Abstracts

Rationale: The vagus nerve stimulator (VNS) is used to treat patients with treatment resistant epilepsy (TRE), but its mechanism of action (MOA) remains poorly understood. However, thalamic activation is thought to play an important role in its MOA. As part of this phase 1 multi-center study (NCT03446664), we investigate the effect of standard VNS stimulation on thalamic fMRI activation. Methods: Up to 40 TRE patients will be implanted with VNS and undergo 3T fMRI (3 scans/day) at 2 weeks (T1), 1 month (T2), 3 months (T3) and 6 months (T4) post-implantation. Prior to undergoing T1 fMRI, the maximum tolerated stimulation output is determined starting at 0.125mA with 0.125mA increments. T2-4 start with stimulation level from the previous visit with 0.125mA increases to determine maximum tolerated setting. Each fMRI scan is a block design where stimulation is provided for 30s (ON), followed by no stimulation for 30s (OFF). Each stimulation step is assessed for 5 min (i.e. 5 ON and 5 OFF blocks). The protocol allows up to 6 stimulation steps during the 30-min fMRI. Scan 1 utilizes standard VNS stimulation parameters. Automated fMRI data processing pipeline in AFNI was developed for individual on-site processing (based on the scanner and output type). In general, data were converted to NIfTI format, then the anatomical MRI skull-stripped, aligned to fMRI and normalized to Talairach space. FMRI underwent slice-timing and motion correction, spatial normalization, Gaussian spatial smoothing to 5mm FWHM, and percent signal change calculation. A general linear model of ON vs. OFF blocks, adjusting for motion and signal drift, was performed to determine VNS-related fMRI activation per stimulation setting. Individual activation map for each setting was thresholded at a significance level of 0.05 (2-tailed) and 2-voxel minimum, and then an anatomical thalamic mask (Fig. 1A) was used to extract mean t-value for each thalamic cluster meeting minimum thresholds. The stimulation setting yielding highest mean thalamic t-value over significant clusters is selected as the Peak. One sample t-tests, covarying for site and visit number, were performed to create group average maps for the Peak VNS-related thalamic activation and for the stimulation settings preceding and following the Peak (Fig. 1B). Group maps were thresholded at a significance level of 0.05 (2-tailed) and 20-voxel minimum. Results: Seven TRE patients from three sites were implanted with VNS and underwent 3T fMRI at T1 (N=7), T2 (N=7), T3 (N=3), and T4 (N=4). Quality data were obtained in all scan sessions, and all participants exhibited thalamic fMRI activation during standard VNS stimulation (Fig. 1B). Group activation map of peak thalamic activation from VNS stimulation (combining 21 datasets) showed corresponding widespread cortical activation (Fig. 1B). Due to the study design, there were fewer data for stimulation settings preceding (n=17 for Peak-1; n=15 for Peak-2) and following (n=11 for Peak+1; n=7 for Peak+2) the Peak. FMRI activation patterns for stimulation settings before and after the Peak are relatively sparse, with a general pattern of increased and decreased activation leading up to, and primarily decreased activation following the Peak. Conclusions: Implementation of a 3T fMRI scanning protocol in TRE patients with VNS for the assessment of fMRI activation changes is feasible. There were individual differences in the stimulation setting that induced peak thalamic fMRI activation. Modulation of thalamic activation consistently observed in TRE patients may be a putative MOA of the VNS. Funding: LivaNova, Inc.