Brain dysfunction in fibromyalgia and somatization disorder using proton magnetic resonance spectroscopy: a controlled study

Abstract

Fayed N, Andres E, Rojas G, Moreno S, Serrano-Blanco A, Roca M, Garcia-Campayo J. Brain dysfunction in fibromyalgia and somatization disorder using proton magnetic resonance spectroscopy: a controlled study.

Objective:  To evaluate the brain metabolite patterns in patients with fibromyalgia (FM) and somatization disorder (STD) compared with healthy controls through spectroscopy techniques and correlate these patterns with psychological variables.

Method:  Design. Controlled, cross-sectional study. Sample. Patients were recruited from primary care in Zaragoza, Spain. The control group was recruited from hospital staff. Patients were administered questionnaires on pain catastrophizing, anxiety, depression, pain, quality of life, and cognitive impairment. All patients underwent Magnetic Resonance Imaging and magnetic resonance spectroscopy (MRS).

Results:  A significant increase was found in the glutamate + glutamine (Glx) levels in the posterior cingulate cortex (PCC): 10.73 (SD: 0.49) for FM and 9.67 (SD: 1.10) for STD 9.54 (SD: 1.46) compared with controls (P = 0.043). In the FM + STD group, a correlation between Glx and pain catastrophizing in PCC (r = 0.397; P = 0.033) and between quality of life and the myo-inositol/creatine ratio in the left hippocampus (r = −0.500; P = 0.025) was found. To conclude Glutamate seems to be relevant in the molecular processes involved in FM and STD. It also opens the door for Proton MRS (¹H-MRS) in STD and suggests that reducing glutamatergic activity through pharmacological treatment could improve the outcome of patients with FM and STD.

Conclusion:  Glutamate seems to be relevant in the molecular processes involved in FM and STD. It also opens the door for Proton MRS (¹H-MRS) in STD and suggests that reducing glutamatergic activity through pharmacological treatment could improve the outcome of patients with FM and STD.

Significant outcomes

• •
   This study demonstrates a significant increase in the levels of Glx, a combined measure of glutamate (Glu) and glutamine (Gln), within the posterior cingulate cortex (PCC) in fibromyalgia (FM) and, to a lesser extent in, somatization disorder (STD) compared with controls.
• •
   Levels of Glx correlates with pain catastrophizing suggesting that elevated levels of Glx in the PCC are associated with increased pain catastrophizing.

Limitations

• •
   The small sample size (N = 10) of each group of patients.
• •
   The proportion of gray matter/white matter and cerebrospinal fluid (CSF) within the corresponding voxel has not been calculated.
• •
   Metabolites were not quantified with the help of segmentation and CSF-correction.

Introduction

Somatoform disorders, according to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (1) are defined by the presence of physical symptoms that suggest a general medical condition but that are not fully explained by a general medical condition, by the direct effects of a substance, or by another mental disorder. The most extreme form of this group is STD, a chronic and polysymptomatic disorder characterized by at least four unexplained gastrointestinal, sexual and pseudoneurological symptoms (1). FM is a disabling disorder characterized by a history of widespread pain for at least 3 months and patient reporting of tenderness in at least 11 of 18 defined tender points when digitally palpated with approximately 4 kg per unit area of force (2). It is associated with other symptoms such as generalized muscular aches, stiffness, fatigue, and non-restorative sleep (2). In both conditions, pain is usually the most disabling symptom, and both may cause clinically significant distress or impairment in social, occupational, or other areas of functioning.

Previous functional neuroimaging studies on FM confirm that patients with FM show augmented neuronal responses to both innocuous and painful stimuli (3). Those studies are consistent with the allodynia and hyperalgesia observed in this condition (4). A Positron Emission Tomography (PET) study in FM showed reduced presynaptic dopaminergic activity in several brain regions in which dopamine plays a critical role in modulating nociceptive processes (5). Previous studies with diffusion tensor imaging in FM showed alterations in the right thalamus and significantly lower fractionated anisotropy in comparison with controls (6). In STD, neuroimaging research has been scarce, with a small number of studies in the field of SPECT describing hypoperfusion, primarily in the non-dominant hemisphere (7), or with functional Magnetic Resonance Imaging (MRI) demonstrating increased activation of the pain processing area (thalamus, basal ganglia, and opercular insular area) during an episode of pain induced by a pin prick (8).

Previous studies with Magnetic Resonance Spectroscopy (MRS) at 4 T have yielded multiple findings: that glutamate (Glu) levels, an excitatory neurotransmitter within the central nervous system, in the anterior cingulate cortex increase in response to an external painful stimulus; that measured changes in glutamine (Gln) levels correlate highly with the perceived intensity of pain; and that these dynamic changes are detectable by Proton MRS (¹H-MRS) (9). In humans, as in animals, an acute hypofunctional N-methyl-d-aspartate receptor (NMDAR) state is associated with increased glutamatergic activity (10). More recently, a growing body of literature suggests that Glx (glutamate + glutamine) may play a role in FM pathology in the insula (11), hippocampus (12), and PCC (13). Otherwise, little is known about the use of MRS in STD.

Magnetic resonance spectroscopy provides a non-invasive method for characterizing chemical and cellular features in vivo. It can be used to measure the chemical composition of tissues, to characterize certain tissue metabolic processes, and to identify unanticipated chemical or metabolic factors in diseases. In brain tissue, the concentrations and mobility of MRS-visible low-molecular-weight chemicals are measured as spectral peaks; these can be used to detect abnormalities in brain regions that seem normal in MRI and to elucidate the pathology underlying MRI-visible abnormalities (14).

Aims of the study

To evaluate brain metabolite patterns through spectroscopy techniques in patients with somatization disorder and in patients with fibromyalgia compared with healthy controls and evaluate these patterns with respect to various psychological variables such as pain catastrophizing, anxiety, depression, pain, quality of life, and cognitive impairment.

Material and methods

Design

Controlled, cross-sectional study.

Patients

The FM and the STD groups were recruited from primary health care centers in the city of Zaragoza, Spain. Patients were required to meet the following inclusion criteria: 18–65 years old; able to understand Spanish; fulfill the criteria for FM or STD (DSM-IV criteria); and no pharmacologic treatment 1 week before the study began. FM was diagnosed according to the American College of Rheumatology (2) by a rheumatologist, and STD was diagnosed by a psychiatrist using the Structured Clinical Interview for the DSM-IV Axis I Disorders (SCID-I). Patients were excluded if they had other Axis I psychiatric disorders, or were pregnant or lactating. Patients with FM were excluded whether they fulfilled STD disorder and vice versa. The healthy control group was recruited among hospital staff, with an adjustment for gender and age (±3 years), to match the STD group. The study was approved by the Aragon Ethics Committee and has, therefore, been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. All persons gave their informed consent prior to their inclusion in the study.

Measurements

Sociodemographic and clinical variables.

Sociodemographic data.  Gender, age, marital status, education, and occupation were collected.

Pain Catastrophizing Scale (PCS).  This construct involves an exaggerated negative orientation toward noxious stimuli. The PCS is a 13-item self-report questionnaire (15).

Hospital Anxiety Depression Scale (HADS).  This is a self-report scale designed to screen for the presence of depression and anxiety disorders in medically ill patients.

It contains 14 items rated on four-point Likert-type scales. Two subscales assess depression and anxiety independently (HADS-Dep and HADS-Anx respectively) (16).

Pain Visual Analogue Scale (PVAS).  A VAS is a 10-cm horizontal line, with perpendicular lines on the edges, defined as the extreme limits of pain experience. Anchoring points at each edge are characterized by verbal expressions such as ‘No pain’ (accompanied by the number 0) at one edge and ‘Maximum pain ever experienced’ (accompanied by the number 100) at the other edge.

Mini-Mental State Exam (MMSE).  This is a fully structured scale that includes seven categories: orientation to place, orientation to time, registration, attention and concentration, recall, language, and visual construction (17).

EuroQol 5D (EQ5D).  This is a standardized instrument for use as a measure of health outcomes. Applicable to a wide range of health conditions and treatments, it provides a simple descriptive profile and a single index value for health status (18). Despite other questionnaires such as Short-Form 36 (19) being more widely used to evaluate quality of life in chronic pain patients, we chose EQ-5D because it provides a single comprehensive measurement of quality of life. It is easier to assess the correlation between quality of life and brain metabolites using a single comprehensive measurement.

All patients underwent the following neuroimaging techniques:

• i)
   Magnetic resonance imaging: Data were acquired using a 1.5 T Sigma HD clinical scanner (GE Healthcare Diagnostic Imaging; Milwaukee, WI, USA). All images were acquired using an eight-channel phased-array head coil (NVHEAD A).
• ii)
   Magnetic resonance spectroscopy: A coronal T2-weighted image [repetition time (TR) = 5350 ms, echo time (TE) = 85 ms, 90° flip angle, number of excitations = 2, matrix size = 320 × 256, field of view = 24 × 24 cm, slice thickness/gap = 5/0 mm] in the plane that goes through inner auditive conducts and the brain peduncle was used to locate volumes of interest (VOI) (2  × 2 × 2 cm) in all brain locations. A midsagittal T1-weighted image (TR = 560  ms, TE  =  12  ms, 90° flip angle, number of excitations  =  1, matrix size = 256 × 160, field of view  =  24 × 24 cm, slice thickness/gap = 5/0 mm) was obtained to locate a voxel in the PCC and a parasagittal T1 image, 30 mm on the right with respect to the plane of the corpus callosum, to locate a voxel in the anterior and posterior insula (Fig. 1). The following areas of exploration were chosen: (i) areas in which the authors have found increased levels of Glx powered by MRS, (11–13) (the insular cortex, hippocampus and PCC), (ii) brain structures that are activated during painful conditions in healthy controls (insular cortex) (20) in patients with FM (21), and (iii) the regions in the previously mentioned reports that have been implicated in cognitive impairment (22) and the brain′s default network (the hippocampus and PCC) (23). In previous studies on FM by our group (13), other pain-related areas such as the thalamus and sensitive-motor region were studied without finding significant differences when compared with healthy control subjects. For this reason, those areas were not analyzed in this study.

Figure 1.  Voxel placement in the both hippocampi (a,b), anterior (c) and posterior insula (d), and the posterior cingulate area (e). Example of a spectrum acquired with the LCModel software in posterior cingulate area of a patient with fibromyalgia showing an increased peak of Glx (glutamate + glutamine) (f).

1H-Magnetic resonance spectroscopy was carried out by means of a short–TE of 35 ms and a TR of 2000 ms and 128 accumulations using a single-voxel with a spin echo technique that uses selective excitation with gradient spoiling for water suppression. The mode of spectral acquisition was the Proton Brain Spectroscopy-Point Resolved Spectroscopy technique (PROBE-PRESS). The PRESS mode is used more than STEAM because it increases the signal/noise ratio and is less sensitive to movement artifacts (24). A short TE (20–40 ms) allows us to increase the signal/noise ratio and to visualize most metabolite peaks, with the inconvenience of some degree of peak overlapping of. Time matters in clinical practice, so short TEs are preferable. In our experience with a 1.5T GE Signa Horizon-clinical scanner (GE Healthcare Diagnostic Imaging, Milwaukee, WI, USA), a TE of 30 ms and a TR of 2500 ms have proven valuable (14).

For the quantification of absolute concentrations of brain metabolites, expressed in mmoles/kg wet weight, we used the user-independent frequency domain–fitting program LCModel, version 6.2-0, (Stephen Provencher, Oakville, ON, Canada), (25), applying an eddy current correction and using an internal water signal reference to calculate absolute metabolite concentrations. Apart from the individual analysis of the glutamate (Glu), creatine (Cr), N-acetyl aspartate (NAA) and myo-inositol (mI) compounds, we studied the summed concentrations of the following three compound pairs: NAA + N-acetyl aspartyl glutamate (NAA + NAAG), referred to as total NAA, glycerophosphocholine + phosphocreatine (GPC), referred to as total Cho, and glutamate + glutamine, referred to as Glx. Absolute metabolite values were only considered when the Kramer–Rao lower bound was below 20%, thus indicating that these metabolites could be reliably estimated.

Concentration values are expressed as arbitrary institutional units and are not corrected for contributions by CSF and by a small reduction in the numerical values from residual T1 and T2 relaxation effects. We also obtained the ratios of the peak amplitude of the metabolites relative to creatine. In clinical practice, metabolic ratios are assessed using Cr, which is considered the most stable metabolite, as an internal reference (26). Before starting this study, we studied the test–retest reliability of metabolite measurements in every area in a sample of patients with other pathologies, with two consecutive studies completed without removing the patient from the scanner. According to the resulting alpha coefficients, we must assume a mean random variation of approximately 8% for mI/Cr and of approximately 10% for NAA/Cr and Cho/Cr (26). We also carried out a second immediate MRS in 16 patients before they were removed from the scanner to check the reproducibility of Glx. The intra-class correlation coefficients were also remarkable (0.80 for the PCC).

Statistical analysis

To describe the quantitative variables, the means and standard deviations were calculated because of the normality of variables. Sociodemographic variables were compared by group using the chi-square or anova test, depending on whether the variable was qualitative or quantitative. To analyze possible differences in brain metabolite levels between patients with FM, STD and healthy controls, a one-factor anova test was calculated using post hoc analysis when differences were found. In addition, we used the non-parametric Spearman’s rho correlation in the FM group, in the FM + STD group and in the whole sample, to study the relationship between brain metabolites for which the levels were significantly different, and for which the clinical variables were studied. To determine the statistical significance in psychological tests between the controls, the patients with FM, and the patients with STD, a one-factor anova test was used. Age was included as a covariate in the statistical models. Statistical analyses were carried out using spss 15.0 SPSS for Windows, Rel. 15.0.1.2006. Chicago: SPSS Inc, and P-values lower than 0.05 were considered statistically significant for all analyses.

Results

Sociodemographics, psychological variables, and pain measurements

During the recruitment period, three patients with FM were excluded (one because she did not want to discontinue her treatment and two because they were diagnosed with depressive disorders) and two patients with STD disorder because they had comorbid psychiatric disorders. The excluded patients did not show significant differences regarding sociodemographic and clinical variables compared with the participants. The final sample was made up of 10 patients with FM and 10 with STD, which were compared with 10 age-matched healthy control subjects (range: 32–54 years).

No significant differences were found in gender or age among the three groups (Table 1). In the control group, rating scores on the psychopathology questionnaires were within the normal range. Psychological ratings of the healthy control group were significantly different from the FM and STD groups, while there were no differences between the FM and STD groups on any psychological variable (Table 1). In the FM group, the mean duration of the disorder was 2.13 (SD = 0.52) years, while in the STD group, it was 3.82 (SD = 0.76). The psychological profiles showed the usual psychological characteristics of FM and STD patients: high scores in anxiety and depression assessed with the HADS; high scores on the PCS and in pain assessed with the PVAS, and low quality of life as measured by the EQ5D. The MMSE scores suggested symptoms of cognitive dysfunction in FM (mean: 32.90; SD: 0.73) and STD (mean: 33.42; SD: 0.69), but at levels less severe than those found in patients with dementia.

Table 1.   Comparison of sociodemographic and psychological variables in healthy controls, fibromyalgia (FM), and somatization disorder (STD) groups

	Healthy controls (HC) (N = 10)	FM (N = 10)	STD (N = 10)	Significance
^aChi-squared test. ^bOne-factor analysis of variance (anova). Post hoc analysis using Bonferroni test when variances are homogeneous (according to Levene’s test) and using Dunnet’s test when variances were not homogeneous. *FM > HC; STD > HC; FM = STD. **HC > FM; HC > STD; FM = STD. HADS, Hospital Anxiety Depression Scale.
Sociodemographic variables
Gender (female, N)	8	9	8	χ^2 = 1.25, P = .787^a
Age (mean, SD)	39.52 (11.13)	38.94 (5.56)	43.92 (9.96)	F = 0.88, P = .425^b
Ethnic group (European, N)	10	10	10	–, P = 1^a
Psychological variables
Anxiety (HADS-anx) (mean, SD)	1.70 (0.94)	4.80 (1.61)	3.60 (1.34)	F = 13.72, P = 0.001^b*
Depression (HADS-dep) (mean, SD)	1.60 (0.69)	4.90 (2.07)	4.20 (1.39)	F = 13.40, P = 0.001^b*
Catastrophization (Pain Catastrophizing Scale) (mean, SD)	13.60 (1.42)	24.80 (4.18)	23.70 (8.21)	F = 13.12, P = 0.001^b*
Pain (Pain Visual Analogue Scale) (mean, SD)	6.50 (5.79)	69.50 (9.55)	61.50 (9.14)	F = 169.17, P = 0.001^b*
Cognitive function (Mini-Mental State Exam) (mean, SD)	35 (0)	32.90 (0.73)	33.42 (0.69)	F = 34.94, P = 0.001^b**
Quality of life (EuroQol 5D) (mean, SD)	92.5 (4.85)	66.0 (8.43)	65.50 (6.43)	F = 52.59, P = 0.001^b**

Spectroscopic results

Table 2 reports the values of the different metabolites for FM, STD, and for healthy controls. Results revealed a significant increase in the Glx levels in the PCC: 10.73 (SD: 0.49) for FM and 9.67 (SD: 1.10) for STD vs. 9.54 (SD: 1.46) in controls (P = 0.043). Decreased Glu levels were found in the left hippocampus in patients compared with controls (P = 0.019). Choline levels in the PCC and the right and left hippocampi, as well as choline/creatine ratios in the left hippocampus were lower in patients compared with controls (P = 0.027, 0.029, 0.002, and 0.02 respectively). The NAA/Cr and NAA + NAAG/Cr ratios of the posterior insula and the NAA + NAG levels of the left hippocampus were lower in patients compared with controls (P = 0.019, 0.048), and significantly lower myo-inositol levels and myo-inositol/creatine ratios were observed in the left hippocampus in patients compared with controls (P = 0.008 and 0.048). We did not find significant differences for the other metabolites and ratios.

Table 2.   Comparison of metabolite levels and metabolite/creatine ratios in posterior cingulate, posterior insula, and hippocampus determined by proton magnetic resonance spectroscopy in healthy controls, fibromyalgia (FM), and somatization disorder (STD) groups

Region and metabolites	Healthy controls (HC) (N = 10)	FM (N = 10)	STD (N = 10)	p^a
	Mean (SD)	Mean (SD)	Mean (SD)
^aOne factor anova. Post hoc analysis using Bonferroni test when variances are homogeneous (according to Levene’s test) and using Dunnet’s test when variances were not homogeneous. ^bNAA + NAAG = N-acetyl aspartate + N-acetyl aspartyl glutamate. *HC > STD; HC = FM; FM = STD. **HC > FM; HC > STD; FM = STD. ***HC > FM; HC = STD; STD = FM. ****HC < FM; HC < STD; STD < FM.
Posterior cingulate
Glutamate + glutamine (Glx)	9.54 (1.46)	10.73 (0.49)	9.67 (1.10)	F = 3.56, P = 0.043****
Choline	1.12 (1.09)	1.05 (0.06)	1.01 (0.07)	F = 4.17, P = 0.027*
Posterior insula
N-acetylaspartate/Creatine	1.40 (0.09)	1.35 (0.14)	1.25 (0.10)	F = 4.62, P = 0.019*
NAA + NAAG/Creatine	1.56 (0.09)	1.47 (0.16)	1.40 (0.14)	F = 3.40, P = 0.048*
Left hippocampus
Myo-inositol/Creatine	1.19 (0.09)	1.10 (0.10)	1.08 (0.09)	F = 3.42, P = 0.048*
Choline/Creatine	0.34 (0.02)	0.32 (0.03)	0.31 (0.01)	F = 4.51, P = 0.020*
Myo-inositol	5.86 (0.77)	4.72 (0.89)	4.79 (0.81)	F = 5.86, P = 0.008**
Choline	1.70 (0.18)	1.40 (0.28)	1.37 (0.11)	F = 7.74, P = 0.002**
NAA + NAAG^b	7.22 (1.12)	5.98 (1.03)	6.09 (0.47)	F = 5.47, P = 0.010**
Glutamate	6.50 (0.92)	5.42 (0.80)	5.80 (0.67)	F = 4.62, P = 0.019***
Right hippocampus
Choline	1.57 (0.24)	1.29 (0.28)	1.29 (0.22)	F = 4.04, P = 0.029**

Pearson’s correlations revealed a strong correlation between pain catastrophizing and quality of life and Glx in the PCC in the FM group. In the FM + STD group, a correlation between Glx and pain catastrophizing in PCC and between quality of life and the myo-inositol/creatine ratio in the left hippocampus was described. When the three groups (FM + STD + healthy controls) were studied together, choline in the posterior cingulate correlates with all psychological variables, and pain correlates with all metabolites studied in the left hippocampus. Other metabolites such as choline in the right hippocampus and N-acetylaspartate in the posterior insula also correlates with many psychological variables, but this is not the case for Glx in the PCC (Table 3).

Table 3.   Correlations among brain metabolites and psychological variables in the three groups

Brain metabolites	HADS –Anx	HADS-Dep	PCS	PVAS	MMSE	EQ5D
HADS–Anx, Hospital Anxiety Depression Scale, Anxiety subscale; HADS-Dep, Hospital Anxiety Depression Scale, Depression subscale; PCS, Pain catastrophizing Scale; PVAS, Pain Visual Analog Scale; MMSE, Mini-Mental State Exam; EQ5D, EuroQol 5D; FM, fibromyalgia; NAA, N-acetyl aspartate; NAAG, N-acetyl aspartyl glutamate; STD, somatization disorder.
FM group
Posterior cingulate cortex (PCC)
Glutamate + glutamine (Glx)	–	–	0.428 (P = 0.028)	–	–	−0.632 (P = 0.041)
FM + STD group
Left hippocampus
Myo-inositol/Creatine	–	–	–	–	–	−0.500 (P = 0.025)
PCC
Glutamate + glutamine (Glx)	–	–	0.397 (P = 0.033)	–	–	–
FM + STD + healthy controls
Posterior cingulate
Choline	−0.377 (P = 0.040)	−0.398 (P = −029)	−0.408 (P = 0.025)	−0.450 (P = 0.013)	0.463 (P = 0.010)	0.528 (P = 0.003)
Posterior insula
N-acetylaspartate/Creatine	–	–	–	−0.390 (P = 0.033)	–	–
NAA + NAAG/Creatine	–	–	–	−0.386 (P = 0.035)	–	–
Left hippocampus
Myo-inositol/Creatine	–	–	–	−0.370 (P = 0.044)	–	–
Choline/Creatine	–	–	−0.377 (P = 0.040)	−0.458 (P = 0.011)	–	–
Myo-inositol	–	−0.457 (P = 0.011)	–	−0.553 (P = 0.002)	0.437 (P = 0.016)	0.424 (P = 0.019)
Choline	–	−0.419 (P = 0.021)	−0.466 (P = 0.009)	−0.638 (P = 0.0001)	0.478 (P = 0.007)	0.571 (P = 0.001)
NAA + NAAG^b	–	–	–	−0.513 (P = 0.004)	0.475 (P = 0.008)	0.470 (P = 0.009)
Glutamate	–	–	–	−0.409 (P = 0.025)	0.442 (P = 0.015)	0.373 (P = 0.042)
Right hippocampus
Choline	–	–	−0.428 (P = 0.018)	−0.498 (P = 0.005)	0.373 (P = 0.043)	0.511 (P = 0.004)

Discussion

The results from this study demonstrate a significant increase in the levels of Glx, a combined measure of glutamate (Glu) and glutamine (Gln), within the PCC in FM and, to a lesser extent in, STD compared with controls (Fig. 1) This factor also correlates with the PCS and the MMSE, suggesting that elevated levels of Glx in the posterior cingulate are associated with increased pain catastrophizing and cognitive impairment. To our knowledge, this is the first spectrometric study on STD and the first to demonstrate the elevation of Glx levels of patients with STD and FM. We recently demonstrated increased Glx levels within the PCC of FM patients (13) and significant correlations with depression, pain, and global function. Previous studies have shown higher concentrations of Glu within the posterior insula (11) and the left hippocampus (12). Studies have also shown that Glu levels in the anterior cingulate cortex increase in response to an external painful stimulus, that changes in Gln levels highly correlate with the perceived intensity of pain, and that these dynamic changes are detectable by ¹H-MRS (9).

Glu has been implicated as an important mediator in the neurotransmission, potentiation, and negative effects associated with pain, and it has been associated with chronic pain sensitization (27). Our data suggest that Glx plays a role in this augmented pain processing in those individuals who have elevated Glx levels. Because higher Glx levels were associated with an elevation of the PCS, it is likely that Glx in the posterior cingulate is related to pain processing. In the neurotransmission process, Glutamate is released into the synapse from neuronal cells and is then taken up by astroglial cells and transformed into Glu and afterward, it is transported back to the neuron to be recycled to the Glutamate pool. It has been suggested that most glucose oxidation and uptake, and therefore energy production, that occurs in the brain is used to support this Glutamate/Glutamine cycling. Thus, it has been proposed that the physiological responses seen on functional MRI and PET (3, 5) that arise from increase energy demand with neuronal activation are directly related to this Glx cycle.

The PCC is an area involved in memory and has been studied extensively in Mild Cognitive Impairment (MCI) and Alzheimer’s Dementia (AD) (22, 28). Standard cognitive measures used in AD diagnostics such as the MMSE and the Clinical Dementia Rating Scale Sum of Boxes (CDR-SB) are correlated with posterior cingulate deactivation induced during an associative memory encoding task (29). In other words, clinically less impaired subjects in terms of higher MMSE or lower CDR-SB scores demonstrated greater task-induced deactivation in the posteromedial regions of the default network, and vice versa. Prominent coactivation of the hippocampus suggests that the default-mode network (DMN) is closely involved with episodic memory processing (30). Patients with AD have shown decreased resting-state activity in the posterior cingulate and hippocampus, suggesting that disrupted connectivity between these regions accounts for the posterior cingulate hypometabolism and hypoperfusion commonly detected in PET studies of early AD (31).

We hypothesized that increases in brain excitatory neurotransmitters could result in neuronal hyperexcitability. As part of its neurotransmitter role, Glx is an excitatory amino acid, and excessive Glx neurotransmission has been implicated in excitotoxic neuronal damage (32). Our study confirms a significant reduction in choline (LH/RH/PCC), myo-inositol (LH), NAA (LH/PI) and lutamate (LH) in the FM and STD groups compared with controls (Fig. 2). Indeed, neurochemical changes that could be indicative of such damage have been reported previously (13, 33, 34). These studies report a decrease in NAA in patients with chronic pain in the dorsolateral prefrontal cortex and the thalamus, respectively, two areas also involved with pain processing and perception, and they attribute this loss of NAA to a neurodegenerative process present in chronic pain. Our findings are consistent with a recent ¹H-MRS study, which showed decreased NAA levels within the hippocampus of individuals with FM (35). In another study, a reduction in the absolute concentration of NAA of the right and left hippocampi was reported in a sample of 15 patients with FM (36). The lower hippocampal and insular NAA levels suggest neuronal or axonal metabolic dysfunction, or some combination of these processes. Some studies found that the persistence of elevated Ca²⁺ levels in hippocampal neurons exposed to glutamate correlated with the extent of neuronal death (37) and that a large increase in Ca²⁺ in cultured hippocampal neurons after glutamate application predicted cell death (38). We suggest that hippocampal dysfunction may be, in part, responsible for some of the phenomena associated with FM and STD. Blocking NMDA receptors in the hippocampal formation reduces nociceptive behaviours; this reduction, in turn, supports the hypothesis that the hippocampal formation is involved in the pain-related neural processing and expression of pain-related behaviours (39).

Figure 2.  Box plot representing the ordinal values for the Glx (glutamate + glutamine) in posterior cingulate rea (a), Choline (b), myo-Inositol (c), N-acetyl aspartate (NAA) + N-acetyl aspartatyl glutamate (d), Glutamate (e) in the left hippocampus and NAA + N-acetyl aspartyl glutamate/creatine ratio in the posterior insula (f) of controls, somatization disorder and fibromyalgia.

In the last few years, the notion that the brain has a default or intrinsic mode of functioning has received increasing attention (23). We propose that high levels of glutamate in certain regions of the brain (in this study we found elevated levels of glutamate in the PCC, a key zone in the default network hypothesis) cause cellular damage and disruptions in circuits involved in pain perception underlying the cognitive and behavioural impairments accompanying chronic pain. The chronic pain condition could cause a sustained lesion in the brain through glutamate toxicity and could explain the structural damage and significant atrophy seen in chronic pain patients. The results of this study seem not to correlate with a reduction in cerebral gray matter, because a decrease in gray matter is generally associated with a decrease in N-acetilaspartate (21), which we found not to be the case in our patients.

Apparently, this study found laterality effects concerning metabolites in the hippocampi of patients with FM and SD, as was also found in a previous study in patients with FM (13). Traditionally, somatization symptoms related to emotional disturbance have shown a predominant left-oriented manifestation of the body among patients with hysterical hemianesthesia, hypochondriasis, and psychogenic pain, weakness, or paralysis (40, 41). It has been suggested that the right (non-dominant) hemisphere of the brain is more often than the left associated with the emotional reaction of somatoform symptoms (42). However, the lateralization theory for psychopathology in patients with somatization symptoms or somatoform disorders is not sufficiently documented by scientific research (7, 43). Maybe neuroimage studies could give some light on this matter.

Other remarkable fact is the correlations of all metabolites in the left hippocampus with pain and the correlation of choline in the PCC with all psychological tests. In this sense, higher Cho levels and lower NAA:Cho ratios in both hippocampi have been reported in patients with FM (36). The finding of metabolic brain differences between patients with FM and healthy controls in neural structures such as the hippocampus and amygdala (both of which pertain to the limbic system and are involved in fear, avoidance, and emotional responses experienced during pain) is compatible with a possible augmented emotional processing in patients with FM, in line with the augmented pain processing proposed by some authors (3). A significant mass effect (shift of intracranial structures) is mainly described in brain mass lesions such as contusions or hematomas. These brain mass lesions have not been described previously in either STDs or FM and, therefore, are understood to have no effect on the correlation between the brain metabolites and psychological variables in the three groups.

One of the limitations of this exploratory research is the small sample size (N = 10) of each group of patients. This shortcoming could explain the limited correlations between brain metabolites and psychological variables found when only one or two groups were studied. However, this study has only proposed a new hypothesis; but larger replication studies are needed. Other limitation is that FM is classified as an undifferentiated somatoform disorder. This condition and STD are both included in the same psychiatric classification as somatoform disorders, so there may be a certain overlap between these two types of patients. In addition, despite the patients who were excluded from the study not differing from the rest of the group in terms of sociodemographic and clinical variables, it cannot be ruled out that the metabolite and psychological profiles may have been different for those patients. Another limitation is that the proportion of gray matter/white matter and CSF within the corresponding voxel has not been calculated, so the ¹H-MRS signal arises from the gray and white matter and is an amalgamation of multiple cell types. As a consequence, differences in the fraction of particular tissue types within the voxel may be due to anatomical differences associated with the conditions under investigation. Finally, metabolites were not quantified with the help of segmentation and CSF-correction, so levels of glutamate are often contaminated in part by glutamine, and glutamate exists in the neurotransmitter pool as well as in the metabolic pool.

This study brings glutamate, a key neurotransmitter, to the forefront of our thinking behind the molecular processes involved in FM and STD. It also opens the door for ¹H-MRS studies on other pain states, such as STD, which may also be mediated by central factors. Finally, these data also suggest that reducing glutamatergic activity through the pharmacological treatment of chronic pain may, therefore, not only alleviate suffering and discomfort, but may also be neuroprotective and effective in both FM and STD, two significantly disabling disorders.

Acknowledgements

We thank Professor M. Sarasa of the University of Zaragoza for the LCModel software acquired through the grant SAF2006-13332 from the Spanish Ministry of Science.

Declaration of interest

Miquel Roca has received grants for research from Almirall, Janssen, Lundbeck, Lilly and Servier. Javier Garcia-Campayo has received grants for research from Lundbeck, Lilly, Pfizer and astra-Zeneca. The remaining authors have none to declare.