Magnetic resonance spectroscopic imaging based detection of genetic alterations in gliomas

Gliomas, the most common primary malignant brain tumors, have been reclassi- fied based on genomic advancements, providing new insights into oncogenic mechanisms and enabling individualized treatments for specific molecular subclasses. The impor- tance of genetic markers in glioma diagnosis and treatment planning was emphasized in both 2016 and 2021 revisions of the World Health Organization (WHO) classifica- tion of central nervous system (CNS) tumors. Specifically, isocitrate dehydrogenase (IDH) and telomerase reverse transcriptase promoter (TERTp) have been indicated as genetic biomarkers, impacting treatment response, and survival rate in gliomas. Preop- erative and noninvasive detection of these mutations has been drawing great interest, and proton magnetic resonance spectroscopy (1H-MRS) has been important for this purpose. This study delves into MRS profiles of 225 glioma patients, unraveling dis- tinct molecular signatures associated with these genetic alterations utilizing short echo time (TE) single-voxel 1H-MRS. A classification approach was developed integrating conventional machine learning algorithms and a one-dimensional convolutional neural network (1D-CNN) to identify IDH and TERTp mutations in gliomas. The 1D-CNN model showed exceeding 90% accuracy in identifying IDH mutation and 75% accu- racy in detecting TERTp mutation. This study also analyzed the metabolic correlates of survival outcomes in the IDH wildtype (IDH-wt), TERTp mutant (TERTp-mut) gliomas, known as the most aggressive form of gliomas. Glutamine-glutamate complex (Glx) and glutathione (GSH) have emerged as significant metabolic correlates influenc- ing patient survival in these patients. To conclude, 1H-MRS could aid in preoperative identification of glioma molecular subgroups, offering insights into glioma biology.