基于术前MRI的列线图模型预测脑膜瘤病理分级  

Preoperative nomogram model based on MRI predicting the pathological grade in patient with meningioma

作　　者：杜涛明[1] 任盈丽 范杰[1] 莫云海[1] 唐烨真[2] 李敏[3] DU Taoming;REN Yingli;FAN Jie;MO Yunhai;TANG Yezhen;LI Min(Department of Radiology,Chengdu Seventh People’s Hospital,Chengdu 610041,China;Southwest Minzu University,Chengdu 610041,China;Department of Radiology,the Second Affiliated Hospital of Guangxi Medical University,Nanning 530007,China)

机构地区：[1]成都市第七人民医院放射科,四川成都610041 [2]西南民族大学,四川成都610041 [3]广西医科大学第二附属医院放射科,广西南宁530007

出　　处：《实用放射学杂志》2023年第10期1569-1573,共5页Journal of Practical Radiology

基　　金：四川省医学会科研基金专项科研课题项目(2021HR28);四川省自然科学基金项目(2022NSFSC0507)。

摘　　要：目的分析术前脑膜瘤MRI征象及表观扩散系数(ADC)值,并基于MR相关参数构建预测脑膜瘤WHO病理分级的列线图模型。方法纳入脑膜瘤患者210例,其中低级别组(WHO 1级)158例,高级别组(WHO 2~3级)52例,并分析其术前临床和MRI资料。使用单因素和多因素回归分析方法进行脑膜瘤MRI特征参数对比分析,并筛选出脑膜瘤WHO病理分级的潜在预测因素和独立预测因素,进而建立WHO病理分级列线图预测模型。采用受试者工作特征(ROC)曲线、曲线下面积(AUC)、校准曲线以及决策曲线分析(DCA)评估模型性能、稳健性和临床获益。结果单因素回归分析发现,肿瘤形态、大小、瘤周水肿、脑膜尾征、强化均匀度、邻近组织侵犯以及ADC值为影响脑膜瘤WHO病理分级的潜在因素(P<0.05);多因素回归分析显示,肿瘤形态、强化均匀度、邻近组织侵犯及ADC值是脑膜瘤WHO病理分级的独立影响因素(P<0.05),然后纳入这4个MRI特征参数构建列线图模型。列线图模型有较高性能,AUC高达0.905[95%置信区间(CI)0.889~0.941]。校准曲线显示列线图模型有较高的稳健性,其预期估量值和实际观察值的吻合度较高。DCA提示模型在脑膜瘤WHO病理分级应用中的临床获益性较佳。结论基于术前MRI建立的列线图模型具有较高的预测性能,可为脑膜瘤治疗策略的制订提供参考依据。Objective To analyze the MRI features and apparent diffusion coefficient(ADC)values of meningioma before surgery,and to construct a nomogram model based on the MR parameters for predicting the WHO pathological grade of meningioma.Methods A total of 210 meningioma patients were collected,including 158 patients in the low grade group(WHO grade 1)and 52 patients in the high grade group(WHO grade 2-3),the clinical and MRI data of all meningioma patients before surgery were analyzed.Univariate and multivariate regression analyses were used to analyze MRI characteristic parameters and screen out the potential and independent predictors of WHO pathological grade of meningioma,and then the nomogram predictive model of WHO pathological grade was established.The receiver operating characteristic(ROC)curve,area under the curve(AUC),calibration curve and decision curve analysis(DCA)were applied to assess the performance,robustness and clinical benefit of the model.Results Univariate regression analysis showed that tumor shape,size,peritumoral edema,meningeal tail sign,enhancement uniformity,adjacent tissue invasion and ADC values were the potential factors affecting the WHO pathological grade of meningioma(P<0.05);Multivariate regression analysis showed that tumor shape,enhancement uniformity,adjacent tissue invasion and ADC values were the independent influencing factors for WHO pathological grade of meningioma(P<0.05),and then the nomogram model was constructed based on the four MRI characteristic parameters.The nomogram model demonstrated a high performance with AUC of 0.905[95%confidence interval(CI)0.889-0.941].The calibration curve reflected the robustness of the nomogram model,and the prediction results of the model were highly matched with the actual observation results.The DCA indicated that the model had a good clinical benefit in the application of WHO pathological grade of meningioma.Conclusion The nomogram model based on preoperative MRI has high prediction performance,which will provide a reference for the form

关 键 词：脑膜瘤 列线图 预测模型 病理分级 

分 类 号：R739.45[医药卫生—肿瘤] R446.8[医药卫生—临床医学]