Abstracts

Rationale: Topiramate (TPM) is indicated for several conditions including partial onset seizures. Patients on TPM manifest varying degrees of cognitive impairment including deficits in generative (phonemic) verbal fluency as measured by the Controlled Oral Word Association (COWA) test. We explored the utility of a pharmacokinetic/pharmacodynamic (PK/PD) model derived from previously collected 100 mg TPM data to simulate subjects' cognitive response to 200 mg. The simulation results are compared to empirical data from individuals given a single 200 mg TPM dose to assess the validity of this approach. Methods: We previously developed linear and exponential decline PK/PD link models to describe the effect of TPM (100 mg) exposure on COWA performance (Ahmed et al, 2012, Abstract LBI-1, ASCPT). In the current study, we administered the COWA test to healthy volunteers (n=9, 18-50 years old) randomized to a 200 mg TPM or placebo arm. COWA was assessed at pre- and post-treatment baselines and at ~2.5 hours after dose. Blood samples were collected immediately after the COWA for measurement of TPM levels. Model-based simulation analyses were conducted where the individual COWA-time profiles were simulated for up to 4 hours after a single dose of 200 mg of TPM. 1000 datasets were simulated and the median, 2.5th, and 97.5th percentiles of the simulated PD profiles were overlaid on the actual observations from the current study. Plots were visually examined to assess the predictability of both link models to the PD profiles, and the percentage of observations falling inside the 95 percentile confidence intervals (CI) was calculated. Results: Overall 25 observations for COWA were collected. The TPM PK after a single 200 mg dose was simulated using a two-compartment linear model. Sources of variability that influenced the simulated PD (i.e., COWA) profiles were: 1) interindividual variability in PK parameters; 2) interindividual variability in parameters of the exposure-response models; and 3) practice effect; an estimated 12% increase in baseline COWA scores after the 3rd time the test was administered. Both models were predictive of the PD profiles after a 200 mg TPM dose. Of the observed COWA scores, 80% and 88% fell within the 95 percentile CI of the simulated COWA-time profiles using the exponential and linear decline models, respectively. Both models simulated a comparable median COWA-time profile. The linear model resulted in a higher simulated variability in PD profiles than the exponential decline model. Conclusions: The developed PK/PD models were useful in simulating both median and interindividual variability of the COWA-time profiles after a single 200 mg TPM dose. These models can be useful in predicting subpopulations of patients who would be more susceptible to impairment of COWA while providing PK/PD evidence-based dose optimization of TPM in such patients with epilepsy.