Secondary Logo

Share this article on:

Dynamic pain connectome functional connectivity and oscillations reflect multiple sclerosis pain

Bosma, Rachael L.a; Kim, Junseok A.a,b; Cheng, Joshua C.a,b; Rogachov, Antona,b; Hemington, Kasey S.a,b; Osborne, Natalie R.a,b; Oh, Jiwonc; Davis, Karen D.a,b,d,*

doi: 10.1097/j.pain.0000000000001332
Research Paper
Global Year 2018

Pain is a prevalent and debilitating symptom of multiple sclerosis (MS); yet, the mechanisms underlying this pain are unknown. Previous studies have found that the functional relationships between the salience network (SN), specifically the right temporoparietal junction a SN node, and other components of the dynamic pain connectome (default mode network [DMN], ascending and descending pathways) are abnormal in many chronic pain conditions. Here, we use resting-state functional magnetic resonance imaging and measures of static and dynamic functional connectivity (sFC and dFC), and regional BOLD variability to test the hypothesis that patients with MS have abnormal DMN-SN cross-network sFC, dFC abnormalities in SN-ascending and SN-descending pathways, and disrupted BOLD variability in the dynamic pain connectome that relates to pain inference and neuropathic pain (NP). Thirty-one patients with MS and 31 controls completed questionnaires to characterize pain and pain interference, and underwent a resting-state functional magnetic resonance imaging scan from which measures of sFC, dFC, and BOLD variability were compared. We found that (1) ∼50% of our patients had NP features, (2) abnormalities in SN-DMN sFC were driven by the mixed-neuropathic subgroup, (3) in patients with mixed NP, dFC measures showed that there was a striking change in how the SN was engaged with the ascending nociceptive pathway and descending modulation pathway, (4) BOLD variability was increased in the DMN, and (5) the degrees of sFC and BOLD variability abnormalities were related to pain interference. We propose that abnormal SN-DMN cross-network FC and temporal dynamics within and between regions of the dynamic pain connectome reflect MS pain features.

The brain networks related to salience and pain was disrupted in patients with multiple sclerosis pain, and related to pain interference and neuropathic pain features.

aDivision of Brain, Imaging, and Behaviour—Systems Neuroscience, Krembil Brain Institute, Krembil Research Institute, Toronto Western Hospital, University Health Network, Toronto, ON, Canada

bInstitute of Medical Science, University of Toronto, Toronto, ON, Canada

cKeenan Research Centre for Biomedical Science, St. Michael's Hospital, Toronto, ON, Canada

dDepartment of Surgery, University of Toronto, Toronto, ON, Canada

Corresponding author. Address: Krembil Research Institute, Toronto Western Hospital, 399 Bathurst St, Room MP12-306, Toronto, ON M5T 2S8, Canada. Tel.: (416) 603-5662. E-mail address: karen.davis@uhnresearch.ca (K.D. Davis).

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.painjournalonline.com).

Received February 10, 2018

Received in revised form June 15, 2018

Accepted June 26, 2018

Back to Top | Article Outline

1. Introduction

Chronic pain is a prevalent and debilitating symptom of multiple sclerosis (MS); however, the mechanisms underlying MS-related pain are not well understood.18,22,23,31,47 Past investigations of MS pain have focused on neuropathic pain (NP) and have used structural magnetic resonance imaging (MRI) to link lesion characteristics to pain symptoms with inconclusive results.50 However, pain is not due to activity in one specific brain area, but arises from the coordination and dynamics across several brain networks that comprise the dynamic pain connectome (ascending nociceptive and descending modulation pathways, salience network [SN] and default mode network [DMN]).33,34 We posit that MS pain, both NP and non-neuropathic pain (non-NP), can arise from functional abnormalities of the dynamic pain connectome; however, we expect these abnormalities to be distinct for patients with NP vs non-NP because their pains have different features.

The right temporoparietal junction (rTPJ) is a key node of the salience network and is known to direct attention to changes in sensory input and coordinates brain activity to facilitate a behavioural response.15,35 In patients with chronic pain, the rTPJ is functionally disrupted7,24 and abnormal communication between the rTPJ and other regions of the salience network with other regions within the dynamic pain connectome has been described.3,24,26,27 Thus, functional deficits of the salience-pain neural systems may be a hallmark of chronic pain. Therefore, chronic pain in patients with MS is likely due to be related to aberrant cross-network functional connectivity between the rTPJ (a node of the SN) and other regions of the dynamic pain connectome. Furthermore, because pain demands attentional resources and causes disability, these brain abnormalities likely also contribute to the pain interference in daily function.

Previous examination of within network brain abnormalities in MS pain used static measures of functional connectivity (sFC) and found dysfunction in the DMN in patients with MS pain.51 However, emerging evidence suggest that brain communication across regions does not remain static across time, but rather is intrinsically dynamic between intermodular regions (ie, SN-ascending nociceptive or SN-descending modulation pathways) and reflects the ability to switch between states.11,28,29,33,34,57 For example, greater salience network–control network dynamics was also associated with the prioritization of a cognitive task over pain.12 Furthermore, we have recently demonstrated that greater regional brain flexibility (BOLD variability) was related to greater pain-coping and less pain sensitivity.49 However, whether dynamic communication between (dynamic functional connectivity [dFC]) or within (BOLD variability) regions subserves optimal function or is a sign of dysfunction is unknown and may depend on the particular context.

In the current study, resting-state functional magnetic resonance imaging (rsfMRI) and measures of sFC, dFC, and BOLD variability were used to test the hypothesis that patients with MS have abnormal: (1) DMN-SN cross-network sFC, (2) SN-ascending and SN-descending pathways' dFC, and (3) BOLD variability in the dynamic pain connectome that relates to pain inference and NP.

Back to Top | Article Outline

2. Materials and methods

2.1. Participants

Participants included 31 patients with MS (20 women; mean ± SD age = 37 ± 10 years, 11 men; mean age = 42 ± 8 years), and 31 age- and sex-matched healthy controls (HCs). Patients with MS were recruited from the St. Michael's Hospital Multiple Sclerosis clinic and HCs were recruited from the community. All participants provided informed written consent to procedures approved by the University Health Network and St. Michael's Hospital Research Ethics Boards in accordance with the Declaration of Helsinki. All patients had a diagnosis of MS and were excluded from the study if they had a painful disease unrelated/in addition to MS-related pain (eg, a diagnosis of arthritis), if they required a mobility aid, or if they had contraindications for MRI. The inclusion criteria for HCs were: (1) no previous history of chronic pain or current experience of pain on a regular basis, (2) no previous diagnosis of a psychiatric disorder, metabolic, or neurological condition, (3) no major surgery in the past 2 years, and (4) no standard contraindications for MRI.

Back to Top | Article Outline

2.2. Clinical measures: Brief Pain Inventory and painDETECT

All patients were assessed at St. Michael's Hospital in Toronto, ON, Canada, and clinical information including scores on the Expanded Disability Status Scale, a highly valid and reliable method of assessing disability, which ranges from 0 to 10 and is scored by a neurologist in 0.5 unit increments in which higher scores indicate higher levels of disability,39,45 disease symptom onset, date of disease diagnosis, and MS subtype were obtained. In one study session, patients underwent the neuroimaging protocol and completed a number of self-administered questionnaires including the Hospital Anxiety and Depression Scale.58 The Hospital Anxiety and Depression Scale assesses the nonphysical symptoms of anxiety and depression, and scores above 8 are considered clinically relevant.5 General pain and pain interference was assessed using the Brief Pain Inventory,48 and the painDETECT20 was used to assess NP. The Brief Pain Inventory measures 2 domains of pain: pain severity (worst pain, least pain, average pain, and pain now) and pain interference (how pain interferes with: general activity, walking, work, mood, enjoyment of life, relations with others, and sleep).16,55 Participants rate each item on a scale from 0 (no pain or does not interfere) to 10 (pain as bad as you can imagine, completely interferes). For each patient, a composite (mean) score was calculated for pain interference measures to reflect quality of life measures. The painDETECT scale has been shown to be a valid and reliable assessment for the presence of NP features.20,54 Scores on this scale range from −1 to 38. Scores below 13 indicate that a neuropathic component is unlikely, scores of 13 to 18 indicate that a neuropathic component may be present (and thus, these patients may have mixed neuropathic and nociceptive pain), and scores above 18 indicate that a neuropathic component is highly likely.20 Consistent with previous studies,53 we categorized patients with MS according to their painDETECT scores into mixed NP (scores above 12) and non-NP groups. In addition, patients with scores between 13 and 18 (mixed) also had sensory loss as determined by their mechanical detection thresholds, which lends support to classifying them as having NP features (Supplementary Fig. 1, available online as supplemental digital content at http://links.lww.com/PAIN/A617). Participants were also asked about the chronicity and location of their pain (Supplementary Table 2, available online as supplemental digital content at http://links.lww.com/PAIN/A617).

Back to Top | Article Outline

2.3. Neuroimaging acquisition

Both HCs and patients with MS underwent an MRI (3T GE) brain imaging session. Data acquisition included a T1-weighted anatomical scan (1 × 1 × 1 mm3 voxels, matrix = 256 × 256, 180 axial slices, repetition time = 7.8 seconds, echo time = 3 ms, inversion time = 450 ms) and a 9-minute, 14-second T2*-weighted rsfMRI scan (3.125 × 3.125 × 4 mm3 voxels, matrix = 64 × 64, 36 axial slices, repetition time = 2 seconds, echo time = 30 ms, flip angle = 85°, 277 volumes). For the resting-state scan, participants were instructed to “close your eyes, do not try to think about anything in particular; do not fall asleep.”

Back to Top | Article Outline

2.4. Preprocessing of resting-state functional magnetic resonance imaging data

The preprocessing of the rsfMRI data was conducted using the fMRI expert analysis tool (FEAT) toolbox in FMRIB Software Library (FSL).30 The following standard procedures were applied: (1) the first 4 volumes of the rsfMRI scan were removed, (2) nonbrain tissues were removed using the Brain Extract Tool (BET) within FEAT, and (3) motion correction was performed using MCFLIRT. The rsfMRI data from each participant were then linearly registered to their high-resolution anatomical image, which was skull-stripped using the opti-BET tool,42 followed by nonlinear registration to MNI152-2 mm space using FNIRT. Scanner-related noise and physiological noise were removed by means of applying aCompCor4,10 as described previously.38 The aCompCor method implements principal component analysis on the thresholded (to prevent the removal of gray matter) white matter and cerebrospinal fluid maps to generate components with the greatest variance. Five components containing the most variance were removed from the resting state data as well as the 6 motion parameters derived from MCFLIRT were regressed out from each participant's rsfMRI data. Finally, spatial smoothing was applied using a 4-mm full-width at half-maximum kernel, and temporal filtering was performed in FSL to retain the signal between 0.01 and 0.1 Hz.

Back to Top | Article Outline

2.5. Definition of seeds

To examine functional reorganization in regions of the dynamic pain connectome, we selected key regions of the salience (rTPJ) and DMN (posterior cingulate cortex [PCC] and medial prefrontal cortex [mPFC]), as well as the ascending nociceptive (postcentral gyrus; S1) and descending modulation (periaqueductal gray region [PAG]) nociceptive pathways. The location of the rTPJ seed (MNI x = 62, y = −36, x = 28) was based on previous studies that identified this region of the anterior rTPJ to highly coactivate with32 and be highly connected to other regions involved in attention and salience.35 The seeds in the PCC (MNI x = −8, y = −50, z = 28) and mPFC (MNI x = −2, y = 56, z = 16) were based on the coordinates from our previous study of the DMN,24 and the seed within the PAG region (MNI x = 0, y = −32, z = −10) was according to previous studies.17 For the connectivity analyses, we created an 8-mm spherical seed at the peak coordinates of the rTPJ, PCC, mPFC, and a 3-mm radius seed in the PAG. The S1 mask was constructed from the Harvard Oxford Cortical Structural Atlas to include all of the gray matter in the postcentral gyrus. A nonlinear transformation of each seed region from standard space to each subject's MRI space was performed and the mean time series from each was extracted.

Back to Top | Article Outline

2.6. Brain communication: static functional connectivity analysis

The sFC was measured between the seeds of salience and DMNs (average sFC between rTPJ-PCC and rTPJ-mPFC) and between the salience-ascending (rTPJ-SI) and salience-descending (rTPJ-PAG) by means of a Pearson correlation between the average time series of each seed pair. Correlation values were then Fisher r-to-z transformed. The sFC was compared between HCs and patients with MS, as well as between pain feature-based subgroups of patients with MS (neuropathic and non-neuropathic) using unpaired sample t tests. Statistical significance was determined at P < 0.05, with false discovery rate (FDR) controlling for multiple comparisons across all P values determined from group (MS vs HC) and subgroup (mixed NP vs Non-NP) comparisons across all pairwise sFC measures (SN-DMN, SN-Asc, and SN-Des).

Back to Top | Article Outline

2.7. Flexibility in communication: dynamic functional connectivity

Dynamic FC was also measured between the same seed pairs by calculating the dynamic conditional correlation method as described previously.12,41 Briefly, to calculate the dynamic conditional correlation, (1) each time course was prewhitened using an autoregressive and moving average (ARMA) (1,1) model, (2) a generalized autoregressive conditional heteroscedastic (GARCH) model was applied to estimate the conditional SD over time, (3) the residuals of the time series were standardized by the conditional SD, and (4) the dynamic conditional correlation was then used to calculate the time-varying correlation between the time series, using an exponentially weighted moving average that is derived from the data using maximum likelihood.41 The SD of each dynamic conditional correlation across the time series was computed and used as the summary metric of dFC.13 High dFC indicates greater fluctuations in the strength of connectivity between 2 regions over time. The dFC was again compared between controls and patients, and between patient subgroups using independent-samples t tests. Multiple comparisons were corrected using FDR across all P values determined from group (MS vs HC) and subgroup (mixed-NP vs Non-NP) comparisons across all pairwise dFC measures (SN-DMN, SN-Asc, and SN-Des).

Back to Top | Article Outline

2.8. BOLD signal variability

The standardized BOLD signal variability was calculated on a voxel-wise basis within a mask containing the key regions described above (DMN, SN, ascending nociceptive regions, and descending modulation regions) as previously reported.49 For each voxel within the mask, we calculated the SD of the BOLD signal, then subtracted the average SD from the gray matter across the whole brain, and then divided it by the SD of the BOLD signal variability across the gray matter.43 Previous studies of low-frequency oscillations related to pain, such as those present in regional BOLD variability time courses, have found distinct patterns of response in different frequency bands.1,25 Therefore, we also used AFNI's 3d Bandpass function to temporally filter the rsfMRI data such that we could examine BOLD variability in the different slow-wave frequency bands (slow-5: 0.01-0.027 Hz; slow-4: 0.027-0.073 Hz; and slow-3: 0.073-0.198 Hz).8,14,59 To compare the BOLD signal variability in each band between controls and patients with MS, we used FSL's Randomise tool for nonparametric permutation inferencing, using positive and negative contrasts with 5000 permutations and a threshold-free cluster enhancement. The group-level results are thresholded at P < 0.05 (family-wise error-corrected for multiple comparisons).

Back to Top | Article Outline

2.9. Relationship between brain functional reorganization and pain interference

The association between brain measures (sFC, dFC, and BOLD variability) within the dynamic pain connectome and pain interference (from the Brief Pain Inventory) was assessed by means of a Spearman correlation. Correlation analyses for both sFC (pain interference vs sFC values) and dFC (pain interference vs dFC values) were corrected for multiple comparisons using FDR (P < 0.05). Regional BOLD variability was calculated for each voxel within the dynamic pain connectome mask; therefore, in addition to assessing the univariate relationship between BOLD variability and pain interference, we also examined the multivariate association between brain patterns of BOLD variability and pain interference scores using the Pattern Recognition for Neuroimaging Toolbox (PRoNTo) (Shrouff et al., 2013) and kernel ridge regression (k-fold cross-validation, k = 5 folds). We first used permutation testing to determine which voxels demonstrated significant group BOLD variability differences between patients and controls, as described above (P < 0.05, familywise error rate corrected). Next, the voxels that demonstrated group differences were entered into the multivariate regression to determine whether these group differences in BOLD variability could be used to make generalizable inferences about pain interference. For the multivariate regression analyses, permutation testing was also implemented (5000 iterations) to determine the statistical significance.

Back to Top | Article Outline

3. Results

3.1. Pain and clinical characteristics of patients with multiple sclerosis

The demographic and clinical characteristics of the patients with MS and corresponding HC information are presented in Table 1. There was no significant differences in age (t(60) = 0.15, P = 0.88) or sex between controls and patients. However, patients with mixed NP features were significantly older than patients with non-NP (t(29) = 3.40, P = 0.002). Patients with mixed NP also had significantly higher Expanded Disability Status Scale scores (t(29) = 3.24, P = 0.003), although there was no difference in disease duration between the 2 subgroups (t(29) = 1.48, P = 0.15). Compared with controls, patients had significantly higher anxiety (t(60) = 3.79, P < 0.001) and depression scores (t(60) = 4.38, P < 0.001). Patients with mixed NP had significantly higher depression (t(29) = 2.39, P = 0.02), but not anxiety scores (t(29) = 1.76, P = 0.09), than patients with non-NP. Patients with mixed NP also had significantly higher pain severity (t(29) = 2.22, P = 0.03) and pain interference (t(29) = 2.24, P = 0.03) scores compared with patients with non-NP (Fig. 1). Common examples of nociceptive pain in our sample were musculoskeletal back pain and migraines, whereas a common example of NP included ongoing extremity pain. Neuropathic pain was assessed using the painDETECT questionnaire and psychophysical testing. Information regarding the chronicity as well as examples of pain in each subgroup is described in Supplementary Table 2 (available online as supplemental digital content at http://links.lww.com/PAIN/A617).

Table 1

Table 1

Figure 1

Figure 1

Back to Top | Article Outline

3.2. Abnormal SN-DMN cross-network static functional connectivity in multiple sclerosis

The time series between the core nodes of the salience network (rTPJ, anterior insula, and midcingulate cortex) in both HC and patients with MS had strong positive functional connectivity as expected between these regions (supplementary materials, available online as supplemental digital content at http://links.lww.com/PAIN/A617). Static FC analysis in controls confirmed the characteristic anticorrelations normally present between nodes of the SN and the DMN (Fig. 2). This cross-network anticorrelation was significantly diminished for the whole MS group (t(60) = 4.08, P < 0.001), but this finding was driven mostly by the patients with mixed NP who had a profound cross-network abnormality (t(30) = 4.43, P < 0.001). The cross-network abnormality in the patients with non-NP was only a trend (P = 0.09) (Fig. 2). However, mixed NP and non-NP groups did not have significantly different SN-DMN sFC (P = 0.50). At the group level, there was no significant association between SN-DMN sFC and painDETECT scores in patients with MS (P = 0.85).

Figure 2

Figure 2

The sFC between the SN and ascending nociceptive pathway did not differ between patients and controls (P = 0.29), nor did it differ between pain subgroups (mixed NP vs HC, P = 0.81; non-NP vs HC, P = 0.22; mixed NP vs non-NP, P = 0.56). Similarly, salience-descending sFC did not differ between patients and controls (P = 0.34), nor did it differ between pain subgroups (mixed-NP vs HC, P = 0.65; non-NP vs HC, P = 0.38; mixed-NP vs non-NP, P = 0.55). At the group level, there were no significant associations between SN-Asc and SN-Des sFC and painDETECT scores in patients with MS (P = 0.90 and P = 0.56, respectively).

Back to Top | Article Outline

3.3. Altered SN-ascending and SN-descending dynamic functional connectivity in multiple sclerosis

In both controls and patients, dFC values in the dynamic pain connectome ranged from approximately 0.2 to 0.65. There was no significant difference in SN-DMN dFC between controls and patients (P = 0.06) or between pain subgroups (mixed-NP vs HC, P = 0.17; non-NP vs HC, P = 0.12; mixed-NP vs non-NP, P = 0.89) (Fig. 3). There was greater SN-ascending dFC for the whole patient group compared with controls (t(60) = 3.10, P = 0.03). Both mixed NP and non-NP groups had higher dFC compared with their respective control groups, although only the non-NP was significantly different (mixed-NP; t(30) = 1.17, P = 0.06, non-NP; t(28) = 2.84, P = 0.03). The SN-ascending dFC was not statistically different for the mixed-NP vs non-NP subgroups (P = 0.72). Finally, the SN-descending dFC did not differ significantly between patients and control groups (P = 0.24); however, it was significantly attenuated in the mixed NP group compared with the non-NP group (t(29) = 2.39, P = 0.02). At the group level, there was no significant association between SN-DMN sFC, SN-Asc, or SN-Des sFC and painDETECT scores in patients with MS (P = 0.45, P = 0.81, and P = 0.60, respectively).

Figure 3

Figure 3

Back to Top | Article Outline

3.4. Increased BOLD variability in the default mode network

BOLD variability values were calculated for each voxel within the regions comprising the dynamic pain connectome described above (SN, DMN, ascending nociceptive pathway, and descending modulation pathway). In controls, the standardized BOLD variability values in the regions of the dynamic pain connectome ranged from −0.68 to 0.96, whereas in patients, these values ranged from −0.37 to 1.96. We next evaluated BOLD variability for specific frequency bands. In slow wave 3 (sw3), there was abnormally high BOLD variability for the whole group of patients in the DMN (t(60) = 2.46, P = 0.01) (Fig. 4). However, when divided into mixed NP and non-NP subgroups, these findings were only marginally significant (mixed-NP vs HC, P = 0.06; non-NP vs HC, P = 0.12) and there was not a significant difference between mixed NP and non-NP subgroups (P = 0.50). In slow wave 4 (sw4), there was greater BOLD variability in DMN in the whole patient group compared with controls (t(60) = 3.42, P = 0.001). Note that the group differences in sw3 and sw4 were in the same brain region (PCC of the DMN); however, the spatial extent of these findings was more restricted in sw4. Both mixed NP (P = 0.05) and non-NP (t(28) = 2.73, P = 0.01) had higher sw4 BOLD variability in the DMN. The sw4 DMN BOLD variability was not statistically different between mixed NP and non-NP subgroups (P = 0.87). There were no other significant BOLD variability differences between patients and controls in other voxels of the dynamic pain connectome in sw3 or sw4 (P > 0.05). In slow wave 5, there were no voxels that survived the threshold of significance when comparing controls and patients.

Figure 4

Figure 4

Back to Top | Article Outline

3.5. Relationship between brain functional reorganization and pain interference

Correlation analysis revealed a significant relationship between SN-DMN sFC and pain interference (Rho = 0.43, P = 0.03) (Fig. 5A). However, there was no significant relationship between pain interference and SN-ascending sFC (P = 0.30), or SN-descending sFC (P = 0.15). Furthermore, there were no significant relationships between pain interference and dFC in SN-DMN (P = 0.47), SN-ascending (P = 0.19), and SN-Des (P = 0.95). Finally, there was no univariate relationship between BOLD variability (in sw3 or sw4). However, results from a multivariate analysis revealed that the pattern of BOLD variability in the PCC (in sw3) was associated with pain interference (r = 0.44, P = 0.01) (Fig. 5B).

Figure 5

Figure 5

Back to Top | Article Outline

4. Discussion

This study has provided novel insight into the NP features in patients with MS through the lens of the relationship between MS pain and cross-network functional reorganization within the dynamic pain connectome. Our key findings were that: (1) approximately 50% of patients with MS experienced pain and sensory features of mixed NP, (2) patients with mixed NP had abnormal cross-network connectivity between the salience and DMNs, (3) all patients exhibited a disruption in brain signal variability between the SN and the ascending nociceptive pathway, but only patients with mixed NP have disrupted brain signal variability between the SN and descending pain-modulation pathway, (4) patients exhibited amplified regional neural variability in the DMN, and (5) abnormalities in sFC and brain signal variability were related to the degree of interference of pain on activities of daily living. Therefore, our study indicates that SN cross-network communication is dysfunctional in patients with MS pain, and the degree of abnormality reflects both features of NP and how much pain interferes in daily functioning.

We examined cross-network sFC to explore pain-related alterations in network organization and found that MS patients with mixed NP have abnormally high SN-DMN connectivity. The right temporoparietal junction, a key node of the SN, detects relevant stimuli from the environment and coordinates other brain networks, such as the DMN, to generate a behavioral response.15,44,56 By contrast, the DMN is active during internally directed, self-referential thought and suppressed when attention is externally engaged.19,38 In healthy individuals, the activity between the SN and the DMN is anticorrelated. Disruptions in this cross-network anticorrelation can been identified in several chronic pain conditions and the degree of sFC abnormality is related to pain symptoms.2,24,46 In the current study, we found that a disruption in SN-DMN anticorrection was present only in the mixed NP group, and across all patients with MS, the degree of abnormality was related to pain interference scores. In mixed NP pain, the relationship between the SN and DMN may be altered due to the experience of salient fluctuating pain and greater pain interference. Acute pain experienced by healthy individuals is salient and able to allocate attention away from internal thought, thereby suppressing the DMN. By contrast, in patients with chronic pain, ongoing mixed NP pain may actually cause thoughts to be more directed towards pain. This is consistent with evidence from patients with temporomandibular disorder in which sFC within the DMN corresponded to rumination about pain.37 Furthermore, salience network-DMN cross-network sFC abnormalities are also evident in patients with posttraumatic stress disorder52,53 and are related to inattention symptoms in attention deficit hyperactivity disorder (ADHD).36 This suggests that SN-DMN sFC abnormalities are not specific to pain, but rather that over time, the continuous experience of salient symptoms become ingrained in internal thought and alters attention. Thus, SN-DMN abnormalities in MS patients with mixed NP likely reflect an alteration in the neural salience–pain relationship.

The organization of brain networks and the relationship between networks have typically been described using static measures of FC, which assume that these measures remain stable in the resting-state brain over time.29,57 However, it is now known that the relationship between brain regions is dynamic and can fluctuate over time.29 Therefore, here we also examine the salience–pain relationship using measures of dFC to understand whether SN-ascending nociceptive or SN-descending modulation pathways are dynamically altered in MS pain. We found that although SN-ascending nociceptive pathway dFC was increased in patients, patients with mixed NP had a decreased SN-descending dFC. It is important to note that our sFC and dFC findings were distinct and that only dFC measures were sensitive to differences in patients vs controls or between mixed NP and non-NP subgroups in these brain regions. Findings of dFC but not sFC differences are consistent with previous studies that have demonstrated that the most dynamic connections are intermodular and those with low sFC.57 Therefore, using measures of dFC provides additional insight into MS pain-related functional reorganization of the salience–pain relationship. Abnormalities in the dynamic relationship between SN-ascending and SN-descending modulation pathway may reflect alterations in “the balance between efficient information processing and metabolic expenditure.”57 In HCs, higher dFC between the salience network and executive control networks has been shown to reflect a greater ability to prioritize a cognitive task over a pain stimulus.12 Thus, in patients with higher pain severity and interference, the SN may “overcoordinate” or “overengage” with the ascending nociceptive pathway as a result of or, alternatively, as a driver of the barrage of painful and salient input experienced by these patients. Conversely, MS patients with mixed NP had a decrease in dFC between the SN-descending modulation pathway. This may reflect an inability of the SN to dynamically engage with the descending modulation system to respond to ongoing pain. Therefore, the relationship between dFC and pain is likely network-specific and context-specific. Previous studies have demonstrated that dFC measures are sensitive to variability in behaviour.12,28 Because pain fluctuates over time and mixed NP is highly dynamic, measures of brain dynamics within the dynamic pain connectome may capture dynamics in pain perception. However, future studies that acquire continuous measures of ongoing pain or more thoroughly measure the temporal dimension of pain are required to resolve the precision with which measures of dFC track pain fluctuations.

Measures of sFC and dFC inform how regions and networks of the brain are communicating with one another; an additional consideration is the baseline neural variability at a regional level. Using measures of BOLD variability, we found that patients with MS had greater BOLD variability (sw3) in the DMN. This is inconsistent with our hypothesis and deviates from findings in HCs in whom greater BOLD variability was associated with lower pain sensitivity and better pain-coping.49 However, it is consistent with previous findings that describe increased power in regions of the DMN in patients with chronic lower back pain.1 BOLD signal variability could reflect regional cortical excitability.1,3,21 Enhanced variability in the DMN could therefore reflect increased engagement in internal thoughts and rumination about pain. In chronic pain, this increase in DMN baseline activity could in part account for the inability to suppress this network when engaging in an external task.2 Default mode network BOLD variability (sw3) was also predictive of pain interference scores; we suggest that the inability to disengage from thoughts on pain likely consumes attentional resources and disrupts the ability to fully engage in day-to-day activities.

There are several limitations to this study. We found that patients had higher depression scores than controls, and patients with mixed NP had slightly higher (although largely overlapping) depression scores compared with non-NP patients. Depression is highly comorbid in chronic pain conditions and the neurobiological underpinnings are overlapping.6 Therefore, as it is difficult to separate the effects of depression vs pain on the brain, the results presented here may also, in part, be influenced by the depression. Age and disease severity are also known to alter brain function. However, our supplementary analyses (available online as supplemental digital content at http://links.lww.com/PAIN/A617) suggested that our results did not change when controlling for age and that disease severity was not related to our findings (Supplementary Table 1, available online as supplemental digital content at http://links.lww.com/PAIN/A617). Another potential confound is the level of drowsiness or vigilance in our resting state scan because these factors are known to influence measures of dFC.40 We did not assess this and therefore are unable to control for or remove the effects of vigilance in our study. However, we also do not have any evidence that there would be systematic group difference in vigilance that could account for our results. Therefore, any effect of vigilance likely adds noise to our data that obscures our results, rather than inflate them. Of note, the variance of the dynamic conditional correlation, which we use as our measure of dFC, has been shown to be scan–rescan reliable,13 which alleviates some of the concerns regarding noise confounds on this measure. Finally, patients were not required to stop taking medications; the potential physiological effects of these medications on the BOLD fMRI signal are unknown in this study.

Altogether, our findings reveal that the relationship between salience and pain is disrupted in patients with MS pain. This is reflected in abnormalities in brain communication of the salience network—specifically SN-DMN sFC, SN-ascending, and SN-descending dFC, and also BOLD variability in the DMN that relate to pain interference and mixed NP pain features. Disruption in the salience–pain relationship is evident in a number of different chronic pain conditions including diabetic neuropathy9 and lower back pain.24 Therefore, these phenomena are a hallmark across chronic pain conditions, suggesting that the neural underpinnings of MS pain can also be described by a functional reorganization in key regions of the dynamic pain connectome.

Back to Top | Article Outline

Conflict of interest statement

The authors have no conflict of interest to declare.

This work was funded by the Multiple Sclerosis Society of Canada. R.L. Bosma was a recipient of a CIHR Postdoctoral Research Award and K.S. Hemington and J.C. Cheng were supported by CIHR Doctoral Research Awards.

Back to Top | Article Outline

Acknowledgements

The authors thank Dr Aaron Kucyi for his help in developing the analysis pipeline and Eugen Hlasny and Keith Ta for expert technical assistance in MRI acquisition.

Back to Top | Article Outline

Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/A617.

Back to Top | Article Outline

References

[1]. Baliki MN, Baria AT, Apkarian AV. The cortical rhythms of chronic back pain. J Neurosci 2011;31:13981–90.
[2]. Baliki MN, Geha PY, Apkarian AV, Chialvo DR. Beyond feeling: chronic pain hurts the brain, disrupting the default-mode network dynamics. J Neurosci 2008;28:1398–403.
[3]. Baliki MN, Mansour AR, Baria AT, Apkarian AV. Functional reorganization of the default mode network across chronic pain conditions. PLoS One 2014;9:e106133.
[4]. Behzadi Y, Restom K, Liau J, Liu TT. A component based noise correction method (CompCor) for BOLD and perfusion based fMRI. Neuroimage 2007;37:90–101.
[5]. Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the Hospital Anxiety and Depression Scale. An updated literature review. J Psychosom Res 2002;52:69–77.
[6]. Boakye PA, Olechowski C, Rashiq S, Verrier MJ, Kerr B, Witmans M, Baker G, Joyce A, Dick BD. A critical review of neurobiological factors involved in the interactions between chronic pain, depression, and sleep disruption. Clin J Pain 2016;32:327–36.
[7]. Borsook D, Edwards R, Elman I, Becerra L, Levine J. Pain and analgesia: the value of salience circuits. Prog Neurobiol 2013;104:93–105.
[8]. Buzsaki G, Draguhn A. Neuronal oscillations in cortical networks. Science 2004;304:1926–9.
[9]. Cauda F, D'Agata F, Sacco K, Duca S, Geminiani G, Vercelli A. Functional connectivity of the insula in the resting brain. Neuroimage 2011;55:8–23.
[10]. Chai XJ, Castanon AN, Ongur D, Whitfield-Gabrieli S. Anticorrelations in resting state networks without global signal regression. Neuroimage 2012;59:1420–8.
[11]. Chang C, Glover GH. Time-frequency dynamics of resting-state brain connectivity measured with fMRI. Neuroimage 2010;50:81–98.
[12]. Cheng JC, Bosma RL, Hemington KS, Kucyi A, Lindquist MA, Davis KD. Slow-5 dynamic functional connectivity reflects the capacity to sustain cognitive performance during pain. Neuroimage 2017;157:61–8.
[13]. Choe AS, Nebel MB, Barber AD, Cohen JR, Xu Y, Pekar JJ, Caffo B, Lindquist MA. Comparing test-retest reliability of dynamic functional connectivity methods. Neuroimage 2017;158:155–75.
[14]. Di Martino A, Ghaffari M, Curchack J, Reiss P, Hyde C, Vannucci M, Petkova E, Klein DF, Castellanos FX. Decomposing intra-subject variability in children with attention-deficit/hyperactivity disorder. Biol Psychiatry 2008;64:607–14.
[15]. Downar J, Mikulis DJ, Davis KD. Neural correlates of the prolonged salience of painful stimulation. Neuroimage 2003;20:1540–51.
[16]. Dworkin RH, Turk DC, Farrar JT, Haythornthwaite JA, Jensen MP, Katz NP, Kerns RD, Stucki G, Allen RR, Bellamy N, Carr DB, Chandler J, Cowan P, Dionne R, Galer BS, Hertz S, Jadad AR, Kramer LD, Manning DC, Martin S, McCormick CG, McDermott MP, McGrath P, Quessy S, Rappaport BA, Robbins W, Robinson JP, Rothman M, Royal MA, Simon L, Stauffer JW, Stein W, Tollett J, Wernicke J, Witter J. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. PAIN 2005;113:9–19.
[17]. Eippert F, Bingel U, Schoell ED, Yacubian J, Klinger R, Lorenz J, Buchel C. Activation of the opioidergic descending pain control system underlies placebo analgesia. Neuron 2009;63:533–43.
[18]. Foley PL, Vesterinen HM, Laird BJ, Sena ES, Colvin LA, Chandran S, MacLeod MR, Fallon MT. Prevalence and natural history of pain in adults with multiple sclerosis: systematic review and meta-analysis. PAIN 2013;154:632–42.
[19]. Fox MD, Raichle ME. Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat Rev Neurosci 2007;8:700–11.
[20]. Freynhagen R, Baron R, Gockel U, Tolle TR. painDETECT: a new screening questionnaire to identify neuropathic components in patients with back pain. Curr Med Res Opin 2006;22:1911–20.
[21]. Garrett DD, Kovacevic N, McIntosh AR, Grady CL. The importance of being variable. J Neurosci 2011;31:4496–503.
[22]. Green R, Cutter G, Friendly M, Kister I. Which symptoms contribute the most to patients' perception of health in multiple sclerosis? Mult Scler J Exp Transl Clin 2017;3:2055217317728301.
[23]. Heesen C, Bohm J, Reich C, Kasper J, Goebel M, Gold SM. Patient perception of bodily functions in multiple sclerosis: gait and visual function are the most valuable. Mult Scler 2008;14:988–91.
[24]. Hemington KS, Wu Q, Kucyi A, Inman RD, Davis KD. Abnormal cross-network functional connectivity in chronic pain and its association with clinical symptoms. Brain Struct Funct 2016;221:4203–19.
[25]. Hodkinson DJ, Wilcox SL, Veggeberg R, Noseda R, Burstein R, Borsook D, Becerra L. Increased amplitude of thalamocortical low-frequency oscillations in patients with migraine. J Neurosci 2016;36:8026–36.
[26]. Hong JY, Kilpatrick LA, Labus JS, Gupta A, Katibian D, Ashe-McNalley C, Stains J, Heendeniya N, Smith SR, Tillisch K, Naliboff B, Mayer EA. Sex and disease-related alterations of anterior insula functional connectivity in chronic abdominal pain. J Neurosci 2014;34:14252–9.
[27]. Hubbard CS, Khan SA, Keaser ML, Mathur VA, Goyal M, Seminowicz DA. Altered brain structure and function correlate with disease severity and pain catastrophizing in migraine patients. eNeuro 2014;1:e20. 14.
[28]. Hutchison RM, Morton JB. It's a matter of time: reframing the development of cognitive control as a modification of the brain's temporal dynamics. Dev Cogn Neurosci 2016;18:70–7.
[29]. Hutchison RM, Womelsdorf T, Allen EA, Bandettini PA, Calhoun VD, Corbetta M, Della Penna S, Duyn JH, Glover GH, Gonzalez-Castillo J, Handwerker DA, Keilholz S, Kiviniemi V, Leopold DA, de Pasquale F, Sporns O, Walter M, Chang C. Dynamic functional connectivity: promise, issues, and interpretations. Neuroimage 2013;80:360–78.
[30]. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM. FSL. Neuroimage 2012;62:782–90.
[31]. Kalia LV, O'Connor PW. Severity of chronic pain and its relationship to quality of life in multiple sclerosis. Mult Scler 2005;11:322–7.
[32]. Krall SC, Rottschy C, Oberwelland E, Bzdok D, Fox PT, Eickhoff SB, Fink GR, Konrad K. The role of the right temporoparietal junction in attention and social interaction as revealed by ALE meta-analysis. Brain Struct Funct 2015;220:587–604.
[33]. Kucyi A, Davis KD. The dynamic pain connectome. Trends Neurosci 2015;38:86–95.
[34]. Kucyi A, Davis KD. The neural code for pain: from single-cell electrophysiology to the dynamic pain connectome. Neuroscientist 2017;23:397–414.
[35]. Kucyi A, Hodaie M, Davis KD. Lateralization in intrinsic functional connectivity of the temporoparietal junction with salience- and attention-related brain networks. J Neurophysiol 2012;108:3382–92.
[36]. Kucyi A, Hove MJ, Biederman J, Van Dijk KR, Valera EM. Disrupted functional connectivity of cerebellar default network areas in attention-deficit/hyperactivity disorder. Hum Brain Mapp 2015;36:3373–86.
[37]. Kucyi A, Moayedi M, Weissman-Fogel I, Goldberg MB, Freeman BV, Tenenbaum HC, Davis KD. Enhanced medial prefrontal-default mode network functional connectivity in chronic pain and its association with pain rumination. J Neurosci 2014;34:3969–75.
[38]. Kucyi A, Salomons TV, Davis KD. Mind wandering away from pain dynamically engages antinociceptive and default mode brain networks. Proc Natl Acad Sci U S A 2013;110:18692–7.
[39]. Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology 1983;33:1444–52.
[40]. Laumann TO, Snyder AZ, Mitra A, Gordon EM, Gratton C, Adeyemo B, Gilmore AW, Nelson SM, Berg JJ, Greene DJ, McCarthy JE, Tagliazucchi E, Laufs H, Schlaggar BL, Dosenbach NUF, Petersen SE. On the stability of BOLD fMRI correlations. Cereb Cortex 2017;27:4719–32.
[41]. Lindquist MA, Xu Y, Nebel MB, Caffo BS. Evaluating dynamic bivariate correlations in resting-state fMRI: a comparison study and a new approach. Neuroimage 2014;101:531–46.
[42]. Lutkenhoff ES, Rosenberg M, Chiang J, Zhang K, Pickard JD, Owen AM, Monti MM. Optimized brain extraction for pathological brains (optiBET). PLoS One 2014;9:e115551.
[43]. Martino M, Magioncalda P, Huang Z, Conio B, Piaggio N, Duncan NW, Rocchi G, Escelsior A, Marozzi V, Wolff A, Inglese M, Amore M, Northoff G. Contrasting variability patterns in the default mode and sensorimotor networks balance in bipolar depression and mania. Proc Natl Acad Sci U S A 2016;113:4824–9.
[44]. Menon V, Uddin LQ. Saliency, switching, attention and control: a network model of insula function. Brain Struct Funct 2010;214:655–67.
[45]. Meyer-Moock S, Feng YS, Maeurer M, Dippel FW, Kohlmann T. Systematic literature review and validity evaluation of the Expanded Disability Status Scale (EDSS) and the Multiple Sclerosis Functional Composite (MSFC) in patients with multiple sclerosis. BMC Neurol 2014;14:58.
[46]. Napadow V, LaCount L, Park K, As-Sanie S, Clauw DJ, Harris RE. Intrinsic brain connectivity in fibromyalgia is associated with chronic pain intensity. Arthritis Rheum 2010;62:2545–55.
[47]. O'Connor AB, Schwid SR, Herrmann DN, Markman JD, Dworkin RH. Pain associated with multiple sclerosis: systematic review and proposed classification. PAIN 2008;137:96–111.
[48]. Osborne TL, Raichle KA, Jensen MP, Ehde DM, Kraft G. The reliability and validity of pain interference measures in persons with multiple sclerosis. J Pain Symptom Manage 2006;32:217–29.
[49]. Rogachov A, Cheng JC, Erpelding N, Hemington KS, Crawley AP, Davis KD. Regional brain signal variability: a novel indicator of pain sensitivity and coping. PAIN 2016;157:2483–92.
[50]. Seixas D, Foley P, Palace J, Lima D, Ramos I, Tracey I. Pain in multiple sclerosis: a systematic review of neuroimaging studies. Neuroimage Clin 2014;5:322–31.
[51]. Seixas D, Palace J, Tracey I. Chronic pain disrupts the reward circuitry in multiple sclerosis. Eur J Neurosci 2016;44:1928–34.
[52]. Sharp TJ, Harvey AG. Chronic pain and posttraumatic stress disorder: mutual maintenance? Clin Psychol Rev 2001;21:857–77.
[53]. Sripada RK, King AP, Welsh RC, Garfinkel SN, Wang X, Sripada CS, Liberzon I. Neural dysregulation in posttraumatic stress disorder: evidence for disrupted equilibrium between salience and default mode brain networks. Psychosom Med 2012;74:904–11.
[54]. Tampin B, Bohne T, Callan M, Kvia M, Melsom Myhre A, Neoh EC, Bharat C, Slater H. Reliability of the English version of the painDETECT questionnaire. Curr Med Res Opin 2017;33:741–8.
[55]. Turk DC, Dworkin RH, Allen RR, Bellamy N, Brandenburg N, Carr DB, Cleeland C, Dionne R, Farrar JT, Galer BS, Hewitt DJ, Jadad AR, Katz NP, Kramer LD, Manning DC, McCormick CG, McDermott MP, McGrath P, Quessy S, Rappaport BA, Robinson JP, Royal MA, Simon L, Stauffer JW, Stein W, Tollett J, Witter J. Core outcome domains for chronic pain clinical trials: IMMPACT recommendations. PAIN 2003;106:337–45.
[56]. Uddin LQ. Salience processing and insular cortical function and dysfunction. Nat Rev Neurosci 2015;16:55–61.
[57]. Zalesky A, Fornito A, Cocchi L, Gollo LL, Breakspear M. Time-resolved resting-state brain networks. Proc Natl Acad Sci U S A 2014;111:10341–6.
[58]. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361–70.
[59]. Zuo XN, Di Martino A, Kelly C, Shehzad ZE, Gee DG, Klein DF, Castellanos FX, Biswal BB, Milham MP. The oscillating brain: complex and reliable. Neuroimage 2010;49:1432–45.
Keywords:

Multiple sclerosis; Chronic pain; fMRI; Dynamic functional connectivity; BOLD variability; Salience network; Default mode network

Supplemental Digital Content

Back to Top | Article Outline
© 2018 International Association for the Study of Pain