Spatial Neurophysiological Gradients at High Resolution Differentiate Tumor Types in the Human Cortex at the Tumor-Brain Interface

Brain tumors interact with adjacent neural circuits, with tumor infiltration altering neural physiology, and aberrant neuronal activity, in-turn, promoting tumor growth. Due to limited patient numbers, variable recording methods, and tumor heterogeneity, direct evidence of electrophysiologic distinctions between specific tumor types and these interactions in the intact human brain is lacking. Here, we analyzed intraoperative recordings from 100 participants undergoing tumor resection, epilepsy surgery, or movement disorder deep brain stimulator implant surgeries using clinical electrocorticography (ECoG), mesoscale ECoG, and high-density micro-ECoG. Tumor cases included IDH-mutant (IDHm) gliomas across grades, IDH-wildtype (IDHwt) glioblastomas, and metastases. Multiscale neurophysiological activity was spatially mapped to patient-specific tumor and brain anatomy, enabling direct assessment of physiological changes across the tumor-brain interface. Our findings show that population-averaged electrophysiological measures alone provide limited diagnostic separation, whereas spatially-resolved analyses reveal robust physiological gradients linked to tumor distance. Measures of excitatory-inhibitory balance, particularly aperiodic exponent and high-gamma activity, varied systematically with tumor proximity and correlated with local tumor volume fraction around individual electrode contacts. These relationships were detectably different between infiltrative glioma (both IDHwt and IDHm) and metastases, even across recording scales. Furthermore, functional connectivity revealed specific patterns of network organization with the infiltrative glioma cohort, clearly separating IDHwt and IDHm. Beta band coherence between the tumor and brain was significantly reduced in IDHm compared to within the tumor or within brain connectivity (p=0.0006; Kruskal-Wallis test ; mean coherence tumor-brain: 0.26, tumor-tumor: 0.44; brain-brain: 0.34), a pattern not seen in IDHwt or metastases.. Physiological abnormalities extended into radiographically normal-appearing brain. Together, these findings reveal spatially organized, tumor type-dependent physiological signatures of tumor-brain interactions in the intact human brain. Electrophysiology may therefore provide a dynamic physiological map of the tumor-brain interface that complements structural imaging.