阴道上皮内病变2及以上的风险预测模型构建及临床验证  

Construction and Clinical Validation of a Risk Prediction Model for Vaginal Intraepithelial Neoplasia Grade 2 or Worse Lesions

作　　者：范子玉 石皆春 顾辰洁 马昕宇 邢燕[1] FAN Ziyu;SHI Jiechun;GU Chenjie(Department of Gynecology,The First Affiliated Hospital of Nanjing Medical University,Nanjing Jiangsu 210029,China)

机构地区：[1]南京医科大学第一附属医院妇科,江苏南京210029

出　　处：《实用妇产科杂志》2025年第1期42-47,共6页Journal of Practical Obstetrics and Gynecology

基　　金：江苏省卫生健康发展研究中心开放课题(编号:JSHD2022024);江苏省妇幼健康研究会课题(编号:JSFY202308);江苏省妇幼保健协会课题(编号:FYX202319)。

摘　　要：目的:构建阴道上皮内病变2及以上(即VaIN 2^(+))病变的风险预测模型,并建立个体诊断VaIN 2^(+)的列线图及风险分层,为阴道病变的治疗提供指导。方法:收集2021年1月至2024年1月在南京医科大学第一附属医院妇科子宫颈病中心经阴道镜活检诊断为VaIN的女性248例,以组织学病理结果为金标准,分为低于VaIN 2组和VaIN 2^(+)组,对两组进行单因素比较分析,采用多因素Logistic回归分析确定VaIN 2^(+)的危险因素并构建诊断模型,应用R语言软件建立列线图模型,采用受试者工作特征(ROC)的曲线下面积(AUC)、校准曲线验证与评价该模型的区分度、校准度和临床实用价值。结果:单因素分析发现,人乳头瘤病毒(HPV)型别、TCT结果、子宫颈上皮内瘤变(CIN)级别、醋白改变、阴道病变时间、阴道病变位置和CIN病变时间为发生VaIN 2^(+)的影响因素(P<0.1);多因素二元Logistic回归分析提示,HPV16/18阳性、CIN级别≥CIN 2、厚醋白改变、阴道病变时间≥5年、阴道病变位置为阴道上1/3是发生VaIN 2^(+)的独立危险因素(OR>1,P<0.05),CIN病变时间<3年为保护性因素(OR<1,P<0.05),其中醋白改变影响最大(OR 4.54)。根据多因素二元Logistic回归分析建立回归模型,该模型的AUC为0.813;构建列线图模型,对其进行内部验证后得到一致性指数(C-index)为0.81;使用X-tile软件将患者进行风险分层,总分越高的患者发生VaIN 2^(+)的风险越高。结论:本研究构建的列线图模型可个体化预测患者发生VaIN 2^(+)的病变风险,准确性和临床实用性较高。Objective:To construct a risk prediction model for Vaginal Intraepithelial Neoplasia Grade 2 or Worse(VaIN 2^(+))lesions,and to establish a nomogram for individual diagnosis of VaIN 2^(+)and risk stratification,so as to provide guidance for the treatment of vaginal lesions.Methods:A total of 248 women diagnosed with VaIN through colposcopic biopsy at the Center for Gynecologic and Cervical Diseases,First Affiliated Hospital of Nanjing Medical University,from January 2021 to January 2024 were included in this study.Based on the gold standard established by histological and pathological findings,these patients were categorized into a lower VaIN 2 group and a VaIN 2^(+)group.Univariate comparative analysis was performed on the two groups.Multivariate Logistic regression analysis was used to determine the risk factors of VaIN 2^(+)and to construct a diagnostic model.The nomogram model was established by using R Studio software.The discrimination,calibration and clinical practical value of the model were evaluated by the area under the receiver operating characteristic(ROC)curve and calibration curve.Results:Univariate analysis identified that HPV type,cervical lesion grade,acetowhite change,vaginal lesion duration,vaginal lesion location,and cervical lesion duration as influencing factors for diagnosing VaIN 2^(+)(P<0.1).Multivariate binary Logistic regression analysis indicated that HPV16/18 positivity,cervical lesion grade≥CIN 2,thick acetowhite change,vaginal lesion duration≥5 years,and vaginal lesion location at the upper 1/3 of the vagina were independent risk factors for diagnosing VaIN 2^(+)(OR>1,P<0.05),while cervical lesion duration<3 years was a protective factor(OR<1,P<0.05),with acetowhite change having the greatest impact(OR 4.54).A regression model was established based on the multivariate binary Logistic regression analysis,with an AUC of 0.813.A nomogram model was constructed and internally validated,yielding a consistency index(C-index)of 0.81.Patients were stratified into risk groups using the X

关 键 词：阴道上皮内病变 风险因素 预测模型 列线图 

分 类 号：R737.33[医药卫生—肿瘤]