川崎病患儿对静脉注射免疫球蛋白治疗无反应预测模型构建

Construction of a prediction model for nonresponse to intravenous immunoglobulin treatment in children with Kawasaki disease

目的基于N末端B型利钠肽原(NT-proBNP)水平并结合川崎病患儿临床特点及常规实验室指标,构建川崎病患儿对静脉注射免疫球蛋白(IVIG)治疗无反应的预测模型。方法纳入268例川崎病患儿,均于发病10 d内接受规律的IVIG治疗,并于治疗前完善NT-proBNP和其他常规实验室检查。将80%的患儿归为训练集(215例),根据对IVIG治疗的反应将患儿分为有反应组和无反应组,比较两组患儿临床资料和实验室检查指标,并通过多因素Logistic回归模型分析影响患儿对IVIG治疗的反应的因素,根据影响因素构建川崎病患儿对IVIG治疗无反应的列线图预测模型。将剩余20%的患儿作为验证集(53例),对模型进行外部验证。结果纳入训练集的215例患儿中,有反应组184例和无反应组31例。无反应组冠状动脉扩张比例和中性粒细胞百分比(Neu%)、红细胞沉降率(ESR)、C反应蛋白(CRP)、总胆红素、间接胆红素、AST、NT-proBNP水平均高于有反应组(均P<0.05)。多因素Logistic回归分析结果显示,Neu%、ESR、CRP、总胆红素、AST和NT-proBNP水平均是川崎病患儿对IVIG治疗无反应的影响因素(均P<0.05)。基于上述影响因素构建的列线图预测模型的一致性指数为0.927,区分度良好。应用验证集的数据对模型进行外部验证,绘制校正曲线,校正曲线显示列线图预测模型的预测可能性绝对误差为0.016,一致性良好。结论NT-proBNP和Neu%、ESR、CRP、总胆红素、AST水平升高的川崎病患儿出现对IVIG治疗无反应的风险更高。基于NT-proBNP水平和其他危险因素建立的列线图预测模型对川崎病患儿的IVIG治疗反应具有较好的预测效果,可以在本地区进一步加以验证后推广。

Objective To construct a prediction model for nonresponse to intravenous immunoglobulin(IVIG)treatment in children with Kawasaki disease based on the N-terminal pro-B-type natriuretic peptide(NT-proBNP)level combined with the clinical features and routine laboratory indicators of children with Kawasaki disease.Methods A total of 268 children with Kawasaki disease enrolled received regular IVIG treatment within 10 days after the disease onset,and the NT-proBNP testing and other routine laboratory examinations were completed before the treatment.About 80%of the children were classified into train set(n=215),who were divided into response group and nonresponse group according to their response to IVIG treatment,and the clinical data and the indicators of laboratory examinations were compared between the two groups.In addition,the factors influencing the children′s response to IVIG treatment were analyzed by the multivariate Logistic regression model.A nomogram prediction model for nonresponse to IVIG treatment was constructed in children with Kawasaki disease based on the influencing factors.The remaining children(20%)served as validation set(n=53)for an external validation performed on the model.Results Among the 215 children enrolled in the train set,there were 184 cases in the response group and 31 cases in the nonresponse group.The nonresponse group exhibited a higher proportion of cases sustaining coronary artery ectasia,a higher percentage of neutrophils(Neu%),a higher erythrocyte sedimentation rate(ESR),as well as increases in the levels of C-reactive protein(CRP),total bilirubin,indirect bilirubin,AST,and NT-proBNP as compared with the response group(all P<0.05).The results of the multivariate Logistic regression analysis showed that the Neu%,ESR,CRP,total bilirubin,AST,and NT-proBNP levels were the influencing factors of nonresponse to IVIG treatment in children with Kawasaki disease(all P<0.05).The nomogram prediction model constructed based on the aforementioned influencing factors exhibited a concordance index of 0.927,which showed a favorable discrimination.The data of the validation set were used to perform an external validation on the model,and the calibration curve was plotted.The calibration curve revealed that the nomogram prediction model harbored an absolute error of prediction probability of 0.016,which showed a favorable concordance.Conclusion Higher risks for nonresponse to IVIG treatment occur in children with Kawasaki disease when they sustain an increase in the NT-proBNP,Neu%,ESR,CRP,total bilirubin,or AST level.The nomogram prediction model constructed based on the NT-proBNP level and other risk factors has favorable prediction efficacy for response to IVIG treatment in children with Kawasaki disease,which can be promoted in the local region after further validation.