Functional and neurochemical disruptions of brain hub topology in chronic pain

Figures

• 

  Figure 1.. Summary of graph theoretical methods. Resting state fMRI data were analyzed for 40 FM patients and 46 HCs. In addition, a separate validation cohort of 11 FM patients was also analyzed. Two hundred sixty-four nodes were defined using the Power atlas30 that reliably segregate into large-scale resting state networks. Pairwise (Fisher Z-transformed Pearson) correlation matrices were created for each participant, which were thresholded and binarized across a range of network densities. Graph theoretical measures were calculated, and between-group differences were assessed using nonparametric permutation testing. Post hoc correlations between hub measures, clinical pain, and glutamate were also performed. FM, fibromyalgia; fMRI, functional magnetic resonance imaging.

• 

  Figure 2.. Altered hub topology in FM. Hubs were defined for each group separately as greater than 1 SD above the mean group eigenvector centrality. FM hubs (A) include the bilateral anterior insulae, mid insula, M1, S1, SMA, and STG. HC hubs, depicted in (B), are mainly in frontal and posterior regions. Ant Ins, anterior insula; FM, fibromyalgia; HC, healthy controls; M1, primary motor cortex; Mid Ins, mid insula; S1, primary somatosensory cortex; SFG, superior frontal gyrus; SMA, supplementary motor area; STG, superior temporal gyrus; V1/V2/V3, primary, secondary and visual association cortices.

• 

  Figure 3.. Rich-club membership is altered in FM. (A) Both FM and HC have significant rich-club organization compared with random networks. The rich-club regime (compared with random networks) for FM was between k = 4 and k = 6 and k = 8 and k = 22 and for HC was between k = 2 and k = 25. There was no significant difference in the rich-club coefficient (ϕ) between FM and HC at any k level. Normalized rich-club curves are also depicted, which show an increasing ϕnorm over a range of k for both groups. However, rich-club membership was different between groups. FM and HC rich-club nodes are depicted in (B and C), respectively, for k = 22 in FM and k = 25 in HC, the highest significant k level. The FM rich club includes the bilateral anterior insulae, mid cingulate, and anterior cingulate cortex. The data depicted are from 5% network density; results are similar for other thresholds (See Supplementary Fig. 3, available at http://links.lww.com/PAIN/A729). ACC, anterior cingulate cortex; Ant Ins, anterior insula; FM, fibromyalgia; HC, healthy control; MCC, mid cingulate cortex. *FM significantly different from random networks; +HC significantly different from random networks.

• 

  Figure 4.. Rich-club membership varies with clinical pain. The high pain FM group (A) had rich-club nodes primarily in bilateral S1, M1, SMA, and the right posterior insula. The rich club in the medium pain FM group (B) contained fewer S1/M1 nodes and more nodes in the default mode network. In the low pain group (C), there was a shift in rich-club membership to posterior default mode and visual regions. The data depicted are from 5% network density; results are consistent across thresholds (See Supplementary Fig. 4, available at http://links.lww.com/PAIN/A729). FM, fibromyalgia; M1, primary motor cortex; S1, primary somatosensory cortex; SMA, supplementary motor area.

• 

  Figure 5.. Correlations with clinical pain in FM. (A) Within FM patients, there were positive correlations between clinical pain and eigenvector centrality (a measure of how connected a brain region is to other highly connected regions) in 14 bilateral S1/M1 nodes, 3 nodes in right STG, 2 nodes in the right posterior insula, and 1 node each in the right SMA and right SMG. Representative correlations are shown in (B) for left M1/operculum and right M1/S1 and regression lines for all 21 correlations are depicted in (C). EC, eigenvector centrality; FM, fibromyalgia; M1, primary motor cortex; S1, primary somatosensory cortex; SMA, supplementary motor area; SMG, supramarginal gyrus; STG, superior temporal gyrus; VAS, visual analog scale.

• 

  Figure 6.. The relationship between clinical pain, eigenvector centrality, and posterior insula Glx. (A) Glutamate + glutamine (Glx; CSF-corrected) was measured in the right posterior insula using 1H-MRS (voxel shown in dark gray). The overlap between the 1H-MRS voxel and 2 nodes from the Power's atlas is shown in the inset. (B) In FM, clinical pain and posterior insula Glx positively correlated with the averaged, standardized eigenvector centrality in these 2 posterior insula nodes. (C) In a mediation analysis, posterior insula eigenvector centrality significantly mediated the relationship between Glx in the posterior insula and clinical pain. Standardized values are shown for ease of interpretation. 1H-MRS, proton magnetic resonance spectroscopy; AIU, arbitrary institutional unit; EC, eigenvector centrality; FM, fibromyalgia; Glx, glutamate + glutamine; SM, sensorimotor network; VAS, visual analog scale.