Abstracts

Rationale: Memory impairment is frequently reported by persons with epilepsy (PWE). The effects of cannabidiol (CBD) on cognition in PWE are uncertain. We examined the effects of adjunct therapy with a pharmaceutical grade CBD (Epidiolex) on working memory in PWE using modified Sternberg fMRI task (mST). Methods: Eleven PWE (5 male) performed mST before receiving CBD and after 2 weeks of reaching a stable intermediate dose (20 or 25 mg/kg/day). Each patient performed 2 runs of a modified mST during each fMRI visit. Patients were presented with a group of either 2 or 6 letters for 5 sec and instructed to remember these letters (encoding). After a black screen for 4 or 12 sec, a single letter (target) appeared and patients had to indicate via button press whether or not the target was part of the group of letters they just saw (retrieval). For each subject, event-related modeling of encoding and retrieval conditions was performed (accounting for head motion, MRI signal drift and serial correlation of noise parameters). Single-subject general linear modeling was also performed to assess the effect of low vs. high load (2 vs. 6 letters) on encoding, and the combined effect of load and delay (low load/short delay vs. high load/long delay) on retrieval. Regression analyses were performed to assess treatment changes in neural recruitment during encoding and retrieval, while accounting for changes in seizure frequency (significance at p < 0.05 uncorrected, cluster threshold 10 voxels). Changes in memory performance were assessed using paired samples t-tests. Results: There was a non-significant (p=0.43) increase in task accuracy from pre-CBD (30.7%) to visit 2 (35.2%); response times were similar at both visits (p=0.89). Compared to pre-CBD fMRI, at visit 2 we observed both increased and decreased prefrontal activity during 6-letter encoding (Fig. 1A) and increased activity during 2-letter encoding (Fig. 1B). There was greater difference in fMRI response to encoding 6 vs. 2 letters in the right uncus and middle frontal gyrus at visit 2 compared to pre-CBD fMRI, while difference in encoding was greater at pre-CBD compared to visit 2 fMRI in the left inferior frontal gyrus and bilateral posterior cingulate (Fig. 1C). For working memory retrieval, recruitment of brain regions was decreased at visit 2 compared to pre-CBD fMRI for both load/delay conditions (Fig. 2A,B), and differences in activation between conditions were greater at visit 2 compared to pre-CBD fMRI in the thalamus, left cerebellar tonsil, and right lingual gyrus, middle temporal gyrus and superior frontal gyrus (Fig. 2C). Conclusions: PWE showed non-significant gains in working memory that were associated with a shift in neural recruitment during the mST compared to pre-CBD fMRI. We posit that the overall greater activation during encoding and decreased activation during retrieval after CBD treatment suggest greater neural engagement to improve encoding processes and correspondingly more efficient retrieval processing that requires less neural demand. These preliminary results in a small sample of patients suggest the potential for changes in fMRI response to a working memory probe while receiving CBD for the treatment of epilepsy. Funding: State of Alabama