基于SEER数据库的阴茎癌预测模型的建立及验证

Development and validation of a nomogram model for penile cancer based on SEER database

目的探讨影响阴茎癌(Penile cancer,PC)患者的预后危险因素,构建可靠的预测模型并进行验证。方法分析2004年至2015年美国监测、流行病学和最终结果数据库(The Surveillance,Epidemiology,and End Results,SEER)中被诊断为PC的患者,应用Cox回归分析来筛选独立预后危险因素,构建列线图并使用受试者工作特征曲线(Receiver-operating characteristic curve,ROC)、校准曲线、决策曲线分析法(Decision curve analysis,DCA)及生存曲线评估列线图预测能力。结果本研究最终纳入1357例PC患者,包括训练集945例和验证集412例。总生存期(Overall survival,OS)的独立危险因素为年龄、肿瘤大小、婚姻状况、Grade分级、N分期、M分期。癌症特异性生存期(Cancer special survival,CSS)相关的独立危险因素为肿瘤大小、Grade分级、T分期、N分期、M分期。结合以上因素,我们构建了1年、3年、5年和8年的预后列线图,在OS和CSS中,发现N分期对PC患者的预后影响最大。ROC曲线、校准曲线、DCA曲线及生存曲线均表明我们的模型具有更好的预测准确性和临床实用性。结论本研究基于SEER数据库建立的预测模型能够较好地预测PC患者预后。

Objective To analyze the risk factors affecting the survival prognosis of patients with penile cancer(PC),construct a reliable predictive model and validate it.Methods We analyzed clinical data of the patients diagnosed with PC in the Surveillance,Epidemiology,and End Results(SEER)database from 2004 to 2015.Then we applied Cox regression to screen independent influencing factors,constructed a nomogram model and assessed the predictive power of the nomogram using the receiver operating characteristic(ROC),calibration curve,receiver-operating characteristic curve(DCA)and survival curve results.Results Total of 1357 PC patients were finally enrolled in the study,including 945 in the training set and 412 in the internal validation set.In multivariate analysis,related factors were age,tumor size,marital status,Grade,N-stage,M-stage,(P<0.05),and related factors were tumor size,Grade,T-stage,N-stage,M-stage,(P<0.05).Combining the above factors,we respectively constructed prognostic nomograms at 1 year,3 years,5 years and 8 years.In OS and CSS,N-stage showed the greatest impact on the poor prognosis of PC patients.The ROC curve,calibration curve,DCA curve and survival curve indicated that our model had better predictive accuracy and clinical utility.Conclusions The predictive model developed in this study based on the SEER database was able to better predict the prognosis of PC patients.