Abbreviations

8.1 Introduction

Major depressive disorder (MDD) and bipolar disorder (BD) are likely to be etiologically diverse, resulting from the contributions of multiple pathophysiologic processes present in affected individuals to varying degrees. In MDD, there is abundant evidence that alterations in serotonin and other monoamines (1) and glutamatergic signaling (2) are implicated in its pathogenesis, and most available treatments target these pathways. Likewise, MDD may result from the effects of inflammatory cytokines (3, 4) or oxidative stress (5) on neuronal activity. One set of processes that may bring together these different etiologies, however, is alterations in brain bioenergetics, which can be studied in vivo using magnetic resonance spectroscopy (MRS).

8.2 Magnetic Resonance Spectroscopy and Brain Bioenergetics

Four major spectroscopic techniques have been used to study MDD and BD: proton magnetic resonance spectroscopy (¹H-MRS), phosphorus magnetic resonance spectroscopy (³¹P-MRS), lithium spectroscopy (⁷Li-MRS), and fluorine magnetic resonance spectroscopy (¹⁹F-MRS). The former two methods allow the measurement of metabolites that are directly and indirectly involved in bioenergetic pathways and will be the focus of this chapter.

As noted elsewhere in this volume (CITE), ¹H-MRS produces a spectrum that encompasses seven major peaks (though more can be identified with special approaches): choline (Cho), N-acetylaspartate (NAA), total creatine (tCr), myoinositol (mI), the amino acids (AA), lipids, and lactate (Lac) (6). Cho includes a variety of choline-containing compounds with similar resonances: choline, acetylcholine, phosphocholine, cytidine diphosphocholine, and glycerophosphocholine (7). Cho represents membrane biochemistry, as choline is a product of myelin breakdown it is increased by rapid cellular proliferation, as in brain tumors; Cho can also be raised by increases in myelin synthesis and may reflect cellular density (7). Cho indirectly represents brain bioenergetics because it indicates cell growth and proliferation (8, 9), but will not be considered further in this chapter.

The tCr resonance reflects both phosphocreatine (PCr) and creatine (Cr), which are in rapid near-equilibrium because of the creatine kinase (CK) reaction, through which ATP and Cr combine, producing adenosine diphosphate (ADP) and PCr; PCr functions as short-term storage for high-energy phosphate when ATP is generated in excess of energy demands (10). tCr is therefore directly related to brain energy metabolism, though early ¹H-MRS studies of MD typically used tCr as a reference, expressing other metabolites as ratios to tCr, on the assumption that tCr levels vary minimally. In fact, tCr can fluctuate under some circumstances, problematizing its use as a reference (11).

The NAA resonance contains N-acetyl aspartate and a small quantity of N-acetyl-aspartyl-glutamate (NAAG) (12). NAA is a direct indicator of mitochondrial activity because it is synthesized in mitochondria (13) by N-acetyltransferase-8 (14) in a fashion correlated with ATP production, oxygen consumption, and glucose utilization (15–17). NAA may serve as a sink for aspartate produced via the mini-citric-acid cycle, which is an alternative pathway for the production of alpha-ketoglutarate for entry into the Krebs cycle (18, 19). NAA is involved in maintaining osmotic homeostasis in cells (20), and is a substrate for the synthesis of NAAG (21). NAAG is a dipeptide composed of NAA and glutamate (22), and may serve as a neurotransmitter (22).

The mI resonance reflects levels of mI, mI–monophosphate, and glycine; mI is the primary constituent of the peak. mI is located primarily in glia such as astrocytes and to be absent from neurons (23), and may be less relevant to the assessment of bioenergetics than other compounds, so will not be further considered.

The AA peak encompasses glutamate (Glu), glutamine (Gln), and gamma aminobutyric acid (GABA) and is designated as “Glx.” Increases in Glx indicate increased cellular destruction or neurotransmission, as Glu is an excitatory neurotransmitter active at the N-methyl-d-aspartate receptor, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor, kainite receptor, and metabotropic glutamate receptor (24, 25). Glu is the most abundant neurotransmitter in the brain (26). It is metabolized to Gln, which is stored in astrocytes; astrocytes pass Gln back to neurons where it is converted to Glu (26) (27). Because Glu is an excitatory neurotransmitter, increases in Glx could indicate increased excitatory neurotransmission and thus increased brain energy consumption; increased neurotransmission is linked to increased Krebs cycle activity and glucose consumption (28, 29). Changes in Glu and related metabolites are, however, explored elsewhere in this volume (CITE) and will not be considered here.

The lipid resonance represents brain levels of acetate and macromolecular proteins and is not directly relevant to the assessment of brain bioenergetics. Finally, the Lac resonance measures cellular energy utilization since Lac is a product of glycolysis, which is used by cells whenever the Krebs cycle and oxidative phosphorylation are not sufficient to support energy demands (30, 31).

³¹P-MRS offers a more direct method of assessing brain bioenergetics because it can measure high-energy phosphate compounds. There are seven chief resonances in the typical ³¹P-MRS spectrum: phosphomonoesters (PME), inorganic phosphate (Pi), phosphodiesters (PDE), phosphocreatine (PCr), γ-ATP, α-ATP, and β-ATP. Of these, PCr is the dominant and central peak and its relevance to assessing brain bioenergetics has already been mentioned. The ATP resonances α, β, and γ reflect the distinct resonances of the three phosphate groups in the compound. Up to 25% of each ATP signal represents other nucleotide triphosphates, such as guanosine triphosphate (GTP), uridine triphosphate (UTP), and cytidine triphosphate (CTP). PME and PDE are indirectly related to metabolism because they reflect rates of lipid synthesis and will not be considered further in this review. ³¹P-MRS also provides an indirect measure of brain pH, which can be estimated from the difference in chemical shift between Pi and PCr (32, 33).

The interpretation of MRS findings in MDD and BD is complicated by multiple factors. We might anticipate that persons with different mood disorders could exhibit different findings even in the same mood state (e.g., depression), that persons with the same disorder might exhibit different findings in different mood states (e.g., mania or depression), and that developmental and degenerative processes associated with age could affect the relative cellular composition and activity of a given brain region, so that adolescents with depression might differ from young adults, who might differ from elderly adults. Similarly, duration of illness and number of mood episodes – both of which affect the clinical manifestations of BD and MDD – could alter metabolite levels, so that persons early in an illness might exhibit different results than persons late in an illness. Women and men may also exhibit different findings because of the effects of sex hormones both concurrently and developmentally. Likewise, medications such as antidepressants (AD), lithium, anticonvulsant (AC) mood stabilizers, and antipsychotics (AP) – many of which have been shown to alter the metabolites studied with both of the major spectroscopic techniques (34) – could alter findings. Moreover, both BD and MDD are likely to result from malfunctions in particular neural circuits, so that abnormalities could exist in brain regions participating in these circuits but not elsewhere. Similarly, tissue type may affect MRS results, as the metabolic activities of gray matter (GM) and white matter (WM) vary widely and are not necessarily coupled. On the other hand, bioenergetic abnormalities might pervade the brain, so that identification of a metabolic abnormality in a given volume of interest (VOI) in persons with a mood disorder does not necessarily show that this region contributes to the condition.

8.3 Major Depressive Disorder

MRS findings in MDD suggest the presence of multiple bioenergetic abnormalities. ¹H-MRS studies (Table 8.1) have focused on multiple VOIs in both adults and adolescents. Some studies have not revealed alterations in NAA in MDD (35–46), but many others indicate that NAA measurements are decreased, including in the right dorsolateral prefrontal cortex (DLPFC) (47), posterior cingulate cortex (PCC) (48), medial prefrontal cortex (MPFC) (49), and left Hippo (50, 51). NAA/tCr ratios have been found to be reduced in the basal ganglia (BG) (52), left prefrontal WM (53, 54), bilateral prefrontal WM (55, 56), bilateral Hippo and Thal (57), left DLPFC WM (58), and left ACC (59).

[NAA] or NAA/tCr ratios in the left DLPFC (60), bilateral VMPFC (61), right Hippo (62) and pregenual ACC (63) have been shown to be negatively associated with duration of illness and with illness recurrence (64). Similarly, lower [NAA] in the right DLPFC (65) and MPFC (49) are inversely correlated with MDD severity. The effect of treatment on NAA measurements appears mixed. In one study (66), there were no differences between controls and MDD subjects who were medication-free for at least three months, though the NAA/tCr ratio in the pregenual ACC fell in MDD subjects after they had been taking AD; the authors hypothesized that this was an effect of illness duration, with which NAA/tCr was correlated. In contrast, another study (67) found that there were no differences in NAA/tCr ratios in the bilateral Hippo between treatment-naive depressed subjects and controls, though these ratios increased significantly in the right Hippo after twelve weeks of treatment with duloxetine. Another, early study (68) found that depressed subjects had lower NAA/Cho ratios at baseline compared to HC, and that these levels increased after two to three months of treatment with the AD nefazodone. Similarly, NAA/tCr ratios were lower in the bilateral Hippo and bilateral Thal of depressed subjects compared to controls at baseline, and increased after six months of treatment with the AD paroxetine (69). In contrast, left DLPFC NAA/tCr ratios did not differ between unmedicated MDD subjects and controls, and did not change appreciably after eight weeks of AD treatment (70). Most recently, subjects treated with adjunctive creatine plus escitalopram exhibited increased [NAA] in the MPFC after eight weeks compared to those received placebo plus escitalopram (49).

Neurostimulation techniques such as electroconvulsive therapy (ECT) and repetitive transcranial magnetic stimulation (rTMS) are often effective treatments for depression (71, 72), and appear to affect brain NAA measurements. Two studies (36, 39) found that ECT had no effect on Hippo [NAA], though another showed that NAA/tCr in the bilateral Hippo fell after ECT, with this change correlation with Hippo volume increases (73). Similarly, dorsal ACC and right Hippo [NAA] fell among subjects treated with ECT (51). In contrast, however, [NAA] in the left Amg increased with response to ECT + AD (74). Similarly, left ACC [NAA] was reduced in MDD subjects taking escitalopram for at least two weeks, though these levels normalized in rTMS responders (75). Psychotherapeutic interventions may affect NAA, as eight weeks of mindfulness-based cognitive therapy was associated with increases in left ACC NAA/tCr ratios (59).

Measures of tCr are less extensively reported in ¹H-MRS studies of MDD, and results are mixed. Multiple studies have reported no difference in [tCr] in several brain regions (35, 37–39, 44, 47, 61, 76). Still, [tCr] in inferior prefrontal WM was increased in MDD subjects compared to controls (53), and a similar finding was demonstrated in the left caudate nucleus (CN), though no differences were found in the ipsilateral Put or Thal (41). Studies have also shown reduced [tCr] in MDD in the L DLPFC (77), PCC (48), and Hippo (51). The last of those studies indicated that treatment with ECT was associated with increases in dorsal ACC and subgenual ACC [tCr]. Likewise, oral creatine supplementation in subjects with MDD significantly increases [tCr] (49).

³¹P-MRS studies in MDD (Table 8.2) provide an additional stream of information about bioenergetically relevant metabolites in MDD. Although some studies have been negative (78, 79), in general there is evidence that MDD is associated with reduced β-NTP. Moore et al. (80) first demonstrated that BG [β-NTP] was reduced in MDD. Later, it was shown that FC (frontal cortex) [β-NTP] was reduced in MDD (81). Renshaw et al. (82) found that although BG [β-NTP] and total purine levels did not differ between MDD and controls, [β-NTP] was 21% lower in fluoxetine responders than nonresponders. In female adolescents with MDD, baseline depression severity is negatively correlated with [β-NTP] (83).

PCr levels have most often been reported to be unchanged in persons with depression, but there are consistent reports that such levels increase with AD treatment. In several studies, [PCr] did not differ in the FC of depressed subjects and controls (81, 84). Still, treatment with acetyl-L-carnitine for twelve weeks was associated with an AD response and [PCr] in the PFC increased in tandem with improvements in depression severity (85). Likewise, Kondo et al. (83) found that adolescent females treated with fluoxetine and adjunctive creatine exhibited increases in [PCr].

A significant barrier to interpreting MRS studies of bioenergetic markers (as well as other metabolites) is differences in GM and WM composition of VOIs. Forester et al. (86) examined thirteen patients with MDD and ten controls and looked at whole-brain metabolites segmented into GM and WM. They found that total tissue (GM+WM) [β-NTP] and the concentration of total NTP ([tNTP]) were lower in depressed subjects and that [tNTP] decreased after twelve weeks of treatment with sertraline. When the authors looked at the two different tissue types, they found that [tNTP] was reduced in WM but not in GM before treatment. In a study of ten older subjects with MDD, with the whole brain segmented into GM and WM, increased WM [β-NTP] was positively associated with the Stroop score, a measure of executive function, while increased GM [PCr] showed the same association (87). The same group later looked at GM versus WM metabolites in a larger study encompassing fifty subjects with MDD. They found that [PCr] was significantly elevated in GM but reduced in WM, and that depression ratings were correlated with GM [PCr], though not with WM [PCr] (88). One way of interpreting these seemingly inconsistent data, which suggest both that AD response is associated with increasing [PCr] and that higher [PCr] and [tCr] are associated with depression, is that increased GM [PCr] is associated with depression but is also a marker of AD response-readiness, such that persons with depression are less able to respond to ADs if they have lower [PCr]. Indeed, this was the interpretation offered by Iosifescu et al. (84), who found that subjects with MDD who responded to triiodothyronine (T3) supplementation exhibited increased [tNTP] but reduced [PCr], while elevated baseline [PCr] predicted response to T3.

There is little data regarding alterations in pH in MDD. In one study, pH was increased in whole-brain GM in unmedicated subjects with MDD, and normalized with treatment with sertraline (86); in another study, pH was increased in MDD in female adolescents in a fashion correlated with depression severity (83). Volz et al. (81) found, however, that pH did not differ significantly between subjects with depression and controls. Reports of alterations in Pi are also limited; in their study of GM versus WM segmentation, Harper et al. (88) found that Pi levels were reduced in the GM of depressed subjects but showed a trend toward being increased in the WM of depressed subjects, correlated with depression severity.

8.4 Bipolar Disorder

Spectroscopic investigations of BD are perhaps more extensive than those of MDD, to date, but interpretation of these studies is rendered more difficult than in MDD because of greater differences in medication exposure as well as the larger number of mood states investigated, including mania (BD-M), hypomania (BD-HM), euthymia (BD-E), depression (BD-D), and mixed (BD-Mx). In general, however, spectroscopic findings are similar across mood states, with exceptions noted later.

¹H-MRS studies in BD (Table 8.3) evince significant perturbations in multiple bioenergetic markers. Most studies in BD have suggested that [NAA] and NAA/tCr are reduced in brain regions implicated in BD, including in the DLPFC (89–92), MPFC (90, 93, 94), Hippo (95, 96), BG (97), PF WM (98), ACC (99), and PCC (48). Even so, some studies have found increased NAA in BD, including in the BG (100), ACC and VLPFC (17), and Thal (101). A few studies have also not demonstrated differences in [NAA] or NAA/tCr, including in the BG (102), frontal lobes (FL) (103), ACC (104–106), and left DLPFC (107–109). Given the number of factors that distinguish these studies, it is difficult to determine whether these differences might be explained by differences in mood state, medication exposure, or other variables, though it should be noted that reduced [NAA] or NAA/tCr has been seen in several brain regions in BD-D, BD-E, and BD-M, as well as in medicated and unmedicated subjects. Still, brain lithium levels have been positively correlated with brain [NAA], suggesting that lithium exposure may be an important confound (110).

BD has also been associated with changes in tCr. Overall, studies appear to suggest that [tCr] is reduced in several brain regions in BD, including in the FL (111), cerebellar vermis (112), Hippo (96), CN (97), MPFC (90, 94), DLPFC WM (90), and PCC (48). Several studies have failed to find alterations in [tCr] in BD, including in the L DLPFC (91, 107, 109) and ACC (105, 113). Dager et al. (114) found that there were no significant differences in [tCr] between unmedicated BD subjects and controls in a variety of GM and WM regions including the MFC, ACC, Put, CN, Ins, Thal, parietal cortex, and OCC. They did, however, show that depression severity in BD was inversely correlated with [tCr]. A few studies have suggested that [tCr] is increased in BD, including in the Thal (114), ACC and MPFC (115), and VLPFC (17). These discrepant findings are not clearly explained by differences in treatment status, mood state, or other factors, suggesting that more research is needed.

PCr constitutes the majority of the tCr signal, so alterations in [tCr] in BD would be expected to coincide with alterations in [PCr]. To date, however, few ³¹P-MRS studies (Table 8.4) indicate that this is true. In BD-E subjects, Kato et al. (79) found that, in a central 30 mm axial slice, [PCr] trended toward being reduced. In a later study, the same group demonstrated lower [PCr] in subjects with BDII compared to controls, though no difference in [PCr] in subjects with BDI compared to controls (116). They also found that L VLPFC [PCr] was significantly reduced in BD-E subjects compared to controls (117). Dudley et al.(118) also showed that [PCr] was reduced in the whole brain as well as right hemisphere GM in BD irrespective of mood state. Multiple other studies, however, indicate that there are no significant differences between BD subjects’ [PCr] and those of healthy controls (119–126). Interpretation of these studies tends, however, to be limited by the fact that they included subjects in different mood states. Several intriguing studies have demonstrated dynamic abnormalities in PCr synthesis in BD. Murashita et al. (123) found that in nineteen subjects with BD, all of whom were treated with lithium, [PCr] fell after photic stimulation (a method of increasing metabolic activity in the visual cortex) in subjects who did not respond to lithium, but remained stable in lithium-responsive subjects and controls. This suggested that subjects with BD have a deficit in PCr production after metabolic stress that is improved by lithium. In a similar study, however, Yuksel et al. (125) compared twenty-three subjects with BD taking various medications to matched controls; they found that [PCr] fell in response to photic stimulation in controls, but not in BD subjects, though PCr/ATP ratios were reduced in BD, and [ATP] fell in BD in response to photic stimulation. Both studies, therefore, suggest some inefficiency in the creatine kinase reaction in BD. Shi et al. (124) used magnetization transfer to estimate the rate constant for the creatine kinase reaction in BD, however, and found that it did not differ significantly between BD-E, BD-D, and controls. In contrast to this, however, Du et al. (121) studied twenty subjects with a first episode of BD-D or BD-M with psychotic features; they found that although PCr/ATP did not significantly differ between BD subjects and controls, there was a 13% reduction in the rate constant for the creatine kinase reaction. The difference between these two studies may be due to the absence of psychosis among the subjects studied by Shi et al., or to the difference in mood states, as many of the subjects studied by Du and colleagues were manic (as well as psychotic).

Table 8.4 3¹P-MRS studies in BD

Publication	Age group	Comparison	VOI	Treatment status	Findings
Kato et al. (79)	Adults	12 MDD (MDD-D + MDD-E), 10 BD (BD-D + BD-E), 12 HC	30 mm frontal axial slice	11 BD-D Li, 3 BD-D Li	↑ pH BD-E vs. BD-D
Kato et al. (127)	Adults	17 BD (BD-M + BD-E)	30 mm frontal axial slice	Li +/- AP	pH ↑ BD-M vs. BD-E; ph ↓ BD-E vs. HC; −Δ pH BD-M vs. HC.
Kato et al. (128)	Adults	31 BDI, 9 BDII, 60 HC	30 mm frontal axial slice	Li	↓ pH BD vs. HC
Kato et al. (116)	Adults	15 BDII, 14 BDI, 29 HC	30 mm frontal axial slice	Various	↓ pH in BDI vs. HC; −Δ PCr in BDI vs. HC; ↓ PCr in BDII vs. HC; −Δ PCr in BDII depressed vs. euthymic
Deicken et al. (146)	Adults	12 BD-E, 14 HC	B TL	UM	PME ↓ BDE vs. HC; otherwise −Δ
Murashita et al. (123)	Adults	19 BD, 25 HC	OCC	Li, then PS	−Δ PCr with PS in Li resp; in Li non-resp; ↓ PCr with PS
Hamakawa et al. (129)	Adults	13 BD-E, 10 HC	B BG	8 Li, 6 AC, 2 AD, 10 AP	pH ↓ BD-E vs. HC; otherwise −Δ
Jensen et al. (122)	Adults	11 BD-D, 9 HC	30 mm axial slice	Various, then triacetyluridine × 6 weeks	−Δ BD vs. HC; ↑ pH triacetyluridine resp vs. non-resp
Brennan et al. (119)	Adults	20 BD-D	WB	Various, then acetyl-L-carnitine or PBO	−Δ pH or other metabolite related to time or treatment arm
Sikoglu et al. (126)	11–20	8 BD, 8 HC	FL	6 Li	pH −Δ BD vs. HC; ↑ pH with age in BD; ↓ Pi in BD vs. HC; −Δ PCr
Weber et al. (117)	11–21	19 BD-M, 14 BD-E, 20 HC	B ACC, L VLPFC	UM >2 weeks	↓ pH + ADP in ACC in BD-M vs. HC; ↓ ADP in L VLPFC in BD-E vs. H; ↓ PCr in L VLPFC in BD-E vs. HC.
Shi et al. (124)	Adults	14 BD-E, 11 BD-D, 23 HC	FL, CorCa, Thal, OL	5 UM, 4 Li, 11 AD, 2 AP, 8 AC	Rate constant for CK reaction −Δ; β-NTP/TP correlated with rate constant for CK reaction.
Yuksel et al. (125)	Adults	21 BD-E, 2 BD-D, 22 HC	OCC	Various, then PS	−Δ PCr, ATP, or pH; ↓ ATP in BD with PS but not HC; ↓ PCr in HC but not BD with PS
Dudley et al. (118)	12–21	16 BD-M, 8 BD-E, 19 HC	WB GM vs. WM	UM	↓ PCr in WB and right hemisphere GM in BD; ↓ ATP in WB and right hemisphere WM in BD
Du et al. (121)	Adults	20 BD + psychosis, 28 HC	FL	Various	PCr/ATP −Δ BD vs. HC; rate constant for CK reaction ↓ 13% in BD vs. HC

Please refer to the Abbreviation list on page 83 for more information if needed.

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

Chapter 1 – Brain Imaging Methods in Mood Disorders Chapter 2 – Neuroanatomical Findings in Unipolar Depression and the Role of the Hippocampus Chapter 7 – Functional Connectome in Bipolar Disorder Chapter 9 – Imaging Glutamatergic and GABAergic Abnormalities in Mood Disorders Chapter 15 – Magnetoencephalography Studies in Mood Disorders Chapter 16 – An Overview of Machine Learning Applications in Mood Disorders

Stay updated, free articles. Join our Telegram channel

Join