妊娠合并系统性红斑狼疮患者妊娠丢失预测模型构建

Construction of a prediction model for pregnancy loss in pregnant women with systemic lupus erythematosus

目的:针对系统性红斑狼疮(SLE)孕妇发生妊娠丢失构建列线图预测模型。方法:回顾分析2012年9月1日至2022年12月1日在广西医科大学第一附属医院住院的327例SLE孕妇,其中妊娠丢失75例,非妊娠丢失252例。经单因素分析、多重共线性检验,LASSO回归筛选,根据AUC指标利用多因素logistic回归分析得到最终模型,对模型区分度、校准度进行检验,通过Bootstrap算法内部验证,最终以列线图的形式呈现。结果:多因素logistic回归构建预测模型变量:国际标准化比值(P=2.48×10~(-8))、SLEDAI-2K(P=1.16×10~(-4))、纤维蛋白原(P=0.001)、C3(P=0.009)。预测公式:Logit(P)=-1.654+1.061×国际标准化比值+0.941×SLEDAI-2K-0.677×纤维蛋白原-0.578×C3。模型的区分度采用ROC曲线下面积(AUC)评估,AUC=0.868(95%CI为0.816~0.919),灵敏度为83.7%,特异度为77.3%。H-L检验结果P=0.327(χ~2=9.186,df=8),Brier得分为0.102。决策曲线提示模型有较好的临床决策价值。Bootstrap内部验证后校正的C指数为0.857,校准曲线显示一致性较好。结论:SLEDAI-2K、国际标准化比值、纤维蛋白原、C3是SLE孕妇发生妊娠丢失的独立预测因素,通过临床常用指标建立了SLE孕妇发生胎儿丢失风险预测模型。

Objective:To establish a nomogram prediction model for pregnancy loss in pregnant women with systemic lupus erythematosus.Methods:A retrospective cohort study was conducted to analyze 327 pregnant women with SLE who were hospitalized in the First Affiliated Hospital of Guangxi Medical University from September 1,2012 to December 1,2022.Relevant clinical data were collected and grouped according to pregnancy loss,divided into pregnancy loss group(n=75)and non-pregnancy loss group(n=252).After single factor analysis,multi-collinearity test,LASSO regression screening,the final model was obtained by using multi-factor logistic regression analysis according to the AUC index,and the model discrimination and calibration were tested,and the internal verification of the Bootstrap algorithm was carried out,and finally in the form of a nomogram presented.Results:Multivariate logistic regression to construct predictive model variables:international normalized ratio(P=2.48×10-8),SLEDAI-2K(P=1.16×10-4),fibrinogen(P=0.001),C3(P=0.009).The prediction formula was:Logit(P)=-1.654+1.061×international normalized ratio+0.941×SLEDAI-2K-0.677×fibrinogen-0.578×C3.The discrimination of the model was evaluated by the area under the ROC curve(AUC),AUC=0.868(95%CI 0.816~0.919),the sensitivity was 83.7%,and the specificity was 77.3%.The result of HL test was P=0.327(χ2=9.186,df=8),and the Brier score was 0.102.The decision curve indicated that the model had good clinical decision value.The corrected C index after Bootstrap internal validation was 0.857,and the calibration curve showed good consistency.Conclusion:SLEDAI-2K,international normalized ratio,fibrinogen,and C3 are independent predictors of pregnancy loss in pregnant women with SLE.A risk prediction model for fetal loss in pregnant women with SLE was established by using common clinical indicators.