Resting-state and task-related neurometabolite levels differentially relate to cortical excitability in the sensorimotor and prefrontal cortex

Abstract Background Normal brain function requires a dynamic balance between inhibition and excitation. While magnetic resonance spectroscopy (MRS) quantifies the chief inhibitory and excitatory neurometabolites, GABA and glutamate-glutamine (Glx), the combination of transcranial magnetic stimulation and electroencephalography (TMS-EEG) provides a complementary measure of cortical inhibition-excitation dynamics via transcranial evoked potentials (TEPs). However, the relationship between neurometabolite concentrations and TEPs is unclear. Objective To characterize the relationship between neurometabolite concentrations and TEPs, as a function of TEP component, TMS pulse type, brain region, neurometabolite, and MRS brain state. Methods Twenty-five young healthy adults completed a 4-day protocol. Sessions 1 and 2 involved screening, anatomical MRI, and functional MRI localization of DLPFC. In session 3, single- and paired-pulse TMS-EEG were applied over SM1 and DLPFC. Session 4 included resting-state and motor-task-related MRS of SM1 and DLPFC. Results In SM1, task-related GABAergic tone strongly predicted early to mid-latency TEPs. In DLPFC, local early to mid-latency TEPs showed no relationship to neurometabolites, whereas late and global TEP outcomes revealed some links. The later N100 TEP was the only component consistently modulated by paired-pulse TMS. Task-related MRS measures consistently outperformed resting-state measures in predicting TEPs for SM1, while the opposite was true for DLPFC. Conclusions Single-pulse TEPs reliably index the local inhibitory tone in SM1, with limited added value from paired-pulse paradigms. These findings support the use of single-pulse TEPs as accessible markers of cortical inhibition, especially for SM1, and may inform biomarkers and strategies for individualized neuromodulation. Key point summary Normal brain function relies on a balance between inhibition and excitation, yet the relationship between local cortical neurometabolite levels and cortical excitability in humans remains poorly understood. We measured inhibitory and excitatory neurometabolites using magnetic resonance spectroscopy and assessed cortical excitability responses to non-invasive stimulation using combined transcranial magnetic stimulation - electroencephalography (TMS-EEG). In the primary sensorimotor cortex (SM1), local cortical responses to stimulation were best explained by inhibitory GABA levels during motor task performance. In the dorsolateral prefrontal cortex (DLPFC), global cortical responses were best predicted by resting-state neurometabolite levels. Overall, paired-pulse TMS paradigms offered little additional value over single-pulse paradigms for informing on neurometabolite levels. By demonstrating region- and state-dependent links between neurometabolite levels and cortical excitability, our findings position TMS-EEG as an accessible biomarker of cortical inhibition, particularly in SM1. Graphical Abstract Abstract figure legend The left panel provides a stepwise overview of the study protocol, including (i) participant screening, (ii) functional and anatomical magnetic resonance imaging (MRI) scanning for individualization purposes, (iii) single- and paired-pulse transcranial magnetic stimulation–electroencephalography (TMS-EEG), and (iv) magnetic resonance spectroscopy (MRS) acquired at rest and during a bimanual motor task. The middle and right panels show the grand-average EEG responses to single-pulse TMS over the sensorimotor cortex (SM1, red) and dorsolateral prefrontal cortex (DLPFC, blue), respectively. The topographical maps below each EEG response depict the averaged scalp distributions corresponding to each canonical transcranial evoked potential. Per region, the top left illustration shows the group-level MRS voxel placement, whereas the top right illustration shows the simulated TMS-induced electric fields. While early to mid-latency local cortical responses were best predicted by task-related GABAergic inhibition in SM1, global cortical responses following DLPFC stimulation were best predicted by resting-state neurometabolite levels.