基于SEER数据库的髓外浆细胞瘤患者生存预测模型构建及其应用价值  

Construction and application value of a survival prediction model for patientswith extramedullary plasmacytoma based on SEER database

作　　者：李雪国 古再丽努尔 韩怡天 王如民 郎涛[1] LI Xueguo;Guzailinuer;HAN Yitian;WANG Rumin;LANG Tao(Department of Hematology,People’s Hospital of Xinjiang Uygur Autonomous Region,Urumqi,Xinjiang 830001,China)

机构地区：[1]新疆维吾尔自治区人民医院血液科,乌鲁木齐830001

出　　处：《重庆医学》2024年第8期1143-1150,共8页Chongqing medicine

基　　金：新疆少数民族科技人才特殊培养计划(2021D03021);新疆维吾尔自治区人民医院院内项目(20220134)。

摘　　要：目的构建髓外浆细胞瘤(EMP)患者的生存预测模型并验证其应用价值。方法从美国国家癌症研究所监测,流行病学和最终结果(SEER)数据库中收集2000-2019年诊断为EMP的患者数据。使用随机森林算法筛选变量,并进一步采用多因素Cox比例风险回归模型确定EMP预后的独立危险因素,基于此构建列线图模型。根据列线图计算患者总分,使用X-tile软件依据最佳截断值将总样本分为低、中、高风险3组,分析3组的Kaplan-Meier生存曲线。使用时间依赖受试者工作特征(t-ROC)曲线的曲线下面积(AUC)、X年N次K折交叉验证AUC、校准曲线及临床决策曲线(DCA)评估模型的预测效能。结果共纳入1458例EMP患者,包括796例删失和662例死亡。删失组、死亡组均以年龄60~<75岁(35.4%、41.2%)、男性(63.1%、66.3%)、白种人(81.3%、80.4%)、已婚(68.2%、59.4%)为主。最终确定年龄(45~<60岁/60~<75岁/≥75岁)、婚姻状况(单身/其他)、非第一原发恶性肿瘤、未手术及未放疗是EPM患者生存的独立危险因素(P<0.05)。基于以上独立危险因素构建生存预测模型并绘制列线图,根据最佳截断值将总样本分为0~96分组(低风险组)、98~135分组(中风险组)、139~191分组(高风险组),Kaplan-Meier生存分析显示3组生存时间有明显差异(P<0.0001)。1、3、5年t-ROC曲线的AUC及交叉验证AUC均>0.7,提示该模型具有较好的区分度;校准曲线提示预测与实际总生存率之间具有较好的一致性;DCA提示该模型可以改善临床收益。结论基于SEER数据库构建的EMP患者生存预测模型可协助临床医生识别预后风险因素并预测患者总生存率。Objective To construct a survival prediction model for the patients with extramedullary plasmacytoma(EMP),and to evaluate its application value.Methods The data of the patients diagnosed as EMP from 2000 to 2019 were collected from the National Cancer Institute’s Surveillance,Epidemiology,and End Results(SEER)database.Random forest algorithm was used to screen variables,multivariate Cox proportional hazards regression model was used to determine the independent risk factors of EMP prognosis,and a nomogram model was constructed based on these factors.The total score of the patients was calculated according to the nomogram model,and the total samples were divided into the low,medium and high risk groups according to the optimal cutoff value by using X-tile software,and the Kaplan-Meier survival curves of the three groups were analyzed.The predictive performance of the model was evaluated using area under the curve(AUC)of time-dependent receiver operating characteristic(t-ROC)curve,AUC after X years N repeated K-fold cross-validation,calibration curve,and decision curve analysis(DCA).Results A total of 1458 patients with EMP were included,including 796 cases of deletions and 662 cases of death.In the deletion and deaths groups,the majority were aged 60-<75 years old(35.4%,41.2%),male(63.1%,66.3%),white race people(81.3%,80.4%),and married(68.2%,59.4%).Finally,age(45-<60 years old/60-<75 years old/≥75 years old),marital status(single/others),non-primary malignant tumor,and without surgery and radiotherapy were determined to be the independent risk factors for the survival of EPM patients(P<0.05).Based on the above independent risk factors,a survival prediction model was constructed,and a nomogram was drawn.According to the optimal cutoff value,the total sample was divided into the score of 0-96 group(low risk group),score of 98-135 group(medium risk group)and score of 139-191 group(high risk group).The Kaplan-Meier survival analysis revealed that there was statistically significant difference in the survival time

关 键 词：髓外浆细胞瘤 SEER数据库 预后 列线图预测模型 

分 类 号：R733[医药卫生—肿瘤]