目的 构建预测乳腺大汗腺癌(breast apocrine carcinoma,BAC)患者总生存情况的列线图.方法 从监测、流行病学和最终结果(Surveillanc,Epidemiology,and End Results,SEER)数据库中选择2010-2016年期间诊断为BAC的患者,按7∶3比例随机分为训练集和验证集;另收集2010年1月1日至2018年12月31日期间国药东风总医院收治的BAC患者对列线图进行外部验证.采用单因素和多因素Cox回归模型来确定影响BAC患者总生存期的风险因素并以此构建列线图预测模型.用C指数和受试者操作特征曲线下面积(area under curve,AUC)评价列线图对BAC患者3、5年总生存情况的区分能力,采用校准曲线评估列线图的预测情况与实际发生情况的接近程度.结果 从SEER数据库中共纳入符合本研究条件的BAC患者649例(其中训练集454例、内部验证集195例),纳入了国药东风总医院的21例BAC患者(外部验证集).多因素Cox回归分析发现,BAC患者的年龄、T分期、M分期、S分期、手术方式及化疗是与BAC患者预后有关的重要影响因素(P＜0.05),基于这些因素构建的列线图预测模型在训练集、内部验证集和外部验证集中的C指数值分别为0.76、0.77和0.88,校准曲线显示列线图预测的3年和5年总生存率与实际3年和5年总生存率均接近理想曲线.列线图模型评估BAC患者的3年和5年总生存期的AUC(95％CI)在训练集分别为0.84(0.78,0.89)和0.76(0.71,0.83),在内部验证集分别为 0.81(0.73,0.91)和 0.84(0.77,0.91),在外部验证集分别为 0.80(0.70,0.91)和 0.84(0.76,0.91).结论 本研究基于SEER数据库构建的预测BAC患者总生存情况的列线图对BAC患者预后预测具有一定的价值.
A nomogram for predicting overall survival among patients with breast apocrine carcinoma:a study based on SEER database and internal and external validation
Objective To develop and validate a nomogram for predicting overall survival among patients with breast apocrine carcinoma(BAC).Methods The patients diagnosed with BAC from 2010 to 2016 were selected from the Surveillance,Epidemiology,and End Results(SEER)database and then randomly divided into a training set and a validation set by a 7∶3 ratio.Additionally,external validation of the nomogram was conducted on BAC patients admitted to the Dongfeng Hospital Affiliated to Hubei Medical College from January 1,2010 to December 31,2018.The risk factors affecting the overall survival of BAC patients were determined by univariate and multivariate Cox regression analyses,which were used to develop the nomogram prediction model.The discriminative abilities of the nomogram for the 3-and 5-year overall survival rates were evaluated by the C-index and area under receiver operating characteristic curve(AUC),and the fit of actual data and nomogram-predicted data for calibrators should be evaluated.Results A total of 649 BAC patients who met the included criterion for this study were enrolled from the SEER database(including 454 in the training set and 195 in the internal validation set),and 21 BAC patients from the Dongfeng Hospital(external validation set)were included.The multivariate Cox regression analysis showed that the age,T stage,M stage,S stage,surgical method,and chemotherapy were the risk factors affecting the overall survival of BAC patients.The C-index values of the nomogram prediction model based on these risk factors was 0.76,0.77,and 0.88 in the training set,internal validation set,and external validation set,respectively.The calibration curves of the actual 3-and 5-year overall survival rates and nomogram-predicted 3-and 5-year overall survival rates were close to the ideal curve.The AUCs(95％CI)of the nomogram prediction model for evaluating the 3-year and 5-year overall survival rates of BAC patients were 0.84(0.78,0.89)and 0.76(0.71,0.83)in the training set,0.81(0.73,0.91)and 0.84(0.77,0.91)in the internal validation set,and 0.80(0.70,0.91)and 0.84(0.76,0.91)in the external validation set,respectively.Conclusion Nomogram based on the SEER database to predict the overall survival of BAC patients has a good predictive effect for BAC patients.
NETL
NSTL
万方数据
