Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jul 1.
Published in final edited form as: Arthritis Rheumatol. 2021 May 31;73(7):1318–1328. doi: 10.1002/art.41620

Greater Somatosensory Afference with Acupuncture Increases Primary Somatosensory Connectivity and Alleviates Fibromyalgia Pain via Insular GABA: A Randomized Neuroimaging Trial

Ishtiaq Mawla 1,2, Eric Ichesco 2, Helge J Zöllner 3,4, Richard A E Edden 3,4, Thomas Chenevert 5, Henry Buchtel 2, Meagan D Bretz 2, Heather Sloan 2, Chelsea M Kaplan 2, Steven E Harte 1,2, George A Mashour 1,6, Daniel J Clauw 2, Vitaly Napadow 7,8,*, Richard E Harris 1,2,*
PMCID: PMC8197768  NIHMSID: NIHMS1654853  PMID: 33314799

Abstract

Objective:

Acupuncture is a complex multi-component treatment that has shown promise for the treatment of Fibromyalgia (FM), however, clinical trials have shown mixed results, possibly due to heterogeneous methodology and lack of understanding of the underlying mechanism of action. We sought to understand the specific contribution of somatosensory afference to improvements in clinical pain, and the specific brain circuits involved.

Methods:

76 FM patients were randomized to receive 8 weeks (2 treatments/week) of electroacupuncture (EA, with somatosensory afference) or mock laser acupuncture (ML, with no somatosensory afference). Brief Pain Inventory (BPI) Severity, resting state functional MRI (rs-fMRI), and proton magnetic resonance spectroscopy (1H-MRS) in the right anterior insula (aINS) were collected at pre- and post-treatment.

Results:

FM patients receiving EA experienced a greater reduction in pain severity compared to ML (mean difference, EA=−1.14, ML=−0.46, Group × Time interaction, p=0.036). Participants receiving EA, as compared to ML, also displayed increased resting functional connectivity between the primary somatosensory cortical representation of the leg (S1leg; i.e. S1 subregion activated by EA) and aINS. Increase in S1leg-aINS connectivity was associated with reductions in BPI severity (r=−0.44, p=0.01) and increases in aINS gamma-aminobutyric acid (GABA+) (r=−0.48, p=0.046) following EA. Moreover, increases in aINS GABA+ was associated with reductions in BPI severity (r=−0.59, p=0.01). Finally, post-EA changes in aINS GABA+ mediated the relationship between changes in S1leg-aINS and BPI severity, bootstrapped CI=[−0.533, −0.037].

Conclusion:

The somatosensory component of acupuncture modulates primary somatosensory functional connectivity in association with insular neurochemistry to reduce pain severity in FM.

Introduction

Fibromyalgia (FM) is a common chronic pain condition afflicting 2–8% of the population, and is characterized by widespread somatic pain, fatigue, poor sleep, negative mood, and cognitive disturbances1. While peripheral factors (e.g., small fiber neuropathy2, immune system3) may play some role in FM, the disorder is thought to be caused primarily by aberrant central nervous system (CNS) physiology which amplifies the perception of pain (also known as centralized or nociplastic pain4). Notably, neuroimaging research has shown that FM patients demonstrate increased levels of the excitatory neurotransmitter glutamate5, decreased levels of the inhibitory neurotransmitter gamma aminobutyric acid (GABA)6, and upregulated GABAA receptor concentration7, within the insula. Moreover, increased functional brain network connectivity to pro-nociceptive brain areas and decreased connectivity to anti-nociceptive brain areas have been found in FM8–10. These results suggest that the CNS is a prime target for therapeutic interventions for FM.

Due to the ongoing opioid public health crisis11, non-pharmacologic interventions for FM such as acupuncture have been gaining attention. However, meta-analyses of acupuncture trials have shown mixed results, with some showing that verum (active) acupuncture is no more effective than sham controls12,13, while others have shown that acupuncture is superior to both sham and no-acupuncture controls in reducing pain14. One reason for the mixed meta-analytic results may be the inclusion of heterogenous treatment paradigms and sham controls across different trials. Acupuncture is a complex procedure that consists of multiple methodological (e.g., needling sensation, location, depth, etc), and contextual components (e.g., expectancy, patient-practitioner rapport, treatment ritual, etc.)15. Importantly, sham controls used in previous acupuncture trials may not have properly accounted for all of these different components of acupuncture.

In this study, we specifically evaluated CNS mechanism(s) of action underlying the somatosensory afferent component of acupuncture, and how such mechanisms may engender an analgesic response in FM. Since verum acupuncture produces somatosensory sensation through needling and palpation, we designed a comparator sham control procedure to lack all aspects of tactile sensation. Many previous trials of acupuncture used sham controls with acupoint palpation and tactile stimulation, mimicking real needle insertion and manipulation, thus confounding verum and sham acupuncture in terms of somatosensory afference12–14. We randomized FM patients into two separate acupuncture therapy groups: electroacupuncture (EA, i.e. with somatosensation) and mock laser acupuncture (ML, without somatosensation). EA has been demonstrated to be clinically effective at reducing pain for FM13. We hypothesized that EA would specifically recruit somatosensory pathways in the CNS in order to produce greater analgesia compared to ML.

Methods

Overall Protocol

This study was a single-center, blinded, sham controlled, randomized non-crossover longitudinal neuroimaging study, pre-registered with ClinicalTrials.gov (NCT02064296). The study took place at the University of Michigan, Ann Arbor, MI from December 2014 to November 2019. All study protocols were approved by the University of Michigan Institutional Review board, and all study participants provided written informed consent in accordance with the Declaration of Helsinki.

Participants and Study Timeline

Participants suffering from FM were recruited for the study. Full details of inclusion and exclusion criteria are provided in Supplementary Methods A. Following screening, participants were invited to complete a baseline behavioral (day 0) and baseline MRI assessment (occurred anywhere between day 1 and 3), and eligible subjects were randomly assigned to one of two parallel study arms (Figure 1A). We used computer generated permuted block randomization (blocks of 4, 6, or 8). An acupuncturist was informed of group allocation of each participant through a sealed envelope, which was not accessible by the principal investigators, study staff, or data analysts. The two intervention arms were (i) Electroacupuncture (EA), with somatosensory afference, and (ii) Mock Laser acupuncture (ML), without somatosensory afference. After treatment, a second behavioral (day 33–40) and MRI assessment (day 34–43) were collected. Patient-reported outcomes were collected before and after therapy during the behavioral session. Whole brain resting state functional MRI (rs-fMRI) and right anterior insula (aINS) proton magnetic resonance spectroscopy (1H-MRS) were collected during the MRI sessions before and after therapy.

Figure 1. Study Overview.

Figure 1.

Open in a new tab

(A) Non-crossover randomized controlled neuroimaging trial in FM with acupuncture intervention. Behavioral session, resting state functional MRI (rs-fMRI) and proton magnetic resonance spectroscopy (1H-MRS) were collected at baseline (Pre-tx) and after therapy (Post-tx). (B) Acupuncture locations for EA and ML. All subjects were blindfolded and in supine position. In EA, LI-4 was administered to the dorsal surface of the right hand, and LI-11 was administered to the crease of the right elbow. Bolt symbols indicate where needles received current through the EA device. For ML, a deactivated laser was hovered over the same acupuncture points as in EA for the same duration of time. Abbreviations: Du=Governor meridian, LI=Large Intestine, ST=Stomach, SP=Spleen, GB=Gall Bladder, LV=Liver.

Acupuncture Treatment

FM participants received 8 EA or ML treatments over four weeks (twice a week). During all treatment sessions, participants were positioned supine on an exam table and blindfolded. Blindfolding ensured masking of the treatments in order to avoid any visual afference, as visual afference can also influence acupuncture-induced analgesia16. All treatments were performed by three (H.B., M.D.B., and H.S.) trained acupuncturists with board certification from the National Certification Commission for Acupuncture and Oriental Medicine.

The EA group received low frequency EA at 3 pairs of acupoints: right side LI-11 to LI-4, left side GB-34 to SP-6, and bilateral ST-36. Needles were also inserted in Du-20, right ear shenmen, and left LV-3 (Figure 1B) but with no electrical current delivered. EA needles were stimulated with low intensity and frequency using a constant-current electro-acupuncture device (AS SUPER 4 Digital Needle Stimulator) which allowed for flexible setting of pulse width (1 ms), frequency (2 Hz), and shape (biphasic rectangular) parameters. The current intensity was set at each session for each patient individually at the midpoint between sensory and pain thresholds, based on common clinical practice and our previous EA study with chronic pain patients17, and stimulation lasted 25 minutes per session. The duration and frequency of treatment is based on common clinical practice and is within the bounds of previous acupuncture trials18. The selection of acupuncture sites was based on predominant FM symptoms including multisite pain; headache, gastrointestinal pain and dysfunction; disrupted sleep; and chronic fatigue.

For the ML acupuncture therapy group, a laser acupuncture device (VitaLaser 650, Lhasa OMS) was manually positioned approximately 1–2 cm over all of the same acupoints used in EA. There was no palpation prior to positioning the device, and there was no physical contact between the device and skin. The laser light was demonstrated to the participants at the first visit to enhance credibility of the intervention; however, the laser was turned off during the actual treatment, thus removing any potential optically induced or thermal sensation, whilst maintaining all treatment rituals, as previously described19,20 (Figure 1B). ML treatments also lasted 25 minutes.

Participants were not informed about a sham or placebo at consent, so all participants were led to believe that both EA and ML are viable treatments for FM. These procedures constituted as IRB authorized deception, and all participants were fully debriefed after the final MRI visit.

The verbal instructions used by acupuncturists were standardized across all treatments (Supplementary Methods B and C). After each treatment, the MGH Acupuncture Sensation Scale (MASS)21 was administered to evaluate de qi and perceived somatosensory afference. The 13-item questionnaire included sensations such as Soreness, Aching, Deep Pressure, Tingling, etc. (0=none; 10=unbearable scale) and weighted summation of these sensations constituted the MASS Index. This measure served as a fidelity check to assess whether FM patients consistently reported increased levels of sensation in response to EA compared to ML. In addition, after the first treatment and after the last treatment, a Credibility Questionnaire (Supplementary Methods D) was administered which assessed the perception of validity and credibility of the treatments. This ensured that any differences in clinical or neuroimaging outcomes were not due to differences in perception of credibility.

Clinical outcomes

The Severity subscale of the Short Form Brief Pain Inventory (BPI) was the primary clinical outcome. BPI-Severity measures worst pain in 24 hours, least pain in 24 hours, pain on average, and pain right now. BPI-Severity was measured at pre- and post-therapy. PROMIS (https://www.healthmeasures.net/explore-measurement-systems/promis) Anxiety and Depression scales were used as secondary clinical outcomes and to assess whether neuroimaging outcomes were influenced by these factors. Furthermore, we collected a series of exploratory outcome measures: BPI Pain Interference, American College of Rheumatology 2011 FM Survey Criteria22, Pain Catastrophizing, and PROMIS (Physical Function, Fatigue, Sleep). A detailed analysis of these exploratory outcomes is outside the scope of the present manuscript; however, descriptive statistics for each are reported in Supplementary Results B.

Mechanistic outcome: rs-fMRI of the primary somatosensory cortex

Resting state functional MRI (rs-fMRI) in an awake, eyes-open state, and anatomical T1-weighted (T1w) MRI were acquired with a 15-channel head coil in a 3.0T MRI system (Philips Ingenia, Best, Netherlands). Minimal preprocessing of rs-fMRI and T1w images were performed using fMRIprep 1.1.823. Full details of MRI acquisition parameters and preprocessing steps are provided in Supplementary Methods E.

Since somatosensory afferent input is encoded in the primary somatosensory cortex (S1), we chose the S1 cortical representation of the legs as the seed region to examine somatosensory circuits (i.e. communication between S1leg and other brain regions). S1leg was the chosen seed as most EA needles were placed on the leg (Figure 1B), and our group has previously localized this S1leg region in FM patients (centroid MNI coordinates x=± 8, y=−38, z=68)24. Bilateral spherical seeds (4 mm radius) were used to extract fMRI timeseries and seed-to-voxel correlation analysis was used to evaluate whole-brain connectivity maps for S1leg. Timeseries from the S1leg seed (fslmeants) were used as a GLM regressor (fsl_glm) to obtain whole-brain parameter estimates and associated variances, for each participant. These parameter estimates and variances were then passed on to group level analysis, conducted on FMRIB’s Local Analysis of Mixed Effects (FLAME 1+2)25 to improve mixed-effects variance estimation. S1leg connectivity was then contrasted between pre- and post-treatment periods using paired sample t-tests for EA and ML separately. Interaction between EA and ML was conducted using an independent samples t-test of the paired post-pre difference images. As age influences neuroimaging outcomes, it was included as a regressor of no-interest in all analyses. Multiple comparisons family-wise error correction was conducted using gaussian random field (GRF) cluster threshold (Z > 2.3) and significance at corrected p < 0.05.

Mechanistic outcome: 1H-MRS measurement of Glx and GABA+ in the right anterior insula

1H-MRS spectra were acquired from automated voxel placement covering the right aINS, as our previous study showed differences between FM and pain-free controls in this region6. The 1H-MRS voxel dimensions were based on our previous study6. Single-voxel point resolved spectroscopy (PRESS) was used to measure Glx. A separate GABA+ edited Mescher-Garwood-PRESS (MEGA-PRESS), which co-edits signal from macromolecules and homocarnosine, was conducted to estimate GABA+ levels26. Conventional PRESS spectroscopy data was analyzed with LCModel27. MEGA-PRESS spectra were processed in Gannet 3.1.528, a MATLAB-based toolbox specifically developed for edited MRS. Full details of PRESS and MEGA-PRESS acquisition parameters, preprocessing and analysis details are provided in Supplementary Methods F. The final GABA+ estimates were reported in institutional units (GABA+ (i.u.), approximating millimolar concentrations) and also as an integral ratio with respect to the creatine signal (GABA+/Cr). Treatment-related change in Glx and GABA+ was computed as the difference between pre- and post-therapy values.

Statistical analyses

Besides the aforementioned image-based statistics, statistical analyses were performed in IBM SPSS Statistics 26 (IBM, Armonk, NY). For the assessment of changes in the primary clinical outcome (BPI-Severity) and secondary outcomes (Supplementary Results B), a 2 (Group: EA, ML) by 2 (Time: Pre, Post) mixed-design ANOVA was conducted. For the assessment of MASS Index, a 2 (Group: EA, ML) by 8 (Time: Pre, Post) mixed-design ANOVA was conducted. Greenhouse-Geisser correction was used to adjust for sphericity assumptions in the repeated measures ANOVA. Mean credibility scores were assessed for group differences using an independent samples t-test. Associations between changes in extracted values from S1leg connectivity, GABA, and BPI-severity were conducted using Pearson’s correlation adjusted for age. To determine whether relationships assessed with Pearson’s r were directionally different for EA compared to ML, the single-tailed Fisher’s z cocor algorithm29 was used. For mediation analyses, bias-corrected bootstrapped (10,000x) mediation was conducted using the Process Macro on SPSS30, and estimates of indirect effects were computed at the 95% confidence level (adjusting for age).

Charts and Figures

All charts were created on GraphPad PRISM Version 8.2.1 (GraphPad Software, San Diego, California USA, www.graphpad.com). Figure 1 and 5B were created with BioRender.com

Results

Clinical characteristics and demographics

Flow of participants in the protocol is described in Supplementary Results A. Full demographic and clinical characteristics are listed in Supplementary Results B and medication usage for each participant is listed in Supplementary Results C.

Post-therapy reduction in BPI Severity is greater in EA versus ML

For BPI Severity, the two-way Group × Time mixed-design ANOVA demonstrated a significant main effect of Time (F(1, 70)=25.09, p<0.001) and no main effect of Group (F(1, 70)=0.03, p=0.861). However, there was a significant Group × Time interaction (F(1, 70)=4.56, p=0.036), such that EA reduced BPI Severity to a greater extent compared to ML (Figure 2A). There was no baseline difference in BPI Severity between EA and ML (t(70)=0.85, p=0.396). Changes in BPI Severity were not related to changes in Depression (EA: r(33)=0.24, p=0.165; ML: r(35)=−0.08, p=0.65) or Anxiety (EA: r(33)=0.07, p=0.71; ML: r(35)=0.17, p=0.31).

Figure 2. BPI Severity and MASS Index response to therapy.

Figure 2.

Open in a new tab

(A) EA demonstrated significantly greater post-therapy reduction in BPI Severity compared to ML (Group × Time interaction, p=0.036). (B) Patients receiving EA reported significantly higher somatosensory afference (MASS Index) compared to those receiving ML (Main effect of Group (p<0.001)). Error Bars in both plots indicate Standard Error of the Mean (SEM).

EA elicits greater somatosensory afference compared to ML

For MASS Index scores, the 2 (Group) × 8 (Time) mixed-design ANOVA demonstrated a significant main effect of Time (F(4.0, 224.9)=2.85, p=0.025), a significant main effect of Group (F(1, 56)=31.01, p<0.001), but no Group × Time interaction effect (F(4.0, 224.9)=0.35, p=0.84) (Figure 2B). Treatment credibility was equal across both groups (Supplementary Results D).

S1leg connectivity increases post-therapy in EA versus ML

A whole-brain seed connectivity analysis of S1leg showed significant post-therapy increases in connectivity for the EA group, notably to the bilateral anterior insula (aINS), posterior insular (pINS), and right non-leg S1 subregions. Conversely, the ML group showed reductions in S1leg connectivity to the left anterior/mid insula (a/mINS). The whole-brain Group × Time interaction effect showed that the magnitude of increase in S1leg connectivity for EA was greater than that of ML, notably showing regions such as the bilateral aINS, pINS, and right non-leg S1. Figure 3A shows relevant contrasts, and full detail of clusters are in Supplementary Results E. We also confirmed that our rs-fMRI results were not confounded by head motion (Supplementary Results F).

Figure 3. S1leg connectivity response to therapy.

Figure 3.

Open in a new tab

(A) In EA, S1leg connectivity to aINS, pINS, and non-leg S1 increased with treatment. In ML, S1leg connectivity to a/mINS decreased with treatment. The EA>ML contrast showed that magnitude of S1leg connectivity increase was higher in EA compared to ML. (B) Within EA, as S1leg-aINS and S1leg-pINS connectivity increased, BPI severity decreased post-therapy. Values have been adjusted for age.

Increases in S1leg connectivity were related to improvements in BPI Severity in EA

For EA, there was a significant relationship between change in S1leg-aINS connectivity and change in BPI severity (r(30)=−0.44, p=0.01), such that the greater the increase in S1leg-aINS connectivity, the greater the reduction in BPI-Severity post-therapy (Figure 3B). Change in S1leg-aINS connectivity was not related to change in BPI severity for ML (r(35)=−0.02, p=0.91). The correlation for EA was significantly stronger than that of ML (Fisher’s z=−1.78, p=0.04). Changes in S1leg-aINS connectivity were not related to post-therapy changes in depression (EA: r(30)=0.02, p=0.93; ML: r(35)=−0.14, p=0.41) or anxiety (EA: r(30)=−0.12, p=0.51; ML: r(35)=0.11, p=0.50).

Similarly, we found that for EA, there was a significant relationship between change in S1leg-pINS connectivity and change in BPI severity (r(30)=−0.43, p=0.01), such that the greater the increase in S1leg-pINS connectivity, the greater the reduction in BPI severity post-therapy (Figure 3B). Change in S1leg-pINS connectivity was not related to change in BPI severity for ML (r(30)=−0.04, p=0.84). The correlation for EA was significantly stronger than that of ML (Fisher’s z=−1.70, p=0.04). Changes in S1leg-pINS connectivity were not related to post-therapy changes in depression (EA: r(30)=−0.19, p=0.29; ML: r(35)=0.18, p=0.29) or anxiety (EA: r(30)=−0.24, p=0.18; ML: r(35)=0.13, p=0.45).

Changes in aINS GABA+ is linked with changes in S1leg-aINS connectivity in EA

Figure 4A shows the average MEGA-PRESS spectrum across all subjects, respectively. We found that the right aINS cluster from the post-pre S1leg connectivity group map in EA overlapped with the MNI-transformed aINS 1H-MRS voxel placement (Figure 4B). There was no main effect of treatment on GABA+ (Supplementary Results G). However, we found that greater increase in S1leg-aINS connectivity was associated with greater increase in aINS GABA+ post-therapy (GABA+(i.u.): r(16)=0.48, p=0.046 (Figure 4C); GABA+/Cr: r(16)=0.46, trending p=0.052). This S1leg-aINS connectivity and aINS GABA+ relationship was not present for ML (GABA+(i.u.): r(23)=−0.17, p=0.43; GABA+/Cr: r(23)=−0.15, p=0.47), and the correlation for EA was significantly stronger than that of ML (GABA+(i.u.): Fisher’s z=2.08, p=0.02; GABA+/Cr: Fisher’s z=1.94, p=0.03). Furthermore, we confirmed that this relationship was specific to inhibitory and not excitatory neurotransmitter changes (Supplementary Results H).

Figure 4. aINS GABA response to EA therapy.

Figure 4.

Open in a new tab

(A) Average of 1H-MRS voxel in the right aINS across all subjects transformed to MNI space, and corresponding mean and SD of the MEGA-PRESS spectra. (B) Intersection of voxels encompassing both the aINS GABA voxel and the aINS cluster from S1leg connectivity map. (C) Greater increase in S1leg-aINS connectivity was associated with greater increase in aINS GABA+(i.u.) concentration post-therapy. (D) Greater increase in aINS GABA+(i.u.) was associated with greater reduction in post-therapy clinical pain in FM. Values have been adjusted for age.

Changes in aINS GABA+ is linked with improvements in BPI Severity in EA

We found that greater increases in aINS GABA were associated with greater reductions in BPI severity (GABA+(i.u.): r(16)=−0.59, p=0.01 (Figure 4D); GABA+/Cr: r(16)=−0.65, p=0.004). This relationship was not found for ML (GABA+(i.u.): r(16)=−0.16, p=0.44; GABA+/Cr: r(23)=−0.13, p=0.53), and the correlation for EA was stronger than that of ML (GABA+(i.u.): Fisher’s z=−1.54, trending p=0.06; GABA+/Cr: Fisher’s z=−1.92, p=0.03). Changes in aINS GABA+ in EA and ML were not related to post-therapy changes in depression (GABA+(i.u.): EA: r(16)=0.12, p=0.63 and ML: r(23)=0.07, p=0.74; GABA+/Cr: EA: r(16)=0.23, p=0.36 and ML: r(23)=0.03, p=0.89) or anxiety (GABA+(i.u.): EA: r(16)=−0.21, p=0.40 and ML: r(23)=0.10, p=0.65; GABA+/Cr: EA: r(16)=−0.06, p=0.82 and ML: r(23)=0.08, p=0.72). Furthermore, we confirmed that this relationship was specific to inhibitory and not excitatory neurotransmitter changes (Supplementary Results H).

aINS GABA+ mediated the effect of S1leg-aINS connectivity on BPI severity in EA

Finally, we conducted a mediation analysis to link S1leg-aINS connectivity (X), BPI severity (Y), and aINS GABA+(i.u.) (Mediator) in one statistical model. Results showed that greater increase in S1leg-aINS connectivity was associated with greater reduction in BPI severity post-therapy indirectly through greater increase in aINS GABA+(i.u.) (β=−0.187, BootSE=0.130, BootLLCI=−0.533, BootULCI=−0.037, Figure 5A). The direct effect of increase in S1leg-aINS connectivity on reduction in BPI severity post-therapy was not significant (Effect=−0.184, SE=0.186, LLCI=−0.581, ULCI=0.212), suggesting that the effect of S1leg-aINS connectivity on BPI severity is transmitted through aINS GABA+(i.u.). The R2 value for BPI Severity in this model was 0.39. This effect was also present when GABA+/Cr estimates were used as the mediator (Supplementary Results I).

Figure 5. Mediation analysis and proposed mechanistic model.

Figure 5.

Open in a new tab

(A) Increase in aINS GABA+(i.u.) mediates the relationship between increased S1leg-aINS connectivity and decreased BPI severity post-therapy. (B) The longitudinally informed mechanistic model proposes that somatosensory afference increases communication between S1leg and the aINS, producing an effect of increased GABAergic inhibition in the aINS, leading to reduced clinical pain in FM.

Discussion

Our randomized neuroimaging trial evaluated the role of somatosensory afference in acupuncture to the reduction of clinical pain in FM. We found that EA (designed to generate sustained somatosensory afferent activity) was more effective than ML acupuncture (designed to generate no somatosensory afference) in reducing clinical pain. As the EA intervention was heavily directed to the patient’s legs, we examined brain connectivity with the primary somatosensory cortical representation of the leg (S1leg). We found that following EA therapy, FM patients demonstrated increased communication of this S1leg region with the anterior and posterior insula (aINS, pINS), as well as non-leg S1 subregions. Greater post-therapy increases in S1leg-aINS and S1leg-pINS connectivity were associated with greater reduction in clinical pain. Moreover, we measured the concentration of the inhibitory neurotransmitter GABA in the insula and found that greater post-therapy increase in S1leg-aINS connectivity was associated with greater increase in aINS GABA+, suggesting that S1leg signaling may increase GABAergic inhibition in the aINS. Furthermore, we found that greater increases in aINS GABA+ were associated with greater reduction in clinical pain. Finally, increased aINS GABA+ mediated the effect of increased S1leg-aINS connectivity on reduced clinical pain in EA. Cumulatively, these results allow us to establish a mechanistic model for the role of somatic sensation in acupuncture therapy: somatosensory afference leads to increased S1leg-aINS signaling, resulting in increased GABAergic inhibition in the aINS, ultimately reducing clinical pain (Figure 5B).

Our research extends previous work demonstrating somatotopically-specific involvement of S1 in acupuncture. Early research found that ST-36 EA produced stimulus-evoked BOLD activation in the contralateral S1leg region31. Later work examined somatotopic specificity of S1 morphology and functional involvement in clinical populations, linking S1-metrics with therapeutic outcomes. Specifically, in Carpal Tunnel Syndrome (CTS), longitudinal electroacupuncture therapy targeting the median nerve at the wrist increased the S1 separation distance between median-nerve innervated digits 2 and 3, and this increase in S1 digit separation predicted long-term clinical improvements17. Another recent study using manual acupuncture in chronic low back pain showed increases in gray matter volume and white matter integrity in back-specific S120. However, these studies were limited to local changes within S1 and did not explore cross-network signaling.

There is some evidence of increased cross-network communication in response to acute EA stimulation. In healthy individuals, acute EA stimulation produced increased connectivity of the Default Mode and sensorimotor network to the anterior cingulate (a key node of the salience network)32. In the current study, we found evidence for increased connectivity between S1leg and right aINS, and the degree of this connectivity increase was linked to improvements in clinical pain. This result may seem counterintuitive as chronic nociplastic pain is often characterized by heightened resting functional connectivity of S1 and the aINS relative to pain-free controls33,34. However, those studies assessed pathology-specific S1 subregions (e.g., S1back for lower back pain). In our study, analyses evaluated connectivity of S1 subregions specifically targeted by EA – i.e. S1leg. Furthermore, recent work has causally shown that GABAergic inhibition is recruited in the aINS to reduce nocifensive behavior35. Therefore, our results suggest that S1leg may be signaling the aINS to reduce clinical pain via GABAergic inhibition. Alternately, acupuncture may temporarily upregulate pronociceptive signaling between S1leg and aINS, which may trigger endogenous descending inhibitory systems to counteract through GABAergic inhibition of the aINS (i.e. healing processes initiated by temporary injury)36. These frameworks need further validation through reverse translational studies.

In FM patients reduced levels of GABA in the aINS6, and a compensatory upregulation of GABAA receptors7 have been reported. Pharmacologic interventions that enhance GABAergic neurotransmission have been found efficacious for FM–a phase-3 randomized trial of sodium oxybate (a GABA agonist) showed improvements in FM symptoms37. Based on these observations, reverse translational research has shown a causal link between aINS GABA and nocifensive behaviors in rats–decreasing endogenous levels of GABA in the agranular insula (rat homolog of the aINS) increased thermal and mechanical sensitivity38. Our study extends this literature by showing that increases in aINS GABA+ were associated with improvements in clinical pain following EA treatment, suggesting that somatosensory afference may modulate GABAergic inhibition to produce analgesia. The aINS is a hyper-reactive locus in FM patients39, and patients that have a post-therapy increase in aINS GABA+ may reduce such hyper-reactivity or hyperactivity, resulting in analgesia. Interestingly, although GABA is a molecular product of glutamate, our study did not note any association between clinical outcomes and Glx, suggesting that specific GABAergic pathways may be involved in somatosensation-enhanced acupuncture analgesia.

Another notable link established in our study was that increased long-range cortico-cortico communication post-therapy may lead to increased GABAergic inhibition. Although GABAergic neurons contribute significantly to local energy consumption40, the relationship between BOLD activity and GABA derived from 1H-MRS is complex–some studies in healthy individuals have shown that greater GABA is related to greater task-based negative BOLD responses41,42 while other research across multiple cortical regions have shown no such relationships43. With regards to BOLD functional connectivity, both positive and negative correlations with GABA have been noted–greater within-primary motor (M1) connectivity has been shown to be negatively correlated with M1 GABA44, whereas dorsal anterior cingulate GABA was not related to salience network GABA45. One recent study in healthy individuals measured GABA in two nodes of traditionally anti-correlated networks, the medial prefrontal cortex (mPFC) and the dorsolateral prefrontal cortex (dlPFC), and showed that mPFC-dlPFC functional connectivity at rest was positively correlated with dlPFC GABA and negatively correlated with mPFC GABA46, suggesting that intrinsic functional connectivity architecture may be associated with varying GABAergic tone across the cortex. Few studies have noted treatment-related changes in GABA and functional connectivity; one study noted that administration of Gamma‐hydroxybutyrate (a GABA agonist) increased right aINS functional connectivity47. Due to the complex relationship between GABA and BOLD functional connectivity across previous studies, our results need further validation. Nevertheless, our longitudinally-informed model (Figure 5B) proposes that increased S1leg-aINS connectivity influenced GABA+ in the aINS to reduce clinical pain. The downstream effects of this S1leg-aINS pathway needs further investigation; one possibility is that S1 taps into aINS regulation of sympathetic outflow as the aINS is part of the central autonomic network48. In fact, our previous study has shown that during experimental pressure pain in FM patients, S1leg-aINS connectivity was associated with reduced cardiovagal modulation24. Additionally, GABA is not the only neurotransmitter regulating aINS function; in a subsample of FM participants from this study, we found that elevated Choline (often involved in neuroinflammation) in FM was related to pain interference via aINS-Putamen functional connectivity49. Future studies should more explicitly examine the role of the autonomic nervous system and/or other neurotransmitters involved in somatosensation-induced acupuncture analgesia.

While our study demonstrates mechanistic links of acupuncture treatment via S1leg-aINS connectivity and aINS GABA, the clinical translation of these brain markers warrants further evaluation. For instance, a possible hypothesis is that aINS GABA and S1leg-aINS connectivity at baseline is predictive of the therapeutic trajectory of acupuncture, which would increase its clinical utility. Future studies should be focused on using neuroimaging markers at baseline to predict acupuncture treatment outcomes.

Our study was designed to specifically examine somatosensory afference, but other contextual factors (patient-clinician rapport, expectations, etc) may have contributed to analgesia as well, particularly in the ML comparator group. Thus, our results highlight the importance of carefully designed controls in acupuncture trials, as various specific and non-specific components contribute towards treatment outcomes. Researchers need a thorough understanding of the various factors that might be contributing towards analgesia while designing an acupuncture trial.

Limitations of our results should be noted. Despite a strong relationship between changes in aINS GABA+ and changes in clinical pain/S1leg-aINS connectivity, we did not observe a main effect of post-therapy GABA+ increase. We reason that aINS may be downstream of our proposed pathway (Figure 5B) and 4-weeks of treatment may not be sufficient to increase aINS GABA+. Future studies should be designed with a longer treatment schedule, including a post-therapy assessment period to examine long-term effects.

In summary, our study found that the somatosensory component of acupuncture specifically modulated functional communication and inhibitory neurochemistry in the somatosensory-insular circuit in order to reduce clinical pain in FM patients. With future rigorous mechanistic studies of acupuncture, we may be able to discover novel CNS pathways involved in non-pharmacologically induced analgesia and design new treatments that modulate CNS pathways in chronic pain pathologies.

Supplementary Material

Supinfo S2
Supinfo S1

Acknowledgments

Funding Sources

This project was supported by R01AT007550 (NIH-NCCIH) awarded to R.E. Harris and V. Napadow and F99DK126121 (NIH-NIDDK) training grant awarded to I. Mawla. This project applies tools developed under NIH R01 EB016089 and P41 EB015909.

Footnotes

Conflicts of Interest

All authors have no conflicts of interest with this manuscript.

References

  • 1.Clauw DJ. Fibromyalgia: a clinical review. JAMA. 2014;311(15):1547–1555. [DOI] [PubMed] [Google Scholar]
  • 2.Harte SE, Clauw DJ, Hayes JM, Feldman EL, St Charles IC, Watson CJ. Reduced intraepidermal nerve fiber density after a sustained increase in insular glutamate: a proof-of-concept study examining the pathogenesis of small fiber pathology in fibromyalgia. Pain Rep. 2017;2(3):e590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Kaplan CM, Schrepf A, Ichesco E, et al. Association of Inflammation With Pronociceptive Brain Connections in Rheumatoid Arthritis Patients With Concomitant Fibromyalgia. Arthritis Rheumatol. 2020;72(1):41–46. [DOI] [PubMed] [Google Scholar]
  • 4.Kosek E, Cohen M, Baron R, et al. Do we need a third mechanistic descriptor for chronic pain states? Pain. 2016;157(7):1382–1386. [DOI] [PubMed] [Google Scholar]
  • 5.Harris RE, Sundgren PC, Craig AD, et al. Elevated insular glutamate in fibromyalgia is associated with experimental pain. Arthritis Rheum. 2009;60(10):3146–3152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Foerster BR, Petrou M, Edden RA, et al. Reduced insular gamma-aminobutyric acid in fibromyalgia. Arthritis Rheum. 2012;64(2):579–583. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Pomares FB, Roy S, Funck T, et al. Upregulation of cortical GABAA receptor concentration in fibromyalgia. Pain. 2020;161(1):74–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.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(8):2545–2555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Ichesco E, Schmidt-Wilcke T, Bhavsar R, et al. Altered resting state connectivity of the insular cortex in individuals with fibromyalgia. J Pain. 2014;15(8):815–826 e811. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Harper DE, Ichesco E, Schrepf A, et al. Resting Functional Connectivity of the Periaqueductal Gray Is Associated With Normal Inhibition and Pathological Facilitation in Conditioned Pain Modulation. J Pain. 2018;19(6):635 e631–635 e615. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Rummans TA, Burton MC, Dawson NL. How Good Intentions Contributed to Bad Outcomes: The Opioid Crisis. Mayo Clin Proc. 2018;93(3):344–350. [DOI] [PubMed] [Google Scholar]
  • 12.Madsen MV, Gotzsche PC, Hrobjartsson A. Acupuncture treatment for pain: systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no acupuncture groups. BMJ. 2009;338:a3115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Langhorst J, Klose P, Musial F, Irnich D, Hauser W. Efficacy of acupuncture in fibromyalgia syndrome--a systematic review with a meta-analysis of controlled clinical trials. Rheumatology (Oxford). 2010;49(4):778–788. [DOI] [PubMed] [Google Scholar]
  • 14.Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med. 2012;172(19):1444–1453. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Langevin HM, Wayne PM, Macpherson H, et al. Paradoxes in acupuncture research: strategies for moving forward. Evid Based Complement Alternat Med. 2011;2011:180805. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Makary MM, Lee J, Lee E, et al. Phantom Acupuncture Induces Placebo Credibility and Vicarious Sensations: A Parallel fMRI Study of Low Back Pain Patients. Sci Rep. 2018;8(1):930. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Maeda Y, Kim H, Kettner N, et al. Rewiring the primary somatosensory cortex in carpal tunnel syndrome with acupuncture. Brain. 2017;140(4):914–927. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.MacPherson H, Maschino AC, Lewith G, et al. Characteristics of acupuncture treatment associated with outcome: an individual patient meta-analysis of 17,922 patients with chronic pain in randomised controlled trials. PLoS One. 2013;8(10):e77438. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Irnich D, Salih N, Offenbacher M, Fleckenstein J. Is sham laser a valid control for acupuncture trials? Evid Based Complement Alternat Med. 2011;2011:485945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kim H, Mawla I, Lee J, et al. Reduced tactile acuity in chronic low back pain is linked with structural neuroplasticity in primary somatosensory cortex and is modulated by acupuncture therapy. Neuroimage. 2020;217:116899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Kong J, Gollub R, Huang T, et al. Acupuncture de qi, from qualitative history to quantitative measurement. J Altern Complement Med. 2007;13(10):1059–1070. [DOI] [PubMed] [Google Scholar]
  • 22.Wolfe F, Clauw DJ, Fitzcharles MA, et al. Fibromyalgia criteria and severity scales for clinical and epidemiological studies: a modification of the ACR Preliminary Diagnostic Criteria for Fibromyalgia. J Rheumatol. 2011;38(6):1113–1122. [DOI] [PubMed] [Google Scholar]
  • 23.Esteban O, Markiewicz CJ, Blair RW, et al. fMRIPrep: a robust preprocessing pipeline for functional MRI. Nat Methods. 2019;16(1):111–116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Kim J, Loggia ML, Cahalan CM, et al. The somatosensory link in fibromyalgia: functional connectivity of the primary somatosensory cortex is altered by sustained pain and is associated with clinical/autonomic dysfunction. Arthritis Rheumatol. 2015;67(5):1395–1405. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Woolrich MW, Behrens TE, Beckmann CF, Jenkinson M, Smith SM. Multilevel linear modelling for FMRI group analysis using Bayesian inference. Neuroimage. 2004;21(4):1732–1747. [DOI] [PubMed] [Google Scholar]
  • 26.Mescher M, Merkle H, Kirsch J, Garwood M, Gruetter R. Simultaneous in vivo spectral editing and water suppression. NMR Biomed. 1998;11(6):266–272. [DOI] [PubMed] [Google Scholar]
  • 27.Provencher SW. Automatic quantitation of localized in vivo 1H spectra with LCModel. NMR Biomed. 2001;14(4):260–264. [DOI] [PubMed] [Google Scholar]
  • 28.Edden RA, Puts NA, Harris AD, Barker PB, Evans CJ. Gannet: A batch-processing tool for the quantitative analysis of gamma-aminobutyric acid-edited MR spectroscopy spectra. J Magn Reson Imaging. 2014;40(6):1445–1452. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Diedenhofen B, Musch J. cocor: a comprehensive solution for the statistical comparison of correlations. PLoS One. 2015;10(3):e0121945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Hayes AF. Introduction to Mediation, Moderation, and Conditional Process Analysis, Second Edition A Regression-Based Approach. In: 2nd ed. New York: Guilford Publications; 2017. [Google Scholar]
  • 31.Napadow V, Makris N, Liu J, Kettner NW, Kwong KK, Hui KK. Effects of electroacupuncture versus manual acupuncture on the human brain as measured by fMRI. Hum Brain Mapp. 2005;24(3):193–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Dhond RP, Yeh C, Park K, Kettner N, Napadow V. Acupuncture modulates resting state connectivity in default and sensorimotor brain networks. Pain. 2008;136(3):407–418. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Kutch JJ, Ichesco E, Hampson JP, et al. Brain signature and functional impact of centralized pain: a multidisciplinary approach to the study of chronic pelvic pain (MAPP) network study. Pain. 2017;158(10):1979–1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Kim J, Mawla I, Kong J, et al. Somatotopically specific primary somatosensory connectivity to salience and default mode networks encodes clinical pain. Pain. 2019;160(7):1594–1605. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Gamal-Eltrabily M, Espinosa de Los Monteros-Zuniga A, Manzano-Garcia A, Martinez-Lorenzana G, Condes-Lara M, Gonzalez-Hernandez A. The Rostral Agranular Insular Cortex, a New Site of Oxytocin to Induce Antinociception. J Neurosci. 2020;40(29):5669–5680. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Zhu H. Acupoints Initiate the Healing Process. Med Acupunct. 2014;26(5):264–270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Russell IJ, Holman AJ, Swick TJ, et al. Sodium oxybate reduces pain, fatigue, and sleep disturbance and improves functionality in fibromyalgia: results from a 14-week, randomized, double-blind, placebo-controlled study. Pain. 2011;152(5):1007–1017. [DOI] [PubMed] [Google Scholar]
  • 38.Watson CJ. Insular balance of glutamatergic and GABAergic signaling modulates pain processing. Pain. 2016;157(10):2194–2207. [DOI] [PubMed] [Google Scholar]
  • 39.Harte SE, Ichesco E, Hampson JP, et al. Pharmacologic attenuation of cross-modal sensory augmentation within the chronic pain insula. Pain. 2016;157(9):1933–1945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Buzsaki G, Kaila K, Raichle M. Inhibition and brain work. Neuron. 2007;56(5):771–783. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Northoff G, Walter M, Schulte RF, et al. GABA concentrations in the human anterior cingulate cortex predict negative BOLD responses in fMRI. Nat Neurosci. 2007;10(12):1515–1517. [DOI] [PubMed] [Google Scholar]
  • 42.Muthukumaraswamy SD, Edden RA, Jones DK, Swettenham JB, Singh KD. Resting GABA concentration predicts peak gamma frequency and fMRI amplitude in response to visual stimulation in humans. Proc Natl Acad Sci U S A. 2009;106(20):8356–8361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Harris AD, Puts NA, Anderson BA, et al. Multi-regional investigation of the relationship between functional MRI blood oxygenation level dependent (BOLD) activation and GABA concentration. PLoS One. 2015;10(2):e0117531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Stagg CJ, Bachtiar V, Amadi U, et al. Local GABA concentration is related to network-level resting functional connectivity. Elife. 2014;3:e01465. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Levar N, Van Doesum TJ, Denys D, Van Wingen GA. Anterior cingulate GABA and glutamate concentrations are associated with resting-state network connectivity. Sci Rep. 2019;9(1):2116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Chen X, Fan X, Hu Y, et al. Regional GABA Concentrations Modulate Inter-network Resting-state Functional Connectivity. Cereb Cortex. 2019;29(4):1607–1618. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Bosch OG, Esposito F, Dornbierer D, et al. Prohedonic properties of gamma-hydroxybutyrate are associated with changes in limbic resting-state functional connectivity. Hum Psychopharmacol. 2018;33(6):e2679. [DOI] [PubMed] [Google Scholar]
  • 48.Beissner F, Meissner K, Bar KJ, Napadow V. The autonomic brain: an activation likelihood estimation meta-analysis for central processing of autonomic function. J Neurosci. 2013;33(25):10503–10511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Jung C, Ichesco E, Ratai EM, et al. Magnetic resonance imaging of neuroinflammation in chronic pain: a role for astrogliosis? Pain. 2020;161(7):1555–1564. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supinfo S2
NIHMS1654853-supplement-Supinfo_S2.doc (50.5KB, doc)
Supinfo S1
NIHMS1654853-supplement-Supinfo_S1.docx (3.9MB, docx)

RESOURCES