脉络膜黑色素瘤患者的预后影响因素分析及临床预测模型构建  

Analysis of prognostic factors and nomogram construction for choroidal melanoma

作　　者：曹贤芬 周小平[1] 欧玉仑[1] Cao Xianfen;Zhou Xiaoping;Ou Yuun(Department of Ophthalmology,the First People's Hospital of Chenzhou,Chenzhou 423000,China)

机构地区：[1]郴州市第一人民医院眼科,郴州423000

出　　处：《中华眼外伤职业眼病杂志》2024年第4期241-248,共8页Chinese Journal of Ocular Trauma and Occupational Eye Disease

基　　金：湘南学院校级项目(2019XJ63)。

摘　　要：目的分析Ⅰ~Ⅲ期脉络膜黑色素瘤患者的预后影响因素,构建临床预测模型。方法横断面研究。收集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2004—2015年确诊为脉络膜黑色素瘤3724例的临床资料。采用简单随机抽样按照7∶3的比例,将患者分成训练集(2606例)与验证集(1118例)。应用多因素Cox回归分析及逐步法筛选Ⅰ~Ⅲ期脉络膜黑色素瘤患者预后的独立危险因素,并构建列线图模型。用一致性指数(C-index)、受试者工作特征曲线(ROC)曲线下面积(AUC)、校准曲线、净重新分类指数(NRI)、综合判别改善指数(IDI)和决策曲线分析(DCA)评价列线图的预测性能。结果多因素Cox回归分析显示影响脉络膜黑色素瘤患者的独立危险因素,分别为年龄(>50~71岁:HR=1.534,95%CI:1.231~1.912,P<0.001;>71岁:HR=2.055,95%CI:1.612~2.618,P<0.001);组织学分型(混合细胞型:HR=1.107,95%CI:0.842~1.456,P<0.001;梭形细胞型:HR=0.471,95%CI:0.335~0.664,P<0.001;上皮细胞型:HR=1.276,95%CI:0.873~1.865,P=0.208);美国癌症联合委员会分期(AJCC)(Ⅱ期:HR=1.792,95%CI:1.462~2.196,P<0.001;Ⅲ期:HR=3.562,95%CI:2.853~4.446,P<0.001);未放疗或未知组(HR=1.384,95%CI:1.024~1.871,P=0.035);未手术或未知组(HR=0.658,95%CI:0.506~0.855,P=0.002);未化疗或未知组(HR=0.516,95%CI:0.342~0.778,P=0.002)。训练集及验证集的C-index分别为0.700、0.705。校准曲线显示在训练集与验证集中3和5年预测模型的预测概率与观测概率基本一致。训练集3年、5年AUC分别为:0.752、0.733,验证集AUC分别为:0.751、0.736。训练集3年、5年的NRI值分别为:0.231(95%CI:0.175~0.381),0.280(95%CI:0.219~0.397);验证集NRI值:3年0.350(95%CI:0.206~0.459),5年0.453(95%CI:0.277~0.549)。训练集3年、5年的IDI值分别为0.026和0.035,验证集IDI值3年为0.038,5年为0.040。DCA显示预测模型在不同阈值下提供的净收益均高于AJCC分期系统。结论基于年龄、AJCC分期、组织学分型ObjectiveTo analyze the prognostic factors affecting stageⅠ-Ⅲchoroidal melanoma patients and construct a clinical prediction model.MethodsThis was a cross-sectional study.Clinical data of 3724 cases diagnosed with choroidal melanoma from 2004 to 2015 in the surveillance,epidemiology,and end results(SEER)database were analyzed.Using simple random sampling in a 7∶3 ratio,patients were divided into a training cohort(2606 cases)and a validation cohort(1118 cases).Multivariate Cox proportional hazards regression analyses and stepwise selection method were used to select independent risk factors for prognosis in patients with stageⅠ-Ⅲchoroidal melanoma.Nomogram was then constructed based on independent variables.The performance of the nomogram were validated by concordance index(C-index),receiver operating characteristic curves(ROC),area under the curve(AUC),calibration plot,net re-classification index(NRI),integrated discrimination improvement(IDI)and decision curve analysis(DCA).ResultsMultivariate Cox regression analyse showed the following independent risk factors for patients with choroidal melanoma:age(>50-71 years old:HR=1.534,95%CI:1.231-1.912,P<0.001;>71 years old:HR=2.055,95%CI:1.612-2.618,P<0.001);histological type(mixed cell type:HR=1.107,95%CI:0.842-1.456,P<0.001;spindle cell type:HR=0.471,95%CI:0.335-0.664,P<0.001;epithelioid cell type:HR=1.276,95%CI:0.873-1.865,P=0.208);American Joint Committee on Cancer(AJCC)staging(stageⅡ:HR=1.792,95%CI:1.462-2.196,P<0.001;stageⅢ:HR=3.562,95%CI:2.853-4.446,P<0.001);no radiotherapy or unknown group(HR=1.384,95%CI:1.024-1.871,P=0.035);no surgery or unknown group(HR=0.658,95%CI:0.506-0.855,P=0.002);and no chemotherapy or unknown group(HR=0.516,95%CI:0.342-0.778,P=0.002).The training and validation cohorts C-index were 0.700 and 0.705.The calibration curve showed that the predicted incidence of 3-and 5-year prediction models in the training and validation cohorts were basically consistent with the actual observed rate.The 3-and 5-year AUC of nomogram were 0.7

关 键 词：脉络膜肿瘤 列线图 预后 肿瘤特异性生存率 预测模型 横断面研究 

分 类 号：R739.7[医药卫生—肿瘤]