Understanding the Cognitive Side Effects of Antiepileptic Drugs: Can Functional Imaging Be Helpful?

Effects of Carbamazepine and Lamotrigine on Functional Magnetic Resonance Imaging Cognitive Networks Xiao F, Caciagli L, Wandschneider B, Sander JW, Sidhu M, Winston G, Burdett J, Trimmel K, Hill A, Vollmar C, Vos SB, Ourselin S, Thompson PJ, Zhou D, Duncan JS, Koepp MJ. Epilepsia. 2018;59:1362-1371.

To investigate the effects of sodium channel-blocking antiepileptic drugs (AEDs) on functional magnetic resonance imaging (fMRI) language network activations in patients with focal epilepsy.

In a retrospective study, we identified patients who were treated at the time of language fMRI scanning with either carbamazepine (CBZ; n = 42) or lamotrigine (LTG; n = 42), but not another sodium channel-blocking AED. We propensity-matched 42 patients taking levetiracetam (LEV) as "patient-controls" and included further 42 age- and gender-matched healthy controls. After controlling for age, age at onset of epilepsy, gender, and antiepileptic comedications, we compared verbal fluency fMRI activations between groups and out-of-scanner psychometric measures of verbal fluency.

Patients on CBZ performed less well on a verbal fluency tests than those taking LTG or LEV. Compared to either LEV-treated patients or controls, patients taking CBZ showed decreased activations in left inferior frontal gyrus and patients on LTG showed abnormal deactivations in frontal and parietal default mode areas. All patient groups showed fewer activations in the putamen bilaterally compared to controls. In a post hoc analysis, out-of-scanner fluency scores correlated positively with left putamen activation.

Our study provides evidence of AED effects on the functional neuroanatomy of language, which might explain subtle language deficits in patients taking otherwise well-tolerated sodium channel-blocking agents. Patients on CBZ showed dysfunctional frontal activation and more pronounced impairment of performance than patients taking LTG, which was associated only with failure to deactivate task-negative networks. As previously shown for working memory, LEV treatment did not affect functional language networks.