Abstracts

Rationale: Status epilepticus (SE) has been known to be resistant to benzodiazepine treatment when administered greater than 1 h after seizure onset. The duration and other properties of electrographic SE are variable, as are the effects of potential therapeutic treatments, and thus a straightforward method to quantify different components of SE would be beneficial in both a clinical and experimental environment. The purpose of this study was to determine if the selection of only γ-band frequencies in the raw EEG signal would allow isolation, and prove to be a simple quantification of electrographic SE. This report aims to evaluate these methods for their effectiveness as automated protocols for quantification of SE such that data from groups of animals can be compared in terms of the duration and severity of the SE or the effects of pharmacological intervention. Methods: An optimum frequency range near to the gamma band was selected for extracting seizure -related activity from the EEG signal. This band-pass (20-70 Hz) filtering reduced artifacts in the recordings caused by animal handling and electrical noise. The changes in gamma power during the first 10 h of SE were smoothed using 5 min windows, quantified with an energy operator, and modeled by an 8-th order polynomial. These models were then averaged across animals to compensate for the inter-animal variability in response to drug treatments. Lithium-pilocarpine-induced SE was analyzed using this method during episodes of seizure activity for several hours, which allowed a quantitative determination of drug effectiveness. Three groups were evaluated: (1) Diazapam delivered 60 min following the first convulsive seizure, (2) Propofol at 60 min, and (3) Saline at 60 min. The model was validated by comparing human evaluation of the various distinct stages of SE, number of spike events, and spike frequency. These electrographic stages have been extensively characterized and have been modified and adapted to allow for analysis of a drug’s effectiveness on treating SE. In an attempt to validate automated SE quantification we visually defined stages of SE and made statistical comparisons with the model. Results: Applying 95% confidence intervals of the baseline threshold to raw EEG spike amplitude did not demonstrate significant differences in SE activity over a 10-h period for any drug tested. Filtering the data in the gamma band indicated a difference in EEG spike rate for only the propofol group from hours 2-5. Applying hand scoring methods to the data also only indicated significant differences in SE between the propofol group. Finally, applying the energy operator and modeling approach to EEG data in the gamma frequencies indicated differences in SE for groups treated with propofol, but not diazepam, thus providing similar results as human scoring. Conclusions: This approach has the potential to allow quantitative assessment of the effects of both short- and long-lasting pharmacological manipulations on electrographic SE. The model provides a quick, objective, and modifiable means to quantify the temporal effects of novel anti-convulsants.