Abstracts

We investigated mechanisms underlying suppression of seizures with focal brain stimulation by analyzing human trial data with a computer model of seizure generation. Data from 31 patients treated with responsive stimulation to suppress seizures during presurgical intracranial monitoring were analyzed. A computer simulation model (F. Wendling et al.) was employed to determine the hippocampal cell population most affected by responsive stimulation. Data were grouped into two classes: Stimulated Events (Stim) and Non-Stimulated Events (NonStim). The Stim class consisted of automated detector-triggered stimulations; NonStim events were detected but not stimulated. There were a total of 3,862 Stim and 1,582 NonStim events recorded.
Simulations determined model parameter changes after reactive stimulation compared to non-stimulated events. A permutation test was conducted to determine the difference in [quot]grand-mean[quot] model parameters for both classes. Intracranial EEG (IEEG) was processed before and after event onsets. Each IEEG segment was matched to synthesized data generated by the model, searching all possible parameter values. Signal matching techniques were used to carry out this task, minimizing the normalized power spectrum. Differences betweem Stim and NonStim classes were tallied. There was a statistically significant change between stimulation and nonstimulation class events in the model[apos]s [quot]G[quot] parameter, which correlates with GABAergic fast inhibitory gain (0.1438 under Stim VS. -0.8072 under NonStim, p= 0.02). Changes in parameters [quot]A[quot] (representing excitatory gain) and [quot]B[quot] (slow inhibitory currents) in the model were not significant ( p= 0.23 and 0.24 respectively). Our results demonstrate that focal brain stimulation may suppress seizures through fast inihibitory GABAergic interneurons. This result may stimulate experiments to isolate function in specific populations of neurons during seizure generation and suppression. They may also help guide electrode placement and methods for tuning stimulation parameters in new implantable devices. Further research recording unit ensembles [quot]in-vivo[quot] in animals and humans, linked to biophysical computational neuronal modeling will provide more insight into mechanisms of seizure generation and suppression. These techiques are moving from traditional experiments in the animal laboratory to clinical practice in Neurology and Neurosurgery. (Supported by Whitaker Foundation, Dana Foundation, National Institutes of Health (grant #5R01NS041811-03), Citizens United for Research in Epilepsy (CURE), the Epilepsy Foundation, and a sponsored research agreement between the University of Pennsylvania and NeuroPace, Inc.)