Direct intracranial EEG evidence for a local breakdown in normal sleep homeostasis at the human seizure onset zone

Animal studies have demonstrated that sleep is crucial for physiological synaptic pruning, and that this pruning is primarily regulated at a local, cellular level. We hypothesized that local alterations in sleep homeostatic regulation may constitute an interictal biomarker of the seizure onset zone (SOZ), and that sleep homeostasis abnormalities may also be present in areas beyond the SOZ, proportional to their secondary recruitment during seizures.

Overnight intracranial EEG (iEEG) recordings were obtained from sixteen well-characterized patients with focal epilepsy. The first and last hour of sleep were analyzed during one seizure-free night with well-organized sleep architecture, selected from each patient. We compared slow-wave activity (SWA; i.e., delta power, 1-4 Hz, a marker of synaptic strength) during wakefulness, the first and last hour of sleep, as well as SWA overnight decline in the SOZ, immediately neighboring regions, and regions located >2 cm away. Sleep-dependent changes in the slope of slow-waves and spike-waves were also considered. Finally, we examined whether sleep SWA and slow-wave slope abnormalities were predicted by ictal phase-locked high-gamma (PLHG), a proxy of ictal recruitment, both in the SOZ and in regions away from it.

Consistent with our hypothesis, the SOZ displayed higher increases in SWA from wake to sleep than other brain regions. During sleep, SWA and slow-wave slope were higher in the SOZ compared to both immediately neighboring regions and regions further apart, with the difference being most pronounced during the last hour of sleep. Additionally, we observed a dampened overnight decline in SWA in the SOZ compared to distant cortical areas, suggesting an escape from physiological synaptic homeostasis. While the slopes of epileptic spike-waves were also higher in the SOZ compared to neighboring and distant areas, they did not significantly decline overnight in any area. Finally, as hypothesized, ictal PLHG (a marker of recruitment) predicted the extent of increases in sleep SWA, sleep slow-wave slope, and spike-wave slopes across all brain areas, with results especially consistent for the last hour of the night.

Our results provide direct intracranial EEG evidence for local alterations in sleep homeostasis in human epileptic brains, which are maximal at the SOZ, and proportional to local ictal recruitment across the whole brain. These results provide a solid mechanistic foundation for interventional studies aiming to enhance local sleep homeostasis in epileptic brains, through e.g. pharmacology or neuromodulation, to decrease local neuronal excitability and improve patient outcomes.