Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Oct 23.
Published in final edited form as: World J Biol Psychiatry. 2023 Apr 24;24(8):730–740. doi: 10.1080/15622975.2023.2197653

Basal glutamate in the hippocampus and the dorsolateral prefrontal cortex in schizophrenia: Relationships to cognitive proficiency investigated with structural equation modelling

Jeffrey A Stanley a, Ana M Daugherty a,b, Claire Richter Gorey c, Patricia Thomas a, Dalal Khatib a, Asadur Chowdury a, Usha Rajan a, Luay Haddad a, Alireza Amirsadri a, Vaibhav A Diwadkar a
PMCID: PMC10591941  NIHMSID: NIHMS1930869  PMID: 36999359

Abstract

Objectives:

Schizophrenia is characterised by deficits across multiple cognitive domains and altered glutamate related neuroplasticity. The purpose was to investigate whether glutamate deficits are related to cognition in schizophrenia, and whether glutamate-cognition relationships are different between schizophrenia and controls.

Methods:

Magnetic resonance spectroscopy (MRS) at 3 Tesla was acquired from the dorsolateral prefrontal cortex (dlPFC) and hippocampus in 44 schizophrenia participants and 39 controls during passive viewing visual task. Cognitive performance (working memory, episodic memory, and processing speed) was assessed on a separate session. Group differences in neurochemistry and mediation/moderation effects using structural equation modelling (SEM) were investigated.

Results:

Schizophrenia participants showed lower hippocampal glutamate (p = .0044) and myo-Inositol (p = .023) levels, and non-significant dlPFC levels. Schizophrenia participants also demonstrated poorer cognitive performance (p < .0032). SEM-analyses demonstrated no mediation or moderation effects, however, an opposing dlPFC glutamate-processing speed association between groups was observed.

Conclusions:

Hippocampal glutamate deficits in schizophrenia participants are consistent with evidence of reduced neuropil density. Moreover, SEM analyses indicated that hippocampal glutamate deficits in schizophrenia participants as measured during a passive state were not driven by poorer cognitive ability. We suggest that functional MRS may provide a better framework for investigating glutamate-cognition relationships in schizophrenia.

Keywords: Schizophrenia, hippocampus, dlPFC, glutamate, MRS

Introduction

Schizophrenia (SCZ), a neurodevelopmental disorder with illness onset typically in teens and young adults (Insel 2010), is one of the most debilitating, life-long mental illnesses (WHO 2002; Kessler et al. 2005) and treatment has had limited impact in restoring function. The pathology of SCZ has been attributed to a combination of brain network and neurotransmitter dysfunction that may also be inter-related and lead to subsequent cognitive dysfunction (Benes 2000; Abbott and Bustillo 2006; Carlsson 2006; Brambilla et al. 2007; Diwadkar 2012; Diwadkar, Bustamante, et al. 2014; Diwadkar, Bakshi, et al. 2014). Cognitive dysfunction in SCZ is highly generalised, cutting across domains such as working memory (WM) (Jansma et al. 2004; Tan et al. 2005), executive function (Sullivan et al. 1994), cognitive control (Carter et al. 2001), and learning and memory (Toulopoulou et al. 2003). Notably, sub-networks associated with these domains intersect in the dorsolateral prefrontal cortex (dlPFC) and hippocampus, two areas that lie at the core of the syndrome. The effects of these regional and network deficits are presumably exacerbated by dysfunctions in the interplay between glutamate (Glu) and γ-aminobutyric acid (GABA), the brain’s major excitatory and inhibitory (E/I) neurotransmitters. Glu and GABA function are tightly integrated and together help to facilitate neural engagement. More importantly this integration mediates the neuroplasticity of microcircuits sub-serving cognitive proficiency across multiple domains (Stephan et al. 2006; Isaacson and Scanziani 2011; Lauritzen et al. 2012; Tatti et al. 2017). It has been presumed that Glu is altered in SCZ participants; however, little evidence has attempted to directly relate Glu levels in key areas such as the dlPFC and hippocampus to cognitive ability in SCZ participants using multivariate analytical approaches such as structural equation modelling (SEM) (Castner and Williams 2007; Eichenbaum 2017).

Over the past three decades magnetic resonance spectroscopy (1H MRS) has been a viable method for estimating in vivo Glu levels from localised brain areas in both health and disease. 1H MRS studies in SCZ have demonstrated somewhat consistent results of lower Glu levels in the medial PFC (mPFC) but less consistent results in the dlPFC and hippocampus (Keshavan et al. 2000; Steen et al. 2005; Marsman et al. 2013; Merritt et al. 2021; Smucny et al. 2021). These inconsistencies have vexed the field (Keshavan et al. 2000; Steen et al. 2005; Marsman et al. 2013; Smucny et al. 2021) and are probably driven by multiple factors including (a) sub-optimal methodological applications in detecting and measuring Glu with low precision and accuracy [e.g. reporting Glu as a summation of Glu plus Gln (Glx) or reporting Gln greater than Glu levels], (b) expressing outcome measures as a ratio of levels between metabolites, (c) partial volume effects due to poor localisation of structures, and (d) using inappropriate/incomplete a priori knowledge when modelling 1H MRS data. Another crucial limitation is that 1H MRS was typically acquired without constraining behaviour (i.e. relaxed and simply keeping the head still), creating interpretational challenges. As we have noted (Stanley and Raz 2018), not only is 1H MRS sensitive in detecting dynamic changes in Glu levels induced by task, but the signal is also sensitive in differentiating Glu levels between varying ‘rest’ conditions (e.g. eye closed vs fixating on a crosshair) (Lynn et al. 2018), or sleep/non-sleep (Bartha et al. 1999). Given that behaviour is a strong modulator of neuronal signals and their emergent signatures (Logothetis 2008), we surmise that a lack of behavioural constraint during 1H MRS acquisitions may increase the variability of Glu, undermining sensitivity for detecting group differences, and leading to inconsistencies across studies (Bartha et al. 1999; Lynn et al. 2018).

Here we used a simple behavioural constraint (attention to a visual flashing checkerboard) to constrain the acquisition of Glu in the dlPFC and the hippocampus. The active state provides a measure of visuo-attentional constraint, but without ‘loading’ on either the dlPFC or the hippocampus. Then, using structural equation modelling (SEM) analyses, Glu measures were related to behavioural data in multiple domains [WM, episodic memory (EM), processing speed, PS)], acquired from the same participants in a separate session. The following questions were addressed: (a) do basal Glu levels from the dlPFC and/or hippocampus mediate differences in cognitive ability between SCZ and HC participants; (b) does diagnosis moderate distinct associations between basal Glu levels and cognitive ability?

Methods

Participants

Forty-four DSM-V diagnosed SCZ participants (35 males and 9 females; mean age ± SD, 31.8 ± 8.4 years; age range: 18.9–49.2 years; 40 right-handed) and 39 HC individuals (29 males and 10 females; mean age ±SD, 28.6 ± 6.7 years; 19.8–49.2 years of age; 33 right-handed) provided informed consent to participate. All procedures were approved by Wayne State University’s IRB. SCZ patients were identified by the treating physicians (AA and LH). A research psychologist (UR) confirmed diagnosis using the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders Axis I (SCID-I) (APA 2013). All SCZ participants were stabilised on antipsychotic medication for at least 3 months. Participants with a schizoaffective disorder were excluded (minimizing confounding effects from mood dysregulation).

Clinical symptom severity ratings were carried out using the Positive and Negative Syndrome Scale (PANSS) instrument (Kay et al. 1987) and general intelligence was estimated using the Wechsler Abbreviated Scale of Intelligence (Psychological Corporation 1999). The most likely date of onset of psychotic symptoms (hallucinations, delusions, or disorganisation of thinking; bizarre or catatonic behaviour) and date of diagnosis for SCZ participants were determined using all clinical information, including medical records, reports by family members or significant others, and the SCID interview. HC participants were free of psychiatric treatment or Axis-I psychopathology (past/present). Participants were screened prior to entering the study to exclude any significant past/current medical and/or neurological illness (e.g. hypertension, thyroid disease, diabetes, asthma requiring prophylaxis, seizures, or significant head injury with loss of consciousness). The two groups did not differ in age or gender distribution though Full-Scale IQ was expectedly lower in the SCZ group. Table 1 provides demographic information.

Table 1:

Subject Group Characteristics.

HC Subjects SCZ Subjects
Sample Size 39 44
Mean Age ± SD (years) 28.6 ± 6.7 31.8 ± 8.4a
Range in Age (years) 19.8 to 49.2 18.9 to 49.2
Sex Distribution 29 males 35 males
10 females 9 females
Handedness 85% right 91% right
Full Scale IQ ± SD 101.1 ± 9.3 86.0 ± 7.1b
Length of Illness ± SD (years) 8.6 ± 7.9
Range in Length of Illness (years) 0.5 to 31.3
PANSS Negative Symptoms ± SD 14.3 ± 3.8
PANSS Positive Symptoms ± SD 14.0 ± 3.5
PANSS Total ± SD 54.8 ± 11.0
Antipsychotic Medication
  - 1st Generation Typical 12
  - 2nd Generation Atypical 32
Open in a new tab
a

Not significantly different when compared to the HC group (t= 1.95, p= .054).

b

SCZ < HC: t= 8.22, p< .0001

1H MRS/MRI protocol

MR data were acquired on a 3 Tesla Siemens Verio system using a 32-channel volume head-coil. The acquisition occurred in the morning (10:00–12:00 h) to reduce circadian confounds. A set of T1-weighted axial images covering the brain was first collected [3D Magnetisation Prepared Rapid Gradient Echo, TR = 2150 ms, TE = 3.5 ms, TI = 1100 ms, flip-angle = 8°, FOV = 256 × 256 × 160 mm3, 1601 mm axial slices, pixel resolution = 1 × 1 × 1 mm3 and acquisition time = 4:59 min]. These images were resampled and used to prescribe in order, the placement of the 1H MRS voxel locations: right hippocampus (anterior and body portion; 1.8 × 3.0 × 1.3 cm3 or 7.0 cm3) followed by the right dlPFC (Brodmann area 9/46; 2.0 × 1.5 × 1.5 cm3 or 4.5 cm3). The placement of both voxels in the right hemisphere was driven by previous studies demonstrating 1H fMRS effects in the right hippocampus (Stanley et al. 2017). Angulation and rotation of the MRS voxel was allowed accordingly to minimise the partial volume effect. Additionally, to ensure consistent and reliable placement of the 1H MRS voxels between participants, the location and orientation of these voxels were systematically derived by using a predefined location mapped on a standard brain as described in (Woodcock et al. 2018).

The behavioural constraint imposed during both 1H MRS measurements involved participants passively viewing a flashing visual checkerboard (4 Hz). As we have shown, the constraint ensures that the measurement condition reflects a ‘non-task-active’ steady-state level of Glu with minimal variability (2018). The single-voxel 1H MRS protocol included: the PRESS sequence (The MRS package was developed by Edward J. Auerbach and Małgorzata Marjańska and provided by the University of Minnesota under a C2P agreement.) with outer volume saturation and Variable Power and Optimised Relaxation Delays for water suppression based on Tkác and Gruetter (2005), TE = 23 ms, TR = 3.37 s, 2048 complex points, 2 kHz bandwidth, nine consecutive measurements for the hippocampus and seven consecutive measurements for the dlPFC, with each measurement consisting of 8 averages. Additionally, a fully relaxed water unsuppressed signal with a TR of 10s and 4 averages was acquired in both locations to eliminate any potential T1 partial saturation effects for absolute quantification.

Prior to combining the individual consecutive measurements for each region, the 0th and 1st order phase and frequency shift were corrected (Zhu et al. 1992) using the LCORAW option in LCModel (Provencher 1993). Also, the first measurement in each scan was removed due to signal not reaching the equilibrium steady state. For each combined 1H MRS spectrum, LCModel with a simulated basis-set (Provencher 1993) was used to quantify Glu and the other 1H metabolites including N-acetyl aspartate (NAA), phosphocreatine plus creatine (PCr+Cr), glycerophosphocholine plus phosphocholine (GPC+PC) and myo-Inositol, glutamine, alanine, aspartate, gamma-aminobutyric acid, glucose, glutathione, lactate, n-acetylaspartylglutamate, scyllo-inositol and taurine. FreeSurfer and FSL tools were used to tissue segment the T1-weighted images and estimate the tissue fraction values within each voxel location (Woodcock et al. 2018). The water-unsuppressed water signal, grey matter, white matter, and CSF voxel content values from the tissue segmentation procedure, along with T1 and T2 relaxation times of metabolites, were utilised to quantify absolute levels (Gasparovic et al. 2006; Posse et al. 2007).

Cognitive test battery

Episodic memory (EM)

(a) The Logical Memory subtest from the Wechsler Memory Scale (Wechsler 2009) measures the ability to verbally recall the reading of two stories (passages) one at a time. A score of total correct items for both stories was calculated for the immediate and delayed (20 min) recall. (b) The Memory for Names task from the Woodcock-Johnson Psychoeducational Battery (Woodcock and Johnson 1989) measures associative memory (learning to associate pictures of an imaginary ‘space creature’ with the creature’s name) and was tested twice (immediately and after a 20 min delay) by identifying the creatures named by the examiner. The total number of correct responses were scored for the immediate and delayed recall modes.

Working memory (WM)

(a) The Listening Span task (Salthouse et al. 1989) requires participants to answer simple questions about a sentence while simultaneously remembering the last word of the sentence. Each block contained three trials and the test item increased by one for each successive block with a maximum of seven test items. A total score of correctly answering the question and correctly recalling the final word of the sentence across all trials was calculated. (b) The Computation Span (Salthouse et al. 1989) requires participants to solve simple arithmetic problems while simultaneously remembering the last digit of each problem. Each block contained three trials and the test item increased by one for each successive block with a maximum of seven test items. A total score of correctly answering the question and correctly recalling the final digit of the problem across all trials was calculated.

Processing speed (PS)

Three two-choice reaction time tasks (numeric: odd vs. even numbers; verbal: consonants vs. vowels; and figural: symmetric vs. asymmetric figures) consisting of 40 stimuli each were presented in a randomised fast or slow condition and each task had two sets of trials (Schmiedek et al. 2010). The key outcome measurements reflecting PS were the mean reaction time for each task.

Statistical analyses and modelling

Prior to analysis, data were evaluated for missing patterns, and screened for assumptions of normality and linearity. Of the available sample, 6% (n = 5) were missing regional metabolite measures, 1.2% (n = 1) missing the logical memory test and 2.4% (n = 2) missing the CSPAN test. Data were missing at random (Little’s χ2 = 42.59, p = .49) and cases missing data were removed by pairwise deletion in primary analysis of regional metabolites. The second analysis of cognitive correlates in the SEM framework included all cases with full information maximum likelihood estimation; a covariance estimation that does not require imputation and introduces no bias under the assumption data missing at random (Little et al. 2014). With the available data, one case presented as a univariate outlier in hippocampal NAA and another case in a reaction time measure; and one case was a multivariate outlier. Because univariate and multivariate normality were reasonably met, these cases were maintained in the analyses.

A series of general linear models (including age and grey matter tissue fraction as covariates), were used to determined group differences in metabolite levels (Glu, NAA, PCr+Cr, GPC+PC and myo-Inositol) from each location. As we subsequently note in the results, the high quality of the 1H MRS spectra allowed us to confidently report absolute Glu levels (obviating the need to embed values in a summation of Glu plus glutamine, as is typically done). A Bonferroni correction was applied for multiple comparisons for the five 1H MRS outcome measurements per voxel locations as well as for the ten outcome measurements from the cognitive tasks (i.e. significant p-value threshold of .01 and .005, respectively).

SEM implemented in MPlus (ver7.4) was applied to evaluate the relation between basal Glu levels and cognitive ability as a function of diagnosis. SEM simultaneously tests hypotheses of group differences in latent cognitive ability, accounting for correlations among cognitive domains, and quantifying the unique relation with basal Glu levels in each region. Two alternative hypotheses were tested: (a) do basal Glu levels partially account for cognitive deficits in SCZ participants as compared to HC (i.e. mediation); and (b) does the relation between basal Glu level and cognition differ between SCZ participants and HC (i.e. moderation). Confirmatory factor analysis specified latent constructs of EP, WM, and PS that were equivalent between groups and used in further hypothesis testing. All models included age and gray matter tissue fraction as covariates. Model reliability was evaluated by a compendium of fit indices: χ2 non-significance, comparative fit index (CFI >0.9 indicates good fit), root mean square error of approximation (RMSEA <0.10 indicates good fit), and standardised root mean residual (SRMR <0.08 indicates good fit). Path coefficients, and indirect effects (mediation), were interpreted for effect magnitude and statistical significance (p < .05); all coefficients were bootstrapped with bias correction (5000 draws)(Hayes and Scharkow 2013) to estimate 95% confidence intervals (BS 95% CI). Moderation was tested with a grouped modelling approach including constraints for equal factor loadings and variances, and freely estimating the intercept and variance of regional Glu measures. Group differences in the magnitude of the path coefficients were tested for statistical significance by an approximate z-test (p-value of threshold <.05).

Results

1H MRS spectral quality and voxel placement

Four 1H MRS spectra were rejected for poor quality (dlPFC: 2 SCZ; hippocampus: 1 HC and 1 SCZ), one to error in voxel placement (dlPFC: 1 SCZ) and there were four incomplete scans (dlPFC: not collected in 1 SCZ; hippocampus: not collected in 1 HC and 2 SCZ). The S/N ratio of NAA was comparable between groups for the dlPFC (X2 = 1.44, p = .23) and hippocampus (X2 = 2.92, p = .087) (Table 2). The full-width-at-half-maximum (FWHM) values of NAA were comparable between groups in the dlPFC (X2 = 3.77, p = .052), but demonstrated broader spectral peak in the hippocampus of SCZ participants compared to HC (X2 = 4.57, p = .033) (Table 2). The correlation between FWHM and Glu in either region was not significant. The Cramer-Rao Lower Bound (CRLB) values of Glu ranged between 3% and 8% (mean±SD; 4.2 ± 0.7) for the dlPFC and between 5% and 14% (6.9 ± 1.6) for the hippocampus. An example of a typical quantified 1H MRS spectrum from the dlPFC and hippocampus is shown in Figure 1. Additionally, regarding the consistency in placing the 1H MRS voxel in the two locations, the grey matter tissue fraction values were not significantly different between groups in the right dlPFC (X2 = 3.35, p = .067), though values were lower in the right hippocampus of SCZ vs HC (X2 = 4.81, p = .028) (Table 2).

Table 2:

Mean absolute metabolite levels (±SEM) expressed in institutional units and spectral characteristics for both subject groups in the dlPFC and hippocampus.

dlPFC Hippocampus
HC
(N=39)		 SCZ
(N=40)		 HC
(N=37)		 SCZ
(N=41)
Glu 11.4 (0.118) 11.0 (0.149) 8.87 (0.148) 8.10 (0.166) a
NAA 12.4 (0.144) 12.1 (.166) 7.91 (0.136) 8.17 (.169)
PCr+Cr 8.13 (0.0761) 8.12 (0.0809) 6.80 (0.133) 6.66 (0.0948)
GPC+PC 2.11 (0.0421) 2.08 (0.0381) 2.25 (0.0432) 2.26 (0.0331)
myo-Inositol 6.30 (0.140) 6.52 (0.140) 6.99 (0.125) 6.51 (0.177) b
S/N of NAA 26.1 (0.664) 25.0 (0.708) 11.9 (0.403) 10.9 (0.326)
FWHM (Hz) 4.9 (0.10) 5.3 (0.13) 7.0 (0.32) 7.9 (0.29) c
CRLB of Glu 4.2 (0.073) 6.9 (0.18)
Range in CRLB of Glu 3 to 8 5 to 14
Grey Matter Tissue Fraction (%) 39.7 (1.35) 36.1 (1.44) 58.9 (1.55) 54.3 (1.27) d
Open in a new tab
a

SCZ < HC: X2= 8.12, p= .0044

b

SCZ < HC: X2= 5.14, p= .023

c

SCZ > HC: X2= 4.57, p=.033

d

SCZ < HC: X2= 4.81, p= .028

Figure 1.

Figure 1.

Open in a new tab

From left to right, voxel location (red box) superimposed on the structural MRI images from the (a) right dlPFC and (b) right hippocampus next to examples of typical modelled 1H MRS spectra (black: acquired; red: modelled; blue: modelled Glu signal; and residual below) and plots depicting mean metabolite levels (±SEM) for Glu, NAA, PCr+Cr, GPC+PC and myo-Inositol between HC (blue bars) and SCZ (red bars) participants. *Signify significantly lower Glu and myo-Inositol levels in SCZ compared to HC participants.

Group differences in Glu and other metabolite levels

There were no significant group differences in the Glu level or in the other four metabolites (NAA, PCr+Cr, GPC+PC and myo-Inositol) in the right dlPFC. However, Glu levels were significantly lower in the SCZ group (X2 = 8.12, p = .0044) in the right hippocampus, an effect in which the grey matter tissue fraction term was not significant in the model (X2 = 2.05, p = .15). Additionally, myo-Inositol levels in the right hippocampus were significantly lower in the SCZ group (X2 = 5.14, p = .023) (Table 2).

Cognitive performance

SCZ participants demonstrated deficits on EP performance with significant impairments on all outcome measurements (i.e. immediate, and delayed recall on the Logical Memory and Memory for Names; all p < .0001) (Table 3). Similarly, performance during the WM tests (Computation and Listening Span) was poorer in SCZ participants (both p < .0001) (Table 3). Finally, performance on PS was significantly poorer in SCZ participants with Numerical tokens (p = .0032) but not with Verbal or Figural tokens (Table 3).

Table 3:

Mean cognitive test scores (±SEM) of both HC and SCZ subjects.

Cognitive Domain HC SCZ
Episodic Memory
    Logical Memory (immediate recall) 25.4 ± 1.1 14.0 ± 0.9 p<.0001
    Logical Memory (delayed recall) 22.1 ± 1.2 9.8 ± 0.7 p<.0001
    Memory for Names (immediate recall) 63.3 ± 1.4 50.7 ± 1.7 p<.0001
    Memory for Names (delayed recall) 29.0 ± 1.3 18.4 ± 1.5 p<.0001
Working Memory
    Computational Span 44.8 ± 2.4 29.8 ± 1.6 p<.0001
    Listening Span 45.3 ± 1.9 28.5 ± 1.6 p<.0001
Processing Speed (reaction time in ms)
    Verbal 1,380 ± 68 1,568 ± 69 n.s.
    Numerical 1,214 ± 54 1,467 ± 51 p=.0032
    Figural 2,763 ± 125 2,888 ± 168 n.s.
Open in a new tab

Mediation and moderation effects between Glu and cognition

Prior to testing hypotheses in relation to SCZ diagnosis, we examined the bivariate relations between Glu levels and latent cognitive ability within the entire sample. Right hippocampal Glu showed a significant positive correlation with WM, although other correlations were weak (Table 4). There was little support for the hypothesis that Glu levels may mediate SCZ-related differences in cognitive ability on any of the three domains. The model had excellent fit: χ2 = 83.64, p = .07, CFI = .97, RMSEA = .05, SRMR = .06, and accounted for 95% of variance in latent EM ability (p < .001), 46% in WM ability (p < .001), and 28% in PS (p < .01) (Figure 2). Independent of diagnosis, Glu levels in the right hippocampus (all β = −.11 to .08, p ≥ .24) and right dlPFC (all β = .00 to .11, p ≥ .36) were not significantly correlated with cognitive performance (Figure 2). The cumulative indirect effect of SCZ diagnosis on cognitive ability via Glu levels and gray matter fraction was not significant for EP (standardised indirect = −.03, p = .52; BS 95% CI: −.12, .04), WM (standardised indirect = −.06, p = .19, BS 95% CI: −.14, .01), or PS (standardised indirect = −.03, p = .60; BS 95% CI: −.15, .07). These results indicate that while SCZ participants demonstrated lower Glu levels in the hippocampus, this reduction did not significantly account for diagnosis-related differences in cognitive ability.

Table 4:

Association between Glu levels and cognitive ability by region.

dlPFC Glu Hippocampal Glu
Bootstrapped
95% CI
 Bootstrapped
95% CI
Cognitive Domain β (p-value) LL UL β (p-value) LL UL
Episodic Memory −0.06 (0.65) −0.26 0.13 0.17 (0.18) −0.06 0.38
Working Memory 0.01 (0.96) −0.19 0.17 0.26 (0.02) 0.08 0.42
Processing Speed 0.13 (0.33) −0.10 0.43 −0.02 (0.88) −0.25 0.22
Open in a new tab

Note: Standardized estimates within the structural equation model are reported with respect to latent cognitive domain constructs. Coefficients were bootstrapped with bias-correction to produce 95% confidence intervals (CI), reported with lower and upper level (LL, UL).

Figure 2.

Figure 2.

Open in a new tab

Latent variable path model assessing the effects of diagnosis (SCZ vs HC) on cognition (Episodic Memory, Working Memory and Processing Speed) mediated by neurochemistry (hippocampal Glu vs dlPFC Glu). Latent factor loadings were fixed to 1 and measurement residuals were freely estimated (denoted with *) to specify latent cognitive constructs of each domain. Estimated paths are illustrated with straight arrows, with solid lines indicating significant paths (p < .05) and dashed lines indicating non-significant paths. Model R2 values and significance for each latent cognitive outcome are reported.

The second alternative hypothesis of SCZ diagnosis moderating the relation between Glu level and cognitive ability had partial support. The relation between right dlPFC Glu and PS was positive among SCZ individuals (b = 1.10, p = .09) and negative among HC individuals (b = −.91, p = .12); neither correlation reached significance within group, however the magnitude (and direction) of effects significantly differed between groups (z = −2.01, p = .02; BS 95% CI: −3.85, −0.45). The relation of dlPFC Glu to EP (z = 1.91, p = .08; BS 95% CI: .12, 4.05) and WM (z = −1.64, p = .53; BS 95% CI: −6.33, 2.58) were equivalent between groups. SCZ diagnosis did not moderate the relation between hippocampal Glu and EP (z = .14, p = .89; BS 95% CI: −1.65, 2.05), WM (z = .39, p = .70; BS 95% CI: −3.07, 5.29), or PS (z = −.76, p = .45; BS 95% CI: −2.29, 0.92). Accordingly, hippocampal measures were removed from the model, following which the model fit data with dlPFC measures moderately well: χ2 = 123.78, p = .03 (SCZ = 77.63, HC = 46.15), CFI=.94, RMSEA=.08, SRMR=.11. Taken together, SCZ differences in hippocampal Glu were on a linear continuum with HC, and the independent effect on cognitive ability was smaller than diagnosis-related differences. In contrast, right dlPFC Glu levels were similar between SCZ and HC, but the correlation with PS was different among patients.

Discussion

To our knowledge, this is the first study to investigate inter-group differences in basal levels of Glu (and other metabolites) from both the dlPFC and hippocampus in SCZ participants and controls, when 1H MRS data were acquired under specific behavioural constraint. These constraints are an important determinant of the stability and reliability of estimate Glu in either region (Lynn et al. 2018). We then used SEM analyses to evaluate inter-relationships between Glu, cognitive ability (WM, EP and PS) and diagnosis. We demonstrated: (1) significantly lower basal levels of Glu and myo-Inositol in the right hippocampus of SCZ participants, combined with non-significant group differences in the neurochemistry from the right dlPFC; (2) evidence of poorer performance on all three cognitive domains, WM, EM and PS, in SZ participants compared to HC; (3) a significant association between right hippocampal Glu levels and WM performance across the full sample; (4) SEM-analyses demonstrated non-significant mediation effects between right dlPFC and right hippocampal Glu levels, and group differences on cognitive ability; and (5) while moderation effects were non-significant, the associations between right dlPFC Glu levels and PS were distinct between the groups.

Implication of Glu and Myo-Inositol in SCZ

Consistent with several (including recent 7 Tesla) 1H MRS studies (Merritt et al. 2016; Wang et al. 2019; Godlewska et al. 2021; Smucny et al. 2021; Wijtenburg et al. 2021), basal Glu levels from the dlPFC were not significantly different between SCZ participants and HC. However, basal Glu and myo-Inositol levels from the hippocampus were significantly lower in SCZ participants. This significant effect is notable because both metabolites were expressed as absolute levels, and stands out amidst a majority of null effects [for review, see (Smucny et al. 2021)]. It is worth noting that two recent studies have reported decreased hippocampal Glu and myo-Inositol ratios relative to PCr+Cr in SCZ participants compared to HC, consistent with our current results (Stan et al. 2015; Singh et al. 2018). Additionally, these biochemical deficits in the anterior and body of the right hippocampus were specific to only Glu and myo-Inositol – i.e. NAA, PCr+Cr and GPC+PC basal levels were not different between groups – and after adjusting for tissue content despite SCZ participants demonstrating lower grey matter content in the right hippocampus (Table 2).

As the central excitatory neurotransmitter, Glu is actively engaged in facilitating the neural activity and plasticity that sub serves cognitive ability across domains and brain regions. Moreover, task-induced changes in Gu levels (indexed using 1H functional MRS) show that changes in Glu are driven by shifts in the E/I balance of microcircuits in response to task-related changes in neural engagement (Stanley and Raz 2018). Therefore, if behaviour is not constrained (absent of any neural perturbation) during 1H MRS acquisition, observed deficits in Glu cannot ‘directly’ support a dysfunction in the E/I balance, or an alteration in the modulation or neurotransmission of the glutamatergic system. Instead, lower basal Glu levels in the right hippocampus of SCZ participants may reflect differences in the tissue morphology such as a reduction in the density of the neuropil associated with the glutamatergic system (i.e. related cell bodies, dendritic arbour and supporting cells or astrocytes). This would also be consistent with post-mortem studies in SCZ demonstrating decreased spine density in the hippocampus (Rosoklija et al. 2000).

The other significant finding is reduced hippocampal myo-inositol level in SCZ, which has been previously reported in SCZ participants but mainly in the mPFC (Das et al. 2018; Jeon et al. 2021). Myo-Inositol, which is generally viewed as a cerebral osmolyte, is an intermediate of several important pathways involving inositol-polyphosphate second messengers (Ross and Bluml 2001; Maddock and Buonocore 2012). More importantly, there is evidence of preferential localisation of myo-Inositol in astrocytes and hence, myo-Inositol is often viewed as a marker of glia (Ross and Bluml 2001; Coupland et al. 2005; Kim et al. 2005) – e.g. myo-Inositol levels are significantly higher in the cerebellum where astrocyte content is greater compared to cortical areas (Pouwels and Frahm 1998). Therefore, the evidence of lower Glu levels, as noted above, is consistent with lower myo-Inositol levels as both may potentially reflect reduced neuropil density including cell bodies/dendritic arbour and astrocytes in the right hippocampus of SCZ vs HC.

Mediation and moderation effects

The SCZ sample demonstrated deficits across three domains (WM, EP, and PS) all of which are commonly reported as core features in SCZ (Gold and Harvey 1993; Heinrichs and Zakzanis 1998; Aleman et al. 1999). Also, while WM performance correlated with hippocampal Glu levels across participants, there was no evidence that the observed inter-group difference in hippocampal Glu was driven (or mediated) by the disparity in cognitive ability between groups in any of the three domains. More generally, there was no evidence that associations between Glu and cognition were moderated by diagnosis group. The only exception was a weak and indirect moderation effect demonstrated by the group interaction between dlPFC Glu and PS association. Here, HC demonstrated increasing right dlPFC Glu levels with improving PS while an opposite trend was observed in SCZ. Because SEM is specifically suited for identifying potential mediation/-moderations effects, the lack of an inter-relationship between Glu, cognition and diagnosis is an important negative finding. Our effects extend recent evidence (Reddy-Thootkur et al. 2020) from conventional 1H MRS studies demonstrating little to no support for associations between hippocampal or dlPFC Glu-related measurements (Glu levels or ratios and Glx) with cognitive measurements. Therefore, the above noted hippocampal Glu and myo-Inositol deficits in SCZ, which are presumed to implicate the tissue morphology of the hippocampal microstructure, are not associated with cognitive ability related to WM, EM and PS. This reemphasizes the utility of task-based 1H fMRS over conventional 1H MRS, given the former’s ability to detect dynamic changes in Glu in response to task perturbation. Thus, task-based 1H fMRS may provide greater insight in probing potential dysfunctions in shifting the E/I balance under specific cognitive processes in SCZ (Stanley and Raz 2018).

Strengths and limitations

In general, the hippocampus is particularly difficult to acquire high-quality 1H MRS data, which tends to lead to unreliable Glu measurements [e.g. reporting glutamine levels greater than Glu levels (van Elst et al. 2005; Olbrich et al. 2008; Rusch et al. 2008)]. In this study, the data quality was relatively poorer from the hippocampus compared to the dlPFC; however, the key metrics, S/N, FWHM and CRLB values of Glu (5–14%), were reasonable for reliable Glu quantification. This also suggests that the acquisition of 1H MRS under behavioural constraint is an important determinant in reducing variability (Lynn et al. 2018). We reiterate several additional strengths including (a) utilising an automated voxel placement procedure for consistent/reliable voxel placement across participants (Woodcock et al. 2018), (b) optimising the voxel dimensions of both locations to ensure minimal partial volume effects, (c) utilising a short TE to minimise potential T2 relaxation effects between groups, as well as a long TR of 10s for the water unsuppressed measurement to ensure the complete water signal is acquired, (d) utilising the appropriate tissue fraction values for estimating absolute levels, and (e) performing both the 1H MRS and cognitive battery assessment within days of each other.

Conclusion

These results are the first to show significant and specific reductions in basal Glu and myo-Inositol from the right hippocampus (though not the right dlPFC) in SCZ participants. The observed Glu deficits in the hippocampus were not associated with cognitive performance, and this notable null effect must shape our understanding of brain-behaviour relationships in the illness. Glu and myo-Inositol reductions in the hippocampus appear to reflect more general pathology potentially related to a loss of cellular processing in the neuropil, a pathology that may be uncoupled from cognitive deficits in the illness. Approaches utilising task-based 1H fMRS would be better positioned to address potentially dysfunctions in Glu related to task condition.

Supplementary Material

supplement table

Funding

The research was supported by the National Institute of Mental Health under the Award Number R01 MH111177 (JAS and VAD) and by the State of Michigan (Joe Young Sr./Helene Lycaki funds). The authors thank Caroline Zajac-Benitez and Jonathan Lynn for their assistance.

Footnotes

Statement of interest

None to declare.

Supplemental data for this article can be accessed online at https://doi.org//10.1080/15622975.2023.2197653.

Data availability statement

The 1H MRS, MRI and cognitive performance data are available on request from the authors.

References

  1. [APA] American Psychiatric Association. 2013. Diagnostic and statistical manual of mental disorders. 5th ed. Washington, DC: APA. [Google Scholar]
  2. [WHO] World Health Organization. 2002. Global burden of disease. Geneva: World Health Organization. [Google Scholar]
  3. Abbott C, Bustillo J. 2006. What have we learned from proton magnetic resonance spectroscopy about schizophrenia? A critical update. Curr Opin Psychiatry. 19(2):135–139. [DOI] [PubMed] [Google Scholar]
  4. Aleman A, Hijman R, de Haan EH, Kahn RS. 1999. Memory impairment in schizophrenia: a meta-analysis. Am J Psychiatry. 156(9):1358–1366. eng.le [DOI] [PubMed] [Google Scholar]
  5. Bartha R, Williamson PC, Drost DJ, Malla AK, Neufeld RW. 1999. Medial prefrontal glutamine and dreaming. Br J Psychiatry. 175:288–289. English. [DOI] [PubMed] [Google Scholar]
  6. Benes FM. 2000. Emerging principles of altered neural circuitry in schizophrenia. Brain Res Brain Res Rev. 31(2–3):251–269. [DOI] [PubMed] [Google Scholar]
  7. Brambilla P, Riva MA, Melcangi R, Diwadkar VA. 2007. The role of glutamatergic pathways in schizophrenia: from animal models to human imaging studies. Clinical Neuropsychiatry. 4:199–207. [Google Scholar]
  8. Carlsson A. 2006. The neurochemical circuitry of schizophrenia. Pharmacopsychiatry. 39 Suppl 1:S10–S14. eng. [DOI] [PubMed] [Google Scholar]
  9. Carter CS, MacDonald AW, Ross LL, Stenger VA. 2001. Anterior cingulate cortex activity and impaired self-monitoring of performance in patients with schizophrenia: an event-related fMRI study. Am J Psychiatry. 158(9):1423–1428. [DOI] [PubMed] [Google Scholar]
  10. Castner SA, Williams GV. 2007. Tuning the engine of cognition: a focus on NMDA/D1 receptor interactions in prefrontal cortex. Brain Cogn. 63(2):94–122. eng. [DOI] [PubMed] [Google Scholar]
  11. Coupland NJ, Ogilvie CJ, Hegadoren KM, Seres P, Hanstock CC, Allen PS. 2005. Decreased prefrontal myo-inositol in major depressive disorder. Biol Psychiatry. 57(12):1526–1534. eng. [DOI] [PubMed] [Google Scholar]
  12. Das TK, Dey A, Sabesan P, Javadzadeh A, Theberge J, Radua J, Palaniyappan L. 2018. Putative astroglial dysfunction in schizophrenia: a meta-analysis of (1)H-MRS studies of medial prefrontal myo-inositol. Front Psychiatry. 9:438. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Diwadkar VA. 2012. Adolescent risk pathways toward schizophrenia: sustained attention and the brain. Curr Top Med Chem. 12(21):2339–2347. [DOI] [PubMed] [Google Scholar]
  14. Diwadkar VA, Bakshi N, Gupta G, Pruitt P, White R, Eickhoff SB. 2014. Dysfunction and dysconnection in cortical-striatal networks during sustained attention: genetic risk for schizophrenia or bipolar disorder and its impact on brain network function. Front Psychiatry. 5:50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Diwadkar VA, Bustamante A, Rai H, Uddin M. 2014. Epigenetics, stress, and their potential impact on brain network function: a focus on the schizophrenia diatheses. Front Psychiatry. 5:71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Eichenbaum H. 2017. Prefrontal–hippocampal interactions in episodic memory. Nat Rev Neurosci. 18(9):547–558. [DOI] [PubMed] [Google Scholar]
  17. Gasparovic C, Song T, Devier D, Bockholt HJ, Caprihan A, Mullins PG, Posse S, Jung RE, Morrison LA. 2006. Use of tissue water as a concentration reference for proton spectroscopic imaging. Magn Reson Med. 55(6):1219–1226. [DOI] [PubMed] [Google Scholar]
  18. Godlewska BR, Minichino A, Emir U, Angelescu I, Lennox B, Micunovic M, Howes O, Cowen PJ. 2021. Brain glutamate concentration in men with early psychosis: a magnetic resonance spectroscopy case-control study at 7 T. Transl Psychiatry. 11(1):367. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gold JM, Harvey PD. 1993. Cognitive deficits in schizophrenia. Psychiatr Clin North Am. 16(2):295–312. eng. [PubMed] [Google Scholar]
  20. Hayes AF, Scharkow M. 2013. The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis: does method really matter? Psychol Sci. 24(10):1918–1927. [DOI] [PubMed] [Google Scholar]
  21. Heinrichs RW, Zakzanis KK. 1998. Neurocognitive deficit in schizophrenia: a quantitative review of the evidence. Neuropsychology. 12(3):426–445. eng. [DOI] [PubMed] [Google Scholar]
  22. Insel TR. 2010. Rethinking schizophrenia. Nature. 468(7321):187–193. [DOI] [PubMed] [Google Scholar]
  23. Isaacson JS, Scanziani M. 2011. How inhibition shapes cortical activity. Neuron. 72(2):231–243. English. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Jansma JM, Ramsey NF, van der Wee NJ, Kahn RS. 2004. Working memory capacity in schizophrenia: a parametric fMRI study. Schizophr Res. 68(2–3):159–171. [DOI] [PubMed] [Google Scholar]
  25. Jeon P, Mackinley M, Theberge J, Palaniyappan L. 2021. The trajectory of putative astroglial dysfunction in first episode schizophrenia: a longitudinal 7-Tesla MRS study. Sci Rep. 11(1):22333. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kay SR, Fiszbein A, Opler LA. 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr Bull. 13(2):261–276. [DOI] [PubMed] [Google Scholar]
  27. Keshavan MS, Stanley JA, Pettegrew JW. 2000. Magnetic resonance spectroscopy in schizophrenia: methodogical issues and findings-Part II. Biol Psychiatry. 48(5):369–380. [DOI] [PubMed] [Google Scholar]
  28. Kessler RC, Chiu WT, Demler O, Merikangas KR, Walters EE. 2005. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry. 62(6):617–627. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kim H, McGrath BM, Silverstone PH. 2005. A review of the possible relevance of inositol and the phosphatidylinositol second messenger system (PI-cycle) to psychiatric disorders–focus on magnetic resonance spectroscopy (MRS) studies. Hum Psychopharmacol. 20(5):309–326. eng. [DOI] [PubMed] [Google Scholar]
  30. Lauritzen M, Mathiesen C, Schaefer K, Thomsen KJ. 2012. Neuronal inhibition and excitation, and the dichotomic control of brain hemodynamic and oxygen responses. NeuroImage. 62(2):1040–1050. English. [DOI] [PubMed] [Google Scholar]
  31. Little TD, Jorgensen TD, Lang KM, Moore EW. 2014. On the joys of missing data. J Pediatr Psychol. 39(2):151–162. [DOI] [PubMed] [Google Scholar]
  32. Logothetis NK. 2008. What we can do and what we cannot do with fMRI. Nature. 453(7197):869–878. eng. [DOI] [PubMed] [Google Scholar]
  33. Lynn J, Woodcock EA, Anand C, Khatib D, Stanley JA. 2018. Differences in steady-state glutamate levels and variability between ‘non-task-active’ control conditions: evidence from 1H fMRS of the prefrontal cortex. Neuroimage. 172:554–561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Maddock RJ, Buonocore MH. 2012. MR spectroscopic studies of the brain in psychiatric disorders. Curr Top Behav Neurosci. 11:199–251. eng. [DOI] [PubMed] [Google Scholar]
  35. Marsman A, van den Heuvel MP, Klomp D, Kahn RS, Luijten PR, Hulshoff Pol HE. 2013. Glutamate in schizophrenia: a focused review and meta-analysis of 1H-MRS studies. Schizophr Bull. 39:120–129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Merritt K, Egerton A, Kempton MJ, Taylor MJ, McGuire PK. 2016. Nature of glutamate alterations in schizophrenia: a meta-analysis of proton magnetic resonance spectroscopy studies. JAMA Psychiatry. 73(7):665–674. eng. [DOI] [PubMed] [Google Scholar]
  37. Merritt K, McGuire PK, Egerton A, Aleman A, Block W, Bloemen OJN, Borgan F, Bustillo JR, Capizzano AA, Coughlin JM, et al. 2021. Association of age, antipsychotic medication, and symptom severity in schizophrenia with proton magnetic resonance spectroscopy brain glutamate level. JAMA Psychiatry. 78(6):667. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Olbrich HM, Valerius G, Rüsch N, Büchert M, Thiel T, Hennig J, Ebert D, Van Elst LT. 2008. Frontolimbic glutamate alterations in first episode schizophrenia: evidence from a magnetic resonance spectroscopy study. World J Biol Psychiatry. 9(1):59–63. English. [DOI] [PubMed] [Google Scholar]
  39. Posse S, Otazo R, Caprihan A, Bustillo J, Chen H, Henry P-G, Marjanska M, Gasparovic C, Zuo C, Magnotta V, et al. 2007. Proton echo-planar spectroscopic imaging of J-coupled resonances in human brain at 3 and 4 tesla. Magn Reson Med. 58(2):236–244. English. [DOI] [PubMed] [Google Scholar]
  40. Pouwels PJ, Frahm J. 1998. Regional metabolite concentrations in human brain as determined by quantitative localized proton MRS. Magn Reson Med. 39(1):53–60. [DOI] [PubMed] [Google Scholar]
  41. Provencher SW. 1993. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med. 30(6):672–679. [DOI] [PubMed] [Google Scholar]
  42. Psychological Corporation. 1999. Wechsler abbreviated Scale of Intelligence (WASI) manual. San Antonio (TX): Author. [Google Scholar]
  43. Reddy-Thootkur M, Kraguljac NV, Lahti AC. 2020. The role of glutamate and GABA in cognitive dysfunction in schizophrenia and mood disorders - a systematic review of magnetic resonance spectroscopy studies. Schizophr Res. 249:74–84. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Rosoklija G, Toomayan G, Ellis SP, Keilp J, Mann JJ, Latov N, Hays AP, Dwork AJ. 2000. Structural abnormalities of subicular dendrites in subjects with schizophrenia and mood disorders: preliminary findings. Arch Gen Psychiatry. 57(4):349–356. eng. [DOI] [PubMed] [Google Scholar]
  45. Ross B, Bluml S. 2001. Magnetic resonance spectroscopy of the human brain. Anat Rec. 265(2):54–84. [DOI] [PubMed] [Google Scholar]
  46. Rusch N, Tebartz van Elst L, Valerius G, Buchert M, Thiel T, Ebert D, Hennig J, Olbrich HM. 2008. Neurochemical and structural correlates of executive dysfunction in schizophrenia. Schizophr Res. 99(1–3):155–163. eng. [DOI] [PubMed] [Google Scholar]
  47. Salthouse TA, Mitchell DR, Skovronek E, Babcock RL. 1989. Effects of adult age and working memory on reasoning and spatial abilities. J Exp Psychol Learn Mem Cogn. 15(3):507–516. English. [DOI] [PubMed] [Google Scholar]
  48. Schmiedek F, Lövdén M, Lindenberger U. 2010. Hundred days of cognitive training enhance broad cognitive abilities in adulthood: findings from the COGITO study. Fronti Aging Neurosci. 2:27. English. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Singh S, Khushu S, Kumar P, Goyal S, Bhatia T, Deshpande SN. 2018. Evidence for regional hippocampal damage in patients with schizophrenia. Neuroradiology. 60(2):199–205. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Smucny J, Carter CS, Maddock RJ. 2021. Medial prefrontal cortex glutamate is reduced in schizophrenia and moderated by measurement quality: a meta-analysis of 1H-MRS studies. Biol Psychiatry. 90(9):643–651. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Stan AD, Ghose S, Zhao C, Hulsey K, Mihalakos P, Yanagi M, Morris SU, Bartko JJ, Choi C, Tamminga CA. 2015. Magnetic resonance spectroscopy and tissue protein concentrations together suggest lower glutamate signaling in dentate gyrus in schizophrenia. Mol Psychiatry. 20(4):433–439. eng. [DOI] [PubMed] [Google Scholar]
  52. Stanley JA, Burgess A, Khatib D, Ramaseshan K, Arshad M, Wu H, Diwadkar V. 2017. Functional dynamics of hippocampal glutamate during associative learning assessed with in vivo 1H functional magnetic resonance spectroscopy. Neuroimage. 153:189–197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Stanley JA, Raz N. 2018. Functional magnetic resonance spectroscopy: the “new” MRS for cognitive neuroscience and psychiatry research. Front Psychiatry. 9:76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Steen RG, Hamer RM, Lieberman JA. 2005. Measurement of brain metabolites by 1H magnetic resonance spectroscopy in patients with schizophrenia: a systematic review and meta-analysis. Neuropsychopharmacology. 30(11):1949–1962. [DOI] [PubMed] [Google Scholar]
  55. Stephan KE, Baldeweg T, Friston KJ. 2006. Synaptic plasticity and dysconnection in schizophrenia. Biol Psychiatry. 59(10):929–939. eng. [DOI] [PubMed] [Google Scholar]
  56. Sullivan EV, Shear PK, Zipursky RB, Sagar HJ, Pfefferbaum A. 1994. A deficit profile of executive, memory, and motor functions in schizophrenia. Biol Psychiatry. 36(10):641–653. [DOI] [PubMed] [Google Scholar]
  57. Tan HY, Choo WC, Fones CS, Chee MW. 2005. fMRI study of maintenance and manipulation processes within working memory in first-episode schizophrenia. Am J Psychiatry. 162(10):1849–1858. [DOI] [PubMed] [Google Scholar]
  58. Tatti R, Haley MS, Swanson OK, Tselha T, Maffei A. 2017. Neurophysiology and regulation of the balance between excitation and inhibition in neocortical circuits. [review. Biol Psychiatry 81(10):821–831. English. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. The MRS package was developed by Edward J. Auerbach and Małgorzata Marjańska and provided by the University of Minnesota under a C2P agreement. https://www.cmrr.umn.edu/spectro/
  60. Tkác I, Gruetter R. 2005. Methodology of 1H NMR spectroscopy of the human brain at very high magnetic fields. Appl Magn Reson. 29(1):139–157. English. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Toulopoulou T, Rabe-Hesketh S, King H, Murray RM, Morris RG. 2003. Episodic memory in schizophrenic patients and their relatives. Schizophr Res. 63(3):261–271. [DOI] [PubMed] [Google Scholar]
  62. van Elst LT, Valerius G, Büchert M, Thiel T, Rüsch N, Bubl E, Hennig J, Ebert D, Olbrich HM. 2005. Increased prefrontal and hippocampal glutamate concentration in schizophrenia: evidence from a magnetic resonance spectroscopy study. Biol Psychiatry. 58(9):724–730. [DOI] [PubMed] [Google Scholar]
  63. Wang AM, Pradhan S, Coughlin JM, Trivedi A, DuBois SL, Crawford JL, Sedlak TW, Nucifora FC Jr, Nestadt G, Nucifora LG, et al. 2019. Assessing brain metabolism with 7-T proton magnetic resonance spectroscopy in patients with first-episode psychosis. JAMA Psychiatry. 76(3):314–323. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Wechsler D. 2009. WMS-IV: Wechsler memory scale-fourth edition. San Antonio (TX): Pearson. [Google Scholar]
  65. Wijtenburg SA, Wang M, Korenic SA, Chen S, Barker PB, Rowland LM. 2021. Metabolite alterations in adults with schizophrenia, first degree relatives, and healthy controls: a multi-region 7T MRS study. Front Psychiatry. 12:656459. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Woodcock EA, Arshad M, Khatib D, Stanley JA. 2018. Automated voxel placement: a linux-based suite of tools for accurate and reliable single voxel coregistration. J Neuroimaging Psychiatry Neurol. 3(1):1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Woodcock RW, Johnson MB. 1989. Tests of psychoeducational achievement-revised. Allen (TX): DLM Teaching Resources. [Google Scholar]
  68. Zhu G, Gheorghiu D, Allen PS. 1992. Motional degradation of metabolite signal strengths when using STEAM: a correction method. NMR Biomed. 5(4):209–211. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement table
NIHMS1930869-supplement-supplement_table.pdf (26.4KB, pdf)

Data Availability Statement

The 1H MRS, MRI and cognitive performance data are available on request from the authors.

RESOURCES