Abstracts

Rationale: Neurophysiologically-based therapy, such as giving drugs only when needed to prevent seizures, could transform epilepsy care. Challenges for the development of devices capable of real-time seizure advisories and responsive pharmacotherapy are: 1.)Suitable animal model with spontaneous seizures 2.)Device for recording long-term intracranial EEG (iEEG) in freely behaving animals 3.)Algorithms for seizure detection and forecasting 4.) Pharmacodynamic models of the time-course of AED effect on iEEG. Methods: Canine video-EEG monitoring units were established at 3 institutions. Intracranial EEGs (iEEGs) were monitored continuously using NeuroVista's implantable Seizure Advisory System (SAS). An automated caregiver alert system was developed to alert a caregiver (via e-mail and pager) when seizures occurred. We evaluated a multi-compartment pharmacokinetic (PK) model for phenobarbital (PB) and found that it correlates to pharmacodynamic (PD) changes extracted from EEG. Results: Ten dogs with naturally-occurring epilepsy were implanted with the SAS device. Four studies have been completed: 1.)Retrospective evaluation of the SAS seizure detection algorithm on 11,671 hrs. of iEEG data collected from six dogs. A total of 202 electrographic seizures were captured from 4 dogs. The SAS seizure detection algorithm detected 100% of the 184 clinically observed seizures and 91.1% of all detected events correlated with focal electrographic seizure activity. 2.)Retrospective evaluation of SAS seizure forecasting system was performed on 3 dogs. Over 1 year of continuous iEEG was analyzed, and 45, 15, and 83 seizures were captured from the 3 dogs, respectively. Utilizing the SAS algorithm, better than chance seizure advisories (p < 5x10-5) were obtained on all 3 dogs with 1 to 2 false-positives per day. 3.)Prospective study utilizing the care-giver alert system in 4 dogs. The SAS alerted a veterinarian via an automated message in the event of 3 or more seizures within 4 hours, or a seizure lasting longer than 5 minutes. A veterinarian confirmed the seizure activity via remote video-monitoring, and initiated diazepam or phenobarbital therapy as a single IV dose. There were 4 episodes requiring emergency therapy. 4.)PK and PD study in two epileptic dogs on a multi-week PB regimen. PB concentration-time data were evaluated using compartmental methods and predicted concentration-time profiles for the dogs. The quantitative iEEG features, beta-gamma band spectra and line-length, correlated with PB dosing. Conclusions: A seizure advisory system (SAS) designed to alert patients and caregivers of seizure activity and provide real-time seizure forecasting was demonstrated in dogs with naturally occurring epilepsy. To date use of the SAS in dogs with naturally occurring epilepsy has demonstrated 1.) Ability to perform highly accurate real-time seizure detections 2.)Ability to deliver seizure alerts via e-mail and pager. 3.)Seizure forecasting at levels significantly better than chance. 4.)Feasibility of constructing PK-PD models of the effect of AEDs on iEEG.