A Randomized, Sham-Controlled Trial of Cranial Electrical Stimulation for Fibromyalgia Pain and Physical Function, Using Brain Imaging Biomarkers

Abstract

Fibromyalgia is a chronic pain syndrome that presents with a constellation of broad symptoms, including decreased physical function, fatigue, cognitive disturbances, and other somatic complaints. Available therapies are often insufficient in treating symptoms, with inadequate pain control commonly leading to opioid usage for attempted management. Cranial electrical stimulation (CES) is a promising non-pharmacologic treatment option for pain conditions that uses pulsed electrical current stimulation to modify brain function via transcutaneous electrodes. These neural mechanisms and the applications of CES in fibromyalgia symptom relief require further exploration.

A total of 50 participants from the Atlanta Veterans Affairs Healthcare System (VAHCS) diagnosed with fibromyalgia were enrolled and then block-randomized into either a placebo plus standard therapy or active CES plus standard therapy group. Baseline assessments were obtained prior to the start of treatment. Both interventions occurred over 12 weeks, and participants were assessed at 6 weeks and 12 weeks after treatment initiation. The primary outcome investigated whether pain and functional improvements occur with the application of CES. Additionally, baseline and follow-up resting state functional connectivity magnetic resonance imaging (rs-fcMRI) were obtained at the 6-week and 12-week time points to assess for clinical applications of neural connectivity biomarkers and the underlying neural associations related to treatment effects.

This is a randomized, placebo-controlled trial to determine the efficacy of CES for improving pain and function in fibromyalgia and further develop rs-fcMRI as a clinical tool to assess the neural correlates and mechanisms of chronic pain and analgesic response.

Introduction

Of the many existing states of chronic pain, one of the most notoriously difficult diseases to diagnose, clinically assess, and treat is fibromyalgia. Fibromyalgia is a debilitating chronic pain syndrome that involves chronic widespread pain, decreased physical function, fatigue, psycho-emotional and sleep disturbances, and various somatic complaints affecting approximately 2-3% of the general population in the Americas (about 8 million people in the U.S.)¹. Diagnosis of the disease is heavily reliant on a patient's understanding of their own symptom profile and pain experience, and without that proper understanding by both clinician and patient of the disease, methods of treatment lose considerable efficacy². A better definition of fibromyalgia's origins and impact as well as a reliable clinical biomarker to guide fibromyalgia diagnosis and treatment are necessary to best serve all patients.

Even with a confirmed diagnosis, difficulties with the treatment process only grow. As a whole, chronic pain affects more individuals than heart disease, diabetes, and cancer combined. The subjective nature of its assessment places it as a primary driver for the opioid epidemic, especially given the difficulty in discerning incompletely treated physical pain from substance use disorder and drug-seeking behavior³. In 2020, 91,799 drug overdose deaths occurred in the United States (a 30% increase from 2019), and opioids were found to be the main cause of these deaths (74.8% of all 2020 drug overdose deaths)⁴. Thus, non-pharmacologic alternatives are needed for chronic pain and fibromyalgia treatment to slow the opioid epidemic, which is particularly important in the veteran population where the risk of suicide and opioid use disorder is higher⁵. Non-pharmacologic and complementary therapies are therefore often used as first-line treatments⁶.

The search for novel, efficacious fibromyalgia interventions has led many researchers and clinicians to methods of noninvasive brain stimulation, including cranial stimulation. Even though the pathophysiologic mechanisms that result in the development of the disease have not been definitively determined, existing evidence supports the idea that fibromyalgia is a disorder of autonomic nervous system dysfunction and central (i.e., brain and spinal cord) pain processing mechanisms^7,8. Stimulation of certain areas of the brain could lead to improved function in those areas of processing. Repetitive transcranial magnetic stimulation (rTMS) and transcranial direct current stimulation (tDCS) have been correlated with reductions in pain but have also been associated with activation site scalp irritation, headaches, and inaccessibility outside of treatment facilities⁹. Noninvasive vagus nerve stimulation (nVNS), which can provide neuromodulation through stimulation over the skin at the neck or at the level of the ear, has the potential for the treatment of chronic pain, and invasive vagus nerve stimulation (VNS) has been shown to improve chronic pain symptoms¹⁰. However, neither invasive nor noninvasive VNS have been sufficiently explored in the literature or fully validated for use in fibromyalgia treatment^11,12,13,14.

Cranial electrical stimulation (CES) is a non-pharmacological, noninvasive brain stimulation treatment that consists of pulsed, alternating microcurrent (less than 0.5 mA) applied via transcutaneous electrodes placed on the earlobes¹⁵. It is remarkably accessible and can be delivered through portable devices used by patients within their own living spaces. In comparison to other cranial stimulation methods, the noninvasive nature and the convenience of patient self-application at home increases the potential of CES as a beneficial option for widespread fibromyalgia treatment use and self-management of pain. It has been cleared by the U.S. Food and Drug Administration (FDA) as a treatment for insomnia, depression, anxiety, and pain¹⁵.

The current study evaluates the efficacy of CES as a fibromyalgia treatment modality by comparing active CES (administered by a true study device) versus sham CES (administered by a sham study device). There is some preliminary evidence to support the use of CES in the treatment of pain conditions such as fibromyalgia^16,17. A 2001 study of 60 participants randomized to active or sham CES for 3 weeks of daily 60 min sessions revealed a 28% improvement in tender point scores, 27% improvement in general pain scores, and no placebo effect¹⁸. CES has not been evaluated in a veteran population, nor has it been adequately evaluated in males with fibromyalgia. A Veterans Affairs (VA)-funded systematic review of CES published in 2018 concluded that evidence is insufficient for CES to have clinically important effects on fibromyalgia, given that most trials had small sample sizes, short durations, and a high risk of bias due to inadequate blinding. However, the review suggests that CES does not cause serious side effects, and there is low-strength evidence to suggest modest benefits in patients with anxiety and depression¹⁹. Therefore, further research is warranted regarding the use of this FDA-cleared, low-risk device, particularly in fibromyalgia.

In order to fully evaluate the efficacy, researchers assessed physical fitness alongside neural biomarkers and pain experience. The purpose of treating chronic pain states is to improve physical function. Fibromyalgia is consistently correlated with negative effects on both physical function and patients' perception of their own physical abilities²⁰. Previous studies have utilized simple physical fitness assessments to determine stamina and mobility, such as the 6 Minute Walk Test (6MWT)^20,21, Five Time Sit to Stand (5TSTS)²⁰, and various measures of carrying capacity and strength in the context of daily activities²². To account for standard measures while also mitigating the amount of strenuous activity required right before an MRI scan, the study team used the 30-Second Chair Sit Stand Test as a measure of stamina and mobility and both bicep curls and a hand grip test as measures of strength²³. The movements required in each of these assessments are very common in everyday activities, so it is a clear measure of how people are physically functioning in their day-to-day lives, both with and without treatment.

Even with subjective pain assessments and physical function measures of efficacy, the mechanisms of CES are not fully understood. Prior neuroimaging studies have sought a better understanding by exploring the direct effect of CES on network connectivity in the brain. Feusner et al.²⁴ found that CES is associated with cortical deactivation for 0.5 Hz and 100 µA stimulation of bilateral frontal, parietal, and posterior midline regions and postulated that frequency of stimulation may have more of an effect than current intensity in relation to cortical deactivation. Their group found significant effects on some but not all nodes of the default mode network (DMN). The authors suggest that based on this data, CES may affect resting state functional connectivity. Fibromyalgia and other chronic pain states have been shown to affect intrinsic brain connectivity in regions associated with pain and perception^25,26, so treatments that alter functional connectivity in response could prove to be both beneficial and effective. Further exploration of the longer-term effects of daily treatment in relation to clinical improvement, as well as how deceased activation in the brain relates to previously observed decreases in electroencephalogram frequencies, is needed to further understand the therapeutic mechanism of action²⁷.

Resting-state functional connectivity magnetic resonance imaging (rs-fcMRI) is the neuroimaging method that allows for the observation of these functional connectivity changes. Longitudinal resting state fMRI allows clinicians and researchers to establish a baseline of resting state connectivity and track alterations over time in response to CES treatment methods. It also helps to determine how changes in functional connectivity are correlated to differences in the experience of pain. Initial studies of neuroimaging for fibromyalgia used positron emission tomography (PET) and single-photon-emission computed tomography (SPECT) to examine the brain, but there are issues with both techniques in this regard: SPECT has a lower resolution than PET, and PET scans are invasive, which is not preferable for patients experiencing chronic pain. Functional magnetic resonance imaging (fMRI) scans have greater resolution than SPECT, but they examine brain activity in response to patients' specific actions or perceptions of stimuli²⁸. It is rs-fcMRI scans that can outline functional connectivity between regions of the brain and may be able to determine where and how fibromyalgia exists as well as the best methods of treatment²⁸.

Evaluating the efficacy of non-pharmacologic treatments for pain conditions such as fibromyalgia is of utmost importance both in the current setting of the opioid epidemic and in examining chronic pain as a risk factor for suicide^29,30, which is substantially increased among the veteran population. Additionally, the lack of adequate clinical biomarkers for pain is a recognized knowledge gap. Using a combination of behavioral measures and neuroimaging at multiple timepoints to assess treatment response is a novel approach to fibromyalgia evaluation, as is the utilization of auricular CES as a treatment.

The protocol aims to address the gap in fibromyalgia research by investigating the effects of CES on pain and physical function outcomes and evaluating neuroimaging as a tool for predictive and response biomarkers related to the clinical outcomes of CES therapy³¹.

Protocol

The study was conducted under the approval of the Emory University (IRB 112768) and Atlanta VA Institutional Review Boards (1585632-2; Internal Reference Number: 003) as well as the Atlanta VA R&D Committee (Board Reference Number: 3881). All subjects gave their informed consent for inclusion before they participated in this study. For a visual representation of the study protocol timeline, see Figure 1).

Figure 1: Study timeline. A visual representation of the timeline for executing the study procedures over the trial period. Please click here to view a larger version of this figure.

1. Recruiting participants via available methods

1. Follow the required respective institutional guidelines for disseminating study recruitment materials (flyers, posts, emails, calls, etc.) and approved recruitment methods.
   NOTE: The current study recruited via recruitment letter mailings to specific cohorts and follow-up phone calls to potential participants. The details regarding ordering the CES devices for the study are included in the supplemental CES device order instructions (Supplementary File 1).
2. When someone expresses interest in study participation, screen them over the phone to ensure they meet all inclusion and exclusion criteria.
   1. Include participants who fit the eligibility criteria: age 20-60 years old (limit set during preliminary study to minimize brain structural changes due to aging)³¹; diagnosis of fibromyalgia by the American College of Rheumatology 2016 criteria³²; right-handedness (in order to provide consistency in brain structure and function); pain score of 4 or greater on the Defense and Veterans Pain Rating Scale (DVPRS)³³ in the 3 months prior to enrollment; consistent and stable FM related medication for at least 4 weeks prior to enrollment³⁴; and ability to safely tolerate MRI.
      1. For the fibromyalgia screening portion, fill out the new clinical fibromyalgia diagnostic criteria³² with the potential participant to obtain their widespread pain index (WPI) score and severity score (SS). Score according to the guidelines listed at the bottom of the page.
      2. For the pain score screening portion, ask potential participants to verbally rate their average, consistent pain over the past 3 months on a scale from zero to ten.
      3. For the MRI screening portion, complete an MRI safety screening form with the potential participant provided by an institution or the scanner facility itself.
   2. Ask participants about all possible exclusion criteria. Exclusion criteria include a history of seizures or neurologic conditions that alter the brain; pregnancy; claustrophobia, MRI-incompatible implants, or other conditions incompatible with MRI; and a history of uncontrolled psychiatric illness or autoimmune disease that leads to pain and can better explain symptoms³¹.
      1. If any of the exclusion criteria apply to them, they fail the screening. Do not include them in the study.
3. When someone passes the screening, schedule them for their consent appointment and baseline MRI scan. Schedule 2 h for the consent and scan time.

2. Administering baseline, mid, and post-participant study appointments (see Table 1)

NOTE: All appointments take place at the MRI scanner location.

	Time	Research Activity
Phase 1	Week 1	Recruitment, Screening, Enrollment
Phase 2	Week 2–3	Baseline Assessments, rs-fcMRI
Phase 3	Weeks 2–14	Intervention
Phase 4	Week 6–10	Acute follow-up, rs-fcMRI (6 weeks into treatment)
Phase 5	Week 12–16	Long-term follow-up, rs-fcMRI (12 weeks into treatment)

Table 1: Research design. A timeline of the phases of individual participation over the 12 weeks of the study.

1. Before each baseline visit, prepare a CES device to bring to each participant.
   1. Open up the case to ensure the device is inside (it should be in bubble wrap). Remove the back of the device and insert two AAA batteries.
   2. Replace the device back. Turn the device on to ensure it works before bringing it to the participant. Hold down the top button for about 2 s until the screen lights up to turn it on, and hold it down for about 2 s until the screen goes blank to turn it off.
   3. Ensure each case contains the device, extra batteries, a lanyard, a small solution applicator, and ear clips. Make sure that the bag to give to participants contains the device case, a bottle of conducting solution, an extra cap to fill the small solution applicator, and extra ear clip electrode pads.
2. Once their consent and baseline MRI scan have been scheduled, meet the participant at the MRI scanner location for the appointment with two copies of the consent form, a CES device and its materials, eligibility testing paperwork, and a questionnaire packet for them to complete. Additionally, bring the data transfer device through which the MRI data will be transferred to the analysis site.
   NOTE: This study utilized an encrypted USB drive.
   1. For the eligibility paperwork, include the new clinical fibromyalgia diagnostic criteria³² and the DVPRS³³.
      NOTE: The DVPRS should be included for each visit.
   2. For the physical function tests, include a form to record the results of each test^20,21,35.
   3. For the questionnaires that participants will fill out, include the appropriate PROMIS measures (PROMIS Scale v1.2 - Global Health, Neuro-QOL Item Bank v2.0 - Cognition Function - Short Form, PROMIS Item Bank v1.0 - Substance Use/Alcohol: Alcohol Use, PROMIS Item Bank v1.0 - Emotional Distress - Anxiety- Short Form 7a, PROMIS Item Bank v1.0 - Emotional Distress-Depression - Short Form 4a, PROMIS Item Bank v1.0 - Fatigue - Short Form 6a, PROMIS Item Bank v1.0 - Pain Interference - Short Form 4a, PROMIS Item Bank v1.2 - Physical Function - Short Form 8b, PROMIS Item Bank v2.0 - Social Isolation - Short Form 4a)³⁶.
3. At the MRI location, administer the baseline visit. Call the participant ahead of time to confirm.
   1. Read through the full consent form with the participant ensuring that they understand the information. Ask them if they would like to participate, and if they agree, have them print their name, sign, and date the form.
   2. After consenting, if the participant chooses to participate in the study, complete the inclusion testing paperwork with them. If they are eligible to continue, complete the steps below; if they are ineligible, terminate their study participation at this point.
      1. Ensure that the eligibility testing paperwork includes the new clinical fibromyalgia diagnostic criteria and the DVPRS.
   3. If a participant is eligible after the inclusion testing, explain to the participant how to use the CES device.
      1. Remove the device from the case and show participants how to turn it on and off with the button at the top. Remind participants that the device will automatically deactivate after 60 min has elapsed.
      2. Show participants how to attach the ear clip cord to the side. The opposite end of the cord from the clips plugs into a small hole on the left side of the device labeled with a "1".
      3. Demonstrate how to remove the old electrode pads from the ear clips and attach new ones. To remove old pads, pull up the adhesive from where it connects to the clips. To attach new pads, align the hole in the spare electrode pads with the raised area in the center of the clip and press down firmly.
         NOTE: Do not pull the electrode pads by the soft portion on top. It is prone to tearing and will leave the adhesive stuck to the clips.
      4. Demonstrate how to clip the ear clips to their earlobes and have them try it once. Squeeze the green tips of the clip to open it, position the earlobe between, and gently close the clip to attach.
      5. Show them how to remove the plastic top from the small solution applicator and fill it with the solution from the larger bottle (see steps 2.3.3.6-2.3.3.8). Remind them that the solution is necessary for the device to function properly.
      6. To remove the top of the small solution applicator, remove the cap and press firmly against the side of the tip. The plastic tip will dislodge from the rest of the bottle.
      7. Attach the alternate cap for filling the small applicator to the top of the larger solution bottle. Push up the small nozzle, slide it into the small applicator, and squeeze the large bottle to fill.
      8. Tell participants that they should only add about 1-2 drops of the solution to each earclip electrode pad. Any excess solution should be dabbed off before use.
      9. Instruct the participants to use the device every day of the week for 60 min in the evening, about 1 h before going to bed. Instruct the participants to use the device while at rest (i.e., sitting still or lying down, not moving around).
         NOTE: For both time and position, participants were allowed normal variations in their home environment of usage to increase the practical nature of this therapy for home use. No parameters were set for sound allowed in the environment or required device storage.
      10. Give the participants a blank device log and explain that they need to enter each date of device use, their pain score before and after device use, and the time of device use.
      11. If participants are not able to use their device for a period of time during the 6 weeks between appointments but are able to make the follow-up appointments, let them remain in the study. If they are not able to use the device or make it to the study visits, end their participation. See the supplemental CES device log (Supplementary File 2) for the device log this study used.
          NOTE: The following steps will be repeated during all three visits.
   4. After the device explanation, have the participant complete three short physical function tests²³. For each test, a maximal number will be recorded for the score (total repetitions for the first two and strength per trial for the third).
      NOTE: The stopwatch used for this study was an Apple iPhone 12.
      1. Administer the 30 s chair sit stand test. Follow steps 2.3.4.2-2.3.4.6.
      2. Place a chair with its back against the wall of the testing room. Have the participant sit on the chair with their back against the chair back.
      3. Instruct the participant to rise to a full standing position and then sit all the way back down with their back against the chair back as many times as they can in 30 s.
         NOTE: If they tap the chair but do not fully put their weight back on it in a sitting position, the repetition does not count.
      4. Tell the participants when to begin. Start the timer when they start moving.
      5. Have participants do arm curls with a dumbbell on each side for 30 s per side. Start with the right arm first and then move to the left.
      6. Record the total number of curls per side. Use a 5 lb weight for women and an 8 lb weight for men. Ensure that the participants are seated for the arm curl test.
      7. Have participants do three grip strength trials with each arm on a dynamometer. Follow steps 2.3.4.8-2.3.4.11. Start with the right arm first and then move to the left.
      8. Place the dynamometer in the participant's hand. They should be squeezing the grip at the bottom and not touching the dial at the top.
      9. Tell participants to squeeze the dynamometer as hard as they can and then release. The needle on the dial will stop at the highest level of grip strength they displayed.
      10. Record their grip strength results by writing down the number the needle reaches on the dial. Reset the dial between repetitions by twisting the small knob on the front of the dial face counterclockwise until the needle rests at zero.
      11. Ensure that the participants are seated for the handgrip test.
          NOTE: Between each series, participants receive about 15-20 s of rest as they shift equipment to their other hand. Between each test, participants will receive about 2 min of rest while the equipment and paperwork for the next is prepared.
   5. After the physical function tests, have participants complete the questionnaire packet with a pen. Check over the packet afterward to make sure they have answered all questions.
   6. After all review, testing, and questionnaires are completed, walk participants to the MRI scanner itself (for all MRI protocol information, see section 3). Review the screening form with the MRI technicians and ensure participants have removed all metal from their person.
      1. Make certain that the study team members have no metal on their persons.
      2. Help MRI technicians get participants into the scanner comfortably. Administer the MRI scan protocol.
      3. Instruct the participants about safety procedures (test call button and speakers).
      4. Remind the participants not to move at all during the scan.
      5. Remind the participants of the total duration of the scan, which should be 60 min.
      6. When the scan has been completed, help the participants out of the scanner.
   7. After the scan, finish participants' baseline visit and schedule their mid-visit for 6 weeks later. Participants will have three study visits total, each 6 weeks apart. Push MRI data to the study’s secure servers for analysis.
4. On the scheduled day, administer the mid-visit following the steps of the baseline visit. Call participant ahead of time to confirm.
   1. Administer the DVPRS. Have participants return their first completed device log and provide them with a second device log.
   2. Perform steps 2.3.4-2.3.6 (same as in the baseline visit).
   3. Finish participants' mid-visit and schedule their post-visit for 6 weeks later. Push MRI data to the study’s secure servers for analysis.
5. On the scheduled day, administer the post-visit following the steps of the mid-visit. Call the participant ahead of time to confirm.
   1. Administer the DVPRS. Have participants return their second completed device log.
   2. Perform steps 2.3.4-2.3.6 (same as in the baseline and mid-visits). No additional visit scheduling is needed for the last visit.
   3. If any information is needed to compensate participants, prepare and request that information on this visit.
      NOTE: This study required their name and address to issue checks from the Atlanta VA for compensation.
   4. Finish participants' post-visit. Push MRI data to the study’s secure servers for analysis.
      NOTE: Throughout the study, be sure to call and check in with the participants between visits to ensure they are using the device, completing the device logs, and aware of when their next visit will be.

3. Setting up MRI scan protocols

1. For the MRI scan protocol, acquire BOLD rs-fMRI on a 3T MRI scanner with a 32-channel phased array head coil using a gradient echoplanar imaging (EPI) sequence. Use the following MR parameters: FOV (Field of View)= 220 mm²; TR (Repetition Time)/TE (Echo Time) = 1500/25 ms, multiband-acceleration factor =3; flip angle = 50˚; 110 x 110 matrix size; slice thickness = 2 mm; GRAPPA factor = 2; Partial fourier of 6/8; 34 phase-encode reference lines, 72 interleaved axial slices covering the whole brain, roughly 350 scan volumes to yield 8 min of resting state fMRI data for stable estimation of connectivity networks.
2. In the protocol for the anatomical T1w magnetization prepared rapid gradient echo (MPRAGE), set TR = 2530 ms, TE = 3 ms, flip angle = 7°, slice thickness = 0.8 mm, 1 mm phase resolution. The T1w acquisition lasts approximately 6 min.
3. Acquire diffusion-weighted imaging (DWI) scans on a 3T MRI scanner with a 32-channel phased array head coil using a diffusion spectrum imaging scheme. Collect a total of 129 diffusion sampling directions with a maximum b-value of 3000 s/mm², an in-plane resolution of 2x2 mm², and a slice thickness of 2 mm.
4. Acquire physiological data (cardiorespiratory data utilizing a respiratory monitor belt and pulse oximetry) simultaneously (time-locked) to the fMRI data.
5. Securely transfer the MRI data from the scanner location to a secure site for preprocessing and analysis. Convert DICOMS to NIFTIs to comply with BIDS formatting utilizing dc2bids v2.1.6.
6. Scrub the data of individual identifiers. Use a study subject number on all data wherever applicable. Conduct a quality check to ensure no anomalous artifacts, such as excessive motion, occurred in the data via MRIQC v21.0.0.

4. Preprocessing and analyses

1. Once the MRI data from the study has been received, utilize two separate pipelines to analyze it: one to analyze the functional connectivity between participants and another to analyze white matter tractography.
2. Preprocess subjects' structural T1w and rsfMRI data via fMRIPrep v20.2.5, including brain extraction, tissue segmentation, and normalization of the T1-weighted (T1w) images as well as reference volume estimation, head-motion estimation, slice timing correction, and registration to the T1w for the functional images. This process results in the T1w and rsfMRI data normalized into MNI152NLin2009cAsym space.
   1. Use the preprocessed MRI data in the functional connectivity analysis (CONN).
   2. See the supplemental fMRIPrep Boilerplate document (Supplementary File 3^37-56). For further details, refer to the  link: https://fmriprep.org/en/stable/
3. Import the preprocessed dataset into CONN Toolbox v22a for further processing.
   NOTE: CONN Toolbox updated from v21a to v22a during this study.
   1. In the setup phase, establish 2^nd level covariates to define the study groups (TRUE vs. SHAM) for analysis later and ensure the quality of T1w and rsfMRI images. Smooth fMRI data via 8 mm Gaussian kernel.
   2. Following preprocessing, denoise the data to remove extraneous and physiological noise.
      NOTE: This study computed the first-level analysis but did not evaluate or use it as the researchers are not interested in single-subject results.
   3. Choose seeds/ROIs and subject covariates and set contrasts. Run a group-level seed-to-voxel analysis.
   4. Once the setup, preprocessing, and denoising steps have been completed for the pipelines, set the cluster and voxel thresholds to view connectivity patterns.
   5. See Supplementary Figure 1 and the supplemental CES CONN Instructions (Supplementary File 4) document.
4. Use correlational tractography⁵⁷ to determine longitudinal changes in white matter integrity correlated with the experiment group to identify tract bundles and regions associated with the CES treatment. Below are the main analysis steps:
   1. Convert the raw DWI images from .dcm (DICOM) into .nii.gz (NIfTI) format.
   2. Pre-process the images to correct for susceptibility-induced distortions using FSL's (version 6.0.6) TOPUP^58,59 and for eddy current distortions using FSL's EDDY tool⁶⁰.
   3. Generate DWI image quality control (QC) metrics related to motion at the single subject and study-wise levels using FSL's EDDY QC tools.
   4. Run a two-way repeated measures ANOVA on these QC metrics to identify any between-group variance that may confound group tractography results. If a metric shows a between-group variance that is significant (p > 0.05), then it should be accounted for as a covariate in the correlational tractography analysis.
   5. Import the preprocessed data into DSI Studio (version "Chen" Nov 21 2022) where they are converted to .src (source) files. For more information regarding DSI Studio, refer to the software's website (https://dsi-studio.labsolver.org). 
   6. Reconstruct the imported diffusion data using Q-Space Diffeomorphic Reconstruction (QSDR)⁶¹ to determine the white matter fiber orientations in MNI template space. A .fib (fiber orientation) file is output for each image. The option selections for the reconstruction are:
      method (reconstruction method selection) - QSDR
      param0 (diffusion sampling length) - 1.25 (this is the default)
      template (which template space to reconstruct data to) - ICBM152
      align_acpc (whether to rotate the image volume to align ac-pc) - 0 (false)
   7. Create a connectometry database from these .fib files, which extracts the quantitative anisotropy (QA) values from the reconstructed diffusion data. Calculate the longitudinal change in QA for each subject in the database. Add demographics such as experiment group, age, sex, along with any covariates identified in the QC step via a .csv file to the database.
   8. Next, load the connectometry database into the Group Connectometry Analysis GUI.
   9. Select covariates to be considered in the analysis. One of these selected covariates is designated as the Study variable.
      1. For this analysis, select Group as the study variable. Tracts with longitudinal changes in QA correlated with Group will be identified while the effect of the other covariates selected will be regressed out.
      2. The option selections for the group connectometry analysis are:
         - FDR Control (False Discovery Rate cutoff, only tracts with a significant correlation below FDR will be output) - 0.05
         - Length Threshold (value in voxels of minimum tract length used as null hypothesis) - 20 voxels (or 40 mm for 2 mm voxel sizes)
         - T threshold (t-stat measurement threshold for correlation effect) - 2.5
         - Study region (this pane allows regions to be included/excluded from the analysis) - select Whole Brain with - Exclude Cerebellum checked
         NOTE: For this study, the cerebellum was excluded due to some of the diffusion scans having portions of the cerebellum cut off during acquisition.
   10. Press the Run Connectometry button to perform the analysis, which outputs several files:
       - A .fib file that stores the t-stats and can be opened in DSI studio to visualize the t-stats of tracts with increasing QA (stored as "inc_t") or decreasing QA (stored as "dec_t"), which correlate with Group.
       - A .fdr_dist.values.txt, which lists the FDR values with respect to tract length
       - A .inc.tt.gz which is a tractography file that stores the tracts with increased longitudinal QA correlated with the study variable. (Group in our case). A .dec.tt.gz file stores the tracts with decreased longitudinal QA correlated with the study variable.
       - A .report.html file which conveniently reports the connectometry results along with imbedded tract plots, pictures, as well as the boilerplate information on the correlational tractography analysis steps for publication.
       NOTE: To see examples of the R code used for this study, see the supplemental CES R Code Plots (Supplementary File 5) and the supplemental R Code CES eddy-qc Anova files (Supplementary File 6).

Representative Results

In terms of recruitment results, participants were primarily recruited via mailing of recruitment letters and follow-up phone calls based on the outlined regulations of the Atlanta VA Healthcare System. The study team recruited a total of 50 participants, proving the effectiveness of the methods used in meeting the recruitment goal (see Figure 2). The use of the new clinical fibromyalgia diagnostic criteria allowed the study team to properly screen out individuals who did not meet fibromyalgia criteria³². Asking possible participants about a diagnosis of fibromyalgia is not as robust a measure as the additional screening and could have led to improperly scheduled baseline visits or study participation. Forty-eight participants were randomized into active and sham groups; two participants were excluded because they did not meet eligibility criteria based on the inclusion testing for the study.

Figure 2: Recruitment flow chart. A report and flow diagram of study recruitment, randomization, and allocation of intervention. Please click here to view a larger version of this figure.

Three outcomes were used in sample size calculations: DVPRS (clinical pain), 30 s chair stand test (function), and DMN-SMN connectivity (rs-fcMRI). All power calculations were based on preliminary data. Clinical pain changes using DVPRS were chosen as the primary outcome of interest. The 30 s chair stand test (30sCST) was chosen as the representative functional outcome since it exhibited the smallest between-group change. DMN-SMN connectivity was chosen as the secondary outcome of interest as a neuroimaging biomarker for clinical pain and treatment response. Sample size analyses were conducted assuming a significance of 1% and 80% power (2-sample, 1-sided).

The seeds for this analysis were chosen based on preliminary data, as well as the literature on fibromyalgia, pain, and CES^{16,24,26,63,64,65,66,67,68,69}. Based on the preliminary data, the mean (SD) of the post-treatment change in left primary sensorimotor cortex (L-S1M1) to left posterior cingulate cortex (L-PCC) connectivity is 0.041 (0.079) for the treatment group and -0.026 (0.049) for the standard treatment group; the observed between-group difference effect size is 1.03 (see Table 2^{16,21,26,62,63,64,70,71,72,73}). The study would need 20 subjects in each group to achieve 80% power to detect the difference between the CES group and the standard treatment group in their post-treatment change in connectivity for L-S1M1 to L-PCC at the significance level of 0.01 using a two-sided t-test, assuming the between-group difference effect size is 1.03 as observed in the pilot data. Though the pilot study observed a 17% attrition rate for 12 subjects in our prior study of auricular neuromodulation (all completed their follow-up MRI, but two were lost to follow-up at 8 weeks and 12 weeks), in order to maintain a conservative estimate for sample size calculations, this study assumed a 20% attrition rate. With an expected 20% attrition at the post-treatment visit, the study needed to recruit 20/0.8 = 25 subjects per group.

DMN seeds (x,y,z)	SMN seeds (x,y,z)	SN seeds (x,y,z)
Medial Prefrontal Cortex62,67	Right Putamen64,71	Right dorsolateral prefrontal cortex62
Right PCC70	Left M170	Left anterior insula62
Left PCC16,67	Right M170	Right anterior insula62
Precuneus71	Right S1-Hand16,72	Left posterior insula64
	Left S1-Hand16,72	Right posterior insula63
	Thalamus21	Dorsal anterior cingulate cortex72,73
		Right temporoparietal junction62

Table 2: Seeds for analysis. DMN, SMN, and SN seeds chosen for analysis based on a priori hypotheses. Each seed is presented with references to prior literature supporting its testing in pain syndromes.

Based on prior research of CES in civilian subjects with fibromyalgia, a total of 50 subjects (n = 25 sham, and n = 25 true) should achieve 80% power to detect a difference in pain scores between the two groups¹⁷ (see Table 3). Sample size calculations for this study were conducted using sealedenvelope.com and were based on preliminary data.

	Control		Intervention		N per group
	Mean Change	SD	Mean Change	SD
Clinical Pain (DVPRS)	0.375	1.493	-1.833	2.229	10
Function (30sCST)	-0.250	1.500	3.000	4.980	14
rs-fcMRI (S1M1 to PCC)	-0.026	0.049	0.041	0.079	20

Table 3: Sample size calculation. Calculations related to study sample size.

Imported fMRIPrep functional data were smoothed using spatial convolution with a Gaussian kernel of 8 mm full width half maximum (FWHM) (see Figure 3) shows the functional output from fMRIPrep normalized into MNI152NLin2009cAsym template space (left) and the smoothed functional image from CONN Toolbox (right). This results in an increase to the signal-to-noise ratio, which in turn improves the detection of blood-oxygen-level-dependent (BOLD) signals.

Figure 3: Single subject comparing an unsmoothed functional image in MNI space (left) to its smoothed counterpart at 8 mm FWHM. Please click here to view a larger version of this figure.

The data were then denoised using a standard denoising pipeline⁷⁴, including the regression of potential confounding effects characterized by white matter time-series (5 CompCor noise components), CSF time-series (5 CompCor noise components), motion parameters and their first order derivatives (12 factors)⁷⁵, outlier scans (below 295 factors)⁴⁸, and linear trends (2 factors) within each functional run, followed by bandpass frequency filtering of the BOLD timeseries⁷⁶ between 0.008 Hz and 0.09 Hz. CompCor^49,77 noise components within white matter and CSF were estimated by computing the average BOLD signal as well as the largest principal components orthogonal to the BOLD average, motion parameters, and outlier scans within each subject's eroded segmentation masks (see Figure 4). From the number of noise terms included in this denoising strategy, the effective degrees of freedom of the BOLD signal after denoising were estimated to range from 33 to 240.6 (average 173.4) across all subjects. The denoising resulted in a reduction of physiological and other extraneous noise from the data that could have resulted in confounding effects.

Figure 4: Quality checks. Quality checks graphs from CONN Toolbox displaying the effects of denoising on functional connectivity (FC), mean global signal and maximum motion. Note (A) upper most graph displays data for a single subject and session, (B,C) graphs B and C are the results of group-level denoising. Please click here to view a larger version of this figure.

Group-level analyses were performed using a General Linear Model (GLM)⁷⁸. For each individual voxel, a separate GLM was estimated, with first-level connectivity measures at this voxel as dependent variables (one independent sample per subject and one measurement per task or experimental condition, if applicable), and groups or other subject-level identifiers as independent variables. Voxel-level hypotheses were evaluated using multivariate parametric statistics with random effects across subjects and sample covariance estimation across multiple measurements. Inferences were performed at the level of individual clusters (groups of contiguous voxels). Cluster-level inferences were based on parametric statistics from Gaussian Random Field theory^62,79. Results were thresholded using a combination of a cluster-forming p < 0.005 voxel-level threshold and a familywise error corrected p < 0.0016 cluster-size threshold⁸⁰. Following these steps results in group-level functional connectivity values comparing the CES and sham conditions based on regions of interest (ROI). These results can be visualized in a multitude of ways through the results explorer GUI within CONN Toolbox. To see a volume display visualization of an example group-level results with the red area indicating regions with greater positive connectivity with the ROIs and the blue are regions with greater negative connectivity with the ROI, see Figure 5.

Figure 5: Volume display of post cingulate seed. The image displays greater positive connectivity with the anterior cingulate gyrus and (red) and greater negative connectivity with the precuneus (blue) for the True condition relative to the Sham, pFWEc < 0.05. This figure represents partial group-level data (n = 34). Please click here to view a larger version of this figure.

Supplementary File 1: CES device order instructions Please click here to download this file.

Supplementary File 2: CES device log Please click here to download this file.

Supplementary File 3: fMRIPrep Boilerplate Please click here to download this file.

Supplementary File 4: CES CONN instructions Please click here to download this file.

Supplementary File 5: CES R code plots Please click here to download this file.

Supplementary File 6: R Code CES eddy-qc Anova Please click here to download this file.

Supplementary Figure 1: CONN workbook.(A) Second-level covariates. (B) Experimental conditions. (C) Denoising. (D) Results. (E) Structural data. Please click here to download this figure.

Discussion

The methods of the current study provide not only the possibility of a highly effective treatment modality for fibromyalgia but also the opportunity to improve the diagnostic process of fibromyalgia from the first instance of its pain symptom profile. Use of both active CES and sham CES, with discovery of the type of each individual device being dependent upon serial numbers and a separate key, allowed for blinding of both subjects and researchers until the end of participation, thereby protecting internal validity during the administration of assessments and presenting a clearer clinical picture of the efficacy of both the device and the stimulation⁸¹. The addition of exploring functional connectivity neuroimaging measures helps to alleviate some of the weight placed on participants' subjective reporting for determining how effective the treatment methods truly are. The flexibility of an at-home pain management device introduces ease of usage and access that reduces participant burden, which is especially beneficial for a population that typically experiences considerable pain during even mundane or basic activities. Reduction of both physical pain and exertion could help increase the efficacy of fibromyalgia and chronic pain treatments even further.

In order to ensure that the methods employed ran smoothly, there were a number of modifications that needed to be made during the process. Periodically during data collection for the rs-fcMRI scans, several subjects became uncomfortable laying in the scanner due to experiencing pain. These difficulties resulted in them either needing to briefly exit the scanner before returning to complete their scan or moving around so much that they caused an unsalvageable motion artifact. To compensate for any unusable data during this process, nine subjects received two T1w anatomical scans, the clearer of which was ultimately used for analysis purposes. MRIQC was used to identify any excessive head motion found in the other portions of the MRI data for which there was no additional scan conducted. Using a framewise displacement threshold of 0.8 mm, it was determined that four subjects had at least one session that displayed excessive motion. These sessions were removed from the analysis. Researchers conducting the functional connectivity analysis in CONN Toolbox had to troubleshoot errors that arose during the analysis. For example, four subjects did not have seed-based connectivity (SBC) maps in their first-level analysis results for either mid or post-scans, nor did they have first-level design matrices even after they were already processed and denoised. It was determined that in a previous iteration of a CONN project, only their baseline scans had been collected and processed. Their mid and post-scans were later imported, but CONN created conditions/variables relating to those sessions even though these subjects did not at the time have data for them. CONN thus ignored the newly imported data. The solution was to re-run these subjects from the preprocessing stage and ensure that previous preprocessed data was overwritten.

In addition to addressing methodological adaptation, the researchers also recognized the current study's methodological limitations. There does exist the possibility that the study devices did not remain in the group to which they were assigned (true versus sham CES). Although it is unlikely, at some point they could have begun or stopped delivering the treatment, thereby inaccurately skewing the results. This limitation will be addressed by the device company itself, who will interrogate the devices upon their return to the facility to ensure that they consistently did or did not administer CES as they were originally intended.  Another limitation of the treatment methods is that the records of device usage are heavily reliant upon accurate participant self-report and understanding. The device records the total amount of usage time, but there is a possibility of participants turning the device on without direct use or misreporting other aspects of usage, such as date and time. The device logs could be affected by response bias, which would mean inaccurate reporting of treatment administered⁸². In response to this issue, there may be some benefit to establishing a more frequent and more structured plan of communication with each participant so that they are consistently reminded of the logs as well as when and how to complete them. It also invites further consideration into the devices themselves, how they work, and how the records they keep might be improved in addition to the stimulation they provide.

The methods explored in this study have the capacity for considerable impact across the fields of fibromyalgia research and chronic pain evaluation as a whole. With respect to existing fibromyalgia treatment modalities, auricular CES and its accessibility as a non-pharmacological, noninvasive pain management option address an important gap in existing methods. If proven to be effective, CES responds to the current climate of opioid analgesic mismanagement by providing a possible non-pharmacological alternative, and it could alleviate undue additional pain caused by invasive measures of both intervention and diagnosis. The neuroimaging aims of this study also serve an important and novel purpose in the implications of the methods. Previously, there have been no laboratory or radioactive findings validated as markers for fibromyalgia, which has led to the difficulties experienced in attempted diagnoses². Examining functional connectivity and its relation to brain activity specific to fibromyalgia could drastically change the diagnostic process as well as the determination of treatment, improving both systems. With more rapid and reliable diagnosis also comes the possibility of a better quality of life for patients with this disease. Their distress is intensified when there are no answers and when the extent of their pain is not understood by anyone else. Objective measures of pain assessment allow clinicians to better understand what they are going through and how to help.

The techniques used in this study open the door for further exploration of better methods of diagnosing and treating fibromyalgia and beyond. Fibromyalgia is far from the only chronic pain state reliant upon patient self-reporting and subjective assessment measures⁸³. The neuroimaging techniques used in this study could have broad applications across the diagnostic and intervention fields if researchers continue to examine the importance of brain activity in understanding and treating chronic pain. Auricular CES itself, if proven effective and made a more common and accessible treatment option, could provide widespread pain management help to a population very much in need. Data analysis is still ongoing, so more is yet to be known about the efficacy of adhering to the outlined methods. Ultimately, the results will continue to provide information and present arguments for how fibromyalgia and similar chronic pain states should be addressed in the future.

Materials

Name	Company	Catalog Number	Comments
3T Siemens MAGNETOM Prisma Scanner	Siemens Healthineers	N/A	From Emory's website: "The Siemens Magnetom Prisma 3T whole-body MR system is equipped with: a state-of-the art gradient system with a maximum (per axis) strength of 80 mT/m and slew rate of 200 T/m/sec 64 independent RF receiver channels capable of 204 receiver connections a 2-channel RF transmitter. Multiple coils are available, including: a 64-channel head/neck coil with 52 channels for imaging of the head region a 32-channel head-only coil a 20-channel head/neck coil with 16 channels for head spine array coil flexible chest coil large and small flexible coil for extremity imaging.
Alpha-Stim AID Kit	Electromedical Products International Inc.	SKU: 500KIT	A total of 50 devices ordered for research purposes. From the site: "A prescription or order from a licensed healthcare professional is required to purchase this device (within the USA). FDA cleared for anxiety, insomnia and pain only, with approval for depression outside of the United States."
CONN Toolbox v21a16 (RRID:SCR_009550)	Whitfield-Gabrieli and Nieto-Castanon	Version v21a16 (RRID:SCR_009550)	CONN is an open-source SPM-based cross-platform software for the computation, display, and analysis of functional connectivity Magnetic Resonance Imaging (fcMRI). CONN is used to analyze resting state data (rsfMRI) as well as task-related designs.
DSI Studio (RRID:SCR_009557)	Fang-Cheng (Frank) Yeh	RRID:SCR_009557	DSI Studio is a tractography software tool that maps brain connections and correlates findings with neuropsychological disorders. It is a collective implementation of several diffusion MRI methods, including diffusion tensor imaging (DTI), generalized q-sampling imaging (GQI), q-space diffeomorphic reconstruction (QSDR), diffusion MRI connectometry, and generalized deterministic fiber tracking.
fMRIPrep 20.2.5 (RRID:SCR_016216)	NiPreps (NeuroImaging PREProcessing tools)	Version 20.2.5. (RRID:SCR_016216)	A functional magnetic resonance imaging (fMRI) data preprocessing pipeline that is designed to provide an easily accessible, state-of-the-art interface that is robust to variations in scan acquisition protocols and that requires minimal user input, while providing easily interpretable and comprehensive error and output reporting. It performs basic processing steps (coregistration, normalization, unwarping, noise component extraction, segmentation, skull-stripping, etc.) providing outputs that can be easily submitted to a variety of group level analyses, including task-based or resting-state fMRI, graph theory measures, and surface or volume-based statistics.
MRIQC	NiPreps (NeuroImaging PREProcessing tools)		MRIQC extracts no-reference IQMs (image quality metrics) from structural (T1w and T2w) and functional MRI (magnetic resonance imaging) data. (not directly used for analyses)
Sammons Preston Jamar Hydraulic Hand Dynamometer	Alpha Med Inc.	SKU SAMP5030J1	From the website: Ideal for routine screening of grip strength and initial and ongoing evaluation of clients with hand trauma and dysfunction. Unit comes with carrying/storage case, certificate of calibration and complete instructions. Warranted for one full year. The warranty does not cover calibration. Latex free.
SPRI 5-Pound Vinyl-Coated Weight	SPRI | Amazon	N/A	Color: (E) Dark Blue | 5-Pound. Appears on Amazon: Dumbbells Hand Weights Set of 2 - Vinyl Coated Exercise & Fitness Dumbbell for Home Gym Equipment Workouts Strength Training Free Weights for Women, Men (1-10 Pound, 12, 15, 18, 20 lb), https://www.amazon.com/stores/SPRI/Weights/page/9D10835A-CFAB-4DA1-BEE9-AE993C6B5BC1
SPRI 8-Pound Vinyl-Coated Weight	SPRI | Amazon	N/A	Color: (H) Black |8-Pound. Appears on Amazon: Dumbbells Hand Weights Set of 2 - Vinyl Coated Exercise & Fitness Dumbbell for Home Gym Equipment Workouts Strength Training Free Weights for Women, Men (1-10 Pound, 12, 15, 18, 20 lb), https://www.amazon.com/stores/SPRI/Weights/page/9D10835A-CFAB-4DA1-BEE9-AE993C6B5BC1