Rationale:
Cognitive effort has been shown to modulate or suppress epileptic activity by engaging distributed brain networks and promoting transitions into more stable, interictal states (Lesser RP, Webber WRS, Miglioretti DL. Clin Neurophysiol., 2022, 136, 130-137). Tasks such as arithmetic and spelling can terminate clinically induced seizures more effectively than direct cortical stimulation (Lesser RP, Webber WRS, Miglioretti DL, et al. Clin Neurophysiol. 2019, 130, 2169–2181), suggesting that mental effort recruits endogenous stabilizing mechanisms. These findings highlight cognitive engagement as a potential non-invasive strategy for seizure control via large-scale network reorganization. Building on this, we use intracranial EEG (iEEG) and dynamical network modeling to examine how cognitive engagement alters seizure dynamics in patients with drug-resistant epilepsy (DRE), with a focus on identifying brain states linked to successful task-induced suppression.
Methods:
We analyzed iEEG data from a DRE patient undergoing task-induced seizure suppression. Linear time-varying dynamical network models were constructed, with each electrode modeled as a node. The sink index (SI) was used to quantify the extent to which each node was influenced by the rest of the network (Gunnarsdottir KM, Li A, Smith RJ, et al. Brain. 2022, 145(11), 3901–3915). SI vectors were clustered to identify brain states across seizure conditions, and their temporal evolution was mapped to assess network reorganization.
Results:
During native seizures, the brain consistently entered one of a few distinct ictal states, characterized by stable patterns of network activity associated with seizure onset and propagation. In the cognitive condition, however, the network exhibited repeated transitions between ictal and interictal states before ultimately stabilizing in the interictal configuration. Notably, this shift in network dynamics aligned with the initiation of a question-answer task, suggesting that cognitive engagement facilitated a reorganization of brain activity toward a more stable, non-seizure state (See Figure 1).
Conclusions:
This proof-of-concept case demonstrates that cognitive engagement can drive transitions from seizure-prone to interictal network states. By comparing native and induced seizures, as well as effective versus ineffective cognitive tasks, future research may uncover biomarkers of successful seizure suppression and inform the development of novel cognitive interventions for seizure control.
Funding:
This work was supported by NIH grants R35NS132228 and R01NS125897.