively guide the individualized prediction of Kawasaki disease complicated with CAL in children.