Abstracts

Rationale: Over 30% of epilepsy patients have incapacitating seizures which cannot be completely controlled with medication. For patients with medically refractory epilepsy (MRE), the most common treatment option is surgical resection of the epileptogenic zone (EZ), regions that trigger and spread seizures. Surgical success rates vary between 30 to 70%, which can be attributed, at least partly, to the inaccurate identification of the EZ. Identification of the EZ remains a challenge because there are no clinically validated biomarkers of the EZ. Here we present a potential intracranial EEG biomarker of the EZ that captures information flow or connectivity between brain regions.

Methods: Stereotactic-EEG (SEEG) recordings of epilepsy patients during rest and seizures were analyzed. Transfer entropy (TE) was used to examine the connectivity between SEEG channels (network nodes) and the roles of nodes in epileptic neural networks during rest, moments before seizure, during seizure, and moments after seizure.

Results: During rest, there is a set of nodes that dominate information flow only to EZ nodes and non-EZ nodes. The TE from the dominant to EZ nodes increases tens of seconds before a seizure event, but decreases sharply seconds before seizure, and reaches a minimum during seizure. During seizure, the dominant nodes cease or only weakly interact with the EZ nodes. The TE from the dominant to the EZ nodes peaks immediately after seizure. An example of the TE dynamics of the epileptic neural networks of a patient is presented in Figure 1. The information flow from the dominant to EZ nodes is different from that to non-EZ nodes. To illustrate, Figure 2 shows two key differences between information flow to EZ nodes (Figure 2a) and non-EZ nodes (Figure 2b) from the same patient in Figure 1: (1) the minimum TE is reached during seizure for EZ nodes whereas the minimum is reached at 10–0 s before seizure for non-EZ nodes; and (2) TE after seizure is significantly higher than that before seizure for EZ nodes, but this is not the case for non-EZ nodes.

Conclusions: Our results suggest that (1) seizure occurs when the dominant nodes cease to dominate the information flow or weakly transfer information to the EZ nodes, supporting the hypothesis that seizure occurs when the EZ nodes are no longer effectively being inhibited; (2) seizure stops when the dominant and other channels strongly increase information flow or communication to the EZ nodes, much more strongly than at any other times throughout the seizure. This suggests that the brain makes a concerted effort to inhibit the EZ nodes to stop the seizure. The brain increases communication to the EZ nodes tens of seconds before seizure, but apparently this level of effort is insufficient to stop seizure. The analysis of TE dynamics entering and exiting seizures may identify more accurately the EZ nodes (Figure 2), which may improve surgical planning.

Funding: Please list any funding that was received in support of this abstract.: S. Wing acknowledges the support of the Sabbatical Fellowship at the Johns Hopkins University Applied Physics Laboratory (JHU/APL).