Lasso-logistic回归模型拟合HPV、FAM19A4、PAX1和ZNF582预测宫颈CIN2+病变的价值  

The value of Lasso-logistic regression models fitted to HPV,FAM19A4,PAX1,and ZNF582 in predicting cervical CIN2+lesions

作　　者：倪换娟 李婷[2] 刘森 伏瑶 白田妹 Ni Huanjuan;Li Ting;Liu Sen(Department of Obstetrics and Gynecology,Northern General Hospital of the Chinese People's Liberation Army,Shenyang 110003;Department of Gynecology and Pediatrics,965th Hospital of Chinese People's Liberation Army Joint Logistic Support Force,Jilin 132011;Gynecology Second Ward,the Second Affiliated Hospital of Shanxi University of Traditional Chinese Medicine(Xianyang Maternity Hospital),Xianyang 712000)

机构地区：[1]中国人民解放军北部战区总医院妇产科,沈阳110003 [2]中国人民解放军联勤保障部队第九六五医院妇儿科,吉林132011 [3]陕西中医药大学第二附属医院(咸阳市妇产医院)妇二科,咸阳712000

出　　处：《现代妇产科进展》2025年第3期211-216,共6页Progress in Obstetrics and Gynecology

摘　　要：目的:探讨人乳头瘤病毒(HPV)联合DNA甲基化标志物FAM19A4、PAX1和ZNF582预测女性宫颈CIN2+病变的价值。方法:回顾性收集2022年9月至2024年5月因宫颈癌筛查可疑或异常于北部战区总医院妇产科行阴道镜下宫颈病理组织检查的女性627例,以7∶3随机分为训练集(472例)和验证集(200例),在训练集中通过Lasso正则化回归筛选出重要特征变量,多因素logistic回归分析确定预测宫颈CIN2+病变的显著影响因素,构建列线图模型,绘制受试者工作特征曲线(ROC)、校准曲线、临床决策曲线(DCA)和临床影响曲线(CIC)对该模型的区分度、校准度及临床适用性进行评价,并通过验证集对模型进行内部验证。结果:(1)Lasso回归显示12种高危型HPV阳性、HPV16/18型阳性、FAM19A4、PAX1和ZNF582与CIN2+病变有关,多因素logistic回归显示HPV、FAM19A4、PAX1和ZNF582是宫颈CIN2+病变的显著影响因素(均P<0.05),以此建立模型。(2)该模型在训练集预测女性宫颈CIN2+病变的曲线下面积(AUC)为0.941,灵敏度和特异度分别为0.904和0.847,验证集AUC为0.939,灵敏度和特异度分别为0.873和0.880,提示模型区分度良好。(3)校准曲线显示预测曲线与理想曲线较为贴合,Hosmer-Lemeshow检验显示训练集χ^(2)=3.876,P=0.694,验证集χ^(2)=3.090,P=0.686,提示模型拟合程度较好。(4)DCA和CIC曲线均显示模型临床适用性良好。结论:HPV、FAM19A4、PAX1和ZNF582是宫颈CIN2+病变的显著影响因素,其联合构建的列线图模型用于评估女性宫颈CIN2+有一定的价值,可为患者个体化诊疗提供指导。Objective:To investigate the predictive value of human papillomavirus(HPV)combined with DNA methylation markers FAM19A4,PAX1,and ZNF582 in identifying cervical CIN2+lesions in women.Methods:Retrospective collection of 627 women who underwent colposcopic cervical pathology histological examination in our hospital from September 2022 to May 2024 due to suspicious or abnormal cervical cancer screening.They were randomly divided into a training set(n=472)and a validation set(n=200)in a 7∶3 ratio.Important characteristic variables were identified using Lasso regularized regression in the training set,and multifactorial logistic regression was then employed to determine the significant influencing factors of cervical CIN2+lesions.A nomogram model was constructed,and the discrimination,calibration,and clinical applicability of the model were evaluated by plotting the receiver operating characteristic curve(ROC),calibration curve,clinical decision curve analysis(DCA),and clinical impact curve(CIC)of the subjects,and the model was internally validated by the validation set.Results:(1)Lasso regression showed that the remaining 12 high-risk HPV positives,HPV16/18 positives,FAM19A4,PAX1 and ZNF582 were associated with CNI2+lesions,and multifactorial logistic regression showed that HPV,FAM19A4,PAX1 and ZNF582 were the significant influences on cervical CIN2+lesions(all P<0.05),and to establish the model.(2)The area under the curve(AUC)of the model for predicting cervical CIN2+lesions in women in the training set was 0.941,with sensitivity and specificity of 0.904 and 0.847,respectively,and that of the validation set was 0.939,with sensitivity and specificity of 0.873 and 0.880,suggesting that the model was well discriminated.(3)The calibration curve showed that the prediction curve was more closely matched with the ideal curve.The Hosmer-Lemeshow test showed that the training setχ^(2)=3.876,P=0.694,and the validation setχ^(2)=3.090,P=0.686,suggesting that the model fit well.(4)The clinical decision curve and the clinical

关 键 词：宫颈癌筛查 HPV DNA甲基化 CIN2+ 列线图 

分 类 号：R737.33[医药卫生—肿瘤]