CIN2+病变风险预测模型构建及临床验证

Development and validation of the risk prediction model for cervical intraepithelial neoplasia grade 2 or worse

摘要:

目的：构建宫颈上皮内瘤变2级及其以上（cervical intraepithelial neoplasia grade 2 or worse，CIN2+）病变的风险预测模型，为宫颈病变治疗提供指导。方法：回顾分析2015年1月到2017年12月在门诊进行宫颈活检的1 266例患者的临床病理资料。以组织学病理诊断结果为金标准，通过logistic回归分析确定预测CIN2+的风险因素，构建患病风险预测模型，建立个体化预测列线图模型，评估模型的预测性能和符合度。结果：①单因素卡方检验显示，吸烟史（χ2=29.92，P=0.000）、液基薄层细胞检测（thinprep cytology test，TCT）（χ2=172.89，P=0.000）、人乳头瘤病毒（human papillomavirus，HPV）（χ2=147.03，P=0.000）、阴道镜印象诊断（χ2=340.85，P=0.000）、病变面积/宫颈面积（χ2=206.22，P=0.000）为检出CIN2+的风险因素。②多因素logistic回归分析分别建立回归模型，计算模型的AUC为0.88。③构建列线图模型，对其进行内部验证后得到的一致性指数（index of concordance，C-index）为0.878。结论：本研究基于吸烟史、TCT、HPV、阴道镜印象诊断及病变面积/宫颈面积这5项风险因素建立的回归模型拟合效果较好，预测CIN2+病变风险准确性较高，有助于制定个体化治疗方案。

Abstract:

Objectives：To develop and validate the risk prediction model for cervical intraepithelial neoplasia grade 2 or worse （CIN2+），and provide guidance for the treatment of cervical lesion. Methods：The clinical and pathological data of 1 266 patients who underwent cervical biopsy from January 2015 to December 2017 were retrospectively analyzed，and histopathological examination was used as the golden standard for the final diagnosis. A logistic regression analysis was used to identify the risk factors for predicting CIN2+，the disease risk prediction model and individualized prediction nomogram were developed，and the predictive performance and concordance of the model and nomogram were evaluated，respectively. Results：①The results of chi-square test showed that the risk factors for CIN2+ were smoking history（ χ2=29.92，P=0.000），ThinPrep cytology test（TCT）（χ2=172.89，P=0.000），human papillomavirus （HPV）（ χ2=147.03，P=0.000），colposcopy results（ χ2=340.85，P=0.000），and the ratio of lesion area to cervical area（ χ2=206.22，P=0.000）. ②The regression model was developed through multivariate logistic regression analysis，and the area under the curve of the model was 0.88. ③The nomogram was developed，and its index of concordance was 0.878 after internal validation. Conclusion：The regression model developed in the study is based on the five risk factors of smoking history，TCT，HPV，colposcopy results，and the ratio of lesion area to cervical area，with good performance and accuracy in predicting CIN2+ risk，and can help to establish individualized treatment plans.