基于经阴道三维超声联合临床特征构建诊断宫腔粘连的列线图

A nomogram prediction model for diagnosing uterine adhesions based on three-dimensional transvaginal ultrasound combined with clinical characteristics

目的:基于经阴道三维超声(3D-TVS)指标联合临床特征构建诊断宫腔粘连(IUA)的列线图预测模型,提高3D-TVS诊断IUA的准确度,减少IUA的误诊率和漏诊率。方法:回顾分析2017年6月至2021年7月因疑诊IUA在中山大学孙逸仙纪念医院行3D-TVS和宫腔镜检查的607例患者的临床特征和3D-TVS结果。将收集到的样本按7∶3随机分为训练集和验证集。训练集中,采用logistic回归分析临床特征、3D-TVS特征与IUA的关系。根据logistic回归分析结果,将具有显著统计学差异的变量用于构建预测诊断IUA的列线图预测模型,并在验证集中分别采用受试者工作特征(ROC)曲线、Hosmer-Lemeshow检验、临床校正曲线、临床决策曲线(DCA)和临床影响曲线(CIC)评价列线图预测诊断IUA的能力。结果:最终纳入607例样本,其中宫腔镜诊断IUA者401例,非IUA者206例。将607例样本分为训练集(420例)和验证集(187例)。训练集中,logistic回归分析结果显示IUA患者与非IUA患者的年龄、既往早孕期宫腔操作次数、既往中晚孕期宫腔操作次数、宫腔面积(cm^(2))、子宫内膜回声中断、左侧或右侧宫壁内聚角≤130°比较,差异有统计学意义(P<0.05)。根据上述变量构建诊断IUA的列线图预测模型。验证集中,对该列线图预测模型诊断IUA的能力进行验证。ROC曲线的曲线下面积(AUC)提示该列线图具有较精准的预测诊断IUA的能力。Hosmer-Lemeshow检验和临床校准曲线均显示该列线图模型具有较高的校准能力。DCA和CIC均显示该列线图模型临床实用价值高。结论:根据临床特征和3D-TVS特征构建诊断IUA的列线图预测模型,可在宫腔镜检查前较准确地诊断IUA。年龄、既往早孕期宫腔操作次数、既往中晚孕期宫腔操作次数、宫腔面积(cm^(2))、子宫内膜回声中断、左侧或右侧宫壁内聚角≤130°是诊断IUA的显著独立预测变量。

Objective:By formulating and validating a nomogram for differentiating intrauterine adhesions(IUA)based on clinical characteristics and three-dimensional transvaginal ultrasound(3D-TVS),to improve the accuracy of 3D-TVS diagnosing IUA,and to reduce the rate of misdiagnosis and missed diagnosis of IUA.Methods:A total of 607 subjects who underwent 3D-TVS and hysteroscopy at Sun Yat-sen Memorial Hospital for suspected IUA were enrolled between June 2017 to July 2021.The clinical characteristics and preoperative 3D-TVS characteristics of all subjects were all collected.Firstly,the collected samples were randomly divided into the training set and the validation set by SPSS at a ratio of 7∶3.In the training set,the logistic regression analysis explored the relationship between clinical characteristics,3D-TVS characteristics and IUA.On the basis of logistic regression analysis,the variables with significant statistical differences were selected to establish a nomogram.In validation set,receiver operating characteristic(ROC)curve,clinical calibration curve,Hosmer-Lemeshow test,decision curve analysis(DCA),and clinical impact curve(CIC)were used to evaluate the performance of the nomogram.Results:We finally included 607 samples,of which 401 were diagnosed with IUA by hysteroscopy and 206 were diagnosed with non-IUA.At the beginning,607 samples were divided into a training set(n=420)and a validation set(n=187).In the training set,age,frequency of previous first-trimester procedure,frequency of previous second trimester and late procedure,uterine cavity area(cm^(2)),interrupted endometrial echogenicity,and left or right lateral indentation angle≤130 degrees differed significantly between IUA and non-IUA groups(P<0.05).In validation set,the area under the curve(AUC)of the ROC curve indicated that the nomogram had the ability to accurately predict and diagnose IUA.Both the Hosmer-Lemeshow test and the clinical calibration curve showed that the nomogram had high calibration ability.Both DCA and CIC showed that the model had high clinical value.Conclusions:The differential nomogram was established based on the clinical characteristics and 3D-TVS characteristics.It was able to feasibly distinguish IUA before hysteroscopy.Age,frequency of previous first-trimester procedure,frequency of previous second trimester and late procedure,uterine cavity area(cm^(2)),interrupted endometrial echogenicity and left or right lateral indentation angle≤130 degree were significant variables for predicting IUA.