基于SEER数据库的食管癌患者特定器官转移的预后模型:列线图  

Prognostic Model of Specific Organ Metastasis in Patients with Esophageal Cancer Based on the SEER Database:Nomogram

作　　者：常金明 朱光[1] CHANG Jinming;ZHU Guang(Department of Cardiothoracic Surgery,the Fifth Affiliated Hospital of Zhengzhou University,Zhengzhou 450000,China)

机构地区：[1]郑州大学第五附属医院胸外科,河南郑州450000

出　　处：《河南医学研究》2021年第36期6726-6731,共6页Henan Medical Research

摘　　要：目的建立一个可靠和通用的列线图用于个性化预测食管癌患者发生特定器官转移的预后情况。方法在美国国家癌症研究所监测、流行病学和最终结果(SEER)数据库中筛查出2010—2015年7329例发生骨、脑、肝、肺转移的食管癌患者为研究对象,使用R软件按7∶3随机分为训练集和验证集,进行单因素及多因素Cox回归分析,筛选出其独立的预后因素,使用R软件建立预测患者生存的列线图。采用受试者工作特征(ROC)曲线评估该列线图预测模型的鉴别能力。比较预测的生存状态与观测到的生存状态,在训练集和验证集分别进行1、2、5 a的校准,并将列线图的预测能力与第8版TNM分期进行比对。结果多因素Cox回归分析结果显示,年龄、种族、肿瘤部位、组织学等级、TNM分期、放疗与手术、化疗、肿瘤大小、肿瘤转移器官为食管癌患者特定器官转移的独立预后因素(P<0.05)。预测食管癌的癌症特异生存(CSS)列线图的1、2、5 a ROC曲线下面积(AUC)分别为0.805、0.803、0.809。预测模型的区分度和一致性经校准图、临床决策曲线(DCA)以及验证集的验证,并且DCA曲线图中,列线图的预测表现优于第8版TNM分期,具有较高的净受益。结论本研究建立的列线图在预测发生相应器官转移食管癌患者的CSS方面表现出了较大的预测准确性和可靠性。并且相对于第8版TNM分期,该列线图的预测功效及价值更高。Objective To establish a reliable and universal predictive model predicting the cancer-specific survival(CSS)of patients with corresponding organ metastasis Esophageal Cancer.Methods In the United States National Cancer Institute Surveillance,Epidemiology,and Final Results(SEER)database,7329 esophageal cancer patients with bone,brain,liver,and lung metastases from 2010 to 2015 were screened as the research objects.Patients were randomly divided into the training and validation set according to 7∶3 by using R software.The univariate and multivariate Cox regression was conducted to select the predictors of CSS.R software was used to build a Nomogram prediction chart that predicted patient survival.The receiver operating characteristic(ROC)curve was used to evaluate the discriminative ability of the Nomogram prediction model.The survival status predicted by Nomogram was compared with the observed survival status.The training and the validation set were calibrated for 1,2,and 5 years respectively,and the prediction ability of the Nomogram chart was compared with the 8th edition of TNM staging.Results Multivariate Cox regression analysis showed that age,race,tumor location,histological grade,TNM stage,radiation and surgery,chemotherapy,tumor size and transfer of organs were independent predictors of CSS(P<0.05).The area under the ROC curve(AUC)of 1,2 and 5 years of the Nomogram for predicting the CSS of esophageal cancer were 0.805,0.803 and 0.809,respectively.The discrimination and consistency of the prediction model have been well verified by the calibration chart,clinical decision curve(DCA)and validation set.And in the DCA curve,the prediction performance of the Nomogram graph is better than that of the 8th edition of TNM stage,and with a higher net benefit.Conclusion The nomogram chart established in this study shows great accuracy and reliability in predicting CSS of patients with corresponding organ metastasis of esophageal cancer.And compared with the 8th edition of TNM staging,the nomogram chart has higher

关 键 词：食管癌 SEER数据库 肿瘤转移 列线图 

分 类 号：R735.1[医药卫生—肿瘤]