基于SEER数据库构建临床局限性前列腺癌根治术后包膜外侵犯的预测模型  

Construction of a predictive model for extracapsular extension after radical prostatectomy in clinically localized prostate cancer based on SEER database

作　　者：黄志恒 许长宝[1] 徐寒 张天贺 魏海洋 高俊峰 樊长晖[1] Huang Zhiheng;Xu Changbao;Xu Han;Zhang Tianhe;Wei Haiyang;Gao Junfeng;Fan Changhui(Department of Urology,The Second Affiliated Hospital of Zhengzhou University,Zhengzhou 450000,China;Zhengzhou University,Zhengzhou 450000,China)

机构地区：[1]郑州大学第二附属医院泌尿外科,郑州450000 [2]郑州大学,郑州450000

出　　处：《中华泌尿外科杂志》2025年第3期180-187,共8页Chinese Journal of Urology

基　　金：河南省医学科技攻关计划联合共建项目(LHGJ20240314)。

摘　　要：目的基于SEER数据库,探讨临床局限性前列腺癌患者根治性前列腺切除术(RP)后发生包膜外侵犯(EPE)的独立影响因素,建立列线图模型并进行外部验证。方法提取2010-2021年SEER数据库中接受RP的20916例临床局限性前列腺癌患者(T_(1~2)N_(0)M_(0))的临床病理资料。患者年龄(61.71±7.09)岁;已婚17835例(85.3%);前列腺特异性抗原(PSA)<4 ng/ml 2243例(10.7%),≥4且<10 ng/ml 14831例(70.9%),≥10且<20 ng/ml 2965例(14.2%),≥20 ng/ml 899例(4.3%);临床分期T_(1)期14870例(71.1%),T_(2)期6046例(28.9%);根治术后病理分期T_(1)期48例(0.2%),T_(2)期15794例(75.5%),T_(3)期5001例(23.9%),T_(4)期73例(0.3%)。采用分层抽样方法,将SEER数据库中的患者按7∶3分为训练组和内部验证组。收集2019年9月至2024年9月郑州大学第二附属医院收治的接受RP的75例临床局限性前列腺癌患者作为外部验证组;年龄(65.39±7.45)岁,已婚73例(97.3%);PSA<4 ng/ml 2例(2.7%),≥4且<10 ng/ml 17例(22.7%),≥10且<20 ng/ml 34例(45.3%),≥20 ng/ml 22例(29.3%)。临床分期T_(1)期47例(62.7%),T_(2)期28例(37.3%);根治术后病理分期T_(1)期7例(9.3%),T_(2)期48例(64.0%),T_(3)期18例(24.0%),T_(4)期2例(2.7%)。根据RP术后病理结果将患者分为器官局限和EPE两组。基于SEER数据库中训练组数据,采用单因素logistic回归和逐步向后法的多因素logistic回归分析EPE发生与临床病理因素间的关系,筛选出独立危险因素并建立列线图模型。在训练组中使用受试者工作特征曲线(ROC)的曲线下面积(AUC)、校准曲线和临床决策曲线分别评估模型的区分能力、校准性能和临床净效益;在内部验证组和外部验证组中再次使用AUC、校准曲线和临床决策曲线验证模型的稳定性和泛化能力。结果训练组、内部验证组和外部验证组分别有3585例(24.5%)、1489例(23.8%)、20例(26.7%)临床局限性前列腺癌患者根治术后发生EPE。多因素logistic回归分析结果显示,术�Objective To explore the independent factors influencing extraprostatic extension(EPE)after radical prostatectomy(RP)in patients with clinically localized prostate cancer by utilizing the Surveillance,Epidemiology,and End Results(SEER)database.A nomogram model was developed and externally validated.Methods Clinical and pathological data of 20916 clinically localized prostate cancer patients(T_(1-2)N_(0)M_(0))who underwent RP between 2010 and 2021 were extracted from the SEER database.The mean age was(61.71±7.09)years old,and a total of 17835 patients(85.3%)were married.There were 2243 patients(10.7%)with prostate-specific antigen(PSA)<4 ng/ml,14831 patients(70.9%)with≥4 and<10 ng/ml,and 2965 patients(14.2%)with≥10 and<20 ng/ml.There were 14870 patients(71.1%)with clinical staging of stage T_(1),and 6046 patients(28.9%)with T_(2).There were 48 patients(0.2%)with pathological staging of stage T_(1),15794(75.5%)with T_(2),5001(23.9%)with T_(3),and 73(0.3%)with T_(4) stage after radical surgery.The patients of SEER database were divided into training and internal validation groups in a 7∶3 ratio by using stratified sampling.Additionally,data were collected for 75 clinically localized prostate cancer patients who underwent RP at the Second Affiliated Hospital of Zhengzhou University from September 2019 to September 2024,serving as the external validation group.The mean age was(65.39±7.45)years old.Among them,73(97.3%)were married.There were 2 patients(2.7%)with PSA<4 ng/ml,17 patients(22.7%)with≥4 and<10 ng/ml,and 34 patients(45.3%)with≥10 and<20 ng/ml.There were 47 patients(62.7%)with clinical staging of stage T_(1),and 28 patients(37.3%)with T_(2).There were 7 patients(9.3%)with pathological staging of stage T_(1),48 patients(64.0%)with T_(2),18 patients(24.0%)with T_(3),and 2 patients(2.7%)with T_(4) stage after radical surgery.All patients were categorized into organ-confined(OC)and EPE groups based on post-surgical pathology.Univariate and multivariate logistic regression analyses,with a stepwise back

关 键 词：前列腺肿瘤 癌 临床局限性 包膜外侵犯 预测模型 SEER数据库 

分 类 号：R737.25[医药卫生—肿瘤]