De Pardieu M, Boucebci S, Herpe G, Fauche C, Velasco S, Ingrand P, Tasu J-P
Department of Diagnostic and Interventional Radiology, Poitiers University Hospital, 86000 Poitiers, France.
Epidemiology and Biostatistics, INSERM CIC 1402, Faculty of Medicine and University Hospital, 86000 Poitiers, France.
Diagn Interv Imaging. 2020 Jul-Aug;101(7-8):451-456. doi: 10.1016/j.diii.2020.04.013. Epub 2020 May 20.
The purpose of this prospective study was to determine whether chemical shift gradient-echo magnetic resonance imaging (MRI) could predict glioma grade.
A total of 69 patients with 69 gliomas were prospectively included. There were 41 men and 28 women with a mean age of 50±(SD) years (range: 16-82years). All patients had MRI of the brain including chemical shift gradient-echo sequence, further referred to as in- and out-of phase sequence (IP/OP). Intravoxel fat content was estimated by signal loss ratio (SLR=[IP-OP]/2IP), between in- and out-of-phase images, using a region of interest placed on the viable portion of the gliomas. Association between SLR and glioma grade was searched for using Wilcoxon and Mann-Whitney U tests and diagnostic capabilities using area under the receiver operating characteristic (AUROC) curves.
A significant association was found between SLR value and glioma grade (P<0.0001). SLR>9‰ allowed complete discrimination between grade III and grade II glioma with 100% specificity (95% CI: 85-100%), 100% sensitivity (95% CI: 78-100%) and 100% accuracy (95% CI: 90-100%) (AUROC=1). A SLR>20‰ allowed discriminating between grade IV and grade III glioma with 75% specificity (95% CI: 57-89%), 73% sensitivity (95% CI: 45-92%) and 72% accuracy (95% CI: 57-84%) (AUC=0.825, 95% CI: 0.702-0.948). The AUROC for the diagnosis of high-grade glioma (grade III and IV vs. grade II) was 1.
Chemical shift gradient echo MRI provides accurate grading of gliomas. This simple method should be used as a biomarker to predict glioma grade.
本前瞻性研究旨在确定化学位移梯度回波磁共振成像（MRI）是否能够预测胶质瘤的分级。
前瞻性纳入了69例患有69个胶质瘤的患者。其中男性41例，女性28例，平均年龄为50±（标准差）岁（范围：16 - 82岁）。所有患者均进行了脑部MRI检查，包括化学位移梯度回波序列，以下简称同相位和反相位序列（IP/OP）。通过在胶质瘤的存活部分放置感兴趣区域，利用同相位和反相位图像之间的信号丢失率（SLR = [IP - OP]/2IP）来估计体素内脂肪含量。使用Wilcoxon检验和Mann-Whitney U检验来寻找SLR与胶质瘤分级之间的关联，并使用受试者操作特征（AUROC）曲线下面积来评估诊断能力。
发现SLR值与胶质瘤分级之间存在显著关联（P < 0.0001）。SLR>9‰能够完全区分III级和II级胶质瘤，特异性为100%（95%可信区间：85 - 100%），敏感性为100%（95%可信区间：78 - 100%），准确性为100%（95%可信区间：90 - 100%）（AUROC = 1）。SLR>20‰能够区分IV级和III级胶质瘤，特异性为75%（95%可信区间：57 - 89%），敏感性为73%（95%可信区间：45 - 92%），准确性为72%（95%可信区间：57 - 84%）（AUC = 0.825，95%可信区间：0.702 - 0.948）。诊断高级别胶质瘤（III级和IV级与II级相比）的AUROC为1。
化学位移梯度回波MRI能够对胶质瘤进行准确分级。这种简单的方法应作为预测胶质瘤分级的生物标志物。
