Abstracts

Macroscopic properties of the EEG are frequently correlated with abnormal behaviors. We observed an apparent correlation of increased generalized spike wave discharge (SWD, absence seizure) amplitude to abnormal behaviors in aged F344 rats during long-term video-EEG recordings. We sought to validate quantitatively whether SWD amplitude could be used as a reliable marker to predict myoclonic arousals and limbic seizures in these animals., SWD amplitude recorded from six EEG channels during the initial 5 hours of a 24 hour recording was observed, and 7 schema were established to predict two major types of abnormal behaviors (myoclonic arousal and limbic seizures) in the following 19 hours of recording. The 7 schema consisted of: 1) event occurrence frequency; 2) myoclonic arousal; 3) head nodding (class II); 4) forelimb or hindlimb clonus (class III); 5) running and rearing (class IV); 6) falling and jumping (class V); and 7) fatality.
Five episodes of SWDs were chosen randomly in the first 5 hours of recording and an average amplitude was generated from 6 EEG derivations: F3-C3, C3-P3, F3-P3, C4-P4, F3-F4, and C3-C4.
Quantitative prediction for different abnormal behaviors were determined based on the amplitude variation of the 6 channels. For example, it would be predicted that there would be no myoclonic arousal when the amplitude of F3-C3 was greater than that of F3-P3 by more than 5%, whereas there would be 1 to 2 events of myoclonic arousal in the subsequent 19 hours when the F3-P3 amplitude was over 10% greater than F3-C3 while the amplitude of C3-P3 was less than or equal to C4-P4. For purposes of simplicity, a standard scheme of quantitative comparison was designed: 1 class III event = 2 class II events, 1 class IV event = 4 class II events. Accuracy was judged by comparing the number of observed events to the predicted results. For example, based on SWD amplitude, the range of actual class II events is predicted between 5 and 8 times. Prediction failure occurs when the actual class II event number is 3 or 10., Twenty two EEG files of 20 month old F344 rats (n=9) were analyzed and 3 prediction categories included: 1) frequency of occurrence of myoclonic arousal; 2) event frequency; and 3) limbic event class and fatality. Analysis indicated prediction accuracy of 80% for myoclonic arousals, 70% for event frequency, and 75% for limbic events +/- fatalities. Taken together, prediction accuracy was [sim]70%., These findings suggest that macroscopic properties of electrocerebral activity in aged F344 rats may be used as predictors of abnormal ictal behaviors in these animals. These findings also suggest the possibility that specific differences in EEG spectral power may be quantifiable and equally predictive of these behaviors., (Supported by NIH R01NS046015 to KMK.)