卵巢上皮性癌患者生存预测模型的构建及验证  

Construction and validation of postoperative survival prediction model for patients with epithelial ovarian cancer

作　　者：孔艺燕 李小钊[1] 应小燕[1] KONG Yiyan;LI Xiaozhao;YING Xiaoyan(Department of Obstetrics and Gynecology,the Second Affiliated Hospital of Nanjing Medical University,Nanjing 210000,Jiangsu,China)

机构地区：[1]南京医科大学第二附属医院妇产科,江苏南京210000

出　　处：《中国现代医生》2023年第24期50-57,共8页China Modern Doctor

摘　　要：目的分析卵巢上皮性癌(epithelial ovarian cancer,EOC)流行病学特征及影响预后的相关因素,并绘制列线图个体化预测患者的总体生存率。方法收集公共数据库2010–2017年诊断的12675例EOC术后患者,按7:3比例随机分为建模组(n=8875)和内部验证组(n=3800)。采用Cox回归分析筛选影响术后患者总体生存时间的独立预后因素,并建立模型,从南京医科大学第二附属医院病案系统收集82例患者作为外部验证组,预测EOC患者的生存率。用校准曲线和一致性指数(C指数)、受试者操作特征(receiver operating characteristic,ROC)曲线估计预测和实际总体生存率之间列线图的准确性和一致性。结果根据Cox多因素筛选影响术后EOC患者总体生存时间的独立预后因素(P<0.05)构建预测模型。校准曲线显示基于预测模型预测患者的生存与实际生存具有较好的一致性。建模组、外部验证组、内部验证组在预后模型显示出中等判别能力,1、3、5年生存率的ROC曲线下面积(area underthe curve,AUC)分别为0.772、0.782、0.799;0.742、0.778、0.795;0.653、0.807、0.827。预测模型的C指数为0.763(95%CI:0.755~0.771)。结论预测模型有较好的预后准确性,能可靠预测EOC患者的1、3、5年生存率,有助于EOC患者进行个性化的预后评估和指导临床决策。Objective To analyze the epidemiological characteristics and prognostic factors of epithelial ovarian cancer(EOC),and to draw a line graph to predict overall survival.Methods A total of 12675 EOC patients diagnosed in a public database from 2010 to 2017 were randomly divided into a modeling group(n=8875)and an internal validation group(n=3800)in a 7:3 ratio.cox regression analysis was used to screen independent prognostic factors affecting the overall survival time of patients after surgery to establish a model.Totally 82 patients were collected from our hospital’s medical record system as an external validation group to predict the survival rate of patients with EOC.Calibration curve and consistency index(C index)and receiver operating characteristic(ROC)curve were used to estimate the accuracy and consistency of the graph between prediction and actual overall survival.Results A prediction model was established based on cox multivariate screening of independent prognostic factors(P<0.05)that affected the overall survival time of postoperative EOC patients.The calibration curve showed that the patients'survival predicted by the prediction model was in good agreement with the actual survival.Modeling group,external validation group,internal validation group showed moderate discriminant ability in the prognostic model,and the AUC values of the area under the curve(AUC)for 1 year,3 and 5 years survival were 0.772,0.782,0.799;0.742,0.778,0.795;0.653,0.807,0.827.The C index of the prediction model was 0.763(95%CI:0.755-0.771).Conclusion The prediction model has good prognostic accuracy and can reliably predict the 1 year,3 and 5 years overall survival of patients with EOC.This predictive model could prove to be a good predictive tool for gynecological clinical practice and help patients with EOC to personalize prognosis assessment and guide clinical decision making.

关 键 词：卵巢癌 预后模型 列线图 生存分析 

分 类 号：R737[医药卫生—肿瘤]