基于免疫因子的HSP患儿肾炎风险预测模型的效能分析  

An Efficacy Analysis of a Model for Predicting the Risk of Nephritis in Children with HSP Based on Immune Factors

作　　者：宋丹阳 赵玉歧[1] 王鹤[1] 张会武[1] 曹畅 SONG Danyang;ZHAO Yuqi;WANG He;ZHANG Huiwu;CAO Chang(Department of Pediatrics, Cangzhou Central Hospital, Cangzhou 061000, China;School of Public Health, Hebei Medical University, Shijiazhuang 050017, China)

机构地区：[1]沧州市中心医院儿内三科,河北沧州061000 [2]河北医科大学公共卫生学院,河北石家庄050017

出　　处：《标记免疫分析与临床》2021年第2期299-303,共5页Labeled Immunoassays and Clinical Medicine

基　　金：2020年度河北省医学科学项目研究课题计划(编号:20200327)。

摘　　要：目的探讨以免疫因子为基础的风险模型对过敏性紫癜(HSP)患儿肾炎发生的预测效能。方法选择2018年6月至2019年6月期间于我院确诊的HSP患儿155例,按照是否并发紫癜性肾炎(HSPN)分为HSP组及HSPN组。所有患儿入院时均对免疫球蛋白(IgA、IgG及IgM)、T细胞亚群分布情况(CD3^(+)、CD4^(+)、CD8^(+))、补体(C3及C4)及IL-6水平进行检测,而后收集所有患儿临床资料及实验室检查资料。t检验和χ^(2)检验对比锁定影响HSPN阳性指标后,以HSPN情况为因变量,各阳性指标为自变量,先验概率分别为0.5建立判别式,后再次对数据进行验证。结果HSPN组患者IL-6、补体C3、IgA、CD3^(+)占比及CD8^(+)占比均显著高于HSP组患儿,而CD4^(+)及CD4^(+)/CD8^(+)则显著低于HSP组患儿(P<0.05)。免疫因子对HSPN预测的贝叶斯判别模型为P2=3.842×IL-6(g/L)^(+)28.986×C3(g/L)^(+)12.084×IgA(g/L)^(+)19.902×CD4^(+)/CD8^(+)(%)^(+)0.615×CD3^(+)(%)^(+)0.876×CD4^(+)(%)^(+)1.522×CD8^(+)(%)。考核判别模型表明本模型灵敏度为96.91%,特异性为96.55%,ROC曲线下面积为0.996(95%CI=0.991~1.000),差异具有统计学意义(P<0.001)。结论免疫因子与对于HSPN发生具有显著的相关性,免疫因子的贝叶斯模型建立对于HSPN的发生具有一定的预测价值。Objective To explore the predictive power of a risk model based on immune factors for nephritis in children with Henoch-Schönlein purpura(HSP).Methods From June,2018 to June,2019,155 children diagnosed with HSP in our hospital were divided into the HSP group and HSPN group according to whether they had purpura nephritis(HSPN).All children were admitted to the hospital for immunoglobulin(IgA,IgG and IgM),T cell subset distribution(CD3^(+),CD4^(+),CD8^(+)),complement(C3 and C4)and IL-6 content detection,and we then collected all clinical data and laboratory examination data among these children.With t-test andχ^(2) test results to identify the positive indexes that affected HSPN,the HSPN status was used as the dependent variable,while each positive index was the independent variable,and the prior probability was 0.5 to establish the discriminant,and then the data was verified for the second time.Results The proportions of IL-6,complement C3,IgA,CD3^(+)and CD8^(+)in the HSPN group were all significantly higher than those in the HSP group,while CD4^(+)and CD4^(+)T/CD8^(+)were significantly lower than those of HSP children(P<0.05).The Bayesian discriminant model for the prediction of HSPN by immune factors was as following:P2=3.842×IL-6(g/L)^(+)28.986×C3(g/L)^(+)12.084×IgA(g/L)^(+)19.902×CD4^(+)/CD8^(+)(%)^(+)0.615×CD3^(+)(%)^(+)0.876×CD4^(+)(%)^(+)1.522×CD8^(+)(%).The assessment discriminant model showed that the sensitivity of this model was 96.91%,the specificity was 96.55%,the area under the ROC curve was 0.996(95%CI=0.991-1.000),and the results were statistically significant(P<0.001).Conclusion Immune factors have a significant correlation with the occurrence of HSPN.The establishment of Bayesian model of immune factors has a certain predictive value for the occurrence of HSPN.

关 键 词：免疫因子 过敏性紫癜 肾炎 风险预测 贝叶斯模型 

分 类 号：R725[医药卫生—儿科]