Abstracts

RATIONALE: Methods derived from the theory of nonlinear dynamics have allowed for the identification of a preictal phase in a generalized animal model of epilepsy in which the H218/AGR16/edg-5/LPB2 sphingosine 1-phosphate gene has been disrupted (Carney et al, 2001). This study tests the hypothesis that specific neurodynamical changes are state specific and can be identified in the EEG signal using the short-term maximum Lyapunov exponent (STLmax) as a measure of stability. After reviewing this abstract, the participants should be able to distinguish state specific nonlinear neurodynamical features in an animal model of generalized epilepsy.
METHODS: Daily bifrontal and bilateral hippocampal intracranial video-EEG recordings were obtained in postnatal days (P) 18-25 H218 mice (n = 8) and in littermate controls (n = 2) in order to determine seizure periods, awake, and sleep states. Seizures were characterized by generalized spike-and-wave discharges (SWD) with categorical behavioral changes. During wakefulness, EEG activity consisted of continuous, well-modulated and organized, 8-10 Hz activity. Sleep was identified by the presence of bifrontal 12-14 Hz spindles lasting between 2-5 seconds. STLmax values were estimated every 10.24 seconds for each electrode site in H218 mice (P18-25) and littermate controls. Random samples of STLmax from the left frontal electrode were selected from both awake and sleep states in the following three cases: (1) seizure period in H218 mice (n = 8), (2) seizure free period in H218 mice (n = 8), and (3) littermate controls (n = 2). In each case, mean values of samples of the awake and sleep states were compared by employing the distribution free Wilcoxon rank-sum test.
RESULTS: Seizures were observed only during the awake state in P18-25 H218 mice (n = 8). During the SWD seizure period, STLmax mean values were 2.52 during the awake state and 2.77 during the sleep state (p-value = 0.0213) in H218 mice (n = 8). During the seizure-free interval, STLmax mean values were 2.55 during the awake state and 3.08 during the sleep state (p-value = 0.0003). STLmax mean values in littermate control mice were 2.16 during the awake state and 1.97 during the sleep state (p-value = 0.4061) (n = 2).
CONCLUSIONS: (1) Mean STLmax values were lower in the awake state than in sleep in H218 mice (P18-25). (2) Mean STLmax values were lower in the sleep state than during the awake state in littermate controls. The results suggest that brain dynamics were more ordered (lower STLmax mean value) during the awake state in H218 seizure-prone mice. Although STLmax values were lower (more ordered state) during sleep than during wakefulness in littermate controls, the difference was not statistically significant. Thus, mean STLmax as a measure of brain dynamics may be a sensitive tool to determine arousal state specificity. These findings may help to distinguish arousal state changes from preictal dynamical transitions that have been observed to precede seizures.
[Supported by: NIH/NINDS NS039687
University of Florida Division of Sponsored Research
Children[ssquote]s Miracle Network
U.S. Veterans Affairs]