开发和验证预测脑肿瘤术后消化道出血的列线图:单中心研究

Development and validation of a novel nomogram for predicting gastrointestinal bleeding after brain tumor surgery:a single-center study

目的构建脑肿瘤术后消化道出血的列线图用于术前精确预测脑肿瘤患者术后消化道出血的可能性。方法选择自2021年1~12月于华中科技大学同济医学院附属协和医院神经外科行颅脑肿瘤切除术的患者为研究对象,按照3∶1分层随机抽样原则将患者分为训练集和测试集。根据训练集患者术后住院期间是否发生消化道出血分为消化道出血组和无消化道出血组,采用差异分析和多因素Logistic回归分析研究脑肿瘤术后继发消化道出血的独立影响因素,基于此构建脑肿瘤术后继发消化道出血的临床预测模型列线图。通过校准曲线、临床有效性以及内部测试集来评估临床模型的性能。结果本组共纳入400例脑肿瘤患者,训练集300例,其中脑肿瘤术后出现消化道出血24例(消化道出血组),未出现消化道出血276例(无消化道出血组);测试集100例,其中消化道出血8例,未出现消化道出血92例。2组患者的年龄、肿瘤位置、消化道疾病史、冠心病史、非甾体药物服用史、饮酒史和手术时间比较,差异均有统计学意义(P<0.05)。多因素Logistic回归分析结果显示年龄、非甾体药物服用史、肿瘤位置和手术时间是脑肿瘤患者术后出现消化道出血的独立影响因素,据此构建列线图模型。该模型在训练集中的AUC值为0.817,测试集中的AUC值为0.806,呈现出良好的预测性能和稳定性能。此外校准曲线证实了列线图拟合效果良好。结论通过年龄、非甾体药物服用史、肿瘤位置和手术时间等因素构建的列线图,可简单、有效地预测脑肿瘤患者术后出现消化道出血,将为颅脑肿瘤患者积极预防术后消化道出血提供依据。

Objective To construct a nomogram for accurately predict the possibility of postoperative gastrointestinal bleeding in brain tumor patients.Methods Patients who underwent brain tumor resection at the Neurosurgery Department of Union Hospital,Tongji Medical College,Huazhong University of Science and Technology from January to December 2021 were selected as the research subjects.According to the 3∶1 stratified random sampling principle,the patients were divided into a training set and a testing set.The patients of testing set were divided into two groups based on whether they experienced gastrointestinal bleeding during postoperative hospitalization:the gastrointestinal bleeding group and the non-gastrointestinal bleeding group.Differential analysis and multivariate Logistic regression analysis were used to study the independent influencing factors affecting secondary gastrointestinal bleeding after brain tumor surgery.A clinical prediction model for secondary gastrointestinal bleeding after brain tumor surgery was constructed,and the model is presented as a nomogram.The performance of clinical models evaluated through calibration curves,clinical validity,and internal test sets.Results A total of 400 patients with brain tumors were included,with 300 in the training set.Among them,24 patients had gastrointestinal bleeding after brain tumor surgery(gastrointestinal bleeding group),and 276 patients did not have gastrointestinal bleeding(non-gastrointestinal bleeding group);There were 100 cases in the testing set,including 8 cases of gastrointestinal bleeding and 92 cases without gastrointestinal bleeding.The age,tumor location,history of gastrointestinal diseases,history of coronary heart disease,history of non-steroidal drug use,history of alcohol consumption,and surgical time of the two groups of patients were compared,and the differences were statistically significant(P<0.05).The results of multivariate Logistic regression analysis showed that age,history of non-steroidal drug use,tumor location,and operation time were independent influencing factors for postoperative gastrointestinal bleeding in brain tumor patients.Based on this,a nomogram model was constructed.The model showed good prediction and stability performance,with the AUC value of 0.817 in the training set and 0.806 in the testing set,respectively.In addition,the calibration curve confirmed that the nomogram fits very well for the real results.Conclusion A nomogram constructed based on factors such as age,history of non-steroidal drug use,tumor location,and surgical time can easily and effectively predict postoperative gastrointestinal bleeding in patients with brain tumors,providing a basis for active prevention of postoperative gastrointestinal bleeding in patients with brain tumors.