Metabolite activity in the anterior cingulate cortex during a painful stimulus using functional MRS

To understand neurochemical brain responses to pain, proton magnetic resonance spectroscopy (1H-MRS) is used in humans in vivo to examine various metabolites. Recent MRS investigations have adopted a functional approach, where acquisitions of MRS are performed over time to track task-related changes. Previous studies suggest glutamate is of primary interest, as it may play a role during cortical processing of noxious stimuli. The objective of this study was to examine the metabolic effect (i.e., glutamate) in the anterior cingulate cortex during noxious stimulation using fMRS. The analysis addressed changes in glutamate and glutamate + glutamine (Glx) associated with the onset of pain, and the degree by which fluctuations in metabolites corresponded with continuous pain outcomes. Results suggest healthy participants undergoing tonic noxious stimulation demonstrated increased concentrations of glutamate and Glx at the onset of pain. Subsequent reports of pain were not accompanied by corresponding changes in glutamate of Glx concentrations. An exploratory analysis on sex revealed large effect size changes in glutamate at pain onset in female participants, compared with medium-sized effects in male participants. We propose a role for glutamate in the ACC related to the detection of a noxious stimulus.


Methods
Subjects. Eighteen healthy participants (9 F/9 M, mean age = 26.28, SD = 3.68, range = 21-36 years) were recruited to our study. Each provided informed consent and completed a general health questionnaire. The experimental protocol was explained to each participant in detail before testing. All procedures conformed to the Declaration of Helsinki and were approved by the Research Ethics Board of the University of British Columbia.

Pain stimulation paradigm.
To reduce novelty effects associated with our model of experimental pain in the scanner subjects underwent a familiarization period one week prior to acquiring MR data. The same experimental pain procedures (outlined below) performed during MRS acquisition were performed in a laboratory setting.
The pain model was developed in preliminary testing and consisted of the application of 0.075% topical capsaicin and heat activation via thermo-pad on the volar surface of the right forearm (~ 8 × 5 cm). Capsaicin was applied immediately after baseline acquisition of MRS alongside a neutral thermo-pad without removing participants from the scanner. The thermo-pad was fixed to the skin and covered the area where capsaicin was applied, and activated by the influx of hot water via plastic tubing conveyed via the penetration panel. A nonheat conducting flexible brace was used to fasten the position of the thermo-pad (Fig. 3). The functional MRS followed by acquiring 16 shots every 1.08 min (total = 22.4 min). After 9 min, the thermo-pad was activated by circulating heated water to reach a temperature of approximately 41 °C at the forearm. Capsaicin was heatactivated for a period of 4.4 min ("heat"). Functional MRS was acquired continuously during heat application and for the reminder of the scan after the heat was removed (i.e., "post heat").
Pain intensity ratings. Participants were visually prompted for a pain intensity rating (using the 0-10 numeric rating scale ) every 2 min during the functional MRS scan. Participants indicated an increase, decrease, or no change in perceived pain via an MRI compatible clicker. Feedback was provided by Presentation software (www.neuro bs.com/prese ntati on), allowing for the collection of pain ratings throughout the fMRS acquisition while avoiding verbal communication with subjects and head motion. Before commencing with the experimental procedure, participants were familiarized with the rating device.
Data pre-processing and quantification of metabolites. 3D T1 data was segmented into white matter, gray matter and cerebrospinal fluid using FSL BET and FAST 54 . Individual FIDs were pre-processed (eddy current correction, phase correction, frequency alignment, and visual inspection-for quality assurance) in MAT-LAB (R2016b) using in-house code. 32 shots were averaged for each 2-min block, yielding a total of 11 spectra analyzed using LCModel (v6.3-1H) (Fig. 1B). The simulated basis set was obtained from Steven Provencher specific for the echo time and field strength of this study. LCModel fitting was optimized between 0.2 and 4.2 ppm. The corresponding interleaved non-water suppressed spectra and each participant's brain tissue water volumes were used to calculate the concentration of NAA, tCr, glutamate and Glx in millimolar (mM) 53   www.nature.com/scientificreports/ water FID's were fitted to a single-exponential decay curve to extrapolate the water amplitude and the FID decay rate (T2*). This analysis was performed to assure the water amplitude and decay constants were unaffected by BOLD effects.
Statistical analysis. As a first step, a paired t-test was performed to determine changes in NAA, tCr, glutamate and Glx at the onset of pain. Pain onset was defined as a NRS score ≥ 2 and was compared to the preceding concentration. To assess the specificity of glutamate and Glx, changes in NAA and tCr concentrations were also examined. Water amplitude and T2* stability during rest and pain onset were examined via a paired t-test and effect sizes were calculated. A Pearson correlation was performed between the change in glutamate and Glx concentrations (rest vs pain-onset) and the corresponding pain ratings. An exploratory analysis examined the effects of sex with paired t-tests. Effect sizes were calculated using Cohen's d 58

Results
One participant (f) that underwent the familiarization phase was unable to be scanned due to MR incompatibilities (metallic implant). Data from two participants (1 m; 1 f) was rejected due to excessive motion in the scanner. For the fifteen remaining subjects, SNR was adequate (19.54, 95% CI 18.24-20.84) and linewidth and mean error estimate of glutamate fit were low (3.80 Hz, 95% CI 3.20-4.42; 4%, 95% CI 3.99-4.00, respectively). Tissue measures were individually estimated for metabolite quantification (Across subjects (mean, SD): WM = 0.27 ± 0.06 range = 0.20-0.45; GM = 0.59 ± 0.04 range = 0.50-0.66; CSF = 0.12 ± 0.05 range = 0.04-0.23). In two subjects, 1 MRS acquisition time point had to be excluded due to excessive motion, as evidenced by a large shot to shot variations in the NAA peak frequency location and linewidth. Figure 4 represents baseline-subtracted spectra quality parameters for the 3 different conditions within the functional scan to illustrate the quality metrics remained the constant throughout testing. Table 1 (Fig. 6). At pain onset, the average NRS pain score was 2.9 ± 0.9 across all subjects, and was not significantly different for females (3.0 ± 0.8) compared to males (

Discussion
In the current study, the relationship between glutamate and Glx in the ACC and perception of noxious heat stimulation was examined using fMRS. Medium to large effect sizes were detected for increases in glutamate, Glx, and tCr at the transition to pain (i.e., pain onset, when NRS pain score ≥ 2 for the first time). NAA, but not creatine remained stable at the onset of pain. There was no overall relationship between pain ratings and glutamate or Glx concentrations, indicating the transient nature of increased concentrations. Our observations suggest a role for glutamate in the ACC in initially detecting but not tracking pain.
Glutamate and Glx concentrations changes in the ACC . Evidence of regional activation in the ACC in response to noxious stimulation dates back to seminal studies applying fMRI and PET 27,59-61 . These are further supported by observations from single neuron recordings, which first raised the notion that a region of the ACC was selective for pain 62,63 . Subsequent investigations have demonstrated a more complex role, with the ACC, critically involved in "salience detection" 64,65 and affective dimensions of pain 59 . On the basis that the ACC represents a prominent area of the "neurological signature of pain" responsive to analgesic interventions 11,66,67 , it remains a high priority area of study for pain researchers. Five studies have previously investigated changes in glutamate and Glx in the ACC during the presentation of a noxious stimulus 16,21,[47][48][49] , of which two have observed an increase in Glx 16,21 and glutamate 16 . Similar to Mullins et.al, we applied a long-duration noxious stimulus and observed a similar percent increase in glutamate compared to rest (+ 8.3% compared to Mullins et al. + 9.4%). To our knowledge, ours is the first study to distinguish the onset of pain from tonic appraisal. Based on our results, glutamatergic activation turns "on" independent of the intensity and "off " in the presence of a stimulus that remains painful. The lack of prolonged increases in glutamate concentrations contrasts that overserved for fMRI in various brain regions, including the mid-ACC, where BOLD remains elevated for the duration of the painful stimulation 68 . In agreement with our observations, a transient increase in extracellular glutamate in the ACC 69 and elsewhere in the CNS (e.g., spinal cord 25,28,[70][71][72][73][74][75] ) has been reported in animal models using microdialysis following noxious stimulation induced by way of formalin injection. Critically, in response to formalin, the second wave of pain behaviour is not matched with an increase in extracellular glutamate, as is observed for the initial wave of pain behaviour. Collectively, this points to the glutamatergic response to noxious stimulation, captured by way of fMRS, representing one of the numerous "other" roles of the ACC in appraising sensory stimuli 64,65,76,77 .
Mechanisms underlying the tendency for glutamate to increase in response to noxious stimulation are, at this time, poorly understood. Glutamatergic synapses play a role in sensory transmission, including pain 25,26,28,[78][79][80] . In the ACC, all major forms of ionotropic and metabotropic glutamate receptors have been isolated 81,82 . Metabolically, glutamate is stored as glutamine in astrocytes, and the balanced cycling between these two neurochemical is essential for normal functioning of brain cells 83 . In addition to functioning as a neurotransmitter, glutamate also serves as a metabolic precursor 83 . Glutamate and glutamine are compartmentalized in neurons and glia, respectively 84 and are directly connected to neurotransmission and energy metabolism 85,86 .
Preliminary evidence suggests that glutamate levels may reflect neurotransmission 19 . One practical example is that NMDA antagonists significantly and paradoxically increase glutamate levels in the brain 40 . This could be explained for by NMDA receptor antagonists acting on post-synaptic neurons, preventing binding of glutamate, which increases the visibility of glutamate molecules to MRS (i.e., molecular tumbling rates). Available evidence suggests that glutamate in different locations in the neuron can be more or less visible to MRS depending on acquisition parameters 87 . A longer echo time (TE > 15 ms) is more sensitive to compartmental shift (i.e., neural www.nature.com/scientificreports/ activity), compared to tightly packed glutamate in presynaptic vesicles 19,87 . The increased sensitivity of a change in compartmentation is due to a slower T 2 relaxation rate-the rate at which MRS signal decays over time, as glutamate is freely floating (i.e., not packed in the vesicle). This theory is highlighted by a recent meta-analysis comparing fMRS studies with short and long echo times; with short echo time experiments demonstrating smaller increases in glutamate in response to sensory stimulation compared to longer echo times (~ 2.7 versus 6.4%) 19 . Our study used an echo time of 22 ms. Based on the aforementioned theory, the MRS signal may be more sensitive to glutamate levels moving from the presynaptic neuron to the postsynaptic neuron (i.e., compartmental shift). www.nature.com/scientificreports/ Other metabolites. Our analysis focused on examining glutamate and Glx. Other metabolites such as NAA and tCr, which have more prominent resonances, were examined as a reference, to establish the specificity of changes in glutamate in response to pain. NAA was selected from other potential metabolites because turnover is slow, and is not responsive to acute metabolic disturbances 41 and as expected, concentrations remained stable, even at the onset of pain. Creatine, interestingly, increased in response to pain and did so similarly to glutamate and Glx. A previous study also reported a dynamic tCr response following noxious stimulation albeit a decrease 51 . Fluctuations in creatine concentrations are a major concern because tCr levels are commonly used as a reference in H-MRS (i.e., creatine ratios) 88 , including in functional studies 21,34,47,50 . While the essential role of creatine in energy metabolism and cell energetics is well-established 89 , emerging evidence suggests a role in neuromodulation and neurotransmission [89][90][91][92][93] . The mechanisms underlying changes in tCr in response to pain require further elucidation, our observations, as well as those previously published 51 , raise concerns about the functional stability of creatine in task-related MRS studies.

Sex effects.
In an exploratory secondary analysis, we investigated the effect of sex on changes in glutamate and Glx in response to pain. This analysis reflects a high priority area in pain research 94 , and a commitment to understanding the role of sex in biomedical research 95 . Various neuroimaging techniques have been employed to probe sex differences in the brain, revealing prominent variations in anatomy and function 96 . However, recent meta-research has suggested that a reporting bias may overestimate the magnitude of sex-related differences in the brain 97 . With regards to MRS, sex-related differences in resting glutamate levels have been reported in various brain areas (e.g., hippocampus) but not in the ACC 98 . To our knowledge, studies applying noxious stimulation during the acquisition of fMRS have not previously examined or reported differences related to sex. Our results confirm similar resting glutamate and Glx concentrations in the ACC between men and women and nominally larger increases in glutamate and Glx in response to pain in women. Sex differences in glutamatergic signalling have also been reported in animal models 99 , with female rats tending to be more sensitive to NMDA receptor modulation (e.g. Ketamine) [100][101][102] . Further fMRS research should incorporate sex-based analyses to gain a better understanding of glutamatergic signalling in humans.
Limitations and future directions. The most notable limitation of our study a small sample size (n = 15).
To account for this, we have reported the results of conventional significance testing, as well as effect sizes for all relevant comparisons. Future studies should include a larger sample of men and women to further distinguish the role of excitation in the ACC in response to pain. Another limitation of our study is that we did not acquire GABA edited spectra. Previous fMRS studies have demonstrated the capacity to acquire estimates of GABA 21,50 , one of which has demonstrated reduced concentrations following the presentation of a noxious stimulus 21 .
Reduced GABA concentrations are in agreement with our observations, insofar as this reflects increased excitation. Finally, the field of MRS is rapidly evolving, and new methods of acquisition have emerged. Recent guidelines suggest the localization sequence semi-LASER offers reduced localization error (reducing chemical shift displacement) compared to PRESS 15 . The incorporation of simulated metabolite basis sets into the routine analysis is recommended for capturing the full spectral detail from short TE acquisitions 15 , future studies should aim to incorporate these suggestions.

Conclusion
In summary, healthy participants undergoing tonic noxious stimulation demonstrated increased concentrations of glutamate and Glx at the onset of pain. Subsequent reports of pain were not accompanied by increased glutamate or Glx concentrations. We propose a role for glutamate in ACC related to the detection of a noxious stimulus.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports/