The insula represents a key neurobiological pain hub in psoriatic arthritis

Abstract

Background: Pain remains a principal complaint for people with psoriatic arthritis (PsA), despite successful mitigation of inflammation. This situation alludes to the co-existence of distinct pain mechanisms. Nociceptive and nociplastic mechanisms are clinically challenging to distinguish. Advances in brain functional magnetic resonance imaging (fMRI) have successfully characterised distinct pain mechanisms across several disorders, in particular implicating the insula. This is the first study to characterise neurobiological markers of pain mechanisms in PsA employing fMRI.

Methods: PsA participants underwent a 6-minutes resting-state fMRI brain scan, and questionnaire assessments of nociplastic pain (2011 ACR fibromyalgia criteria) and body pain, assessed using the Numeric Rating Scale (NRS, 0-100). Functional connectivity between insula seeds (anterior, mid, posterior), and the whole brain was correlated with the above pain outcomes correcting for age and sex, and false discovery rate (FDR) for multiple comparisons.

Results: A total of 46 participants were included (age 49 ± 11.2; 52% female; FM score 12.5 ± 5.7; overall pain 34.8 ± 23.5). PsA participants with higher fibromyalgia scores displayed increased connectivity between: (1) right anterior insula to DMN (P < 0.05), (2) right mid and left posterior insula to parahippocampal gyri (P < 0.01 FDR); and (3) right mid insula to left frontal pole (P = 0.001 FDR). Overall pain was correlated with connectivity of left posterior insula to classical nociceptive regions, including thalamus (P = 0.01 FDR) and brainstem (P = 0.002 FDR).

Conclusion: For the first time, we demonstrate objectively that nociceptive and nociplastic pain mechanisms co-exist in PsA. PsA pain cannot be assumed to be only nociceptive in origin and screening for nociplastic pain in the future will inform supplementary analgesic approaches.

Keywords: Chronic pain; Fibromyalgia; Insula; Neuroimaging; Psoriatic arthritis.

Fig. 1

Functional connectivity pipelines. Preprocessing panel displays the default MNI pipeline in the CONN toolbox that requires both structural (T1 weighted) and functional (T2* weighted) MRI images. The preprocessed smoothed (swau) images were then used for. independent component analysis (ICA), which along with a default mode network (DMN) template identified the DMN. Six insula subregions were also extracted as spheres of 6 mm surrounding a peak voxel. Montreal Neurological Institute (MNI) coordinates for each voxel include left anterior insula (LantIC): x =– 32, y = 16, z = 6; left mid insula (LmidIC): x =– 38, y = 2, z = 8; left posterior insula (LpIC): x =– 39, y =– 15, z = 1; right anterior insula (RantIC): x = 32, y = 16, z = 6; right mid insula (RmidIC): x = 38, y = 2, z = 8; right posterior insula (RpIC): x = 39, y =– 15, z = 8. The timeseries of the DMN and insula subregions were extracted from the preprocessed unsmoothed (wau) images and denoised for motion, physiological noise (band-pass filtering) and white matter (WM) and cerebrospinal fluid (CSF) signals. Functional connectivity between the DMN and the six insula regions were used in a region-of-interest (ROI) analysis to look for associates with FMness score. A seed to voxel analyses was also run with all seven regions for FMness score but only the posterior insula regions for the overall body pain NRS. Adapted from Scheinost et al. (2017), while the figure was partly generated using Servier Medical Art, licensed under a Creative Commons Attribution 3.0 (unported license) and edited using Inkscape (2020).

Fig. 2

The default mode network (DMN) in patients with psoriatic arthritis. Visualisation of the DMN in volumetric space (panel A) in an axial view at different slices (z:– 14,– 6, 2, 10, 18, 26, 34, 42). Panel B displays the same network in surface space in left lateral and medial view. Both visualisations were created in the CONN toolbox (Nieto-Castanon, 2020)

Fig. 3

DMN-to-insula connectivity is associated with nociplastic pain. Panel A The left image visualises the seed of the right anterior insular cortex (RantIC) and the default mode network (DMN), which is pictured mostly in the posterior cingulate cortex for visualisation purposes. The images on the right visualise a scatterplot (95% confidence intervals) of the functional connectivity (Fisher z-transformed r values) between the DMN and RantIC and nociplastic pain scores and their Pearson correlation coefficient (R). The table above the plot also displays the degrees of freedom (df), test statistic (t statistic) and p value of the general linear model between functional connectivity and nociplastic pain while controlling for age and sex. Panel B The top images visualise the right mid-insular cortex (RmidIC) and left posterior insular cortex (LpIC) seeds and the clusters of voxels within the left and right parahippocampal gyri as well as the left frontal pole, with which functional connectivity was associated with nociplastic pain while controlling for age and sex in a seed‒to-voxel analysis. The images below display the scatterplots, statistical tables, peak voxel coordinates in MNI space (x, y, z) and cluster size, with the p values after false discovery rate (FDR) correction for multiple comparisons of the general linear models.

Fig. 4

Left posterior insula connectivity to the brainstem and thalamus is associated with current pain. The panels above display the clusters in the brainstem and left thalamus whose connectivity with the left posterior insular cortex (LpIC) was associated with current overall body pain while controlling for age and sex. The panels below display the scatterplots (95% confidence intervals) with Pearson correlation coefficients (R) between the functional connectivity (Fisher z transformed r values) of the seeds and clusters with current overall body pain. The tables describe the peak voxel coordinates in MNI space (x, y, z), the cluster size, the degrees of freedom (df), the T statistics, and the p values after false discovery rate (FDR) correction for multiple comparisons of the general linear models (GLMs)