Glutamate in Schizophrenia: A Focused Review and Meta-Analysis of ¹H-MRS Studies

Abstract

Schizophrenia is a severe chronic psychiatric illness, characterized by hallucinations and delusions. Decreased brain volumes have been observed in the disease, although the origin of these changes is unknown. Changes in the n-methyl-d-aspartate (NMDA)-receptor mediated glutamatergic neurotransmission are implicated, since it is hypothesized that NMDA-receptor dysfunction in schizophrenia leads to increased glutamate release, which can have excitotoxic effects. However, the magnitude and extent of changes in glutamatergic metabolites in schizophrenia are not clear. With ¹H magnetic resonance spectroscopy (¹H-MRS), in vivo information about glutamate and glutamine concentrations can be obtained in the brain. A systematic search through the MEDLINE database was conducted to identify relevant ¹H-MRS studies that examined differences in glutamate and glutamine concentrations between patients with schizophrenia and healthy control subjects. Twenty-eight studies were identified and included a total of 647 patients with schizophrenia and 608 healthy-control subjects. For each study, Cohen’s d was calculated and main effects for group analyses were performed using the random-effects model. Medial frontal region glutamate was decreased and glutamine was increased in patients with schizophrenia as compared with healthy individuals. Group-by-age associations revealed that in patients with schizophrenia, glutamate and glutamine concentrations decreased at a faster rate with age as compared with healthy controls. This could reflect aberrant processes in schizophrenia, such as altered synaptic activity, changed glutamate receptor functioning, abnormal glutamine-glutamate cycling, or dysfunctional glutamate transport.

Introduction

Schizophrenia is a severe chronic psychiatric disease, characterized by hallucinations and delusions, starting in late adolescence or early adulthood. Structural magnetic resonance imaging (sMRI) studies have established that schizophrenia is a brain disease with approximately a 3% tissue loss.¹ Moreover, this loss of brain tissue appears to be progressive as suggested by numerous longitudinal MRI studies.^2–7 It may be explained by reduced neuropil rather than by neuronal loss, implying changes in synaptic, dendritic, and axonal organization in schizophrenia. Reduced neuropil has been observed mostly in the prefrontal region, hippocampus, and thalamus.⁸ This observation, together with the progressive brain tissue loss found in patients suggests that synaptic plasticity and cortical microcircuitry may be abnormal in schizophrenia. Interestingly, progressive brain changes are also present in discordant cotwins of patients with schizophrenia, suggesting that these changes are familial, and possibly genetic, and can therefore not solely be explained by antipsychotic medication intake.⁹

The progressive loss of brain tissue in schizophrenia may represent an ongoing pathophysiological process, which could be an important target for therapeutic intervention. One of the possible mechanisms that may be involved is dysfunction of the glutamatergic system (figure 1), which might affect synaptic plasticity and cortical microcircuitry, in particular (n-methyl-d-aspartate) NMDA-receptor signaling.¹⁰ NMDA-receptors are glutamate-gated ion channels, which play an important role in excitatory neurotransmission, plasticity, and excitotoxicity.^11,12 Indeed, NMDA-receptor antagonists, such as ketamine and phencyclidine (PCP), produce symptoms that mimic psychosis as seen in schizophrenia.^13–16 Depending on the severity and duration of the NMDA-receptor hypofunction state, postsynaptic neurons can develop morphological changes and may cause chronic psychosis and structural brain changes.^17–19 However, in vivo measurement of NMDA-receptor function has been challenging. Although its agonist glutamate can be measured in vivo, evidence of aberrant glutamate levels in schizophrenia is inconsistent. It has been suggested that glutamate levels decrease with age in healthy individuals,²⁰ but it is not known whether glutamate levels change with longer illness duration. Glutamate can be measured using ¹H magnetic resonance spectroscopy (¹H-MRS). ¹H-MRS allows in vivo assessment of the chemical composition of tissues in a noninvasive manner, by using the magnetic resonance signal of hydrogen to determine the concentrations of metabolites. With ¹H-MRS, both physiologically active and inactive glutamate are measured. The majority of physiologically active glutamate is derived from glutamine; therefore, high levels of glutamine may suggest high glutamatergic activity. The glutamatergic transmission and metabolism are strongly coupled, and findings on glutamate and glutamine levels in schizophrenia could give an insight in possible changes in neuronal activity during the course of the disease.²¹ The aims of this structured review and meta-analysis were therefore to determine the extent to which glutamatergic changes occur in patients with schizophrenia compared with those of healthy controls and whether, if present, these changes become more pronounced with increasing age. For this purpose, we evaluated ¹H-MRS studies investigating glutamatergic metabolites in patients with schizophrenia.

Fig. 1.

The glutamatergic synapse. Glutamate is an amino acid, a building block for proteins, therefore, it is abundant in all cells of the body. It is also the most important excitatory neurotransmitter in the central nervous system. Glutamate is synthesized in axon terminals of glutamatergic neurons. It can be produced from α-ketoglutarate -a tricarboxylic acid (TCA)-cycle intermediate- or from glutamine. For glutamate synthesis from glutamine, the enzyme glutamine synthase is transported to the axon terminal. In the cytosol, it converts glutamine into glutamate. Transporters then concentrate glutamate in vesicles. Release of glutamate is triggered by influx of calcium (Ca²⁺) into the presynaptic neuron. The synaptic vesicles fuse with the cell membrane and release glutamate into the synaptic cleft. Glutamate is taken up by the postsynaptic neuron, by glia, or it is recycled in the presynaptic neuron.

Methods

Data Sources

¹H-MRS studies that examined differences in glutamate levels between patients with schizophrenia and healthy control subjects were obtained through a MEDLINE search, using the keywords “glutamate,” “spectroscopy,” and “schizophrenia.” Titles and abstracts of the articles were examined to see whether they fulfilled the inclusion criteria.

Study Selection

Studies were included if they (1) used ¹H-MRS to examine glutamate concentrations in schizophrenia, (2) compared patients with a healthy control group, (3) did not use any interventions (ie, were not subjected to certain actions or imposed to certain behaviors at the time of the study; patients taking antipsychotic medication were included), and (4) were published in the English language. Twenty-eight articles, published between 1994 and 2011, met these criteria. The magnetic field of the MRI-scanners used varied between 1.5 and 4.0 Tesla. The numbers of patients varied between 9 and 40 and the number of healthy controls varied between 11 and 46. Studies were divided based on the stage of illness of their patient populations into high risk, first-episode (FE), and chronic patient studies.

Three studies reported on individuals with a high genetic risk of schizophrenia.^22–24 One study reported on individuals with prodromal symptoms of psychosis.²⁵ Twelve studies reported on FE patients.^26–37 Eight of these studies reported on a patient group of which the majority was antipsychotic-naive.^{26,27,30–35} In 3 studies on antipsychotic-naive FE patients, (part of) the patient group was rescanned after receiving antipsychotic treatment.^32,34,35 In 4 studies on FE patients, the majority of patients had received antipsychotic medication.^28,29,36,37 Four studies report on FE patients and chronically ill patients.^{30 – 32 , 37}

Thirteen studies reported on chronically ill patients,^{30–32,37–44} of which 1 study only included antipsychotic-naive patients,³⁹ 1 study also included unaffected cotwins of patients,⁴⁰ and 1 study also assessed gender differences.⁴² One study reported on a sample consisting of both FE and chronically ill patients.⁴⁵

For 2 studies, the stage of illness of the patient population could not be determined^{46 , 47} (table 1). These 2 studies did provide all the information that was necessary for the meta-analysis and were thus included.

Table 1.

Study Overview

Source	Field Strength	Controls	Patients	Area	Disease Stage	Medication
Bartha et al26	1.5T	10	14	Frontal	first-episode (FE)	none
Bartha et al27	1.5T	11	11	Temporal	FE	none
Bustillo et al28	4.0T	10	14	Frontal	FE	atypical
Bustillo et al37	4.0T	28	30	Whole brain (1 slice)	FE + chronic	?
Chang et al38	4.0T	22	23	Frontal, temporal, occipital	Chronic	atypical
Choe et al39	1.5T	22	23	Frontal	Chronic	none
Galińska et al29	1.5T	19	30	Frontal, temporal, thalamus	FE	atypical
Keshavan et al22	1.5T	46	40	Frontal, temporal, parietal, occipital, basal ganglia	High risk	none
Lutkenhoff et al40	3.0T	21	9	Frontal, hippocampus	Chronic	?
			12		Cotwins	none
Ohrmann et al30	1.5T	20	18	Frontal	FE	none
			21		Chronic	atypical
Ohrmann et al31	1.5T	20	15	Frontal	FE	none
			20		Chronic	atypical
Olbrich et al36	2.0T	32	9	Frontal, hippocampus, amygdala	FE	atypical
Öngür et al41	4.0T	21	17	Frontal, occipital	Chronic	atypical
Reid et al48	3.0T	23	26	Frontal	Chronic	atypical
Rowland et al49	3.0T	11	20	Frontal, parietal	Chronic	atypical
Rüsch et al45	2.0T	31	29	Frontal, hippocampus	FE + chronic	atypical
Shirayama et al46	3.0T	18	19	Frontal	?	atypical
Stanley et al32	1.5T	24	11	Frontal	FE drug-naive	none
			10		FE medicated	?
			11		Chronic	?
Stone et al25	3.0T	27	27	Frontal, hippocampus, thalamus	Prodromal	none
Tayoshi et al42	3.0T	25	30	Frontal, basal ganglia	Chronic	?
Tebartz van Elst et al43	2.0T	32	21	Frontal, hippocampus	Chronic	atypical
Théberge et al33	4.0T	21	21	Frontal, thalamus	FE	none
Théberge et al44	4.0T	21	21	Frontal, thalamus	Chronic	atypical/typical
Théberge et al34	4.0T	16	16	Frontal, thalamus	FE drug-naive	none
			16		FE 10M medicated	atypical
			16		FE 30M medicated	atypical
Tibbo et al23	3.0T	22	20	Frontal	High risk	none
Wood et al47	3.0T	14	15	Frontal	?	atypical
Wood et al35	3.0T	19	15	Temporal	FE drug-naive	none
			19		FE medicated	atypical
Yoo et al24	1.5T	22	22	Frontal, thalamus	High risk	none

Two studies reported insufficient information to calculate Cohen’s d, and these studies were excluded from the meta-analysis.^22,32

In total, 19 studies report on medial (pre) frontal brain areas (including the anterior cingulate cortex),^{22,24–29,33,34,38–42,44,46–49} 7 studies report on the dorsolateral prefrontal cortex,^{24,30–32,36,43,45} 5 studies report on temporal brain areas,^{22,27,29,35,38} 5 studies report on the hippocampus,^{25,36,40,43,45} 8 studies report on the thalamus,^{22,24,25,28,29,33,34,44} 3 studies report on occipital brain areas,^22,38,41 2 studies report on basal ganglia,^22,42 2 studies report on parietal brain areas,^22,49 and 1 study reports results from a whole-brain slice.³⁷ Glutamatergic metabolites that entered the meta-analysis were glutamate, the major excitatory neurotransmitter in the central nervous system, glutamine, a glutamate precursor, and glx, the sum of glutamate and glutamine. Meta-analysis was performed if 3 or more studies on glutamate, glutamine, or glx were available in a particular brain region. Thus, a meta-analysis was performed on glutamate, glutamine, and glx levels in the medial (pre)frontal region and only on glutamate levels in the thalamus and hippocampus. Two studies reported on the same group of subjects^{33 , 34} and to prevent potential bias of overlap, we decided to include only 1 study³³ in the meta-analysis.

Data Extraction

Within a meta-analysis, one defines an effect size statistic, representing the quantitative findings of a set of research studies in a standardized form that permits meaningful comparison and analyses across the studies.⁵⁰ For each study in this meta-analysis, the effects size statistic Cohen’s d was calculated.⁵¹ The Cohen’s d is the difference between the mean of the experimental group and the mean of the control group divided by the pooled SD. The Cohen’s d was calculated as follows. The mean concentration of glutamate, glutamine, or glx for patients were subtracted from mean concentration for comparison subjects and divided by the pooled SD of both. When means and SDs were not available, d-values were calculated from exact P values, t values, or F values. After computing individual effect sizes for each study, meta-analytic methods were applied to obtain a combined effect size, which indicated the magnitude of the association across all studies.⁵² Meta-analyses and meta-regressions were performed with a random-effects model using the statistical package Comprehensive Meta-Analysis V2.⁵³ A random-effects model assumes that the effect size estimated by different studies varies among studies not only because of coincidence but also because of differences in samples or paradigms and that these effect sizes have a normal distribution.⁵⁴ A t test was subsequently performed on the null hypothesis that the d value is 0.00, i.e., indicating no significant difference between the patient and control populations. According to Cohen, absolute d-values of 0.2 represent small effects, d-values between 0.4 and 0.6 moderate effects, and d values of 0.8 or higher large effects.⁵¹In addition, heterogeneity was tested using Cochran’s Q and I ².^55,56 When significant heterogeneity was found, meta-regressions of experimental variables were performed to investigate potential sources of heterogeneity and to identify possible outliers.

When data were available, we also performed a meta-analysis and meta-regression on N-acetyl aspartate (NAA), a potential marker of neuronal viability, to see if changes in glutamate, glutamine, or glx could be a consequence of a decrease in functioning of neurons. A meta-analysis and meta-regression, when data were available, were performed on the gln/glu ratio to explore the changes of glutamate and glutamine relative to each other.

Results

Meta-Analysis of the Medial Frontal Region

Glutamate.

The meta-analysis on glutamate (glu) included 9 studies with a total of 166 patients and 171 healthy controls.^{25,26,28,33,40–42,44–46} The glutamate level in the frontal region was lower in patients than in controls (combined effect Cohen’s d = −0.391, P = .006) (figure 2A).

Fig. 2.

(A). Meta-analysis on frontal region glutamate. Displayed are the Cohen's d and 95% CI per study. The size of the circles is proportional to the study’s weight in the meta-analysis. Studies are ordered according to mean age of the patient group, from young to old. Data were recorded from the following anatomical regions: anterior cingulate cortex (Stone et al²⁵, Théberge et al³³, Bustillo et al²⁸, Öngür et al⁴¹), medial prefrontal cortex (Bartha et al²⁶, Lutkenhoff et al⁴⁰, Shirayama et al⁴⁶). The combined Cohen's d is −0.391 (P = .006). (B). Meta-regression of age of patients on effect size in frontal region glutamate. Each circle represents one study; the size of the circles is proportional to the study’s weight in the meta-regression. (Slope, −0.04; SE of the slope, 0.015; P = .008).

Glutamine.

The meta-analysis on glutamine (gln) included 8 studies with a total of 140 patients and 135 healthy controls.^{25,26,28,33,41,42,44–46} The glutamine level in the frontal region was higher in patients than in controls (combined-effect Cohen’s d = 0.403, P = .045) (figure 3A).

Fig. 3.

(A) Meta-analysis of frontal region glutamine. Displayed are the Cohen's d and 95% CI per study. The size of the circles is proportional to the study’s weight in the meta-analysis. Studies are ordered according to mean age of the patient group, from young to old. Data were recorded from the following anatomical regions: anterior cingulate cortex (Stone et al²⁵; Théberge et al³³; Bustillo et al²⁸; Öngür et al⁴¹), medial prefrontal cortex (Bartha et al²⁶; Shirayama et al⁴⁶). The combined Cohen’s d is 0.403 (P = .045). B. Meta-regression of age of patients on effect size in frontal region glutamine. Each circle represents one study; the size of the circles is proportional to the study’s weight in the meta-regression. (Slope, –0.1; SE of the slope, 0.02; P = .0005).

Glx (Glutamate+Glutamine).

The meta-analysis on glx included 8 studies with a total of 186 patients and 144 healthy controls.^{24,25,29,38,39,47–49} The glx level in the frontal region was lower in patients than in controls but this finding was not significant (combined-effect Cohen’s d = 0.122, P = .393) (see online supplementary figure 1).

Gln/glu Ratio.

The meta-analysis on the gln/glu ratio included 6 studies with a total of 112 patients and 116 healthy controls.^{28,33,41,42,44,46} The gln/glu ratio in the frontal region was higher in patients than in controls but this was not significant (combined-effect Cohen’s d = 0.308, P = .062) (see online supplementary figure 2A)

N-acetyl aspartate.

The meta-analysis on NAA based on the studies included in this meta-analysis, included 19 studies with a total of 401 patients and 378 healthy controls.^{24–26,28–31,33,38–42,44–49} NAA in the frontal region was lower in patients than in controls (combined-effect Cohen’s d = −0.320, P = .019) (see online supplementary figure 3A). The meta-analysis on the glu/NAA ratio included 7 studies.^{26,28,33,40–42,44} The glu/NAA ratio in the frontal region was lower in patients than in controls (combined-effect Cohen’s d = −0.357, P = .038).

Meta-Regression of Age on Effect Size in the Medial Frontal Region

Glutamate.

The meta-regression of age on effect size across 9 studies with a total of 166 patients and 171 healthy controls showed a progressive decrease with age of glutamate in the frontal region in patients as compared with healthy controls^{25,26,28,33,41,42,44–46} (P = .008) (figure 2B).

Glutamine.

The meta-regression of age on effect size across 8 studies with a total of 140 patients and 135 healthy controls showed a progressive decrease with age of glutamine in the frontal region in patients as compared with healthy controls^{25,26,28,33,41,42,44–46} (P = .0005) (figure 3B).

Gln/glu Ratio.

The meta-regression of age on effect size across 6 studies with a total of 112 patients and 116 healthy controls showed a progressive decrease with age of the gln/glu ratio in the frontal region in patients as compared with controls^{28,33,41,42,44,46} (P = .02) (see online supplementary figure 2B).

N-acetyl aspartate.

The meta-regression of age on effect size across 19 studies with a total of 401 patients and 378 healthy controls showed a progressive decrease with age of NAA in the frontal region in patients as compared to controls^{24 – 26 , 28 – 31 , 33 , 38 – 42 , 44 – 49} (P = .04) (see online supplementary figure 3B). The meta-regression of age on effect size across 7 studies with a total of 121 patients and 126 healthy controls showed a progressive decrease with age of the glu/NAA ratio in the frontal region in patients as compared with controls^{26 , 28 , 33 , 40 – 42 , 44} (P = .049).

Meta-Analysis of the Hippocampus

Glutamate.

The meta-analysis of glutamate in the hippocampus included 3 studies, with a total of 47 patients and 60 healthy controls.^25,40,45 No differences in glutamate between patients and controls were observed (combined Cohen’s d = 0.031, P = .92) (see online supplementary figure 4). There were not enough data available to perform a meta-analysis of glutamine in the hippocampus.

Meta-Analysis of the Thalamus

Glutamate.

The meta-analysis of glutamate in the thalamus included 3 studies, with a total of 64 patients and 64 controls.^{25 , 33 , 44} No significant difference was observed for glutamate between patients and controls (combined Cohen’s d = −0.286, P = .20) (see online supplementary figure 5). There were not enough data available to perform a meta-analysis of glutamine in the thalamus.

Outlier and Sensitivity Analysis

We performed an outlier and sensitivity analysis and found 3 potential confounding variables. First, lower magnetic field strengths account for more variation in Cohen’s d. Second, the spectroscopic acquisition method STimulated Echo Acquisition Mode (STEAM) shows less variation in Cohen’s d then the acquisition method Point-RESolved Spectroscopy (PRESS) does. Third, shorter echo times account for more pronounced Cohen’s d effects. For the meta-analyses on the medial frontal cortex, we did not find any outliers.

Discussion

In this meta-analysis of ¹H-MRS studies in schizophrenia, 24 studies were analyzed for changes in glutamate and glutamine. We found that frontal region glutamate is lower, and glutamine is higher in patients as compared with healthy controls. Interestingly, both glutamate and glutamine levels in the frontal region decrease progressively with age in patients with schizophrenia, which could suggest a progressive loss of synaptic activity. This is supported by the gln/glu ratio, which is increased in patients and declines with age. Also, we found decreasing NAA-levels in patients, which may be associated with the progressive brain volume reductions observed in patients with schizophrenia.^2–7 However, the altered frontal glutamate concentration in patients as compared with controls could not be solely explained by the changing NAA levels (with age). Thus, the findings from this meta-analysis suggests a decrease glutamate and an increased glutamine contentration in the brain of patients with schizophrenia, particularly in older patients compared with older controls.

Several limitations have to be taken into account while interpreting the results of this meta-analysis. The most important limitation is that there is only a small number of studies that reported on glutamatergic changes in schizophrenia, as measured with in vivo ¹H-MRS of which most focused on frontal brain areas. It is of course possible that glutamatergic alterations only take place in the frontal region and only in older patients, but the hippocampus and thalamus have been examined in only a few studies, as are the basal ganglia and temporal, parietal, and occipital lobes. Differences in paradigms, such as magnetic field strength, acquisition mode, quantification method, and frontal brain region, also have to be taken into account because this could have an impact on the between-study variability.⁵⁷ Moreover, the progressive decline of glutamatergic levels observed has not been corrected for brain volume changes in this meta-analysis. Several, but not all, of the included studies did take into account partial voluming effects. Therefore, we cannot completely exclude that some of the age-related decline in glutamate concentration in schizophrenia may have been due to brain volume changes. A longitudinal study which assessed patients when medication-naive and after treatment found reductions in precuneal gray matter after 10 months and in frontal, temporal, parietal, and limbic lobes after 30 months of treatment, which was correlated with thalamic glutamine reductions.³⁴ It is therefore too early to make definitive conclusions about the involvement of glutamate in schizophrenia.

Secondly, this meta-analysis only includes cross-sectional data. The conclusion that glutamatergic concentrations progressively decrease with age in schizophrenia is therefore preliminary. Also, the effect size statistic Cohen’s d displays the deviation of patients as compared with healthy controls, which means that possible age-related changes in healthy controls are not taken into account in the meta-analysis. There is evidence though that glutamate changes with age in the healthy brain.²⁰ Unfortunately, medication effects could not be taken into account in this analysis because a major part of the studies did not report sufficient information on antipsychotic intake. The effects of antipsychotic medication cannot be discounted because it has been suggested that haloperidol, clozapine, and olanzapine cause reductions in glutamatergic levels in the rat brain.⁵⁸ However, long-lasting effects of antipsychotic medication have not been found in human studies so far.^28,34

Thirdly, in this meta-analysis, we focused on glutamatergic neurotransmission, however, of course, the measurement of glutamatergic levels alone is not sufficient to draw conclusions about possible neurochemical alterations in patients with schizophrenia. Also, changes in glutamatergic levels may not be detectable with ¹H-MRS. Glutamate and glutamine are difficult to distinguish at lower magnetic field strengths, which makes the quantification of glutamate and glutamine problematic.⁵⁹ Also, ¹H-MRS only provides information about the concentration of metabolites. Other metabolite levels, such as compounds involved in energy metabolism and cell proliferation, should also be taken into account. γ-aminobutyric acid (GABA) is also likely to play an important role. An earlier meta-analysis made clear that N-acetyl aspartate levels are reduced in patients with schizophrenia in frontal lobe gray and white matter.⁶⁰

In conclusion, glutamatergic levels appear to decrease progressively with age in patients with schizophrenia as compared with healthy controls. It is possible though that these progressive reductions are, at least partly, caused by the decrease of brain volume or accumulative intake of antipsychotic medication. In addition, we find an overall increased glutamine level in patients with schizophrenia also in the early (medication-naïve) phase of the disease. Indeed, from the meta-regression, it appears that glutamine levels drop below the healthy control level after patients have reached the age of 35 years, when the majority has reached the chronic phase of the illness. Thus, while we can only draw conclusions with great caution, increased glutamine levels could possibly represent an early marker for glutamate changes in schizophrenia. In fact, the gln/glu ratio is most prominently increased in patients with schizophrenia as compared with the controls at a young age and seems to normalize in older patients This may reflect a deficiency in glutaminase, the enzyme that converts glutamine into glutamate, which results in reduced glutamate levels and increased glutamine levels in the frontal region.⁶¹ In line with this, the amount of glutamate that is released into the synaptic cleft and taken up by astrocytes would decrease. In astrocytes, diminishing levels of glutamate are converted back into glutamine, which in the long term could result in reduced glutamine levels. However, we should be aware that while glutamine represents the major precursor for neuronal glutamate (and GABA), glutamate can also be synthesized from the tricarboxylic acid cycle⁶² and changes in glutamate may also present changes in this second pathway.

There are several possible causes that may explain the altered glutamate levels in schizophrenia. In fact, diminished glutamate receptor activation, in particular of the NMDA-type of glutamate receptor, may be a possible underlying mechanism of reduced synaptic activity because NMDA-receptor antagonists produce the same symptoms as those seen in schizophrenia.^17,18,25 The NMDA-receptor hypofunction hypothesis was evolved from studies with NMDA-receptor antagonists, such as PCP and ketamine.¹⁷ Injection of these NMDA-receptor antagonists leads to increased glutamine levels and decreased glutamate levels^15,63,64 suggesting that NMDA-receptor blockade results in a shift in the glutamine-glutamate cycle.⁶³ Animal studies show that the absence of NMDA-receptor subunits can cause alterations at the molecular and behavioral level and produce schizophrenia-like symptoms.^65–78 This might suggest that changes in NMDA-receptor functioning play an important role in schizophrenia, supporting our current meta-results of age-related decreased glutamatergic levels in patients.

Diminished activation of NMDA-receptors results in insufficient excitatory activity, disrupting the monitoring function of GABAergic inhibitory neurons. In response, these inhibitory neurons may downregulate their own activity.^79,80 This results in disinhibition and hyperactivity of excitatory pathways, which will eventually cause neuronal damage.¹⁷ GABA could thus be an important target for future human in vivo ¹H-MRS research.

Knockout models suggest that other types of glutamate receptors may be involved in the pathophysiology of schizophrenia. There is evidence that knockout of α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA) receptor subunits is implicated in psychosis-like behavior.⁸¹ Deletion of subunits of metabotropic glutamate receptors induce glutamatergic function deficits^{82 , 83} and schizophrenia-like symptoms.^84,85

Interestingly, mouse models have shown that a number of proteins associated with glutamate receptors, some of which are directly linked with schizophrenia susceptibility genes, have been suggested to show modifications in patients, which in turn cause schizophrenia-like phenotypes.^86,87 Another cause of NMDA-receptor hypofunction could be altered expression of intracellular and extracellular glutamate transporters.^88–91 Dysfunction of glutamate transport in the synaptic cleft, and also glutamate transport inside presynaptic neurons, can play a role in the altered glutamatergic levels. Changes in the membrane metabolism in schizophrenia may also be implicated in the changed concentrations of glutamate in schizophrenia. ³¹Phosphorus magnetic resonance spectroscopy has revealed alterations in membrane metabolism in schizophrenia. An increase in membrane breakdown products has been measured in FE schizophrenia, while a reduction of membrane breakdown products has been measured in chronic schizophrenia.^92–94 Together with increased glutamatergic metabolites in early schizophrenia followed by decreased glutamatergic metabolites in chronic schizophrenia, this is consistent with an excitotoxic process.⁹⁴

More information is essential to make solid conclusions on glutamatergic changes over time in schizophrenia. Up until now, studies were mostly cross-sectional and investigated different brain areas. Also, studies did not control adequately for medication intake, duration of illness, and severity of symptoms. It is thus important that future studies take into account the abovementioned limitations and are carried out in a more uniform manner in order to obtain accurate information about alterations in the glutamatergic system in schizophrenia.

Supplementary Material

Supplementary material is available at http://schizophreniabulletin.oxfordjournals.org.

Funding

Netherlands Organization for Scientific Research (NWO) VIDI Grant 917-46-370 (to H.H.); Utrecht University High Potential Grant (to H.H.).