Magnetic Resonance Imaging of Neuroinflammation in Chronic Pain

Research Paper

Magnetic resonance imaging of neuroinflammation in chronic pain: a role for astrogliosis? Changjin Junga,b, Eric Ichescoc, Eva-Maria Rataia,d, Ramon Gilberto Gonzaleza,d, Tricia Burdoe, Marco L. Loggiaa,d, Richard E. Harrisc, Vitaly Napadowa,d,f,*

Abstract Noninvasive measures of neuroinflammatory processes in humans could substantially aid diagnosis and therapeutic development for many disorders, including chronic pain. Several proton magnetic resonance spectroscopy (1H-MRS) metabolites have been linked with glial activity (ie, choline and myo-inositol) and found to be altered in chronic pain patients, but their role in the neuroinflammatory cascade is not well known. Our multimodal study evaluated resting functional magnetic resonance imaging connectivity and 1H-MRS metabolite concentration in insula cortex in 43 patients suffering from fibromyalgia, a chronic centralized pain disorder previously demonstrated to include a neuroinflammatory component, and 16 healthy controls. Patients demonstrated elevated choline (but not myo-inositol) in anterior insula (aIns) (P 5 0.03), with greater choline levels linked with worse pain interference (r 5 0.41, P 5 0.01). In addition, reduced resting functional connectivity between aIns and putamen wasassociatedwithbothpaininterference(wholebrainanalysis,pcorrected , 0.01) and elevated aIns choline (r 520.37, P 5 0.03). In fact, aIns/putamen connectivity statistically mediated the link between aIns choline and pain interference (P , 0.01), highlighting the pathway by which neuroinflammation can impact clinical pain dysfunction. To further elucidate the molecular substrates of the effects observed, we investigated how putative neuroinflammatory 1H-MRS metabolites are linked with ex vivo tissue inflammatory markers in a nonhuman primate model of neuroinflammation. Results demonstrated that cortical choline levels were correlated with glial fibrillary acidic protein, a known marker for astrogliosis (Spearman r 5 0.49, P 5 0.03). Choline, aputativeneuroinflammatory1H-MRS-assessed metabolite elevated in fibromyalgia and associated with pain interference, may be linked with astrogliosis in these patients. Keywords: Chronic pain, Pain interference, Fibromyalgia, Neuroinflammation, Astrogliosis

1. Introduction with neuroinflammation,13 choline and myo-inositol, are Prior studies have suggested that magnetic resonance altered in chronic pain patients suffering from fibromyalgia (FM). Brain neuroinflammation has been noted in animal imaging (MRI) methods such as proton magnetic resonance 29 spectroscopy (1H-MRS)3,13 may be promising for noninvasive models of chronic pain, and FM is a chronic centralized evaluation of neuroinflammation. Our aim was to explicitly pain disorder that has recently been shown to display evaluate whether 1H-MRS metabolites previously associated a neuroinflammatory component assessed with more invasive positron emission tomography (PET).2 We further sought to evaluate possible mechanisms by which levels of the above- Sponsorships or competing interests that may be relevant to content are disclosed 1 at the end of this article. mentioned H-MRS metabolites are linked with neuroinflammation and contribute to chronic pain pathology. R.E. Harris and V. Napadow contributed equally to this work. Several neuroimaging studies have suggested that neuro- a Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA, United States, b Clinical inflammation, mediated by activation of microglia or astrocytes, is Research Division, Korean Institute of Oriental Medicine, Daejeon, South Korea, pronounced in chronic pain patients. For instance, PET imaging c Department of Anesthesiology, Chronic Pain and Fatigue Research Center, using the glia-linked radioligand [11C] PBR28 demonstrated d University of Michigan, Ann Arbor, MI, United States, Department of Radiology, increased binding for chronic low back pain patients and Massachusetts General Hospital, Harvard Medical School, Charlestown, MA, 2,4,31 2 United States, e Department of Neuroscience, Temple University, Philadelphia, FM, with elevated levels linked with fatigue. In contrast to 1 PA, United States, f Department of Anesthesiology, Pain Management Center, PET, H-MRS is a noninvasive neuroimaging technique that has Brigham and Women’s Hospital, Harvard Medical School, Chestnut Hill, MA, suggested neuroinflammation by quantifying levels of metabolites United States such as choline, which shows greater concentration in glial cells *Corresponding author. Address: Center for Integrative Pain Neuroimaging, Martinos than neurons, and has been linked with glial cell membrane Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, 13 1 #2301 149 Thirteenth St, Charlestown, MA 02129, United States. Tel.: 11-617-724-3402; turnover. Myo-inositol, another H-MRS metabolite, plays a role fax: 11-617-726-7422. E-mail address: [email protected] (V. Napadow). in regulating glial cell volume and may also be a marker for 13 Supplemental digital content is available for this article. Direct URL citations appear neuroinflammation. Myo-inositol is believed to be an essential in the printed text and are provided in the HTML and PDF versions of this article on requirement for cell growth, an osmolyte, and a storage source of the journal’s Web site (www.painjournalonline.com). glucose.50 Furthermore, this metabolite is thought to serve as PAIN 00 (2020) 1–10 marker of neuroinflammation/gliosis10,46; however, its role as © 2020 International Association for the Study of Pain a glial marker has also been questioned.44 Choline and/or myo- http://dx.doi.org/10.1097/j.pain.0000000000001815 inositol are elevated in the anterior cingulate cortex in spinal cord

Month 2020· Volume 00· Number 00 www.painjournalonline.com 1

injury patients with neuropathic pain57 and rheumatoid arthritis usage of all subjects has been included in Supplementary Table 1 patients.18 Choline has also been found to be elevated in the (available at http://links.lww.com/PAIN/A952). anterior cingulate cortex in chronic pelvic pain patients and We enrolled 43 (N 5 43) female FM patients and 16 (N 5 16) associated with higher levels of negative affect.24 However, the health controls (HCs). All subjects were right-handed and there precise role that choline and myo-inositol play in neuroinflamma- was no statistically significant age difference between FM and HC tion and how they are linked with other markers and maladaptive groups (FM: 41.07 6 11.64 years old, HC: 44.94 6 16.29 years brain changes in FM are unknown. old, mean 6 SD; P 5 0.31, 2-sample Student t-test). Twelve In this study, analyses focused on whether purported 1H-MRS patients reported using pregabalin or gabapentin for treatment, markers of neuroinflammation (ie, choline and myo-inositol) are which was controlled for in analyses due to our prior publications linked with functional brain connectivity and clinical variables for demonstrating significant neuroimaging changes within the insula FM. Our prior neuroimaging studies with chronic pain patients after pregabalin treatment.25 demonstrated that anterior insula (aIns) cortex connectivity was increased to the default mode network37 and, specifically, aIns to 2.2. Image acquisition and preprocessing medial prefrontal cortex connectivity was associated with 1H-MRS Glx metabolite concentration and clinical pain intensity in pelvic 2.2.1. Proton magnetic resonance spectroscopy pain patients.7 In a separate study, FM patients’ response to 1H-MRS acquisition was conducted on a 3.0T scanner Philips pharmacotherapy was associated with reduced 1H-MRS Glx Ingenia 3.0T (Philips Healthcare, Best, the Netherlands) using metabolite concentration and insula connectivity to default mode a 15-channel receive head coil. Anatomical T1-weighted images network brain regions.25 Given prior evidence of physiological (Axial 3D-MP-RAGE, repetition time (TR)/echo time (TE) 5 8.2/3.7 disruption in the insula cortex, we assessed 1H-MRS-derived ms, matrix 5 240 3 240, number slices 5 154) were acquired to metabolite levels in both anterior and posterior insula (pIns), and guide region of interest (ROI) definition for 1H-MRS. PRESS single functional connectivity between insula and the rest of the brain. An voxel spectra (TR/TE 5 2000/35 ms) were acquired from a 2 3 2 exploratory analysis evaluated the fractional amplitude of low- 3 3-cm sized voxel covering the right aIns region. A second voxel frequency fluctuations (fALFF) of the resting functional MRI (fMRI) was acquired in the right pIns region. Values for different signal, which has also been associated with neuroinflammation in metabolites were computed from the raw spectra data based chronic pain.6,26 Finally, to better understand the mechanisms on a linear combination of individual spectra using the LCModel supporting altered choline or myo-inositol in neuroinflammation, software42 (Version 6.3; Stephen Provencher, Inc, Oakville, ON, we directly linked levels of these metabolites with tissue pathology Canada). Because our study focused on potential neuroinflam- markers for astrogliosis and microglia activation in a nonhuman mation, analyses evaluated choline and myo-inositol concentra- primate model of widespread viral neuroinflammation.48 tion. For the sake of transparency, we wanted to report both of the conventionally reported measures for 1H-MRS concentration: 2. Methods the metabolite ratio values (dividing by total creatine [tCr]) and absolute concentrations, using the water peak as reference from 2.1. Human participants a separate scan. For the former, tCr concentration did not differ We enrolled female patients with FM diagnosis for at least 1 year, between FM and HC (aIns: P 5 0.15, pIns: P 5 0.95, analysis of meeting Wolfe criteria,59 who were recruited from the Chronic variance with age correction), allowing for interpretable FM vs HC Pain and Fatigue Research Center Registry at the University of comparisons using ratios of choline and myo-inositol to tCr. For Michigan. At screening visits, all participants completed an the latter, the concentrations were adjusted by an estimate of informed consent interview and were screened for exclusion cerebrospinal fluid (CSF) content in the voxel (hence, reported as criteria: (1) presence of a known coagulation abnormality, Choline:CSF in the Table 2), with volume computed using thrombocytopenia, or bleeding diathesis; (2) presence of statistical parametric mapping 8 (SPM8; Wellcome Trust Centre a concurrent autoimmune or inflammatory disease; (3) history of for Neuroimaging) software. head injury with substantial loss of consciousness; (4) peripheral neuropathy; (5) routine daily use of narcotic analgesics, mari- 2.2.2. Resting-state functional magnetic resonance imaging juana, or history of substance abuse; (6) stimulant medications; (7) pregnant or nursing; (8) severe psychiatric illnesses (current Resting, whole brain blood-oxygen-level dependent (BOLD) schizophrenia, major depression with suicidal ideation, consis- images were acquired on the same 3.0T system using a T2 3 tent with current medications), or (9) MRI contraindications. weighted echo-planar sequence (TR/TE 5 2000/30 ms, flip angle Female healthy adults were also recruited to form an age- 5 90˚, matrix 5 80 3 80, number slices 5 38, slice thickness 5 matched control group, with the following exclusion criteria: (1) 3.5 mm, voxel size 5 2.75 3 2.75 mm2, number of volumes 5 180, diagnosis of FM and (2) exclusion criteria noted for patients. For all total scan time 5 360 seconds). Subjects were instructed to rest subjects, nonsteroidal anti-inflammatory drugs (NSAIDs) were and keep their eyes open during the 6-minute scan. allowed, as were Food and Drug Administration-approved Respiratory artifacts were corrected using the RETOICOR21 medications for fibromyalgia (pregabalin and duloxetine). How- algorithm, with respiratory belt data collected during the fMRI scan. ever, use of over-the-counter pain medications (NSAIDs) on the Head motion correction included linear realignment with FSL- day of MRI scan was excluded as was the use of narcotic pain MCFLIRT. Brain extraction was completed with FSL-BET, and medications 48 hours before MRI scan. Subjects were also cortical surface reconstruction was completed with T1-weighted excluded for routine daily use of narcotic analgesics, marijuana, anatomical images using FreeSurfer’s recon-all for coregistration or history of substance abuse. Subjects taking stimulant between anatomical and functional images using FreeSurfer’s medications, such as those used to treat ADD/ADHD (eg, bbregister tool. Spatial smoothing with 5 mm full-width-at-half- amphetamine/dextroamphetamine [Adderall], methylphenidate, maximum Gaussian kernel was applied to fMRI data, and further dextroamphetamine), or the fatigue associated with sleep apnea motion/cardiorespiratory artifact correction was completed with or shift work (eg, modafinil), were also excluded. Medication ICA-AROMA.43 High-pass filtering was completed for functional

connectivity analysis (fc 5 0.008 Hz). In addition, the 6 translation/ We assessed spectral power within regions of interest using rotation motion parameters from FSL-MCFLIRT and motion outlier fALFF,63 a measure of total power normalized ALFF. This metric confound matrices (FSL_motion_outliers) were regressed out of has been subdivided into 3 frequency bands12 (slow5 5 0.01- the fMRI data using generalized linear model (GLM) analysis for 0.027 Hz; slow4 5 0.027-0.073 Hz; and slow3 5 0.073-0.192 additional motion censoring. After preprocessing, we excluded 5 Hz), thought to relate to different functional process, with slow4 patients due to excessive head movement (over 1 mm displace- possibly linked with neuroinflammation.6 Computation of fALFF ment between successive image volumes). Thus, a total 38 slow4 was carried out using MATALB (The Mathworks, Inc., (N 5 38) FM patients and 16 HC (N 5 16) were included for Natick, MA). This metric was computed using the same resting- analysis. There were no differences in head motion (framewise state BOLD fMRI data as was used for connectivity analyses. The displacement root mean square) between FM and HC groups (P 5 fALFF metric assessed the amplitude of the fluctuating BOLD 0.26, two-sample Student t-test). fMRI signal within a specific frequency band. Preprocessing included motion censoring, which was performed using FSL_REGFIT in native space. Further preprocessing for fALFF 2.3. Seed-based functional connectivity analyses was similar to preprocessing for seed-based function The aIns seed was defined as gray matter voxels extracted from connectivity analyses. However, instead of frequency-based the larger 1H-MRS aIns voxel from each subject’s native space in temporal filtering, linear detrending was performed before fALFF conjunction with cortical surface reconstructions from the recon- calculation. After preprocessing, the fMRI data were fast-Fourier all Freesurfer tool. These ROIs in native space were registered to transformed, and the sum of amplitudes within the slow4 the Montreal Neurological Institute (MNI) space, and common frequency band was calculated and normalized by dividing by overlapping voxels across subjects was defined as the aIns seed total power. Averaged fALFF values for the gray matter portion of for functional connectivity and power analyses (Fig. 1A), re- the aIns 1H-MRS ROI were then calculated. gionally linked to 1H-MRS data. Preprocessed resting BOLD fMRI images in native space were also spatially normalized to standard 2.5. Clinical measures MNI space using FSL-FNIRT. Time-series averaged over the seed ROI were used as regressor in a GLM with the head motion Clinical pain was assessed with the short form of the Brief Pain nuisance regressors noted above (FSL-FEAT). Estimated param- Inventory15 (BPI), which consists of separate pain severity and eter weight maps and their variance were then passed up to pain interference subscales. Patients also completed the a group level GLM (FSL-FEAT, mixed-effect model). Fibromyalgia Impact Questionnaire11 (FIQ), whereas depression To link connectivity with clinical, imaging, and spectroscopy and anxiety severity were assessed using the Hospital Anxiety metrics, a multivariable regression model was performed with and Depression Scale61 (HADS). functional connectivity metrics and metabolites/behavioral measures, correcting for age and pregabalin usage (yes/no). Maps 2.6. Statistical analysis from this covariate analysis were threshold on the voxel level (z . 2.81) and cluster corrected for multiple comparisons (familywise Clinical assessment scores, aIns choline and myo-Inositol, which error rate, P , 0.01). were normally distributed (Lilliefors test, P . 0.05), were compared between FM and HC using an analysis of variance controlling for age and pregabalin status (yes/no). Subsequent 2.4. Fractional amplitude of low-frequency fluctuations analysis with functional connectivity metrics used z-statistics The fractional amplitude of low-frequency fluctuations of the averaged over a 4-mm diameter sphere centered on the peak resting fMRI signal,64 which is increased in multiple brain areas in voxel of the significant cluster. Correlations between 1H-MRS, chronic pain patients, has been recently suggested to play a role aIns functional connectivity, fALFF, and the clinical assessment in neuroinflammation and associated with calcium waves in scores noted above were calculated using a partial correlation activated astrocytes.6,26 Because low-frequency power of the analysis, controlling for age and pregabalin status. fMRI time series has been tangentially linked with glial processes, Mediation analysis was then performed with 3 modal variables we also included the fALFF metric in this study. using the R mediation package.54 We evaluated whether the

Figure 1. Metabolite levels in anterior insula (aIns). (A) 1H-MRS voxel placement for aIns (red mask represents overlapping voxels across all subjects). (B) Choline ratio to creatine was increased in FM compared to HC (P 5 0.02, red circle denotes mean). *Significant level of P , 0.05, age corrected (mean 1 residual), error bars denote SD. HC, health control.

direct association between dependent and independent varia- was maintained throughout the scans through a saphenous vein bles was significantly reduced by including a mediator variable, catheter. Heart rate, oxygen saturation, end-tidal CO2,and using nonparametric bootstrap estimation with 5000 iterations respiratory rate were monitored continuously. A heated water and random sampling. Confidence intervals (CIs) for a causal blanket was used to prevent hypothermia. At study endpoints, mediation effect, direct effect, and total effect were computed at animals were anesthetized with ketamine-HCl and sacrificed by the 95% confidence level (age and pregabalin status corrected). intravenous pentobarbital overdose. All statistical analyses were performed with the R Statistical All MRI/MRS scans were performed on a 3 T whole-body Package (R-Core-Team, 2012). imager (Magnetom TIM Trio; Siemens AG, Erlangen, Germany). Because neuroinflammation is known to be diffuse with this model, single voxel 1H MR spectroscopy was performed in an 2.7. Nonhuman primates and neuroinflammation model accessible cortical region, precuneus within the medial parietal To gain insights into the cellular contributions to the imaging cortex, using a point-resolved spectroscopy sequence (PRESS) changes observed in the FM patients, and in particular to further with water suppression enhanced through T1 effects, and the validate the status of choline and myo-inositol as neuroinflam- following parameters: voxel size 1.2 3 1.2 3 1.2 cm 5 1.7 cm3, matory markers, we directly linked these 1H-MRS metabolites TE 5 30 ms, TR 5 2500 ms, spectral bandwidth 5 1.2 kHz, 1024 with immunohistochemical measures in an animal model of complex data points, and 192 acquisitions resulting in an widespread viral neuroinflammation. Because no universally acquisition time of 8 minutes. All spectra were processed offline accepted animal model for FM exists, we used the Simian using the LCModel software package42 to determine the immunodeficiency virus (SIV) infection model, which reliably quantities of the brain metabolites choline and myo-inositol. produces extensive neuroinflammation. Evidence that HIV in- Absolute metabolite concentrations in institutional units were fection results in neuroinflammation originates predominantly derived from the same voxel using the water signal as reference. from patients with HIV-associated neurocognitive disorders and We could not report the metabolite ratio values (dividing by tCr) from animal models of NeuroAIDS.62 HIV-associated neuro- because tCr was significantly variable over the longitudinal cognitive disorder have been linked to altered 1H-MRS metrics, timeframe for H-MRS scans.47 To guide placement of the by our group and others,1,45 as well as in animal models of 1H-MRS volumes of interest, sagittal and axial turbo spin echo NeuroAIDS.49 Specifically, neuroinflammation with activated images were obtained using the following parameters: 140 3 astrocytes and microglia is often associated with elevated myo- 140 mm2 field of view, 512 3 512 matrix, echo time of 16 ms; inositol and choline containing compounds. slice thickness was 2 mm for sagittal images and 1.2 mm for axial The SIV model of NeuroAIDS was also specifically chosen images; TR was 4500 ms for sagittal and 7430 ms for axial because it exhibits dynamic changes in choline and myo-inositol images. during SIV progression, allowing for excellent sensitivity for At necropsy, animals were exsanguinated and perfused with correlating 1H-MRS markers of neuroinflammation with estab- 4L of chilled saline. A complete set of brain and peripheral tissues lished tissue and CSF markers. These associations for wide- were collected in 10% neutral buffered formalin, embedded in spread viral neuroinflammation may then be extended to better paraffin, and stained with hematoxylin and eosin. The degree of understand the role of 1H-MRS markers of neuroinflammation for reactive astrogliosis was assessed in 16 animals with the conditions that may not have well-established animal models, monoclonal antiglial fibrillary acidic protein (GFAP, 1:1000; such as fibromyalgia. Boehringer Mannheim, Indianapolis, IN). Paraffin sections Data from 19 adolescent/adult (4- to 5-year old) rhesus (5-mm thick) from the precuneus were immunolabeled overnight monkeys (Macaca mulatta) were included for analysis. Detailed with the monoclonal antibody followed by biotinylated horse methods for our model have been previously published.47,48 antimouse immunoglobulin G, avidin-horseradish peroxidase Briefly, all animals were infected with SIVmac251 virus. To (Vectastain Elite kit; Vector, Burlingame, CA), and reacted with accelerate AIDS progression, CD81 T lymphocytes were de- diaminobenzidine tetrahydrochloride and peroxide (0.03%). pleted using an anti-CD8 antibody that permits the virus to Levels of GFAP were estimated by means of computer-aided replicate unchecked, leading to neuroinflammation and SIV image analysis, as previously described.48 The units of all encephalitis in 85% of animals.58 Because we were interested measurements for GFAP are in arbitrary optical density units in linking neuroinflammation with MRS metabolites, dynamic and range from 0 to 500. range in neuroinflammation was experimentally increased. Microglial activation was assessed by quantifying ionized Specifically, 12 animals remained untreated and were sacrificed calcium binding adaptor protein 1 (IBA-1) in 8 animals. Parietal at 4 weeks (4 animals), 6 weeks (4 animals), and 8 weeks (4 cortical brain sections were incubated with rabbit anti-IBA-1 animals) after inoculation, respectively. Seven SIV-infected CD8- (Wako Corp., Nasushiobara, Tochigi, Japan). Images of tissue depleted animals were treated with an agent known to reduce sections were captured without manipulation using an Olympus neuroinflammation, minocycline, orally starting at 4 weeks after 3-CCD T60C color video camera mounted on an Olympus inoculation, and at daily dosage of 4 mg/kg b.i.d for 4 weeks. The Vanox-SI microscope and analyzed using NIH Image J software studies were approved by the Massachusetts General Hospital (see Ref. 48 for more detail). Subcommittee on Research Animal Care and by the Institutional In addition, CSF (#1.0 mL) was collected at 4 time points from Animal Care and Use Committee of Harvard University. All studies the cerebellomedullary cistern or lumbosacral junction from 8 were performed in accordance with federal laws and regulations, animals while they were anesthetized for MRI/MRS scans. CSF international accreditation standards, and institutional policies. samples were frozen immediately until assay. Levels of monocyte Magnetic resonance imaging and MRS was performed preino- chemoattractant protein-1 (MCP-1), a chemokine biomarker of culation and biweekly until sacrifice. During MRI sessions, each response to inflammation,16 was determined with ELISA (R and D animal was tranquilized with 15 to 20 mg/kg intramuscular ketamine Systems using the manufacturers protocol). hydrochloride and intubated to ensure a patent airway. Atropine All statistical analyses for nonhuman primate data were (0.4 mg/kg injected intravenously) was administered to prevent performed using the R Statistical Package. Associations bradycardia. A continuous infusion of propofol (0.25 mg/kg/min) between neuroinflammatory markers (GFAP, MCP-1, and

IBA-1) and both choline and myo-inositol levels were assessed and FIQ score (r 520.39, P 5 0.02). A whole brain analysis within with the non-parametric Spearman correlation coefficient, due HC found no significant correlation between aIns functional to nonnormal distribution of the data. connectivity and aIns choline/tCr levels. Mediation analysis demonstrated a significant mediation effect of aIns connectivity to putamen on the association between aIns 3. Results choline/tCr and pain interference score (mediation effect: b 5 3.1. Pain-related clinical assessments 15.78, CI: 2.64-35.72, P 5 0.02; total effect: b 5 28.72, CI: 11.55-47.59, P , 0.01) (Fig. 3). Thus, the direct effect of aIns For demographic and clinical assessments, FM patients reported choline on pain interference was significantly reduced (becoming significantly higher BPI pain severity, pain interference, as well as nonsignificant) by aIns connectivity to left putamen (b 5 12.94, CI: HADS depression and anxiety scores (all P , 0.001) compared to 27.4-30.06, P 5 0.22). HC (Table 1). Within the patient group, the FIQ score was highly correlated with BPI pain interference (r 5 0.76, P , 0.001), whereas there was no correlation between the FIQ and pain 3.4. Metabolite associations with blood-based and tissue severity scores (r 5 0.21, P 5 0.21). pathology markers in non-human primates The nonhuman primate neuroinflammation model demonstrated that GFAP in precuneus was positively correlated with the 3.2. 1H-MRS metabolite levels in anterior insula and pIns choline level change in this same brain region induced by SIV Choline levels in aIns were significantly higher in FM patients infection (r 5 0.49, P 5 0.03, Fig. 4A). Furthermore, MCP-1 compared to HC (Choline/tCr: P 5 0.03; trending for Choline/CSF: demonstrated a trending correlation with the change in choline P 5 0.09) (Table 2 and Fig. 1B), adjusted for age. Furthermore, level (r 5 0.32, P 5 0.07, Fig. 4B). By contrast, IBA-1 was not pain interference was correlated with aIns choline/tCr (r 5 0.41, P 5 correlated with choline level change in the same brain area (r 5 0.01), demonstrating that FM patients with high aIns choline also 0.34, P 5 0.27). Moreover, change in myo-inositol levels was not reported high pain interference. Hospital Anxiety and Depression correlated with GFAP (r 5 0.34, P 5 0.14), IBA-1(r 5 0.14, P 5 Scale depression and anxiety scores were not significantly 0.65), nor MCP-1 (r 5 0.10 P 5 0.57). correlated with aIns choline (Choline/tCr vs depression: r 5 0.14, P 5 0.42; Choline/tCr vs anxiety: 0.08, P 5 0.64). By contrast, aIns myo-inositol was not different between FM and HC (myo-inositol/ 3.5. Linking anterior insula fractional amplitude of low- frequency fluctuations with choline levels in FM patients tCr: P 5 0.41; myo-inositol/CSF: P 5 0.19) and there was no correlation between choline and myo-inositol in aIns (tCr: r 5 0.12, The slow4 band of fALFF was negatively correlated with choline/ P 5 0.48; CSF: r 5 0.01, P 5 0.56). Posterior insula showed no tCr levels in aIns (r 520.36, P 5 0.03, Fig. 5), and a trending significant metabolite level difference between FM and HC correlation was found for aIns myo-inositol/tCr levels (r 520.32, (Table 2). There were no outliers (Grubbs’ test) for aIns choline/ P 5 0.06). Furthermore, aIns fALFF slow4 was also inversely tCr, pain interference, pain severity, depression, or anxiety levels. correlated with HADS depression scores (r 520.37, P 5 0.03), We have also included results for concentrations of other relevant but not pain interference scores (r 520.13, P 5 0.44). metabolites, such as N-acetylaspartate (NAA), which has been Interestingly, in the left putamen (the brain region implicated in previously linked with neurodegeneration and/or neuroinflamma- aIns functional connectivity analyses above), fALFF slow4 power tion (Supplementary Table 2, available at http://links.lww.com/ was inversely correlated with pain interference scores (r 520.35, PAIN/A952). P 5 0.04) and HADS depression scores (r 520.44, P , 0.01). There was also trending negative correlation between putamen fALFF slow4 and aIns choline/tCr (r 520.32, P 5 0.06), and 3.3. Functional anterior insula connectivity putamen fALFF slow4 was correlated with aIns fALFF slow4 (r 5 Although resting aIns connectivity was not different between FM 0.46, P 5 0.004). and HC, within FM patients, aIns connectivity to left putamen (peak voxel MNI coordinates: 228, 10, 22 mm) was inversely correlated 4. Discussion with pain interference score (Fig. 2B). Thus, greater pain interference was associated with weaker connectivity between Neuroinflammation is an important target in many disease aIns and left putamen. Connectivity between aIns and left putamen pathologies including pain, and noninvasive measures of neuro- was also anticorrelated with aIns choline/tCr (r 520.37, P 5 0.03) inflammatory processes in humans could substantially aid

Table 1 Clinical and demographic data on the study subjects. Domains Group mean values (SE) Statistics FM (n 5 38) Control (n 5 16) FM vs control, P Age (y) 42.11 (11.67) 44.94 (16.29) 0.47 Symptom (FIQ) 50.18 (18.55) NA NA Pain Pain severity (BPI) 5.37 (1.50) 0.47 (0.83) P , 0.001* Pain interference (BPI) 5.06 (2.05) 0.15 (0.44) P , 0.001* Mood Depression (HADS) 7.26 (3.55) 0.81 (2.26) P , 0.001* Anxiety (HADS) 7.68 (4.26) 2.75 (2.74) P , 0.001* * Statistical significance. BPI, Brief Pain Inventory; FIQ, Fibromyalgia Impact Questionnaire; FM, fibromyalgia; HADS, Hospital Anxiety and Depression Scale.

Table 2 Metabolite differences between fibromyalgia patients and healthy control. Regions Metabolites Group mean values (SE) ANOVA statistics Fibromyalgia Control F P Anterior insula Choline:tCr 0.273 (0.006) 0.249 (0.010) 5.01 0.03* Choline:CSF 1.655 (0.038) 1.506 (0.065) 2.97 0.09 Myo-inositol:tCr 0.647 (0.017) 0.623 (0.029) 0.86 0.36 Myo-inositol:CSF 4.804 (0.149) 4.458 (0.255) 1.38 0.25 Posterior insula Choline:tCr 0.248 (0.007) 0.261 (0.012) 0.13 0.72 Choline:CSF 1.432 (0.044) 1.151 (0.076) 0.47 0.50 Myo-inositol:tCr 0.645 (0.017) 0.638 (0.029) 0.24 0.63 Myo-inositol:CSF 4.089 (0.113) 3.986 (0.193) 0.24 0.63 Age corrected in ANOVA. * P , 0.05. ANOVA, analysis of variance; CSF, cerebrospinal fluid; Choline:tCr, choline ratio to total creatine; Choline:CSF, CSF adjusted choline concentration; Myo-inositol:tCr, myo-inositol ratio to total creatine; Myo-inositol:CSF, CSF adjusted myo-inositol concentration.

diagnosis and therapeutic development. Magnetic resonance turnover.13,22 In previous studies, FM patients showed higher imaging–based metrics have been suggested in the past, and our choline in right dorsolateral prefrontal cortex compared with study assessed both 1H-MRS metabolites (ie, choline and myo- healthy controls, and choline was also correlated with pain inositol) and resting fMRI-based metrics previously linked with levels.40 Urologic chronic pelvic pain syndrome patients have glial activity. Our study found elevated choline (but not myo- demonstrated elevated choline levels in anterior cingulate inositol) levels in aIns for FM compared to healthy adults, with cortex,24 whereas other chronic pain disorders, including greater choline levels linked with worse pain interference. In temporomandibular disorder, have also showed higher choline addition, reduced connectivity between aIns and putamen was in pIns cortex.23 However, we should note that increases in associated with both elevated aIns choline and pain interference. choline are commonly seen with diseases that have alterations in In fact, aIns/putamen connectivity mediated the link between aIns membrane turnover9 in addition to inflammatory and gliotic choline and pain interference. processes.34,35 Interestingly, we did not find altered myo-inositol Choline levels, assessed by 1H-MRS, have been associated levels in FM suggesting that this marker may not be as important with neuroinflammation because elevated choline levels have as choline in tracking neuroinflammation in FM. been linked to glial activation or proliferation leading to abnormal How neuroinflammation is linked with pain and pain-related phospholipid metabolism and accelerated cell membrane morbidity also needs more clarification. Our mediation analysis

Figure 2. Seed-based functional connectivity for the 1H-MRS voxel placed in right anterior insula (aIns) in FM patients. (A) A group map of aIns connectivity in FM patients (corrected P , 0.01) demonstrated expected connectivity to other salience network regions such as midcingulate cortex, as well as posterior insula, visual cortex, basal ganglia, and cerebellum. (B) Correlation between aIns connectivity and pain interference score yielded a cluster in left putamen, which was negatively correlated with pain interference (age and pregabalin status corrected).

Figure 3. Mediation modeling found that aIns functional connectivity to left putamen mediates the association between aIns choline and pain interference. For mediation modeling, (A) the correlation between aIns choline/tCr and pain interference (total effect) and (B) the correlation between aIns choline/tCr and aIns connectivity to left putamen (path a) were both significant (P , 0.05). (C) A significant mediation effect of aIns connectivity to putamen was noted (CME: b 5 15.78, CI: 2.64-35.72, P 5 0.02; proportions mediated: b 5 0.55, CI: 2.11-1.37, P 5 0.02). The model corrected for the effects of age and pregabalin status (mean 1 residual). aIns, anterior insula.

revealed that although aIns choline was associated with pain globus pallidus, and substantia nigra, as well as somatosensory interference, this linkage was in fact mediated by functional aIns and motor areas. Such connections strongly influence the connectivity to the putamen. Pain interference is a function of planning, learning, preparation, and execution of motor behav- motivation and reduced bodily activity rather than the sensory iors.5,32 Indeed, FM patients have shown prior evidence of aspect of clinical pain itself. The putamen is closely intercon- disruptions in putamen structure and signaling, such as reduced nected with other basal ganglia regions such as caudate nucleus, gray matter volume and functional connectivity with insula

Figure 4. In the nonhuman primate neuroinflammation model, postinfection choline level change in precuneus was correlated with (A) GFAP (Spearman r 5 0.49, P 5 0.03) and (B) trending for MCP-1 (Spearman r 5 0.32, P 5 0.07). (C) Medial parietal cortex (precuneus) voxel placement for the nonhuman primate model. GFAP, glial fibrillary acidic protein.

Figure 5. Fractional amplitude of low-frequency fluctuations (fALFF) slow4 correlations with choline and HADS depression score in FM patients. fALFF slow4 in aIns was inversely correlated with (A) aIns choline ratio to total creatine and (B) HADS depression scores. (C) fALFF slow4 in left putamen was also inversely correlated with HADS depression scores. Data are adjusted for age and pregabalin status. aIns, anterior insula; HADS, Hospital Anxiety and Depression Scale.

cortex.14,51 Other FM studies have found reduced cerebral blood degenerative changes in the neuronal population.17 Interestingly, flow in putamen, which was correlated with pain disability,52 and our primate model demonstrated no linkage between choline reduced putamen response to a motor inhibition task,20 levels and IBA-1, which is an immunohistochemical marker suggesting that attenuation of putamen activity in the motor involved in membrane ruffing of microglia—an essential morpho- initialization process. The results of these previous studies are logical change for microglial activation.27 Thus, our results supported by our findings linking neuroinflammation and reduced suggest that 1H-MRS-assessed choline levels do indeed track aIns/putamen connectivity in FM with pain interference, a mea- with neuroinflammation and are more likely associated with sure that compared to pain intensity is more closely related to astrogliosis than microglial activation. motor activity in these patients. Furthermore, the nonhuman primate results found a positive In fact, our study also found a relationship between fALFF correlation between choline and both GFAP and MCP-1, but not slow4, another putative MRI marker for neuroinflammation, and between myo-inositol and these inflammatory markers. The closer choline levels in aIns. However, these 2 markers were inversely link between choline and neuroinflammation is also supported by related—greater choline was linked with reduced fALFF slow4 our recent analysis showing that choline but not myo-inositol power. Furthermore, reduced slow4 power was also correlated correlated with 1H-MRS-assessed brain temperature in the same with depression, corroborating previous studies.30,53 The con- regions.36 Although myo-inositol has been implicated in other nection between fALFF slow4 and neuroinflammation is based on studies as a potential neuroinflammatory marker, these data do not studies suggesting that astrocytic Ca21 oscillations occur in both support its role in either FM neuroinflammation or SIV-induced the slow4 and slow5 frequency range in the thalamus.38,39 In fact, neuroinflammatory processes. a microglial mediator has also been linked with Ca21 oscillations, Several limitations of our study should be noted. For instance, and modulates synaptic neurotransmission.28 Moreover, astro- previous studies have noted that anxiety and depression can cytic Ca21 waves have also been shown to regulate blood flow also impact 1H-MRS metabolites.24,33 In our study, although through a gliovascular interaction,8,41 leading to BOLD fMRI anxiety and depression demonstrated marked differences signal changes.56 Such previous studies support the linkage between FM and HC, we did not find any association between between fALFF slow4 power and astrocytic Ca21 oscillations. aIns choline and mood measures, suggesting that choline levels However, the inverse correlation in our study between slow4 were more associated with pain-related neuroinflammatory power and choline was surprising and needs further clarification, processes. In addition, alcohol consumption, which can also which might come from a clarification of the role for choline in impact 1H-MRS metabolites in the brain,55,60 was not consid- astrocyte and microglia activity. These results suggest that ered in the exclusion criteria. Future studies should further although neuroinflammation may be linked with both choline and consider alcohol consumption in exclusion criteria and as fALFF slow4, the specific pathways reflected by these 2 metrics regressors in the general linear models being tested. Also, our may be distinct, and may be due to different phases of the waxing ROI for 1H-MRS was limited to aIns and metabolite levels from and waning neuroinflammatory state in FM. the putamen would, in retrospect, have shed more light on the To better understand the mechanisms supporting altered mechanisms by which neuroinflammation impacts pain varia- choline or myo-inositol levels in neuroinflammation, we directly bles in FM. However, our mediation analysis did reveal the linked levels of these metabolites with blood-based and tissue importance of functional connectivity in mediating the link pathology markers for neuroinflammation (ie, astrogliosis and between choline levels and pain interference. Finally, our microglial activation). We used data collected from a previously widespread viral neuroinflammation nonhuman primate model reported nonhuman primate model of neuroinflammation caused was not specific to fibromyalgia or chronic pain, and mecha- by SIV.48 Our nonhuman primate model demonstrated that 1H- nisms underlying neuroinflammation in viral SIV (ie, macrophage MRS choline levels were linked with GFAP and monocyte infiltration leading to glial activation) may differ from those that chemoattractant protein 1 (MCP-1). GFAP is the principal underlie neuroinflammation in fibromyalgia. However, we intermediate filament in mature astrocytes. Astrocytes rapidly contend that the association between choline and astrogliosis synthesize GFAP in response to a neurologic insult, and GFAP is tissue markers (ie, GFAP) would generalize across neuro- commonly used as marker of astrogliosis.19 MCP-1 is a chemo- inflammatory states irrespective of mechanistic pathways. kine that plays a role in the recruitment of monocytes from the In summary, our results suggest that neuroinflammation can be blood into the brain and is involved in the activation processes of assessed noninvasively by 1H-MRS and resting BOLD fMRI- inflammatory and neural cells that lead to development of based metrics. Our nonhuman primate data suggest that choline,

which was an important metabolite in differentiating FM from astrocytes: studies by multinuclear magnetic resonance spectroscopy. healthy adults and linked with pain interference, may be Neurochem Res 1997;22:123–31. [11] Burckhardt CS, Clark SR, Bennett RM. The fibromyalgia impact specifically associated with astrogliosis. questionnaire: development and validation. J Rheumatol 1991;18: 728–33. [12] Buzsaki ´ G, Draguhn A. Neuronal oscillations in cortical networks. Science Conflict of interest statement 2004;304:1926–9. The authors have no conflicts of interest to declare. [13] Chang L, Munsaka SM, Kraft-Terry S, Ernst T. Magnetic resonance spectroscopy to assess neuroinflammation and neuropathic pain. J Neuroimmune Pharmacol 2013;8:576–93. Acknowledgments [14] Cifre I, Sitges C, Fraiman D, Mun˜ oz M, Balenzuela P, Gonzalez-Rold ´ an ´ A, Martınez-Jauand´ M, Birbaumer N, Chialvo DR, Montoya P. Disrupted The authors thank Eliezer Masliah, PhD, for support in analysis of functional connectivity of the pain network in fibromyalgia. Psychosom nonhuman primate GFAP data. The authors also acknowledge Med 2012;74:55–62. [15] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain the following organizations for funding support: US National Inventory. Ann Acad Med Singapore 1994;23:129–38. Institutes for Health (NIH), Office of the Director (OT2-OD023867 [16] Deshmane SL, Kremlev S, Amini S, Sawaya BE. Monocyte to V.N.); National Center for Complementary and Integrative chemoattractant protein-1 (MCP-1): an overview. J Interferon Cytokine Health (NCCIH), NIH (P01-AT009965, R61-AT009306, Res 2009;29:313–26. [17] Dhillon NK, Williams R, Callen S, Zien C, Narayan O, Buch S. Roles of R33-AT009306 to V.N., R01-AT007550 to V.N. and R.E.H.); MCP-1 in development of HIV-dementia. Front Biosci 2008;13:3913–18. National Institute for Arthritis and Musculoskeletal and Skin Diseases [18] Emmer BJ, van der Bijl AE, Huizinga TW, Breedveld FC, Steens SC, Th (NIAMS), NIH (R01-AR064367 to V.N.); the National Institute for Bosma GP, van Buchem MA, van der Grond J. Brain involvement in Neurological Disorders and Stroke (R01-NS094306-01A1 and rheumatoid arthritis: a magnetic resonance spectroscopy study. Arthritis R01-NS095937-01A1 to M.L.L.); NIH (R21-NS059331 to EMR); NIH Rheum 2009;60:3190–5. [19] Eng LF, Ghirnikar RS, Lee YL. Glial fibrillary acidic protein: GFAP-thirty- (R01-NS050041 to RGG); USPHS (P51RR00168 to New England one years (1969-2000). Neurochem Res 2000;25:1439–51. Primate Research Center); KIOM (K18051 to V.N.), and The [20] Glass JM, Williams DA, Fernandez-Sanchez ML, Kairys A, Barjola P, Department of Defense (DoD-W81XWH-14-1-0543 to M.L.L.). Heitzeg MM, Clauw DJ, Schmidt-Wilcke T. Executive function in chronic pain patients and healthy controls: different cortical activation during response inhibition in fibromyalgia. J Pain 2011;12:1219–29. Appendix A. Supplemental digital content [21] Glover GH, Li TQ, Ress D. Image-based method for retrospective correction of physiological motion effects in fMRI: RETROICOR. Magn Supplemental digital content associated with this article can be Reson Med 2000;44:162–7. found online at http://links.lww.com/PAIN/A952. [22] Glunde K, Jie C, Bhujwalla ZM. Molecular causes of the aberrant choline phospholipid metabolism in breast cancer. Cancer Res 2004;64:4270–6. [23] Harfeldt K, Alexander L, Lam J, Mansson ˚ S, Westergren H, Svensson P, Article history: Sundgren PC, Alstergren P. Spectroscopic differences in posterior insula Received 17 October 2019 in patients with chronic temporomandibular pain. Scand J Pain 2018;18: Received in revised form 18 December 2019 351–61. Accepted 23 December 2019 [24] Harper DE, Ichesco E, Schrepf A, Halvorson M, Puiu T, Clauw DJ, Harris Available online 17 February 2020 RE, Harte SE; Network MR. Relationships between brain metabolite levels, functional connectivity, and negative mood in urologic chronic pelvic pain syndrome patients compared to controls: a MAPP research network study. Neuroimage Clin 2018;17:570–8. References [25] Harris RE, Napadow V, Huggins JP, Pauer L, Kim J, Hampson J, [1] Alakkas A, Ellis RJ, Watson CW, Umlauf A, Heaton RK, Letendre S, Collier Sundgren PC, Foerster B, Petrou M, Schmidt-Wilcke T, Clauw DJ. A, Marra C, Clifford DB, Gelman B, Sacktor N, Morgello S, Simpson D, Pregabalin rectifies aberrant brain chemistry, connectivity, and functional McCutchan JA, Kallianpur A, Gianella S, Marcotte T, Grant I, Fennema- response in chronic pain patients. Anesthesiology 2013;119:1453–64. Notestine C; Group C. White matter damage, neuroinflammation, and [26] Hodkinson DJ, Wilcox SL, Veggeberg R, Noseda R, Burstein R, Borsook neuronal integrity in HAND. J Neurovirol 2019;25:32–41. D, Becerra L. Increased amplitude of thalamocortical low-frequency [2] Albrecht DS, Ahmed SU, Kettner NW, Borra RJH, Cohen-Adad J, Deng oscillations in patients with migraine. J Neurosci 2016;36:8026–36. H, Houle TT, Opalacz A, Roth SA, Melo MFV, Chen L, Mao J, Hooker JM, [27] Imai Y, Kohsaka S. Intracellular signaling in M-CSF-induced microglia Loggia ML, Zhang Y. Neuroinflammation of the spinal cord and nerve activation: role of Iba1. Glia 2002;40:164–74. roots in chronic radicular pain patients. PAIN 2018;159:968–77. [28] Ji RR, Berta T, Nedergaard M. Glia and pain: is chronic pain a gliopathy? [3] Albrecht DS, Granziera C, Hooker JM, Loggia ML. In vivo imaging of PAIN 2013;154(suppl 1):S10–28. human neuroinflammation. ACS Chem Neurosci 2016;7:470–83. [29] Ji RR, Chamessian A, Zhang YQ. Pain regulation by non-neuronal cells [4] Albrecht DS, Normandin MD, Shcherbinin S, Wooten DW, Schwarz AJ, and inflammation. Science 2016;354:572–7. Zurcher NR, Barth VN, Guehl NJ, Akeju O, Atassi N, Veronese M, Turkheimer [30] Liu CH, Ma X, Song LP, Tang LR, Jing B, Zhang Y, Li F, Zhou Z, Fan J, F, Hooker JM, Loggia ML. Pseudoreference regions for glial imaging with (11) Wang CY. Alteration of spontaneous neuronal activity within the salience C-PBR28: investigation in 2 clinical cohorts. J Nucl Med 2018;59:107–14. network in partially remitted depression. Brain Res 2015;1599:93–102. [5] Alexander GE, Crutcher MD. Preparation for movement: neural [31] Loggia ML, Chonde DB, Akeju O, Arabasz G, Catana C, Edwards RR, Hill representations of intended direction in three motor areas of the E, Hsu S, Izquierdo-Garcia D, Ji RR, Riley M, Wasan AD, Zurcher NR, monkey. J Neurophysiol 1990;64:133–50. Albrecht DS, Vangel MG, Rosen BR, Napadow V, Hooker JM. Evidence [6] Alshelh Z, Di Pietro F, Youssef AM, Reeves JM, Macey PM, Vickers ER, for brain glial activation in chronic pain patients. Brain 2015;138:604–15. Peck CC, Murray GM, Henderson LA. Chronic neuropathic pain: it’s [32] Marchand WR, Lee JN, Thatcher JW, Hsu EW, Rashkin E, Suchy Y, about the rhythm. J Neurosci 2016;36:1008–18. Chelune G, Starr J, Barbera SS. Putamen coactivation during motor task [7] As-Sanie S, Kim J, Schmidt-Wilcke T, Sundgren PC, Clauw DJ, Napadow execution. Neuroreport 2008;19:957–60. V, Harris RE. Functional connectivity is associated with altered brain [33] Mathias LK, Monette PJ, Harper DG, Forester BP. Application of chemistry in women with endometriosis-associated chronic pelvic pain. magnetic resonance spectroscopy in geriatric mood disorders. Int Rev J Pain 2016;17:1–13. Psychiatry 2017;29:597–617. [8] Attwell D, Buchan AM, Charpak S, Lauritzen M, Macvicar BA, Newman EA. [34] Menon DK, Ainsworth JG, Cox IJ, Coker RC, Sargentoni J, Coutts GA, Glial and neuronal control of brain blood flow. Nature 2010;468:232–43. Baudouin CJ, Kocsis AE, Harris JR. Proton MR spectroscopy of the brain [9] Bendszus M, Warmuth-Metz M, Klein R, Burger R, Schichor C, Tonn JC, in AIDS dementia complex. J Comput Assist Tomogr 1992;16:538–42. Solymosi L. MR spectroscopy in gliomatosis cerebri. AJNR Am J [35] Meyerhoff DJ, Bloomer C, Schuff N, Ezekiel F, Norman D, Clark W, Neuroradiol 2000;21:375–80. Weiner MW, Fein G. Cortical metabolite alterations in abstinent cocaine [10] Brand A, Leibfritz D, Wolburg H, Richter-Landsberg C. Interactions of and cocaine/alcohol-dependent subjects: proton magnetic resonance triethyltin-chloride (TET) with the energy metabolism of cultured rat brain spectroscopic imaging. Addict Biol 1999;4:405–19.

[36] Mintzopoulos D, Ratai EM, He J, Gonzalez RG, Kaufman MJ. Simian [50] Ross BD. Biochemical considerations in 1H spectroscopy. Glutamate immunodeficiency virus transiently increases brain temperature in rhesus and glutamine; myo-inositol and related metabolites. NMR Biomed 1991; monkeys: detection with magnetic resonance spectroscopy 4:59–63. thermometry. Magn Reson Med 2019;81:2896–904. [51] Sawaddiruk P, Paiboonworachat S, Chattipakorn N, Chattipakorn SC. [37] Napadow V, LaCount L, Park K, As-Sanie S, Clauw DJ, Harris RE. Alterations of brain activity in fibromyalgia patients. J Clin Neurosci 2017; Intrinsic brain connectivity in fibromyalgia is associated with chronic pain 38:13–22. intensity. Arthritis Rheum 2010;62:2545–55. [52] Shokouhi M, Davis KD, Moulin DE, Morley-Forster P, Nielson WR, Bureau [38] Parri HR, Crunelli V. Pacemaker calcium oscillations in thalamic Y, St Lawrence K. Basal ganglia perfusion in fibromyalgia is related to pain astrocytes in situ. Neuroreport 2001;12:3897–900. disability and disease impact: an arterial spin labeling study. Clin J Pain [39] Parri HR, Gould TM, Crunelli V. Spontaneous astrocytic Ca21 oscillations in 2016;32:495–505. situ drive NMDAR-mediated neuronal excitation. Nat Neurosci 2001;4:803–12. [53] Tadayonnejad R, Yang S, Kumar A, Ajilore O. Clinical, cognitive, and [40] Petrou M, Harris RE, Foerster BR, McLean SA, Sen A, Clauw DJ, functional connectivity correlations of resting-state intrinsic brain activity Sundgren PC. Proton MR spectroscopy in the evaluation of cerebral alterations in unmedicated depression. J Affect Disord 2015;172: metabolism in patients with fibromyalgia: comparison with healthy 241–50. controls and correlation with symptom severity. AJNR Am J [54] Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. Mediation: R package Neuroradiol 2008;29:913–18. for causal mediation analysis, 2017. [41] Petzold GC, Murthy VN. Role of astrocytes in neurovascular coupling. [55] Tunc-Skarka N, Weber-Fahr W, Ende G. Recreational alcohol use Neuron 2011;71:782–97. induces changes in the concentrations of choline-containing compounds [42] Provencher SW. Automatic quantitation of localized in vivo 1H spectra and total creatine in the brain: a (1)H MRS study of healthy subjects. with LCModel. NMR Biomed 2001;14:260–4. MAGMA 2015;28:503–10. [43] Pruim RHR, Mennes M, van Rooij D, Llera A, Buitelaar JK, Beckmann CF. [56] Wang M, He Y, Sejnowski TJ, Yu X. Brain-state dependent astrocytic Ca. ICA-AROMA: a robust ICA-based strategy for removing motion artifacts Proc Natl Acad Sci U S A 2018;115:E1647–56. from fMRI data. Neuroimage 2015;112:267–77. [57] Widerstrom-Noga E, Pattany PM, Cruz-Almeida Y, Felix ER, Perez S, [44] Rae CD. A guide to the metabolic pathways and function of metabolites Cardenas DD, Martinez-Arizala A. Metabolite concentrations in the observed in human brain 1H magnetic resonance spectra. Neurochem anterior cingulate cortex predict high neuropathic pain impact after spinal Res 2014;39:1–36. cord injury. PAIN 2013;154:204–12. [45] Ratai E. Magnetic Resonance Spectroscopy in HIV-Associated [58] Williams K, Westmoreland S, Greco J, Ratai E, Lentz M, Kim WK, Fuller Neurocognitive Disorders: HAND. Cham, Switzerland: Springer, 2016. RA, Kim JP, Autissier P, Sehgal PK, Schinazi RF, Bischofberger N, Piatak [46] Ratai EM, Alshikho MJ, Zurcher NR, Loggia ML, Cebulla CL, Cernasov P, M, Lifson JD, Masliah E, Gonzalez RG. Magnetic resonance Reynolds B, Fish J, Seth R, Babu S, Paganoni S, Hooker JM, Atassi N. spectroscopy reveals that activated monocytes contribute to neuronal Integrated imaging of [(11)C]-PBR28 PET, MR diffusion and magnetic injury in SIV neuroAIDS. J Clin Invest 2005;115:2534–45. resonance spectroscopy (1)H-MRS in amyotrophic lateral sclerosis. [59] Wolfe F, Hauser ¨ W. Fibromyalgia diagnosis and diagnostic criteria. Ann NeuroImage Clin 2018;20:357–64. Med 2011;43:495–502. [47] Ratai EM, Annamalai L, Burdo T, Joo CG, Bombardier JP, Fell R, [60] Zahr NM, Carr RA, Rohlfing T, Mayer D, Sullivan EV, Colrain IM, Hakimelahi R, He J, Lentz MR, Campbell J, Curran E, Halpern EF, Masliah Pfefferbaum A. Brain metabolite levels in recently sober individuals with E, Westmoreland SV, Williams KC, Gonzalez RG. Brain creatine elevation alcohol use disorder: relation to drinking variables and relapse. Psychiatry and N-Acetylaspartate reduction indicates neuronal dysfunction in the Res Neuroimaging 2016;250:42–9. setting of enhanced glial energy metabolism in a macaque model of [61] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta neuroAIDS. Magn Reson Med 2011;66:625–34. Psychiatr Scand 1983;67:361–70. [48] RataiEM,BombardierJP,JooCG,AnnamalaiL,BurdoTH,CampbellJ,Fell [62] Zink MC, Uhrlaub J, DeWitt J, Voelker T, Bullock B, Mankowski J, R, Hakimelahi R, He J, Autissier P, Lentz MR, Halpern EF, Masliah E, Williams Tarwater P, Clements J, Barber S. Neuroprotective and anti-human KC, Westmoreland SV, Gonzalez RG. Proton magnetic resonance immunodeficiency virus activity of minocycline. JAMA 2005;293: spectroscopy reveals neuroprotection by oral minocycline in a nonhuman 2003–11. primate model of accelerated NeuroAIDS. PLoS One 2010;5:e10523. [63] Zou QH, Zhu CZ, Yang Y, Zuo XN, Long XY, Cao QJ, Wang YF, Zang YF. [49] Ratai EM, Pilkenton S, He J, Fell R, Bombardier JP, Joo CG, Lentz MR, An improved approach to detection of amplitude of low-frequency Kim WK, Burdo TH, Autissier P, Annamalai L, Curran E, O’Neil SP, fluctuation (ALFF) for resting-state fMRI: fractional ALFF. J Neurosci Westmoreland SV, Williams KC, Masliah E, Gilberto Gonzalez ´ R. Methods 2008;172:137–41. CD81 lymphocyte depletion without SIV infection does not produce [64] Zuo XN, Kelly C, Adelstein JS, Klein DF, Castellanos FX, Milham MP. metabolic changes or pathological abnormalities in the rhesus macaque Reliable intrinsic connectivity networks: test-retest evaluation using ICA brain. J Med Primatol 2011;40:300–9. and dual regression approach. Neuroimage 2009;49:2163–77.