基于SEER数据库膀胱腺癌预后多因素分析及列线图构建

  目的  通过研究膀胱腺癌(adenocarcinoma of the bladder，ACB)患者独立预后危险因素，建立可以个体化预测膀胱腺癌患者生存预后的列线图模型。  方法  通过收集SEER数据库中2004—2015年诊断为膀胱腺癌患者的临床资料进行筛选，共纳入符合研究条件的腺癌患者659例进行回顾性分析研究，采用Empower Stat软件对所有变量进行COX单因素回归分析，将P < 0.05的变量纳入COX多因素回归分析，使用R语言将COX多因素回归分析中的独立预后危险因素构建成生存列线图模型，运用Bootstrap法进行1 000次等量有放回重复采样对模型进行验证，采用C-index和绘制列线图校准曲线来验证模型的准确性。  结果  年龄>70岁、肿瘤直径>3 cm、未行手术治疗、T3期、T4期、N1期、N2期、M1期为影响患者生存时间的独立危险因素。建立的列线图校准曲线与理想曲线一致性较高，并采用C-INDEX(一致性指数)进行验证，C-INDEX指数为0.74，经内部验证后具有较高的适用性与准确性。  结论  本研究通过SEER数据库确定了ACB患者预后的独立危险因素，并且建立了国内外首个可以进行个体化预测ACB患者生存预后的列线图模型，从而指导临床医师选择合理的治疗方案，并对不同患者预后进行精准的个体化评估。 

Multivariate analysis of prognosis and nomogram construction of cystadenocarcinoma based on SEER database

Objective This study aimed to establish a nomogram model that can individually predict the prognosis of patients with bladder adenocarcinoma(ACB)by studying the independent prognostic risk factors.Methods Clinical data of patients diagnosed with ACB from 2004 to 2015 in the SEER database were collected for screening.A total of 659 patients with adenocarcinoma who met the study conditions were included for retrospective analysis.COX univariate regression analysis was performed on all variables using Empower stat software,and P<0.05 was analysed.Variables of 0.05 were included in COX multivariate regression analysis.Independent prognostic risk factors determined in COX multivariate regression analysis were constructed into a survival nomogram model using R language.The bootstrap method was used to perform 1000 times of equal number of repeated sampling to verify the model.Results Age>70 years,tumour size>3 cm,T3,T4,N1,N2 and M1 were independent risk factors affecting the survival time of patients.The established alignment curve of the nomogram has a high consistency with the ideal curve and high applicability and accuracy after internal verification.Conclusion In this study,the seer database was used to determine the independent risk factors for the prognosis of ACB patients,and the first nomogram model was established to individually predict the survival prognosis of ACB patients and guide clinicians in selecting reasonable treatment options and conducting accurate individualised assessment of the prognosis of different patients.