基于SEER数据库的口腔鳞状细胞局部晚期癌预后模型构建

A prognostic model of locally advanced oral squamous cells carcinoma constructed based on SEER database

目的:构建口腔鳞状细胞癌(OSCC)局部晚期癌患者列线图预后模型并进行验证。方法:纳入SEER数据库中2010~2015年OSCC局部晚期癌患者7118例临床病例资料,按7∶3比例随机分为训练集和验证集。在训练集中采用单因素及多因素Cox比例风险模型筛选并确定OSCC局部晚期癌的预后因素,构建列线图模型预测患者3年和5年的特异性生存率(CSS)。在训练集和验证集中,采用一致性指数(C-index)和ROC曲线评估模型的区分度,采用校准曲线评估模型的预测准确度,使用决策曲线分析法(DCA)评估模型的临床适用度。结果:训练集和验证集分别包括患者4984例和2134例,两集中各项统计指标差异无统计学意义(P>0.05),在训练集中,确定了10个影响OSCC局部晚期癌患者预后的风险因素,包括年龄、性别、人种、婚姻状态、发病部位、组织学分化程度、T分期、N分期、是否手术、是否放疗(P<0.05)。训练集和验证集中,列线图模型C-index分别为0.702和0.704,预测3年和5年CSS的AUC在0.727~0.736之间。校准曲线显示模型预测的3年和5年CSS与实际CSS吻合较好。DCA曲线显示列线图模型净收益优于第7版AJCC TNM分期系统。结论:本研究构建的列线图模型能较为准确地预测OSCC局部晚期癌患者生存率。

Objective:To construct and verify a prognosis model of nomogram for the patients with locally advanced oral squamous cell carcinoma(OSCC).Methods:7118 clinical cases with locally advanced OSCC from 2010 to 2015 in SEER database were included and randomly divided into training set(TS)and validation set(VS)by a 7∶3 ratio.In TS,univariate and multivariate Cox proportional risk models were used to screen and determine the prognostic factors of locally advanced OSCC,and a nomogram prediction model was constructed to predict the cancer specific survival rate(CSS)of patients in 3 and 5 years.In TS and VS,the consistency index(C-index)and ROC curves were used to evaluate the differentiation of the model,the calibration curves were used to evaluate the prediction accuracy of the model,and the decision curve analysis(DCA)method was used to evaluate the clinical applicability of the model.Results:4984 and 2134 cases were respectively included in TS and VS.No statistical difference of the analysis indexes was found between the 2 sets(P>0.05).10 risk factors affecting the prognosis of the patients with locally advanced OSCC were identified in TS,including age,sex,race,marital status,location of onset,degree of histological differentiation,T stage,N stage,with or without operation and radiotherapy(P<0.05).In the TS and VS,the C-index of the nomogram was 0.702 and 0.704 respectively,and the AUC of the 3-year and 5-year CSS was between 0.727 and 0.736.The calibration curves showed that the 3-and 5-year CSS predicted by the nomogram model was in good agreement with the actual CSS.The DCA curves showed that the clinical benefit of the nomogram was better than that of the 7th edition of the AJCC TNM staging system.Conclusion:The nomogram constructed in this study can accurately predict the survival rate of OSCC patients with locally advanced tumor.