基于SEER数据库的下咽癌预后模型的建立与分析

Establishment and analysis of prognosis model for hypopharyngeal carcinoma based on SEER database

目的分析影响下咽癌预后的因素,建立基于监测、流行病学和结果(SEER)数据库的合理预后模型。方法回顾性分析SEER数据库2010至2015年被诊断为下咽癌的165例患者的临床病理资料,通过单因素和多因素Cox比例风险回归模型筛选出与预后相关的影响因素,采用R语言构建绘制列线图,并分析其预测价值。结果165例下咽癌患者的1、3、5 a的肿瘤特异生存率分别为69%(95%CI:61%~77%)、46%(95%CI:38%~54%)和44%(95%CI:36%~52%)。单因素和多因素Cox回归分析显示,对下咽癌预后产生影响的独立危险因素是年龄、术后放疗情况、T分期、M分期和肿瘤直径,基于此5项影响因素,成功构建了下咽癌的列线图预后模型(C-index=0.761,95%CI:0.708~0.814),优于TNM分期系统(C-index=0.739,95%CI:0.682~0.796),且校准曲线显示患者1、3、5a的预测生存率与实际生存率之间一致性尚可。此外,受试者工作特征曲线结果显示,该列线图预后模型[曲线下面积(AUC)=0.863,95%CI:0.799~0.928]预测能力优于TNM分期系统(AUC=0.789,95%CI:0.712~0.866)。结论本研究基于SEER数据库建立的下咽癌患者预后列线图模型有利于临床评估和指导临床决策。

Objective To analyze the factors affecting the prognosis of hypopharyngeal carcinoma and establish a reasonable prognostic model based on surveillance,epidemiology,and end results(SEER)database.Methods The clinical data of 165 patients with hypopharyngeal carcinoma diagnosed in SEER database from 2010 to 2015 were analyzed retrospectively.The prognostic factors were screened by univariate and multivariate Cox proportional hazard regression models.The nomogram was constructed and drawn by using R language,and its predictive value was analyzed.Results The tumor-specific survival rates of 165 patients with hypopharyngeal carcinoma were 69%(95%CI:61%-77%),46%(95%CI:38%-54%)and 44%(95%CI:36%-52%),respectively.Univariate and multivariate Cox regression analysis showed that age,postoperative radiotherapy,T stage,M stage and tumor size were independent risk factors affecting the prognosis of hypopharyngeal carcinoma.Based on these five factors,a nomogram prognosis model for hypopharyngeal carcinoma was successfully constructed(C-index=0.761,95%CI:0.708-0.814),which was superior to TNM staging system(C-index=0.761,95%CI:0.682-0.796),and the calibration curve showed that the predicted survival rate of patients in 1,3 and 5 years was consistent with the actual survival rate.In addition,the results of the receiver operator characteristic curve showed that the nomogram prognosis model[area under the curve(AUC)=0.863,95%CI:0.799-0.928]was superior to TNM staging system(AUC=0.789,95%CI:0.712-0.866).Conclusion In this study,the nomogram prognosis model of patients with hypopharyngeal carcinoma based on SEER database is helpful for clinical evaluation and guidance of clinical decision-making.