基于孕早期胎儿超声软指标和代谢因素的先天性心脏畸形预测模型的构建分析

Construction of a Predictive Model for Congenital Heart Malformations Based on Fetal Ultrasound Soft Indicators and Metabolic Factors in the First Trimester

目的:探讨基于孕早期胎儿超声软指标和代谢指标构建的预测模型在预测先天性心脏畸形(Congenital heart disease,CHD)中的价值。方法:纳入2018年4月至2020年5月在我院妇产科接受产前检查存在CHD高危因素的孕妇200例,采集所有孕妇在孕11~13+6周产检时的尿液,采用液相色谱-质谱(LC-MS)法检测尿液中4-羟基苯乙酸、丙烯酸、丙二酸及尿酸水平;采用胎儿超声心动图检查胎儿颈项透明层(Nuchal translucency,NT)厚度、心脏三尖瓣反流、静脉导管血流异常情况;根据后期随访胎儿是否确诊为CHD,将孕妇分为CHD组和对照组;对比两组孕妇临床资料,采用多因素Logistic回归方程分析影响孕早期胎儿心脏畸形的相关因素;随机选取60%(120例)的病例作为训练集,40%(80例)作为测试集,使用R软件基于训练集建立列线图预测模型,并在测试集评估模型效能。结果:观察组孕妇尿液4-羟基苯乙酸、丙烯酸、丙二酸及尿酸水平高于对照组,NT厚度、三尖瓣反流比例、静脉导管血流异常比例高于对照组,差异具有统计学意义(P<0.05);多因素Logistic回归分析显示,尿液中高4-羟基苯乙酸水平(OR=2.740)、高丙烯酸水平(OR=3.242)、高尿酸水平(OR=1.013)以及NT值增大(OR=20.411)、三尖瓣反流(OR=9.028)是孕早期胎儿出现先天性心脏畸形的独立危险因素(P<0.05);通过构建列线图预测模型,结果显示,模型预测胎儿CHD风险的一致性指数(C-index)为0.915。外部验证显示敏感性为88.00%,特异性为96.36%,准确率为93.75%,阳性预测值为91.67%,阴性预测值为94.64%。结论:孕早期超声检查NT厚度、三尖瓣反流以及孕妇尿液中4-羟基苯乙酸、丙烯酸及尿酸水平与胎儿CHD发生风险相关,基于超声软指标和代谢指标的预测模型对胎儿CHD风险有较好的预测效能。

Objective:To explore the value of the predictive model constructed by fetal ultrasound soft index and metabolic index in the prediction of congenital heart malformations.Methods:Totally 200 cases of pregnant women with high risk factors for CHD who received prenatal examinations in the Department of Obstetrics and Gynecology of our hospital from April 2018 to May 2020 were included.The urine of all pregnant women at the 11~13+6 weeks of pregnancy was collected during the obstetric examination,and liquid chromatography-Mass spectrometry(LC-MS)method was used to detect the levels of 4-hydroxyphenylacetic acid,acrylic acid,malonic acid and uric acid in urine;fetal echocardiography is used to check the thickness of fetal neck transparent layer(Nuchal translucency,NT)and cardiac tricuspid regurgitation Abnormal flow and venous catheter blood flow.According to whether the fetus was diagnosed with CHD in the later follow-up,the pregnant women were divided into CHD group and control group;the clinical data of the two groups of pregnant women were compared,and the multivariate logistic regression equation was used to analyze the relevant factors affecting the fetal heart malform R software was used to build a nomogram prediction model based on the training set,and the effectiveness of the model was evaluated on the test set.Results:The urine levels of 4-hydroxyphenylacetic acid,acrylic acid,malonic acid,and uric acid of the observation group were higher than those of the control group,and the NT thickness,the proportion of tricuspid regurgitation,and the proportion of abnormal blood flow in the venous catheter were higher than those of the control group.The difference was statistically significant(P<0.05);Multivariate logistic regression analysis showed that urine levels of 4-hydroxyphenylacetic acid(OR=2.740),high acrylic acid levels(OR=3.242),high uric acid levels(OR=1.013)and NT values increased(OR=20.411)and tricuspid regurgitation(OR=9.028)are independent risk factors for fetal congenital heart malformations in the first trimester(P<0.05);by constructing a nomogram prediction model,the results show that the model predicts fetal CHD risk with the consistency index(C-index)0.915.External verification showed that the sensitivity was 88.00%,the specificity was 96.36%,the accuracy rate was 93.75%,the positive predictive value was 91.67%,and the negative predictive value was 94.64%.Conclusion:Early pregnancy ultrasound examination of NT thickness,tricuspid regurgitation,and the levels of 4-hydroxyphenylacetic acid,acrylic acid and uric acid in the urine of pregnant women are related to the risk of fetal CHD.The prediction model based on ultrasound soft indicators and metabolic indicators has a good predictive performance for higher risk of fetal CHD.