Abstracts

RATIONALE: Prior cross sectional studies in epilepsy patients have shown significant correlations between subjective measures but poor correlation between subjective and objective neuropsychological measures. The effect of antiepileptic drugs (AEDs) on these associations is unknown. In this study, we compared the correlation of subjective (S) and objective cognitive (C) neuropsychological measures in healthy adults in 3 drug conditions: carbamazepine (CBZ), lamotrigine (LTG), and non-drug (ND). METHODS: Neuropsychological evaluations were performed at 2 non-drug baselines, at the end of two 10 week AED treatments (separated by 4 week washout), and at a final non-drug baseline (4 weeks post last AED) in 25 healthy adults using a randomized, double-blind, crossover design. The battery had 19 instruments yielding 40 variables including 20 subjective and 20 objective cognitive measures. Pearson correlations (2 tailed; p<.0001) were calculated across the neuropsychological variables for each condition: CBZ, LTG and average non-drug (ND). Results were segregated based on comparison of subjective (S) and cognitive (C) categories (S to S, S to C, C to C), and the distribution across the 3 drug states compared. Performance in each AED state was also compared to ND using paired t-tests. RESULTS: Mean dosages on AED test days were LTG = 150 mg/day and CBZ = 696 mg/day (range = 400 - 1000 mg/day). Mean CBZ level = 7.6 g/ml (range = 5.1-10.5). The number of statistically significant correlations in each condition were as follows. Non-drug: S to S = 35, S to C = 7, C to C = 35; CBZ: S to S = 56, S to C = 7, C to C = 8; LTG: S to S = 29, S to C = 28, C to C = 22. These distributions differed for ND vs CBZ (p<.00002), ND vs LTG (p<.0003), and CBZ vs LTG (p<.000001). Compared to ND, performance on CBZ was worse for 24 (60%) of the variables, and LTG was worse on 1 (2.5%) and better on 2 (5%). CONCLUSIONS: Compared to ND, CBZ impairs both S and C measures, altering interrelations of S to S and C to C. In contrast, LTG has less effect on both S and C measures, but may increase S to C correlations via its psychotrophic effects. Further studies are needed to confirm this observation and delineate the underlying mechanisms.