动态对比增强磁共振成像影像组学评估胶质瘤异柠檬酸脱氢酶1突变与微血管生成

Evaluation of IDH1 mutation and angiogenesis in gliomas by radiomics analysis based on DCE-MRI

目的:探讨动态对比增强磁共振成像(DCE-MRI)影像组学在评估胶质瘤异柠檬酸脱氢酶1(IDH1)突变与微血管生成中的应用价值。方法:回顾性分析105例经病理证实的胶质瘤患者,按7:3分为训练组和验证组,采用3DSlicer软件半自动勾画感兴趣区,依次从T 1WI,T 2FLAIR,ADC,容积转运常数(K trans)及血管外细胞外容积比(V e)图像中提取影像组学特征。利用最小绝对收缩和选择算子算法进行特征选择并计算影像组学评分(Radscore),基于筛选出的影像组学特征构建预测胶质瘤IDH1突变的支持向量机模型。对纳入的临床放射学特征进行逻辑回归分析,结合Radscore建立综合列线图。应用免疫组化检测胶质瘤IDH1基因分型、血管内皮生长因子(VEGF)表达、CD105阳性标记的微血管密度(CD105-MVD)。采用受试者工作特征(ROC)曲线评价模型的诊断效能。采用Spearman相关性检验分析Radscore与VEGF表达、CD105-MVD的相关性。P<0.05为差异具有统计学意义。结果:K trans与V e联合模型评估胶质瘤IDH1突变训练组AUC=0.889,验证组0.890;综合列线图训练组AUC=0.938,验证组0.914。各模型的Radscore与VEGF表达、CD105-MVD均呈负相关,其中联合模型的相关性最高(r=-0.570,P<0.001;r=-0.665,P<0.001)。结论:DCE-MRI影像组学可有效评估胶质瘤IDH1突变,且在分析胶质瘤IDH1突变与血管生成相关性方面具有重要临床价值。

Objective:To investigate the value of radiomics analysis based on dynamic contrast enhanced magnetic resonance imaging(DCE-MRI)in evaluating isocitrate dehydrogenase 1(IDH1)mutation and angiogenesis in gliomas.Methods:105 glioma patients with pathologically confirmed were retrospectively analyzed.They were divided into training and validation groups by 7:3.The regions of interest were semi-automatically delineated by 3D slicer software.The radiomics features were extracted from T 1WI,T 2 FLAIR,ADC,volume transfer constant(K trans)and volume of extravascular extracellular space per unit volume of tissue(V e)images.Least absolute shrinkage and selection operator was conducted to obtain the optimized radiomics features and calculate Radscore.The support vector machine models for evaluating glioma IDH1 mutation were developed based on the optimized radiomics features.The selected clinico-radiological features were analyzed by logistic regression.Finally,a nomogram including Radscore and the clinico-radiological features was established.IDH1 genotypes,vascular endothelial growth factor(VEGF)expression and CD105-MVD were detected by immunohistochemistry.Receiver operating characteristic(ROC)curve was used to evaluate the diagnostic efficiency of the models.Spearman correlation test was used to analyze the correlation between Radscores and VEGF expression and CD105-MVD.P<0.05 was considered statistically significant.Results:The AUCs of the model based on K trans and V e were 0.889 and 0.890 in the training and validation groups.The AUCs of nomogram were 0.938 in the training group and 0.914 in the validation group.Radscores of models were negatively correlated with VEGF expression and CD105-MVD,and the Radscores of combined model achieved the highest(r=-0.570,P<0.001;r=-0.665,P<0.001).Conclusion:Radiomics analysis based on DCE-MRI can effectively predict IDH1 mutation,and plays a vital role in evaluating the correlations between IDH1 mutation and angiogenesis in gliomas.