Introduction

Transcutaneous vagal nerve stimulation (tVNS) has been applied to a number of disease areas including visceral pain, depression, migraine and epilepsy¹. However, there is marked heterogeneity in these studies pertaining to the anatomical site of stimulation (eg. cervical or auricular), and the waveform parameters of the stimulating impulse (eg. fixed or variable frequency).

This exploratory cross-sectional study will address these knowledge gaps by comparative functional neuroimaging of the neuronal response to tVNS with disparate anatomical sites and electrical waveform characteristics during rest in healthy participants. Subsequently, functional connectivity changes in the central autonomic network (CAN) (ie. insular cortex, anterior cingulate cortex, thalamus, amygdala, and the nucleus tractus solitarii (NTS) of the brain stem)² will be quantified during the different stimuli.

Methods

MRI Acquisition
To study the effect of anatomical stimulation site and waveform characteristic of the tVNS, fifteen healthy subjects (all females, mean age 24) were scanned with auricular stimulation and cervical stimulation using a fixed wave stimulation (25Hz with pulse width of 200µs) and a variable wave stimulation (oscillate between 10 to 50Hz with pulse width between 50 and 250µs)^3,4. Cervical and auricular stimulation was performed in separate scan sessions with a washout period of 2 weeks. Additionally, a scan with a sham (stimulating the hand) stimulation design was added. For each stimulation, a block design was used with five alternating periods of rest and stimuli lasting 60 seconds each.

For each stimulation, BOLD fMRI data were collected on a Siemens Magnetom 7T scanner (Siemens Healthineers, Erlangen, Germany) using a 32-channel Head Coil. Functional MRI data were acquired with a gradient echo EPI sequence using a multi-band factor of 2. This sequence used the following parameters: 1.25mm pixel size, 1.25mm slice thickness, repetition time (TR) = 2s, echo time (TE) = 19ms, flip angle = 80°, and 300 dynamics. Slices were tilted to also fully incorporate the brainstem⁵. Five additional volumes with opposite phase encoding were acquired for susceptibility distortion correction.
Additionally, for anatomical reference and segmentation, magnetization-prepared 2 rapid-acquisition gradient-echoes (MP2RAGE) MRI (TR = 5000ms, TE = 2.47ms, TI1/T12 = 900/2750ms, 0.7mm cubic voxel size).

Analysis
The fMRI data was first corrected for susceptibility-induced distortion (FSL’s topup⁶). Next, they were slice-time and motion corrected using FSL, and normalized to MNI space. Subsequently, the dataset was denoised using the anatomical component based correction method (aCompCor)⁷. Using aCompCor, the BOLD signal is cleaned using nuisance regressors related to motion and non-neuronal BOLD signals as described previously⁵. Finally, a band-pass filter of 5-100 mHz, and a 6 mm full-width at half maximum (FWHM) Guassian smoothing filter were applied to all datasets.

To visualize the effect of stimulation site and waveform characteristic, the BOLD response was correlated with the block design of the stimulation (convolved with the hemodynamic response function). The resulting Pearson's correlation coefficients were transformed to z-scores using Fisher's r-to-z transformation. Subsequently, a dependent sample student t-test was used to determine whether the BOLD response of the stimuli was significantly different from the sham stimulation, on a voxel-wise basis.

Next to visualization of the BOLD response, a quantitative region-wise analysis was performed. From the AAL-atlas⁹, the insular cortex, anterior cingulate cortex, thalamus and amygdala regions were extracted. Furthermore, the NTS of the brainstem was manually drawn in the MNI-space based on a prior study¹⁰. Mean BOLD response of the stimuli were calculated for these regions, and compared to the sham stimulation using dependent sample t-tests.

Results

The BOLD response of the stimuli with respect to the sham stimulation is shown in Figure 1.

Region-wise analysis revealed that auricular stimulation with a variable frequency resulted in a reduced BOLD connectivity in the amygdala (-.042 vs .031, p<0.01) and insular cortex (-.053 vs .008, p=0.02). Auricular stimulation with a fixed frequency, as well as cervical stimulation did not reveal significant differences on a region-wise basis.

Discussion & Conclusion

In this explorative study, the effects of stimulation site and waveform characteristics of tVNS are investigated. Interestingly, the fixed frequency tends to result in an increased BOLD response, while variable frequency stimulation showed mostly decreases in BOLD activity with respect to sham stimulation.
Moreover, auricular stimulation with a variable frequency was shown to result in a significant decrease of the BOLD response in the amygdala and insular cortex compared to sham stimulation. These preliminary results already provide valuable insights into the neuronal effects of tVNS stimulation site and waveform characteristics. However, further work should also focus on the role of the CAN regions in the cerebral functional network, for example using nodal importance metrics such as the betweenness centrality.

Acknowledgements

No acknowledgement found.