Abstract
Pioneering work in cancer neuroscience revealed that glioma development and infiltration occur in part via novel, functional neuronal-tumor synapses. Resultant hyperexcitability, which perpetuates disease progression, can be detected using standard and high resolution electrophysiology techniques. However, differences across glioma subtypes and the topographic distribution of these alterations are not well characterized in existing literature. To interrogate these perturbations and their extent, we obtained intraoperative intracranial electrophysiology recordings in 72 patients (n=23 IDH-wildtype glioma, 15 IDH-mutant glioma, 28 non-tumor epilepsy cases, and 6 movement disorder cases) using standard (e.g., 6-8 contacts) clinical electrodes (ECoG, n=40), microelectrodes with 128-1024 channels that conform to the brain surface (n=46), and Neuropixels probes (n=29). All recordings were from the frontal, temporal, or parietal lobe. Electrode location relative to the enhancing tumor and associated T2 hyperintensity was determined using a 3D localization pipeline, coregistered to segmented pre-operative imaging. Endpoints include spectral power band analysis with modeling of periodic and aperiodic components and interictal discharge rates (IIDs). Euclidean and geodesic distance, with the latter following anatomic geography, to the enhancing tumor boundary was calculated per participant and channel. Clinical ECoG grids did not discriminate regional or overall differences in high gamma power (HGP), aperiodic (E/I) slope or exponent, or IIDs. However, using high resolution grids, increasing geodesic distance was associated with a linear decrease in high gamma power for both IDH-mutant and -wildtype glioma cases. Similarly, the E/I slope of 30-50 Hz activity increased linearly with distance from tumor boundary, implicating a relative shift toward peritumoral excitatory activity. IDH-wildtype glioma was associated with the highest HGP and IID rate, and geodesic distance was most precise in distinguishing region differences; both distance calculations were relevant for IDH-mutant tumors. Comparison to non-tumor cases highlighted mechanistic similarities across neurologic diseases that may be identifiable using neuroelectrical signatures.