Abstracts

RATIONALE: Antiepileptic drug (AED) neurotoxicity may not be apparent on physical examination, and objective methods for detecting neurotoxicity would be useful. Cognitive tests may reveal AED effects in groups of subjects beginning AEDs, but may not be sensitive to change in individuals. This study tested the hypothesis that EEG measures would be more sensitive to individual change than would cognitive measures.
METHODS: The study compared subjects beginning an AED ([ssquote]AON[ssquote]; N=72) with non-medicated healthy controls ([ssquote]NCO[ssquote]; N=74). AON subjects included 52 healthy volunteers participating in blinded AED trials, and 20 patients beginning conventional AED therapy. All subjects underwent structured EEG recording and cognitive testing prior to initiation of an AED, and 12-16 weeks later. Cognitive measures included Digit Symbol (DSMT), Digit Cancellation, Stroop, tapping speed, visual reaction time (VRT), Selective Reminding, Wonderlic Personnel Test, and Story Recall. Subjective measures included the Profile of Mood States (POMS: Fatigue and Confusion scales), and the Portland Neurotoxicity Scale (PNS). Quantitative EEG measures (occipital) included peak frequency (by power), median frequency (by power), percentage theta power, and percentage delta power. Test-retest changes for all measures were scored against test-retest regression equations derived from NCO subjects, and reported as Z-scores. Z scores from the 2 groups were compared using the Wilcoxon test. We also counted the number of individuals exceeding a Z-score of 2.0 (~95th percentile).
RESULTS: Five of the eight target cognitive measures (DSMT, Stroop, finger tapping, VRT, Story Recall) and all EEG and subjective measures revealed significant differences between AON and NCO subjects. EEG peak frequency change correlated with a cognitive summary score change, POMS-Confusion, and the PNS. AON subjects exceeding a Z-score of 2.0 ranged from 5% (tapping) to 17% (Stroop) for cognitive measures, and reached 31% for the PNS. In contrast 62% exceeded a Z of 2.0 for the EEG peak frequency measure, with similar results for other EEG measures (p[lt]0.01 vs. Stroop; Chi-square).
CONCLUSIONS: Amongst subjects beginning an AED, EEG measures were considerably more sensitive than cognitive measures in detecting statistically significant individual change. The findings indicate that EEG measures can detect AED related neurophysiologic dysfunction not apparent on cognitive testing or physical examination. EEG measures may be useful in the longitudinal assessment of AED neurotoxicity in individuals.
[Supported by: U.S Department of Veteran[ssquote]s Affairs, Pfizer Inc. and Novartis Inc.]