Effects of highly purified cannabidiol (CBD) on fMRI of working memory in treatment-resistant epilepsy

Abstract

ObjectiveWe aimed to determine changes in working memory and functional connectivity via functional magnetic resonance imaging (fMRI)-modified Sternberg task after treatment with highly purified cannabidiol (CBD, Epidiolex®; 100 mg/mL) in patients with treatment-resistant epilepsy (TRE). MethodsTwenty patients with TRE (mean age: 35.8 years; 7 male) performed fMRI Sternberg task before receiving CBD (“PRE”) and after reaching stable dosage of CBD (15–25 mg/kg/day; “ON”). Each patient performed 2 runs of the modified Sternberg task during PRE and ON fMRI. Twenty-three healthy controls (HCs; mean age: 25 years; 11 M) also completed the task. All were presented with a sequence of 2 or 6 letters and instructed to remember them (encoding). After a delay, a single letter was shown, and participants recalled if letter was shown in sequence (retrieval). Paired t-tests were used to analyze accuracy/response times. For each subject, event-related modeling of encoding (2 and 6 letters) and retrieval was performed. Paired t-tests controlling for seizure frequency change and scanner type were performed to assess changes in neural recruitment during encoding and retrieval in key regions of interest. ResultsThere was nonsignificant increase in mean modified Sternberg task accuracy from PRE to ON-CBD (28.6 vs. 32.1%). PRE and ON accuracy was worse than HCs (75.5%, p < 0.001). ON-PRE comparison revealed increased activation in the right inferior frontal gyrus (IFG) during 6-letter encoding. ON-HC comparison revealed increased activation in bilateral IFG and insula during 2-letter encoding. PRE-HC comparison revealed decreased activation in the left middle frontal gyrus during 6-letter encoding. None of these activations were associated with working memory performance. SignificanceTreatment-resistant epilepsy results in poorer working memory performance and lower neural recruitment compared with HCs. Treatment with CBD results in no significant changes in working memory performance and in significant increases in neural activity in regions important for verbal memory and attention compared with HCs during memory encoding.