Classifying adult-type diffuse glioma using 7T magnetic resonance spectroscopic imaging

Abstract

Established clinical MR imaging offers limited accuracy for the precise classification of adult-type diffuse glioma. The purpose of this study was to identify robust 7T MRSI metabolic markers that distinguish glioblastoma, IDH-wildtype (GBM); astrocytoma, IDH-mutant (AST) and oligodendroglioma, IDH-mutant, 1p/19q-codeleted (ODG), and to assess their predictive value alongside established radiological markers. We prospectively evaluated patients with suspected adult-type diffuse glioma. Final diagnoses were established according to the 2021 WHO classification. Routine 3T MRI scans included: T1w-CE; T2; FLAIR; and DWI-ADC. 7T MRSI was performed with a 3D-FID-CRT sequence and median values of metabolites in tumors were extracted from 16 metabolite maps. Group differences and predictive performance were assessed using non-parametric statistics and ROC-curve analysis. Fifty patients (15 GBM, 24 AST, 11 ODG) were included. On 7T MRSI, the metabolite ratios that discriminated between subtypes were: mI/tNAA; Glu/tCr; Gln/tCr; Glx/tCr; and mI/tCr for GBM vs AST (AUCs 0.78–0.86); eight ratios distinguished GBM from ODG (AUC < 0.93), comparable and slightly more robust than the clinical standard. tCho/tNAA and GSH/tNAA were the most predictive for AST vs. ODG (AUC < 0.90) with GSH/tNAA exceeding the predictive abilities of CE, the T2-FLAIR mismatch sign, and nADC. 7T MRSI enables non-invasive metabolic profiling of gliomas and shows potential for the differentiation of GBM, AST, and ODG. Several metabolite ratios demonstrated discriminatory performance comparable to that of conventional MRI markers in this cohort. These findings provide preliminary evidence supporting the utility of 7T MRSI for glioma subtype characterization and suggest further validation in larger studies.