Imaging of increased peritumoral glutamate and glutamine in gliomas using 7T MRSI

ObjectivesDiffuse gliomas, due to their infiltrative properties, still lack effective treatment options. Recent research indicates that infiltration, malignancy, and symptoms such as epilepsy are related to synaptic connections between infiltrating glioma cells and cytotoxic levels of glutamate release. We previously showed that high-resolution 7T magnetic resonance spectroscopic imaging (MRSI) can resolve metabolic heterogeneities in gliomas. With this study, we evaluated 7T MRSI-derived glutamate (Gln) and glutamine (Glu) ratio maps for their use in defining infiltrative tumor activity in the peritumoral region. Materials and MethodsWe analyzed 7T MRSI scans of 36 patients with low- and high-grade gliomas. Within the visible tumor and a peritumoral shell, we calculated medians and Dice similarity coefficients (DSC) for nine metabolic ratios with and without hotspot thresholding and evaluated their correlation to and statistical significance between clinical parameters (e.g., tumor-associated epilepsy, IDH status, grade). ResultsThe Glu/tCr (total creatine) median was significantly higher in the peritumoral VOI (1.13) compared to the tumor (0.92) and normal-appearing white matter (NAWM, 0.87), while the Gln/tCr median was highest in the tumor (0.77, vs 0.44 peritumoral and 0.33 in NAWM, all significantly different). Glu/tCho (total choline) was significantly higher in the peritumor as well (3.44 vs 2.23 tumoral and 2.06 in NAWM). Peritumoral DSCs for Glu/tCr and Gln/tCr hotspots were comparable (0.53 to 0.51). Peritumoral Gln/Glu was significantly different between patients with and without tumor-associated epilepsy, and intratumoral (Glu+Gln)/tCr was significantly different between IDH mutation and wildtype. IDH mutation correlated negatively with the intratumoral (Glu+Gln)/tCr median (-0.53) and high grade correlated with intratumoral Glx/tNAA, Glx/tCr, and Gln/tCr medians (0.50/0.53/0.58). Conclusions7T MRSI can not only map relevant metabolic information in the structurally visible tumor volume, but also detect infiltration in the peritumoral area. Gln and Glu are candidates for the development of presurgical imaging and treatment monitoring.