Proton magnetic resonance spectroscopy (1H-MRS) is a useful tool for the noninvasive study of the brain’s chemistry. 1H-MRS has an inherently low signal-to-noise ratio (SNR); thus, traditional studies have assumed constant metabolite concentrations during the long scan durations required to acquire a spectrum with sufficient SNR for accurate quantification. Advances in MRI technology such as higher magnetic field strengths, improvements in radiofrequency head coils, preamplification, and signal detection systems have considerably increased the available SNR for 1H-MRS. This has made it feasible to examine the brain in a dynamic state using functional 1H-MRS (fMRS). It has been shown previously with fMRS that alterations in the local neurochemistry occur when a particular area of the brain becomes engaged in task transaction 1–8. One metabolite of interest is glutamate (Glu), the main excitatory neurotransmitter in the brain, which has been shown to be related directly to the brain’s consumption of glucose for the production of energy 9 and has been implicated in the pathophysiology of psychiatric disorders, such as schizophrenia 10,11.
Few studies have used fMRS to date. Three studies have shown small (2–4%) but significant increases in the Glu concentration in the occipital cortex upon functional activation with a visual stimulus, such as a flickering checkerboard 1,2,5. Recent results showed a 2% increase in the motor cortex upon functional activation with a finger-tapping paradigm 6. Two other studies have used pain as a stimulus and have identified 9.3 and 18.1% increases in the mean Glu concentration in the anterior cingulate cortex (ACC) and the anterior insular cortex, respectively, when the participants are exposed to the pain 3,4.
Cognitive tasks are generally less sensitive for detection when assessed through standard BOLD fMRI than visual (flickering checkerboard) or motor (finger tapping) tasks 12. To the best of our knowledge, there has been only one study to date that has used fMRS to examine metabolite changes using a cognitive task as the activation paradigm 8. This study did not find any changes in Glx [Glu+glutamine (Gln)], but they did find that GABA increased in concentration upon functional stimulation of the left dorsolateral prefrontal cortex and then decreased in subsequent runs of the task. Although Glx, in this case, combines Glu with Gln, it would be ideal to use a cognitive task to examine the response of Glu independent of Gln as both are crucial to brain function. As Glu abnormalities have been found in neuropsychiatric disorders in the ACC 10,11, a cognitive task that robustly activates the ACC is desirable for the investigation of abnormalities in the dynamic regulation of Glu levels in this area.
In addition to influencing our perception of pain, the ACC is implicated in many other cognitive functions, including selective encoding of stimulus properties 13, general attention 14, and organization of conflicting stimuli 15. The aim of this exploratory study is to show that with fMRS of the ACC, it is feasible to detect Glu concentration changes using a cognitive task as the activation paradigm. The Colour Stroop Task is a common psychological task that involves differentiating potentially conflicting word and color stimuli under congruent (i.e. the word ‘RED’ written in red ink) and incongruent (i.e. the word ‘BLUE’ written in red ink) conditions 16. The Stroop Task was chosen for this study because it activates the ACC and has been implicated previously in schizophrenia 15, a potential application of this technique.
It has been proposed that Glu concentrations will increase upon functional activation of the ACC with a cognitive task because of the increase in the cycling rate of neuronal Glu with Gln with an approximate 1 : 1 stoichiometry to the neuronal glucose oxidation rate, which is directly related to functional MRI signal changes upon stimulation 9. No other specific hypotheses have been made in terms of other metabolites.
Seven healthy control individuals (age 39.8±3.8 years) provided informed written consent according to the guidelines of the Review Board for Health Sciences Research Involving Human Subjects at Western University. The participants included five men and two women.
In this study, the Stroop Task involved four conditions, incongruent, congruent, word only (the name of the color written in white), and color only (written as ‘XXXX’), with each condition represented in 25% of the trials. Four colors were chosen for the task: red, green, blue, and yellow. The participants were asked to respond as quickly and accurately as possible on a four-button keypad. They were allotted 2 s to respond, and stimuli were separated by 1 s of cross fixation. The task was explained to each participant and rehearsed outside of the scanner until it could be performed consistently with 80% accuracy or greater. An angled mirror attached to the head coil allowed the participants to visualize the words of the Stroop Task projected on the screen in the scanner synchronously with the fMRS acquisition. Participants were initially positioned in the scanner with the button press keypad in their hands ready to perform the task. This minimized additional movement at the onset of the Stroop Task. The fMRS acquisition began with a 4-min resting period in which the participants were asked to fixate on a cross in the center of the projection screen. The final 3 s of the resting period were used to prompt the participant that the Stroop Task was about to start with the word ‘Ready’ projected onto the screen. The Stroop paradigm then lasted for 4 min and was followed by a 4-min recovery period. The procedure was written using PsychoPy 17. The accuracy and response times were recorded to ensure participants’ compliance.
All measurements were acquired on a 7.0 T Agilent/Magnex head-only MRI (Agilent, Inc, Walnut Creek, California, USA) with a Siemens AC84 head gradient coil (Siemens, Erlangen, Germany), located at the Center for Functional and Metabolic Mapping at Western University. A transmit-only receive-only head coil with 15 transmitters and 23 receivers 18 was used for all scans. A map of the transmit field for each transmitter was acquired at the beginning of the session to facilitate optimized homogeneity correction of the transmit field for each scan using a B1-shimming approach developed in-house 19. The magnetic field uniformity (B0-shim) was adjusted automatically over the field of view with first-order and second-order shims using RASTAMAP 20 before all acquisitions.
The MRS voxels were 2.0×2.0×2.0 cm (8 cm3) in size. In every individual, a voxel was centered medially and encompassed the bilateral ACC using two fast low-angle shot 2D anatomical imaging sequences in the sagittal [45 slices, repetition time (TR)=950 ms, echo time (TE)=5.23 ms, flip-angle (α)=30°, gap between slices=1 mm, thickness=2 mm, field of view (FOV) 220×220 mm, and matrix size of 220×200] and axial (20 slices, TR=500 ms, TE=5.23 ms, α=30°, gap between slices=1 mm, thickness=2 mm, FOV=220×220 mm, and matrix size 220×220) directions, both using lipid saturation. Voxel positions were prescribed by the scanner operator (R.T.) using anatomical landmarks as trained by a neuroanatomist (N.R.) to ensure consistent voxel placement (Fig. 1).
MRS spectra were acquired individually during the performance of the Stroop paradigm using an ultra-short echo time stimulated echo acquisition mode sequence with outer volume suppression 21 [TR=3 s, TE=10 ms, mixing time (TM)=32 ms, 4000 complex pairs, four steady state scans, 1 s acquisition time, eight-step phase cycle] with 16 water unsuppressed spectra and 240 water suppressed spectra, 80 spectra for each 4 min section of the Stroop paradigm (Rest, Stroop, Recovery). An eight-pulse VAPOR preparation sequence, with an additional water suppression pulse during the TM period 21, provided efficient water suppression. A separate metabolite suppressed spectrum was acquired to assess the macromolecule content 22. Each acquisition outputted 23 spectra, one for each receiver, which required channel combination before use 23. After the channels had been combined, each spectrum was frequency and phase corrected before being averaged together. Quality Eddy Current Correction 24 reduced linewidth distortions before spectral fitting. Spectra were quantified using fitMAN, a time-domain fitting algorithm 25. All spectra were inspected visually for quality.
Because of our directional hypotheses for Glu concentrations, a three-level repeated-measures analysis of variance (rmANOVA) design using the metabolite concentrations at each 4 min (80 spectral averages) section of the functional paradigm was examined using SPSS v.20 (IBM Corp, Armonk, New York, USA) to determine significant variations over time for Glu (α=0.025, one tailed with Bonferroni correction). Because of the exploratory nature of this paper, the remaining metabolites were also tested for significance using a repeated-measures multivariate analysis of variance (rmMANOVA) (α=0.025, two tailed with Bonferroni correction). Only metabolites with Cramer–Rao lower bounds less than 20% were included in the analysis.
The average linewidths of the unsuppressed water signal were 12.8±2.0 Hz after first-order and second-order shim adjustments. The average metabolite linewidth in the resting condition was 8.6±0.8 Hz and no significant narrowing of the linewidths was observed in the activated condition compared with the resting condition. The water signal was suppressed sufficiently as the residual water signal was less than the height of the NAA peak in every spectrum.
A total of 15 metabolites were quantified consistently with Cramer–Rao lower bounds less than 20% for each of the resting, activated, and recovery spectra (Table 1). The time course of group-averaged Glu activation in 1 min increments is shown in Fig. 2 (all stats were performed on the 4 min, 80 averaged spectra). The concentration appears to increase upon onset of the Stroop Task, plateau, and then decrease over 2 min toward the baseline. When the Glu concentrations between the resting, Stroop, and recovery periods were compared, there was a significant 2.6±1.0% (0.24±0.09 µmol/g) increase in the activated state relative to the initial resting period (P=0.02, one tailed), followed by a nonsignificant −2.4±1.2% (−0.23±0.13 µmol/g) trend for the Glu signal to return to baseline after the activation period had ended (P=0.06, one tailed). However, when Glu concentrations were normalized to their baseline value in the resting period, the relative decrease in the recovery period became significant before Bonferroni corrections (0.025<P<0.05). Six of the seven participants showed this increase in Glu upon activation (1.0, 1.3, 5.3, −0.4, 6.0, 0.3, and 5.0%) and six of the seven participants showed the subsequent decrease (1.2, −2.0, −8.8, −1.5, −1.1, −2.9, and −1.6%).
Other than Glu, the remaining metabolites were not significant in the rmMANOVA. It should be noted, though, that in the univariate pair-wise comparisons, ascorbate (Asc) increased 33.1±17.5% (0.20±0.06 µmol/g) during the 4-min recovery period relative to the activation period (P=0.03, two tailed). However, this is only a trend because of the Bonferroni correction.
All participants recorded at least 90% correct responses. Group-averaged response times were 725±278 ms for the congruent condition, 976±261 ms for the incongruent condition, 750±270 ms for the word-only condition, and 721±230 ms for the color-only condition.
Using a Stroop Task as the functional paradigm for an fMRS examination of the dynamic metabolic response to activation in the ACC, an observed significant increase of 2.6±1.0% was found in Glu upon onset of the task, followed by a trend to return to the baseline in a small sample size of healthy controls (n=7). The high percentage of correct responses shows that there was high compliance with the Stroop Task among the participants.
The increase in Glu is comparable with recent fMRS studies of the occipital cortex 1,2,5 and the motor cortex 6, which found Glu increases from 2 to 4% over the course of minutes, with a subsequent decrease upon cessation of the stimuli. Figure 2 indicates that the concentration increased to its new steady state within the first minute of activation. This is relatively faster than that in the above studies, which show that it takes 1–2 min before the concentration reaches its new state. One possible explanation is that because of the involvement of ACC in attention networks 14, and, specifically, selective encoding of stimulus properties 13 the ACC had already been in a state of minor activation. This is consistent with the final data point of the first resting condition in Fig. 2 that is seemingly already ascending toward the peak concentration. The resting Glu concentration is slightly lower than what was observed in other fMRS studies at 7 T 1,2,5,6; however, different scanners, techniques, and brain areas studied led to natural variation in the observed values between studies.
One departure from previous fMRS studies at 7 T 1,2,5,6 is that a significant difference in lactate or glucose was not observed. These are difficult metabolites to quantify reliably; thus, perhaps more participants would be needed to observe significant differences.
It is conceivable that the ACC accumulates Glu faster than other brain regions upon stimulation as can be extrapolated from a study of the ACC using a painful stimulus 3. Mullins and colleagues used event-related fMRS at 3 T, in which one resting spectrum and one activated spectrum are acquired per bolus of pain. After many boluses of pain, there are enough resting and activated spectra that can be averaged to detect a rapid metabolic response. A 9.3% increase in Glu and an 11.4% increase in Glx (Glu+Gln) were observed. The Glu increases observed by Mullins et al.3 are markedly higher than the response detected in the present study. This could be because of the nature of the experimental design (event-related design vs. block design). It is possible that with an event-related Stroop Task, larger similar increases in Glu in the ACC could be observed. However, with the current block design, the magnitude of Glu changes agrees more with the previous work in the visual cortex 1,2,5 and the motor cortex 6.
One study at 3 T presented GABA changes with a working memory task in the dorsolateral prefrontal cortex, but did not detect any changes in Glx (Glu+Gln) 8. The study used a dedicated pulse sequence to detect GABA specifically, which can otherwise be a difficult metabolite to quantify as it was not consistently quantified reliably in this study. They observed an initial increase in GABA concentrations, but subsequent decreases with repeated runs of the task. In a future study, it would be interesting to observe how Glu would respond to repeated runs of the Stroop Task.
Another study that was carried out at 3 T has used an fMRS technique to study the ACC using visually evoked sexual arousal 7 and found significant 10–20% increases in Glx with activation. At lower field strength, it is often necessary to measure a combination of Glu and surrounding metabolites that can include Gln, glutathione, GABA, and sometimes macromolecules together as Glx. It is not clear which metabolites were included in their Glx definition, making it harder to compare these results. A common definition of Glx is simply Glu plus its precursor Gln, as was used in the pain study 3 and the working memory study 8 mentioned above. Using this definition, Glx was examined in this study, but no significant differences were observed.
The increase in Asc in the recovery period, although not significant after Bonferroni correction or at the multivariate level, is intriguing. It has been reported previously by Wilson et al.26 that Glu triggers the release of Asc from astrocytes and increases the extracellular concentration of Asc. Asc is an antioxidant. This is important because Glu neurotransmission leads to increased mitochondrial activity and as a result increased free radicals 27. Wilson et al.26 showed that this effect takes some time to observe; thus, it is possible that the increase in Glu during the activated phase led to the delayed increase in Asc concentration that was observed in the recovery phase.
Glu is the main excitatory neurotransmitter in the brain. It is synthesized from glucose in the mitochondria, released into the synaptic cleft through exocytosis, and then binds with the postsynaptic membrane to induce neurotransmission 28. It is then taken up into the adjacent glial cell, where it is converted into Gln before being transferred back into the neuron and converted back into Glu 28. There is an increase in glucose consumption during functional activation that specifically supports the increased turnover rate of the Glu/Gln cycle 9. Therefore, the increased number of Glu molecules flowing within the Glu/Gln cycle during the functional activation of the ACC with the Stroop Task could explain the increased Glu observed in bulk brain tissue using MRS. The lack of change in Gln concentrations may indicate that the conversion of Gln back into Glu in the neuron is not a rate-limiting step in the Glu/Gln cycle.
This proof-of-concept study shows that the Stroop Task is a robust task that can be used for fMRS studies in the ACC. This introduces a new method to study the abnormalities of Glu modulation in brain disorders. Schizophrenia research may particularly benefit from this method as it is a population that has been shown to have abnormal resting levels of glutamatergic metabolites in the ACC compared with healthy controls 10,11 and impaired ACC activation during performance of the Stroop Task 15. Dysfunction of glutamatergic metabolism may manifest itself as an inability to dynamically upregulate Glu cycling or excessive upregulation of Glu, or even Gln, concentrations during performance of the Stroop Task. The limitations of the study include its small sample size and the possibility that the Glu changes observed are because of movement-induced phase shifts from the participant responding to the task, although it is unlikely that a phase shift would induce consistent increases in Glu rather than decreases. Future studies should have a larger sample size and focus on fMRS in neuropsychiatric disorders.
This work was supported by a New Investigator Fellowship grant (J.T.) from the Ontario Mental Health Foundation (OMHF). Salary support for R.T., B.S., and M.D., was funded in part by operating funds from the Canadian Institute of Health Research (P.C.W., MT-12078).
Conflicts of interest
There are no conflicts of interest.
1. Mangia S, Tkáč I, Gruetter R, Van de Moortele PF, Maraviglia B, Uğurbil K. Sustained neuronal activation raises oxidative metabolism to a new steady-state level: evidence from 1
H NMR spectroscopy in the human visual cortex. J Cereb Blood Flow Metab 2007; 27:1055–1063.
2. Schaller B, Mekle R, Xin L, Kunz N, Gruetter R. Net increase of lactate and glutamate concentration in activated human visual cortex detected with magnetic resonance spectroscopy at 7 tesla. J Neurosci Res 2013; 91:1076–1083.
3. Mullins PG, Rowland LM, Jung RE, Sibbitt WL Jr. A novel technique to study the brain’s response to pain: proton magnetic resonance spectroscopy. NeuroImage 2005; 26:642–646.
4. Gussew A, Rzanny R, Erdtel M, Scholle HC, Kaiser WA, Mentzel HJ, et al.. Time-resolved functional 1
H MR spectroscopic detection of glutamate concentration changes in the brain during acute heat pain stimulation. NeuroImage 2010; 49:1895–1902.
5. Lin Y, Stephenson MC, Xin L, Napolitano A, Morris PG. Investigating the metabolic changes due to visual stimulation using functional proton magnetic resonance spectroscopy at 7T. J Cereb Blood Flow Metab 2012; 32:1484–1495.
6. Schaller B, Xin L, O’Brien K, Magill A, Gruetter R. Are glutamate and lactate increases ubiquitous to physiological activation? A 1
H functional MR spectroscopy study during motor activation in human brain at 7 Tesla. NeuroImage 2014; 93:138–145.
7. Kim TH, Kang HK, Jeong GW. Assessment of brain metabolites change during visual sexual stimulation in healthy women using functional MR spectroscopy. J Sex Med 2013; 10:1001–1011.
8. Michels L, Martin E, Klaver P, Edden R, Zelaya F, Lythgoe DJ, et al.. Frontal GABA levels change during working memory. PLoS One 2012; 7:e31933.
9. Rothman DL, Behar KL, Hyder F, Shulman RG. In vivo NMR studies of the glutamate neurotransmitter flux and neuroenergetics: implications for brain function. Annu Rev Physiol 2003; 65:401–427.
10. Poels EM, Kegeles LS, Kantrowitz JT, Slifstein M, Javitt DC, Lieberman JA, et al.. Imaging glutamate in schizophrenia: review of findings and implications for drug discovery. Mol Psychiatry 2014; 19:20–29.
11. Théberge J, Bartha R, Drost DJ, Menon RS, Malla A, Takhar J, et al.. Glutamate and glutamine measured with 4.0 T proton MRS in never-treated patients with schizophrenia and healthy volunteers. Am J Psychiatry 2002; 159:1944–1946.
12. Drobyshevsky A, Baumann SB, Schneider W. A rapid fMRI task battery for mapping of visual, motor, cognitive, and emotional function. Neuroimage 2006; 31:732–744.
13. Neufeld RWJ, Boksman K, Vollick D, George L, Carter JR. Stochastic dynamics of stimulus encoding in schizophrenia: theory, testing, and application. J Math Psychol 2010; 54:90–108.
14. Posner MI, Petersen SE. The attention system of the human brain. Annu Rev Neurosci 1990; 13:25–42.
15. Ungar L, Nestor PG, Niznikiewicz MA, Wible CG, Kubicki M. Color Stroop and negative priming in schizophrenia: an fMRI study. Psychiatry Res 2010; 181:24–29.
16. Eidels A, Townsend JT, Algom D. Comparing perception of Stroop stimuli in focused versus divided attention paradigms: evidence for dramatic processing differences. Cognition 2010; 114:129–150.
17. Peirce JW. PsychoPy – psychophysics software in Python. J Neurosci Methods 2007; 1621–28–13.
18. Gilbert KM, Curtis AT, Gati JS, Klassen LM, Menon RS. A radiofrequency coil to facilitate B1+
shimming and parallel imaging acceleration in three dimensions at 7 T. NMR Biomed 2011; 24:815–823.
19. Curtis AT, Gilbert KM, Klassen LM, Gati JS, Menon RS. Slice-by-slice B1+ shimming at 7 T. Magn Reson Med 2012; 68:1109–1116.
20. Klassen LM, Menon RS. Robust automated shimming technique using arbitrary mapping acquisition parameters (RASTAMAP). Magn Reson Med 2004; 51:881–887.
21. Tkáč I, Gruetter R. Methodology of 1
H NMR spectroscopy of the human brain at very high magnetic fields. Appl Magn Reson 2005; 29:139–157.
22. Penner J, Curtis A, Gilbert K, Klassen M, Gati S, Smith M, et al.. Metabolite nulling to measure the macromolecule baseline for quantitative 1
H magnetic resonance spectroscopy at 7 Tesla [abstract]. Proc Intl Soc Mag Reson Med 2010; 18:911.
23. Brown MA. Time-domain combination of MR spectroscopy data acquired using phased-array coils. Magn Reson Med 2004; 52:1207–1213.
24. Bartha R, Drost DJ, Menon RS, Williamson PC. Spectroscopic lineshape correction by QUECC: combined QUALITY deconvolution and eddy current correction. Magn Reson Med 2000; 44:641–645.
25. Bartha R, Drost DJ, Williamson PC. Factors affecting the quantification of short echo in-vivo 1H MR spectra: prior knowledge, peak elimination, and filtering. NMR Biomed 1999; 12:205–216.
26. Wilson JX, Peters CE, Sitar SM, Daoust P, Gelb AW. Glutamate stimulates ascorbate transport by astrocytes. Brain Res 2000; 858:61–66.
27. Patel M, Day BJ, Crapo JD, Fridovich I, McNamara JO. Requirement for superoxide in excitotoxic cell death. Neuron 1996; 16:345–355.
28. Hertz L, Dringen R, Schousboe A, Robinson SR. Astrocytes: glutamate producers for neurons. J Neurosci Res 1999; 57:417–428.