The Value Of Serum Ferritin In Predicting IVIG-resistant Kawasaki Disease And Establishment Of A New Predictive Model

Objective:In order to find new predictors and improve the existing early prediction model of intravenous immunoglobulin-resistant Kawasaki disease(IVIGKD),in order toreduce the occurrence of complications such as coronary artery damage and improve the treatment effect and prognosis,this study investigated the predictive value of serum ferritin(SF)on IVIGKD,and to explore other clinical value of SF in Kawasaki disease.This paper analyzes the clinical characteristics of IVIGKD,looks for independent risk factors,establishes a new risk prediction model,tests and compares its predictive efficacy with four commonly used international models,so as to study whether SF can be used as a new indicator to improve the risk prediction scoring system.Methods:Clinical data of 422 Kawasaki disease cases and 142 out-patient healthy children who had received serum ferritin test in a single center of our hospital from January2017 to December 2019 were collected retrospectively,then we set it as the Kawasaki disease(KD)group and the healthy children group,according to the effect of initial IVIG treatment,the KD group was further divided into 57 cases of IVIG-resistant group and 365 cases of IVIG-sensitive group,at the same time,KD group was also divided into 103 cases of coronary artery lesion(CAL)group and 319 cases of coronary artery normal group(NCAL)depending on whether the coronary artery is dilated,the serum ferritin levels of Kawasaki disease and healthy children group,IVIG-resistant and IVIG-sensitive group,CAL and NCAL groups were respectively compared,ROC curve was used to analyze the predictive efficiency of SF to IVIGKD.The general information,clinical symptoms and laboratory test results of children in the IVIG-resistant group and the IVIG-sensitive group were compared and analyzed to understand the clinical characteristics of IVIGKD.Binary Logistic regression analysis was used to explore the independent risk factors of IVIGKD,and a new risk prediction scoring model was established.The predictive efficacy of the new scoring method and four commonly used prediction scoring systems were tested and compared with the case data of this group.Results:(1)Analysis of SF comparative results of each subgroupand:SF level154(101-250)ng/ml in KD group was significantly higher than that in healthy children group37(21-64)ng/ml(P<0.05).SF level of IVIG resistant group 231(117-451)ng/ml was significantly higher than that of IVIG sensitive group 148(94-238)ng/ml](P<0.05).There was no statistical difference in SF level between CAL group and NCAL group[163(100-307)ng/ml vs 151(101-239)ng/ml,Z=-1.309,P=0.190,P>0.05].(2)ROC curve analysis of SF predictive probability of IVIGKD:ROC curve analysis showed that when the cut-off value of 403.55ng/ml was taken,SF’s sensitivity,specificity and Youden index were 0.386,0.94,0.326,greater than 0.3,and AUC(area under the curve)0.71for predicting IVIGKD,showing good predictive efficacy.(3)IVIGKD clinical features and independent risk factors:Compared with IVIG sensitive group,IVIG resistant group significantly higher index has 15,including the average total cost in hospital,average hospitalization days,the incidence of coronary artery damage,Cervical lymphadenopathy,perianal redness and descination,erythema multiforme-like,level of UREA,CRP,WBC,NE%,APTT,FDP,ALT,AST,CREA;Five indicators were significantly reduced,including the course of disease at the initial IVIG treatment,the proportion of patients presenting with incomplete Kawasaki disease,and level of HGB,PLT,and hematohein.According to univariate analysis,15 variables were correlated with IVIGKD,and multivariate Logistic regression analysis showed that 9 variables,including serum ferritin(SF),course of disease at the time of initial IVIG treatment,Cervical lymphadenopathy,erythema multiforme-like,WBC,CRP,APTT,ALT and CREA,were independent risk factors for IVIGKD.(4)Establish a new prediction model:Finally,7 indicators were included,and cut-off values were calculated by ROC curve.According to the assigned values of regression coefficient,a new scoring system was established:erythema multiforme-like(2 points),Cervical lymphadenopathy(1 point),SF≥403.5ng/ml(1 point),WBC≥18.3×10~9/L(1 point),CRP≥83.1mg/L(1 point),APTT≥25.3(SEC)(1 point),ALT≥37.5U/L(1 point),and high-risk value≥4 points.(5)Compared the predictive efficacy with four commonly used scoring systems:The newly established scoring system and four previously used scoring systems were used to predict Kawasaki disease cases in this group,respectively,and the analysis of the prediction results showed that among the Kawasaki disease children in this group,the Kobayashi score had the highest specificity(up to 96%),but the sensitivity was low(only 35%),and the Youden index was the lowest(only 0.31).The sensitivity,specificity and Youden index of Egami score were 53%,92%and 0.45 respectively.Sano score had the highest sensitivity(79%),but the lowest specificity(72%)and Youden index was 0.51.The sensitivity of Fu score was 60%,specificity was 92%,and Youden index was 0.52.The sensitivity and specificity of the new scoring system are 77%,92%and the Youden index is 0.7,which is the highest among the five.The Youden index of the five scoring systems is 0.31-0.7(all>0.3),the predictive efficacy of the new scoring system and the four commonly used scoring systems was satisfactory.ROC curve was used to calculate and improve critical values of Kobayashi score,Egami score and Fu score(SANO score did not need to be improved),after the improvement,their prediction efficiency has been improved.the Youden index and AUC of the five scoring systems were0.44-0.7 and 0.79-0.903 respectively,showing good predictive efficiency.Conclusion:(1)The serum ferritin level of Kawasaki disease children is significantly higher than that of healthy children,and the increase is more obvious in IVIGKD,and SF has a good predictive effect on IVIGKD.(2)IVIGKD children with average hospitalization days and total hospitalization expenses were significantly elevated,Cervical lymphadenopathy,perianal redness and descination,erythema multiforme-like performance relative to see more,are more likely to merge coronary artery damage,hyponatremia,anemia and moderate subclinical renal injury,blood appears as high coagulation state,inflammatory response and liver function damage is more obvious;Independent risk factors for IVIG-resistant include increased test indexes such as SF,WBC,CRP,APTT,ALT and CREA,as well as increased incidence of symptoms such as Cervical lymphadenopathy and erythema multiforme-like.Early treatment of IVIG would increase the risk of IVIG-resistant.(3)The newly established prediction models were as follows:erythema multiforme-like(2points),Cervical lymphadenopathy(1 point),SF≥403.5ng/ml(1 point),WBC≥18.3×10~9/L(1point),CRP≥83.1mg/L(1 point),APTT≥25.3(SEC)(1 point),ALT≥37.5U/L(1 point),and high-risk value≥4 points,showing good predictive efficacy.Four scoring systems,including Kobayashi method,Egami method,Sano method and Fu method,all showed good predictive efficacy for IVIGKD children in our region.The improved critical value was used to obtain better predictive efficacy,and the prediction efficiency of the new model is not inferior to the above scoring system.(4)SF was significantly increased in patients with IVIGKD,The cut-off value SF≥403.5ng/ml had a good predictive effect and was an independent risk factor,and could be used as a new indicator to participate in the improvement of the risk prediction scoring system for unresponsive Kawasaki disease.