基于SEER数据库构建肺癌骨转移预测模型

目的探讨影响肺癌骨转移的相关因素,建立基于SEER数据库的肺癌骨转移预测模型,并对此模型预测价值进行验证方法从美国医疗保险监督、流行病学和最终结果(SEER)数据库中收集2010年1月至2015年12月肺癌骨转移患者的相关资料,利用卡方检验对建模组中相关变量进行单因素分析,同时利用Lasso回归分析进行变量筛选,将有意义的变量纳入多因素logistic回归模型进行分析,以此为基础构建Nomogram图,并对Nomogram模型进行内部和外部验证结果从SEER数据库中共提取到74,466例患者信息,分成建模组(n=52,128)和验证组(n=22,338)建模组的多因素分析结果显示年龄、性别、婚姻状况、病理类型、组织学等级、手术情况、放疗情况、T分期、N分期、转移情况是肺癌骨转移发生的独立影响因素(P<0.05),这些因素被用于构建Nomogram图该Nomogram模型的内部和外部验证结果显示:建模组和验证组的一致性指数(Oindex)分别为0.835(95%CI:0.831〜0.839),0.836(95%CI:0.829〜0.843),建模组和验证组的受试者工作特征曲线(ROC)曲线下面积(AUC)为0.819,0.820结论年龄、性别、婚姻状况、病理类型、组织学等级、手术情况、放疗情况、T分期、N分期、转移情况是肺癌骨转移发生的独立影响因素,本研究构建的预测肺癌骨转移的Nomogram模型具有良好的预测价值。

Objective To investigate relevant factors which effect hone metastasis of lung cancer and constnict a nomogram for prediction based on Surveillance,Epidemiology,and End results database(SEER),and verify the nomogram's predictive values.Methods The relative data of patients who had been diagnosed as bone metastasis of lung cancer from 2010-2015 in database(SEER)were collected and divided into modeling group and validation group.The relative variables were analyzed by the chi-square test in the modeling group,meanwhile,they were selected by Lasso regression analysis.The meaningful factors were incorporated into multiple factors logistic models,and b«ised on the those,a nomogram was constructed.Internal and external validation were taken to test the nomogram.Results The clinical data of 74,466 patients were extracted from SEER database and divided into modeling group(n=52,128)and validation group(n=22,338).The multivariate an.ilysis showed that age,sex,marriage,histology,grade,surgery,radiation,T-staging,N-staging,metastasis were independent factors which effected bone metastasis of lung cancer(P<0.05),and these factors were incorporated to constnict a nomogram.The results of internal and external validation of the nomogram showed th.it,the C-index of modeling group and validation group were 0.835(95%CI:0.831~0.839),0.836(95%CI:0.829~0.843)respectively.The area under the curve(AUC)of modeling group's and validation group's receiver operating characteristic curve(ROC curve)were 0.819 and 0.820,respectively.Conclusion Age,sex,marriage,histology,grade,surgery,radiation,T-staging,N-staging,metastasis are independent factors which effect bone metastasis oflung cancer.The nomogram constmcted in this study has a good forecasting.ibility.