肺腺癌并发脑转移危险因素分析及预测模型构建

Risk factors analysis and prediction model construction of adenocarcinoma of the lung complicated with brain metastasis

目的探究肺腺癌患者并发脑转移的独立危险因素并构建列线图的预测模型。方法选取浙江省医疗健康集团杭州医院2017年1月至2022年1月收治的肺腺癌患者235例,其中145例患者诊断为肺腺癌并发脑转移,为转移组,剩余未出现脑转移的90例患者为未转移组。对两组患者的一般资料进行统计学分析并构建Nomogram预测模型。结果两组患者的性别、T分期、肿瘤直径、进行手术治疗、Karnofsky功能状态(Karnofskyperformancestatus,KPS)评分比较差异均有统计学意义(均P<0.05)。将患者一般临床资料比较差异有统计学意义的连续变量进行受试者操作特征(receiver operating characteristic,ROC)曲线分析。结果显示肿瘤直径、KPS评分的曲线下面积(area under the curve,AUC)分别为:0.866、0.910;最佳截断值分别为:>6.36cm、≤75分。Logistic多元回归分析显示:肿瘤直径>6.36cm、KPS评分≤75、化疗(否)、手术治疗(否)、T分期T_(3)~T_(4)、性别、出现其他部分转移是肺腺癌患者合并脑转移发生的独立危险因素(均P<0.05)。将是否进行化疗、KPS评分、是否出现其他部位转移、T分期、性别、肿瘤直径、是否接受手术治疗作为构建Nomogram模型的预测因子。内部验证结果显示,Nomogram模型预测肺腺癌患者并发脑转移风险C指数为0.694(95%CI:0.213~1.858)。校准曲线显示观测值与预测值间一致性较好。Nomogram模型预测肺腺癌患者并发脑转移风险阈值>0.14,Nomogram模型提供临床净收益。结论本研究构建的列线图预测模型与实际观测数据的结果具有良好的一致性,可用于预测肺腺癌患者并发脑转移的发生。

Objective To explore the independent risk factors of patients with lung adenocarcinoma complicated with brain metastases,and construct the prediction model of the graph.Methods A total of 235 patients with lung adenocarcinoma admitted to Zhejiang Medical and Health Group Hangzhou Hospital from January 2017 to January 2022 were selected,of which 145 patients were diagnosed with lung adenocarcinoma complicated with brain metastasis,and were included in the metastatic group.The remaining 90 patients without brain metastasis were in the non-metastatic group.The general clinical data of the study samples were analyzed statistically and an Nomogram prediction model was established.Results There were statistically significant differences in gender,T stage,tumor size,surgical treatment and Karnofsky performance status(KPS)scores between the two groups(all P<0.05).Receiver operating characteristic(ROC)was analyzed with statistically significant continuous variables in the comparison table of general clinical data of patients.The results showed that the area under the curve(AUC)of tumor size and KPS score were 0.866 and 0.910,respectively.The optimal truncation values were>6.36cm,≤75,respectively.The single factor difference index was incorporated into the Logistic multiple regression model,and the results showed that:Tumor size>6.36cm,KPS score≤75,chemotherapy(no),surgical treatment(no),T stage(T3-T4),gender,and other partial metastases were independent risk factors for lung adenocarcinoma patients with brain metastases(all P<0.05).Whether chemotherapy was performed,KPS score,metastasis to other sites,T stage,sex,tumor size,and surgical treatment were used as predictors to construct a Nomogram model.The results of internal verification showed that the C-index of Nomogram model was 0.694(95%CI:0.213-1.858).The calibration curve shows a good agreement between the observed and predicted values.This Nomogram model predicted that the risk threshold of brain metastases in patients with lung adenocarcinoma was>0.14.In addition,the net clinical benefits of Nomogram models were higher than whether chemotherapy was performed,KPS score,other site metastasis,T stage,sex,tumor size,and whether surgery was performed.Conclusion The prediction model of line graph constructed in this study was in good agreement with the results of actual observation data.It can be used to predict the occurrence of brain metastases in lung adenocarcinoma patients.