Abstracts

Rationale: Topiramate (TPM) is indicated for partial onset seizures and migraine prophylaxis, and is used to manage obesity, bipolar disorder, and pain. Unfortunately, individual patients on TPM manifest varying degrees of cognitive impairment, often reporting significant language and speech problems. In order to characterize TPM-induced effects on spontaneous speech and their relationship to individual drug pharmacokinetics, we studied the effects of a single, 100 mg dose of TPM on a brief neuropsychological battery (NP) in healthy adults. A novel semi-automated language and speech analysis (SALSA) was used to detect drug-induced changes. Methods: Eleven healthy volunteers were included in the analysis of a crossover of TPM (100mg), lorazepam (LOR: 2mg), and placebo (PL). LOR, a benzodiazepine, was chosen as a comparator drug. The NP included the COWA, a test of generative verbal fluency, a picture description (PD) test, used to elicit spontaneous speech, and the MCG Paragraph Memory (MCG) test, where the subject is asked to recall a short story presented verbally twice during the NP (MCG1&MCG2). Baseline measures were collected on the first and last study sessions and averaged to account for practice effects. Subjects were randomized to TPM, LOR, or PL with at least a one-week washout period between each session. The NP, recorded for speech analysis, was given 1hr after dosing, and lasted approximately 1hr. A single blood draw was taken immediately after NP testing; TPM and LOR concentrations were measured using a validated LCMS method. Change in individual test performance was calculated by subtracting the average baseline test score from the score during drug (or placebo) treatment divided by the average baseline. Results: Individual TPM plasma levels varied from 0.23-2.81ug/mL. The number of dysfluencies generated from the PD during the TPM arm and measured using SALSA was highly correlated with TPM levels (Spearman s r=0.65, p=0.03). COWA failed to reveal any relationship between verbal fluency and drug plasma levels. SALSA also discriminated between the effect of TPM and LOR on individual performance on MCG2. The number of correct words recalled while on TPM was correlated negatively with TPM levels (Spearman s r=-0.75 p=0.007), but positively with LOR levels (r=0.76 p=0.007). Conclusions: This is the first study to demonstrate a significant association between TPM plasma levels and individual measures of verbal fluency and recall. SALSA, in combination with a PD task proved more sensitive than traditional methods of analysis to TPM-induced changes in verbal fluency. These data facilitate our understanding of the heterogeneity of the cognitive response to a single dose of TPM and underscore the limitations of only taking into account drug dosage when clinically assessing severity of TPM s effects on speech. Given its ability to differentiate between the effects of TPM and LOR on story recall, we expect SALSA to yield further insights into drug-induced effects on cognitive function.