Abstracts

Rationale: Topiramate (TPM) is known to cause language impairment with word finding difficulties and working memory disturbances in both healthy volunteers and epileptic patients. We investigated whether fMRI can predict the cognitive side-effects of TPM by comparing cognitive task related activations and de-activations in patients on and off this drug.Methods: In a longitudinal pilot study, we examined 4 patients with epilepsy (1 temporal lobe epilepsy (TLE) and 3 extra-TLE) using a verbal fluency fMRI paradigm with alternating periods of rest and generation of words beginning with different letters. FMRI was acquired either prior to starting TPM (TPM-off), and repeated once patients were taking a working dose (TPM-on), or before withdrawal (TPM-on) and again after complete cessation (TPM-off). Two healthy male controls were studied for comparison before (TPM-off) and 3 hours after a single oral dose of 200mg TPM (TPM-on). In addition, subjects underwent a brief neuropsychological evaluation including verbal fluency and digit span. We analyzed the fmri data with SPM8 applying a fixed-effect statistical model to compare the differences between the two acquisitions (TPM-on versus TPM-off) for each subject. SPM maps were thresholded with p<0.001, uncorrected.Results: Both chronic administration and a single dose of TPM led to a reduction of category verbal fluency (%reduction: median 27% range 18%-70%). Four of the six subjects showed a reduced backwards Digit Span Score (TPM-off: median 6.5, range 4-14% versus TPM-on: median 4, range 2-11%). FMRI analysis showed a non-significant decrease in activation of normal language areas during the TPM-on condition. Task-related deactivations revealed the typical default mode network (DMN) during the TPM-off condition (areas in blue in Figure A (patients 1-4, controls 1 and 2), p<0.001 for display) but not during the TPM-on condition (red in Figure A). The differences between the two states resulted in a statistical map that includes precuneus and some frontal and parietal areas of the DMN (Figure B).Conclusions: : The failure to deactivate the DMN during the verbal fluency task on TPM might explain the specific cognitive (language) impairment. Similar neuropsychological and fMRI findings in patients (TPM add-on) and healthy controls (after single dose) provide evidence for a specific effect of TPM in default mode areas that may be essential components of a wider language comprehension network, regardless of other AEDs and underlying pathology. Our preliminary data suggest that fMRI may be useful to detect subtle drug effects on cognitive networks not obvious on neuropsychological evaluation, but larger longitudinal studies in both patients and controls will be necessary to assess the specific influence of TPM and other anti-epileptic medication on cognitive function.