基于SEER数据库和多因素逻辑回归的结直肠癌肝转移预测研究  

Development of a prediction model for liver metastases in colorectal cancer using the SEER database and multivariate logistic regression

作　　者：鲁明[1] 李娟[2] Lu Ming;Li Juan(General Surgery Department,the Second Affiliated Hospital of Nanjing Medical University,Jiangsu Province,Nanjing 210029,China;Oncology Department,the Second Affiliated Hospital of Nanjing Medical University,Jiangsu Province,Nanjing 210029,China)

机构地区：[1]南京医科大学第二附属医院普外科,南京210029 [2]南京医科大学第二附属医院肿瘤科,南京210029

出　　处：《中国肿瘤临床与康复》2025年第3期182-190,共9页Chinese Journal of Clinical Oncology and Rehabilitation

基　　金：江苏省卫生健康委科研项目(K2023024)。

摘　　要：目的 利用SEER数据库,对结直肠癌肝转移患者的临床特征进行描述,分析肝转移发生的危险因素,并构建多因素logistic回归模型分析肝转移患者的预后影响因素,为结直肠癌肝转移的预防和治疗提供科学依据。方法 纳入2015—2017年SEER数据库中结直肠癌患者,收集患者人口学和临床病理学特征,采用logistic回归进行多因素分析,根据多因素分析结果构建列线图预测模型并进行受试者工作特征(ROC)曲线评估。结果共纳入13 908例结直肠癌患者,包括结肠癌患者1 363例(98.17%),直肠癌255例(1.83%),共发生肝转移1379例(9.91%)。肝转移组与非肝转移组的原发肿瘤位置、年龄、婚姻状态、病理类型、Grade分级、T分期、N分期、肿瘤最大直径、癌胚抗原(CEA)和肿瘤沉积物比较,差异有统计学意义(均P<0.05)。多因素logistic分析显示,年龄>60岁(OR=1.398,95%CI:1.235~1.582)、病理类型为黏液性腺癌(OR=1.853,95%CI:1.471~2.333)、T分期越晚(OR=1.734,95%CI:1.561~1.925)、N分期越晚(OR=2.091, 95%CI:1.921~2.276)、 CEA阳性(OR=5.187, 95%CI:4.487~5.996)和有肿瘤沉积物(OR=1.905,95%CI:1.652~2.195)是影响结直肠癌发生肝转移的危险因素。基于此构建的列线图预测模型,ROC曲线下面积为0.837(95%CI:0.827~0.847),提示模型预测性能良好。预测模型的校正曲线与理想曲线接近。结论构建包含多个常规临床指标的结直肠癌肝转移预测列线图模型,预测效能较好,可为结直肠癌患者的个体化诊疗决策提供参考。未来需在前瞻性队列中进行大规模外部验证和进一步优化,为结直肠癌精准治疗提供支持。Objectives This study utilized the SEER(Surveillance,Epidemiology,and End Results)database to describe the clinical characteristics of colorectal cancer patients with liver metastases,identify risk factors for liver metastases,and construct a multivariate logistic regression model to investigate prognostic factors.The findings aim to provide a scientific foundation for the prevention and management of colorectal cancer liver metastases.Methods Data from 13908 colorectal cancer patients in the SEER database(2015-2017)were analyzed.Demographic and clinicopathological characteristics were collected.Univariate analyses were conducted using the x?test and t-test,and multivariate logistic regression was employed for further analysis.A predictive nomogram was developed based on the results of multivariate analysis and evaluated using the area under the receiver operating characteristic(ROC)curve(AUC).Results The incidence of liver metastasis in colorecta1lcancer patientswas9.91%.Primaryy tumor location,age,marital status,pathological type,Tumor grade,T stage,N stage,maximum tumor diameter,CEA and tumor deposits were statistically different between the liver and non-liver metastasis groups(P<0.05).Multivariate analysis indicated that age>60(OR=1.398,95%CI:1.235-1.582),pathological type was mucinous adenocarcinoma(OR=1.853,95%CI:1.471-2.333),T stage(OR=1.734,95%CI:1.561-1.925),N stage(OR=2.091,95%CI:1.921-2.276),Positive CEA status(OR=5.187,95%CI:4.487-5.996)and the presence of tumor deposits(OR=1.905,95%Cl:1.652-2.195)were risk factors for liver metastasis in colorectal cancer.TThe nomogram prediction model constructed based on these factors demonstrated excellent predictive performance,with an AUC of 0.837(95%CI:0.827-0.847),andthe calibration curve of the prediction model was close to the ideal curve.Conclusions This study developed a nomogram prediction model that integrates routine clinicopathological indicators to effectively predict liver metastases in colorectal cancer patients.The model exhibited robust predictiv

关 键 词：结直肠癌 肝转移 SEER数据库 危险因素 预测模型 

分 类 号：R735.3[医药卫生—肿瘤]