Skip to main content
Dryad

Functional representation of trigeminal nociceptive input in the human periaqueductal gray

Cite this dataset

Mehnert, Jan; Alexandra, Tinnermann; Hauke, Basedau; May, Arne (2024). Functional representation of trigeminal nociceptive input in the human periaqueductal gray [Dataset]. Dryad. https://doi.org/10.5061/dryad.mw6m90642

Abstract

The periaqueductal gray (PAG) is located in the mesencephalon in the upper part of the brainstem and as part of the descending pain modulation is considered a crucial structure for pain control. Its modulatory effect on painful sensation is often seen as a systemic function affecting the whole body similarly. However, recent animal data suggest some kind of somatotopy in the PAG. This would make the PAG capable of dermatome-specific analgesic function. We electrically stimulated the peripheral dermatomes of the trigemino-cervical complex consisting the three branches of the trigeminal nerve and the greater occipital nerve in sixty-one humans during optimized brainstem fMRI. We provide evidence for a fine-grained and highly specific somatotopic representation of nociceptive input in the PAG in humans and a descending functional connectivity between the individual representations of the peripheral nerves in the PAG and the brainstem nuclei of these nerves. Our data suggest that the downstream antinociceptive properties of the PAG may be rather specific down to the level of individual dermatomes. This study was preregistered at clinicaltrials.gov: NTC03999060 and primary outcomes already published elsewhere.

README: Functional representation of trigeminal nociceptive input in the human periaqueductal gray

https://doi.org/10.5061/dryad.mw6m90642

Description of the data and file structure

This dataset consists of processed images from functional magnetic resonance imaging (fMRI) in NIFTI-format readable by common MRI viewing and analyzing software like Mango (https://mangoviewer.com/, MRIcron (https://www.nitrc.org/projects/mricron or SPM12 (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/). The images of this dataset were created using the SPM12 toolbox.

For each of the 61 subjects there is a folder containing 8 images:

·        con0003.nii refers to neuronal activity occurring in the brain by stimulating the first trigeminal branch (V1),

·        con0004.nii refers to neuronal activity occurring in the brain by stimulating the second trigeminal branch (V2),,

·        con0005.nii refers to neuronal activity occurring in the brain by stimulating the third trigeminal branch (V3),

·        con0006.nii refers to neuronal activity occurring in the brain by stimulating the greater occipital nerve (GON).

These four images are additionally uploaded in a masked version, where the periaqueductal grey (PAG) is masked, with the suffix ‘_masked’.

As the dataset stems from two independent cohorts, the naming convention of the folders differs. We initially measured 25 participants for hypothesis generation. These are the folders include the phrase ’PRISMA’. Power calculations revealed that 36 participants were needed to reproduce the hypothesized results generated with the first cohort. Folders from this cohort start with the phrase ‘subject’. For the results in the PAG, we combined both groups for a robust outcome resulting in 61 participants. Missing numbers in the series stem from either pilots or dropouts due to technical problems during imaging.

These images were used for the group statistics. We used a one-way repeated-measure ANOVA implemented in aforementioned software package SPM12.

Methods

Sixty-three healthy, right-handed volunteers participated in our study on repetitive, randomize, peripheral, painful electrical stimulation of the three trigeminal branches (V1, V2, V3) innervating the facial dermatomes and the greater occipital nerve (GON) which innervates the back of the head (Figure 1B in the related publication). For the primary outcomes, (confirmation of earlier studies showing somatotopic representation for the insula, thalamus etc) which are not part of this manuscript, we initially measured 25 participants for hypothesis-generation. Power calculations revealed that 36 participants were needed to reproduce the hypothesized results. As the results in the PAG, which we present here, are preregistered as secondary outcome we combined both groups for a robust outcome resulting in 63 participants. Two volunteers of the second cohort had to be excluded due to technical problems, leaving 61 (27 male, age: 28.51 +/- 9.4) for the further analysis. All participants were free from psychiatric and neurological diseases and neither they nor their first-degree relatives suffered from headache disorders.

Electrical stimulation was delivered with a MR-compatible Digitimer DS7A Current Stimulator (Digitimer Ltd., Welwyn Garden City, UK), which was coupled to four WASP electrodes (Specialty Developments, Bexley, UK) via a D188 Remote Electrode Selector (Digitimer Ltd., Welwyn Garden City, UK) and custom-build MR-compatible cables. The cables were build using a published 1 and MR-safety tested design to prevent tissue damage due to currents induced by electro-magnetic wave coupling. While the subject was sitting on the MR-bed, the four electrodes were positioned on the left side according to the three branches of the trigeminal nerve and the GON (Figure 1B in the related publication). The GON was located by palpation according to validated procedures2 and the electrode positioned immediately above. V1 was stimulated by means of an electrode placed on an arbitrary vertical line between the medial and lateral quarter of the face, corresponding to the middle of the eyebrow and approximately 1 cm above. The V2 was stimulated 1 cm lateral of the same arbitrary vertical line on the level of a horizontal line through the inferior part of left ala of the nose. V3 was stimulated along the same vertical line approximately 1 cm caudal from the corner of the mouth. Figure 1B in the related publication sketches the location of the electrodes. Electrode 3 stimulating the third branch of the trigeminal nerve proved to be the most painful of all 3 trigeminal dermatomes. To make sure that all four sites received robust but bearable pain with the same standardized input, we used this site as orientation for the stimulus intensity.

After fixing the electrodes, the subjects were moved into the scanner and the electrical detection thresholds (EDT) by means of the QUEST-procedure 3 at all electrode sites were determined. The final current was set to 10 times the EDT of the electrode above V3 for both experiments but was not allowed to exceed 5 mA nor a pain rating above 50 (with levels from 0 to 100) for a single pulse. The actual stimulation consisted of a small train of three pulses separated by 100 ms each with 400 V and 2 ms duration. Each stimulus was followed by a break of 3 s (jittered between 2 and 4 seconds), a pain intensity rating on a visual analogue scale (VAS) with levels between 0 and 100 using a button box with the right hand, and another break before the next trial started. For technical reasons, the VAS rating are only available for the second cohort, i..e. 36 participants. The inter-trial interval was set to 15 s (jittered between 12 and 18 s). The stimulation site was randomized and each site was repeated 10 times per session. The volunteers participated in 3 sessions resulting in 30 trials per stimulation site and subject during approximately 30 minutes of fMRI scanning. The experimental design is sketched in Figure 1A in the related publication.

All MR data was recorded with a Siemens 3T PRISMA scanner (Siemens, Erlangen) using a 64-channel head coil. During the actual experiment we recorded 3 sessions with 230 images each using an EPI protocol (repetition time 2.93 s, echo time 33 ms, 1.3x1.3x2.0 m3 spatial resolution, GRAPPA acceleration, flip angle 80°, 72 slices with a multiband factor of 2, FOV 215 mm, no gap, flow rephasing) with a field of view covering the brainstem as low as C2/3, cerebellum, midbrain and the insula cortices. In each session the first five images were removed to avoid scanner saturation effect. Afterwards we recorded fieldmaps (repetition time 0.792 s, echo times 5.51 and 7.97 ms, 3x3x2 mm3 spatial resolution, flip angle 20°, 72 slices, FOV 222 mm, no gap) covering the same volume as the EPIs to attenuate the inhomogeneity of the magnetic field. Pulse and breathing were recorded simultaneously to attenuate extra-cerebral (i.e. cardio-vascular) artifacts. Last we acquired high resolution (1 mm3) anatomical images (MPRAGE, repetition time 2.3 s, echo time 2.98 ms, flip angle 9°, 240 slices, FOV 256 mm).

All MRI data was first filtered using the spatially adaptive non-local means filter implemented in the CAT12 toolbox. The fMRI data was then corrected for movements and for distortions of the homogeneity of the magnetic field (fieldmaps) using the realign and unwarp algorithm as implemented in SPM12. Additionally, slice time correction was performed using the onsets of the single slices as suited for our multiband protocol. We then calculated a subject-wise GLM including condition-wise onsets of each stimulus as stick functions, which were then convolved with a hemodynamic response function (HRF). The button box responses as well as the onset and duration of the visual analogue scale were modelled as regressors of no interest. Additional regressors of no interest were included to correct for (uncorrelated) movement, cardiovascular influence, using the algorithms proposed by Deckers and colleagues 4, and changes in the spinal fluid extracted from the 4th ventricle. The co-registered structural images were segmented with the unified segmentation approach algorithm implemented in SPM12 but using the templates provided by Blaiotta 5, which are optimized for the brainstem and spinal cord, to gain deformation fields used to warp the contrast images of the subject-wise GLM into MNI space. Each step was carefully controlled by visual inspection. We further calculated a group template, and gray and white matter masks from the warped structural images.

Group statistics were calculated by a one-way within subject ANOVA. Results from the effect of the individual stimulation locations were masked with a PAG mask stemming from Faull and colleagues 6 and had to pass a statistical threshold of pFWE<10-4 (voxel-wise family wise error (FWE) corrected, T>6.5, n=61, df=180). This resulted in four statistical parametric maps, one for each stimulation site, including their voxel-wise t-values. For each voxel within this PAG mask (Figure 1C in the related publication), we then searched for the stimulation site with the maximal t-value which resulted in clusters specific for the individual stimulation site within the PAG.

References

1.           Schmidt K, Forkmann K, Sinke C, Gratz M, Bitz A, Bingel U. The differential effect of trigeminal vs. peripheral pain stimulation on visual processing and memory encoding is influenced by pain-related fear. Neuroimage. 2016;134:386-395. doi:10.1016/j.neuroimage.2016.03.026

2.           Loukas M, El-Sedfy A, Tubbs RS, et al. Identification of greater occipital nerve landmarks for the treatment of occipital neuralgia. Folia Morphol (Warsz). 2006;65(4):337-342.

3.           Taesler P, Rose M. Psychophysically-anchored, Robust Thresholding in Studying Pain-related Lateralization of Oscillatory Prestimulus Activity. J Vis Exp. 2017;(119). doi:10.3791/55228

4.           Deckers RHR, van Gelderen P, Ries M, et al. An adaptive filter for suppression of cardiac and respiratory noise in MRI time series data. NeuroImage. 2006;33(4):1072-1081. doi:10.1016/j.neuroimage.2006.08.006

5.           Blaiotta C, Freund P, Cardoso MJ, Ashburner J. Generative diffeomorphic modelling of large MRI data sets for probabilistic template construction. Neuroimage. 2018;166:117-134. doi:10.1016/j.neuroimage.2017.10.060

6.           Faull OK, Jenkinson M, Clare S, Pattinson KTS. Functional subdivision of the human periaqueductal grey in respiratory control using 7 tesla fMRI. NeuroImage. 2015;113:356-364. doi:10.1016/j.neuroimage.2015.02.026

Funding

Deutsche Forschungsgemeinschaft, Award: 178316478/A5 to AM, SFB936

Deutsche Forschungsgemeinschaft, Award: TI 1110/1

Max Planck Society