Evaluating metabolites in patients with major depressive disorder who received mindfulness-based cognitive therapy and healthy controls using short echo MRSI at 7 Tesla

Major depressive disorder (MDD) affects about 6.7 % of the adults in the United States per year and is the leading cause of disability for people from 15 to 44 years old. The most common treatments include antidepressant medications and psychotherapy. Most patients with MDD have recurrent episodes and often experience treatment failure under current therapies, which severely impact their quality of life. Mindfulness-based cognitive therapy (MBCT), a method for integrating cognitive behavioral therapy with mindfulness meditation, teaches patients how to disengage from habitual “automatic” dysfunctional cognitive routines, in particular depression-related ruminative thought patterns. MBCT was first used for preventing the relapse of depression [1] and has shown its efficacy in reducing depressive recurrence [2] and treating acute [3], chronic [4], and treatment-resistant depression [5] as an augmentation. These findings make MBCT treatment a potential valuable stand-alone treatment for MDD.

Although the etiology and neuropathology of MDD is still not fully understood, it is thought to be associated with chemical changes that involve multiple brain circuits. Proton magnetic resonance spectroscopy (MRS) is a powerful tool for noninvasively investigating brain metabolism. Spectra acquired with long echo times (TEs) provide estimates of metabolite levels, such as total choline-containing compounds (tCho, reflecting membrane synthesis), total creatine (tCr, reflecting cellular bioenergetics), and N-acetyl aspartate (NAA, a neuronal marker). At short TEs, metabolites such as myo-inositol (mI), glutamate (Glu), and glutamine (Gln) also appear, but overlap between peaks can make it difficult to obtain accurate quantification. Spectral editing methods using J-coupling differences allow unobstructed detection of glutathione (GSH) and γ-aminobutyric acid (GABA) but require acquisition from a relatively large region and/or relatively long acquisition time. High field MR systems, such as 7 Tesla (7T) scanners, offer advantages in higher signal-to-noise ratio (SNR) and enhanced spectral quantification for all of these metabolites [6]. Other brain metabolites, such as N-acetyl-aspartyl glutamate (NAAG), glycine (Gly), and glucose that have much lower concentration and/or overlap with the major peaks, can also be detected at high field using ultrashort TE (≤10 ms) MRS or TE-optimized sequences [7‐9].

Multiple studies using single-voxel and spectroscopic imaging have been performed at lower field strengths in patients with MDD and have reported regional and/or global metabolite differences. Although the overall consensus is that there is hypometabolism in patients compared to healthy controls, there was some variability in the results obtained [10]. Efficacy of other MDD treatments was evaluated using MRS in different brain regions. Changes in metabolites, such as Glu, NAA, and tCho, were found to be correlated with response to pharmacotherapy or antidepressant stimulation techniques [11]. Regions of interest (ROI) for these studies involved locations in the basal ganglia, limbic system, frontal cortex, and occipital cortex. Multivoxel 3D magnetic resonance spectroscopic imaging (MRSI) is of interest because it allows characterization of the spatial extent and metabolic properties within multiple regions of the brain.

The purpose of this study was to compare relative levels of metabolites between unmedicated patients with MDD and healthy controls, to evaluate the relationship between metabolite levels and disease severity, and to examine the association between changes in metabolite levels and the outcome of completed MBCT treatment for these patients using 3D short echo-time MRSI at 7T. The patient population comprised a subset of individuals who participated in a previously published project designed to evaluate the efficacy of treatment based upon standard clinical criteria [12]. ROIs within the basal ganglia, limbic system, and frontal cortex were selected for analysis because they have been identified as being relevant for studying psychiatric diseases previously.

This study has successfully demonstrated the application of 7T H-1 MRSI to healthy controls and to patients with MDD before and after treatment with MBCT. As reported by Eisendrath et al. [12] for the larger study that our population was drawn from, MBCT greatly reduced the severity of depression. Changes in metabolite levels after completing MBCT treatment, and the correlation between metabolite levels and depression severity are of interest for future use in evaluating individuals who may benefit from this or other, similar therapies. The main finding in this respect is that differences in metabolite ratios observed in patients at baseline were normalized after treatment and became similar to the levels in healthy controls. To our knowledge, this is the first study showing differences in metabolites in patients with MDD who were treated with MBCT alone.

The improved sensitivity and spectral resolution of the 7T scanner is important for optimizing data quality using in vivo MRSI [8, 22]. The method used in this study to estimate metabolite levels was demonstrated previously and has been applied in other patient populations [15]. To avoid the time penalty of acquiring an additional 3D MRSI acquisition with unsuppressed water for scaling metabolite signals, we chose to reference levels of the metabolites of interest to the levels of tCr in the same voxel. Prior studies have shown that these acquisition and postprocessing methods are able to provide uniform and reliable quantification in the entire spectral array for metabolite ratios, including tCho/tCr, NAA/tCr, Glu/tCr, Gln/tCr, GSH/tCr, mI/tCr, Gly/tCr, and GABA/tCr, within a clinically reasonable total acquisition time of ~10 min [15].

Previous studies evaluating metabolites in MDD used a single-voxel MRS acquisition with a large voxel size and produced variable results [10]. This may be due to differences in the patient population or in tissue components studied and the degree of partial volume that was present. Multi-voxel 3D MRSI has the advantage of providing an assessment of spatial distribution of metabolites. In this study, 3D MRSI was prescribed within a 4-cm excitation slab that was parallel to the AC-PC line and had full coverage of the thalamus. Five types of ROIs were defined within frontal and limbic lobes, insular and subcortical regions. The apriori selection of ROIs was limited by the position of the excitation slab such that the cortex was partially suppressed by the eight automatically prescribed outer-volume lipid suppression and the spatial resolution (1 cm³) used for data acquisition (see Fig. 1). In an attempt to provide robust measures, we used median metabolite levels from voxels that overlapped by at least 40 % with the anatomic ROIs being considered.

The tCho peak resonates at 3.22 ppm at in vivo spectrum and includes signals from Cho, phosphocholine, glycerophosphocholine, and acetylcholine. Previous studies showed that the level of tCho/tCr was increased in the basal ganglion in patients with MDD [23, 24] and that it decreased with antidepressant treatment (nefazodone) [23]. This is consistent with our findings of significantly increased tCho/tCr in the right caudate in MDD, which decreased to levels in controls after completing MBCT treatment. Although an elevation of tCho/tCr in the basal ganglia has been reported after 8 weeks of fluoxetine treatment in patients who responded to the therapy [25], no controls were included in that study. Elevated tCho is thought to be associated with increased cell density and/or membrane turnover. The alteration in tCho observed in our study suggests abnormalities in myelination [26] and/or glial function [27], which are associated with MDD, and/or altered cellular signal-transduction pathways.

NAA is found only in the brain and spinal cord and is synthesized and stored in neurons but hydrolyzed in oligodendrocytes. Since NAA is a neuronal marker, any pathology with a loss of neurons or function results in reduced NAA levels. Vythilingam et al. [24] reported significantly decreased NAA/tCr in the caudate in patients with MDD. NAA has also been studied in the ACC and the frontal lobe in patients with MDD [28‐30], and prior observed that lower NAA in the ACC and decreased NAA/tCr in the MFG at baseline were normalized after antidepressant treatment [30, 31]. This is similar to our results for patients who completed MBCT treatment. Normalization of NAA/tCr within an 8-week treatment period indicates that MDD is more likely to be associated with neuronal dysfunction rather than neuronal loss [32, 33]. We found that significant correlations with percent changes of HAMD-17 scores between baseline and after MBCT treatment were found only for NAA/tCr levels in the left ACC, suggesting that NAA/tCr is a good marker for evaluating response to therapy.

The mI is predominately located within astrocytes and is a precursor for the phosphatidylinositol second-messenger system, which is also presumed to act as an osmoregulator. Significantly decreased mI/tCr levels were previously reported in the frontal lobe in patients with MDD [29, 34], but there have been limited studies of these chemicals in the basal ganglia and insular cortex. The insula plays a role in consciousness and subjective feelings [35], and studies in MDD using functional MRI (fMRI) have detected abnormal interoceptive activity within this region [36]. In our study, decreased mI/tCr was found in the insula in MDD relative to controls, which may indicate glial dysfunction and/or a reduced number of glial cells in patients with MDD [32, 33]. HAMD-17 scores at baseline were significantly correlated with mI/tCr in the insula and putamen, suggesting that mI/tCr may be associated with the severity of depression.

Glu is a main excitatory neurotransmitter in the brain and is converted to Gln by the reabsorption by neurons or reuptake by astrocytes in order to avoid excitotoxicity. It is difficult to separate Glu from Gln in the short TE spectra at 3T, and Glx is typically designated as the sum of Glu and Gln peaks, with Glu as the majority component. A previous study found that Glx was lower in the dorsolateral prefrontal cortex (DLPFC) of MDD patients compared to healthy controls [37] and that Glu was increased in responders after treatment with 10 days of high-frequency transcranial magnetic stimulation (TMS) [38]. Although Glu/tCr in the left caudate at baseline was not significantly different than that in controls in our study (p = 0.051), it returned to normal levels after the 8-week treatment with MBCT. Gln is also a precursor for GSH, an antioxidant that prevents damage from reactive oxygen species. The role of GSH in MDD has not been widely studied. A previous study showed no significant correlation between GSH and depression or mania for patients with bipolar disorder [39]. Our findings of decreased GSH/tCr in insula and putamen indicate that GSH may play an important role in MDD.

GABA is a main inhibitory neurotransmitter within the brain. Previous studies reported lower GABA in the occipital cortex in MDD patients [40] and an increase in GABA with antidepressant treatment [41]. Reduction in GABA is possibly due to impaired GABAergic function and/or decreased GABA synthesis [42]. The occipital cortex was not analyzed in our study because it was not covered by the excitation slab. Although decreased GABA was previously reported in ACC [43], there were no significant differences in GABA/tCr in ACC in our study.

This study supports the effectiveness of the MBCT treatment as a monotherapy for acute MDD and the effectiveness of H-1 MRSI to detect metabolite changes associated with successful MBCT treatment for MDD. Although the number of patients was relatively small, a number of key results that are consistent with our understanding of brain chemistry were obtained. Another limitation is that the significant changes in metabolite ratios could have been caused by alteration in tCr level rather than in other metabolites. The tCr, which has peaks at 3.0 and 3.9 ppm, includes Cr and phosphocreatine, which are involved in adenosine triphosphate (ATP) metabolism. Although some previous studies found no significant differences in tCr in patients with MDD [28], others reported a significant increase in tCr compared to controls [34]. Implementing and adding a fast MRSI acquisition for unsuppressed water may be useful to better understand the roles of tCr in MDD, as well as the other metabolites that were observed different in this study. However, our results suggest that using 3D MRSI at 7T in conjunction with metabolite ratios may be a time-efficient strategy for differentiating patients from healthy controls and following changes due to treatment.

In conclusion, results of our study demonstrate differences in tCho/tCr, NAA/tCr, Glu/tCr, mI/tCr, GABA/tCr, and GSH/tCr in patients with MDD compared to healthy controls and demonstrated changes in these ratios associated with MBCT treatment and reduction in depression severity. Regional differences in metabolite ratios also existed in patients with MDD. These results reflect the complexity of biochemical changes associated with disease processes and treatment effects. Future studies in a larger randomized controlled study population will be helpful for interpreting changes in metabolite ratios and understanding how therapeutic interventions impact MDD. Such information could also be valuable for determining at an early stage whether patients are responding to therapy or whether a new treatment strategy should be considered.