Schizophrenia Research Forum - A Catalyst for Creative Thinking

Kegeles LS, Mao X, Stanford AD, Girgis R, Ojeil N, Xu X, Gil R, Slifstein M, Abi-Dargham A, Lisanby SH, Shungu DC. Elevated prefrontal cortex ?-aminobutyric acid and glutamate-glutamine levels in schizophrenia measured in vivo with proton magnetic resonance spectroscopy. Arch Gen Psychiatry. 2012 May ; 69(5):449-59. Pubmed Abstract

Comments on News and Primary Papers
Comment by:  Dost Ongur
Submitted 19 January 2012
Posted 19 January 2012

This news story by Allison Curley cogently and succinctly describes the current state of affairs in studies of parenchymal GABA levels in schizophrenia. Measuring GABA in vivo in the human brain has been challenging because this metabolite exists in relatively low concentration and its signal overlaps with that of other, more abundant metabolites. The literature has grown recently with the advent of higher-field MRI scanners and reliable MRS approaches for GABA measurement.

As outlined in the story, the several papers on parenchymal GABA levels in schizophrenia are about evenly split, with reductions and elevations both being reported. Although MRS is characterized by a relatively low signal-to-noise ratio and high variance in most datasets, all the recent studies used reliable MRS techniques such as MEGAPRESS.

In my opinion, the current state of the literature offers two insights:

1. If there was a significant and consistent abnormality in parenchymal GABA levels in schizophrenia, we would have found it and the studies would agree. Rather, it appears that there may be patient and treatment factors leading to differential GABA patterns. For example, to speculate: elevations in early illness may be replaced by reductions with chronic disease, or anticonvulsants may elevate GABA levels while antipsychotics reduce them. Larger datasets with more detailed phenotypic analyses may provide leads for developing a clearer picture. Alternatively, and less interestingly, there may be no or minor abnormalities which result in conflicting findings due to sampling error, technical differences, etc.

2. As a corollary to any of the possibilities above, it is clear that abnormal GABAergic neurotransmission is not necessarily associated with consistently reduced parenchymal GABA levels as measured by MRS. Postmortem and other lines of evidence are quite convincing of abnormalities in GABAergic interneurons in schizophrenia. However, the in-vivo MRS studies are much less consistent, suggesting a disconnect between the two lines of inquiry. Just to describe one possibility, it is possible that GABA is inappropriately stored in synaptic vesicles instead of being released into the synapse and subsequently metabolized, setting up elevated GABA levels but reduced GABAergic neurotransmission.

Although confusing at the moment, the optimistic view is that MRS studies of brain GABA levels in schizophrenia will ultimately offer a more sophisticated understanding of the relationship between metabolite levels measured using MRS and the brain functions we all care about.

View all comments by Dost Ongur

Primary Papers: Elevated prefrontal cortex ?-aminobutyric acid and glutamate-glutamine levels in schizophrenia measured in vivo with proton magnetic resonance spectroscopy.

Comment by:  Reiji Yoshimura
Submitted 23 January 2012
Posted 25 January 2012
  I recommend this paper

The GABA theory of schizophrenia is very attractive. I read with much interest the paper from the labs of Lawrence Kegeles and Dikoma Shungu. The authors demonstrated a 30 percent elevation in GABA levels in the medial prefrontal cortex. We previously investigated brain GABA levels in three regions in early-stage, first-episode medicated and unmedicated schizophrenia patients (Goto et al., 2009). GABA was decreased in the left basal ganglia but unchanged in the frontal lobe. We consider the inconsistency of reports regarding brain GABA in schizophrenia to be mainly attributable to methodological issues. An easier and more accurate way to measure brain GABA might confirm the GABA hypothesis of schizophrenia.


Goto N, Yoshimura R, Ueda N, et al. Reduction of brain gamma-aminobutylic acid (GABA) concentrations in early-stage schizophrenia patients: 3T Proton MRS study. Schizophr Res 2009; 112: 192-3. Abstract

View all comments by Reiji YoshimuraComment by:  Jong H. YoonRichard J. Maddock
Submitted 8 February 2012
Posted 8 February 2012

The study by Kegeles et al. has added unique and important findings to the small but rapidly growing literature assessing in-vivo GABA levels in schizophrenia using MRS. In the context of these studies, the Kegeles publication also raises several challenging questions regarding the potential relevance and reliability of in-vivo GABA studies. Here, we would like to comment on two of these questions. The first pertains to the lack of convergence with the consistent postmortem studies. The second is the apparent lack of consistency across the recent in-vivo GABA studies in schizophrenia.

A starting point in the discussion of the first issue is to recognize the differences in what we are measuring with in-vivo spectroscopy as opposed to the postmortem studies. The latter have consistently demonstrated decreased mRNA levels for GAD67, one of the major synthetic enzymes for GABA, in a subset of GABAergic interneurons in the neocortex of schizophrenia. Based on this postmortem work and the important role GAD67 plays in determining whole cell content of GABA (Asada et al., 1997), many, including Kegeles and coauthors, had predicted MRS measurements of GABA would be decreased in schizophrenia. Spectroscopy measures bulk GABA, the largest fraction of which is cytoplasmic and not vesicular. The cytoplasmic fraction of GABA is synthesized by GAD67 (abnormal in postmortem studies of schizophrenia), while the vesicular fraction is synthesized in part by GAD65 (not apparently abnormal in schizophrenia) (Waagepetersen et al., 2007). While vesicular GABA is the source of GABA for synaptic neurotransmission, cytoplasmic GABA may play a role in both tonic and phasic inhibition mediated by extrasynaptic GABAergic signaling (Wu et al., 2007). One of the major limitations of MRS measurements of GABA, therefore, is that we currently do not really understand to what extent this bulk measurement relates to neural signaling. However, there are a growing number of studies (Edden et al., 2009; Sumner et al., 2010), including one by our group (Yoon et al., 2010), that suggest that bulk GABA measurement is a functionally meaningful measure. These studies have shown high correlations between MRS estimates of GABA and performance on tasks presumably dependent on the magnitude of GABA-mediated inhibition. In addition, animal studies have suggested that the concentrations of vesicular and non-vesicular pools of GABA appear to be in equilibrium (Waagepetersen et al., 1999), implying that bulk GABA levels reflect, to some degree, the vesicular fraction. Nonetheless, as others have pointed out, the diverse components of the GABA MRS measurements leave open a number of potential explanations as to why bulk GABA levels may not be decreased in schizophrenia in the setting of decreased GAD67 mRNA levels.

The second set of questions concerns the apparent lack of consistency among the recent set of in-vivo GABA studies. The potential reasons for this are many and diverse, and include clinical and neuroimaging-related factors that may have varied across the spectroscopy studies, including differences in illness severity, length of illness, brain regions assessed, and methods for GABA quantification. The Kegeles paper has identified medication status as an important clinical variable for which future studies should attempt to account. In-vivo GABA spectroscopy using MEGA PRESS is a relatively new method, particularly as applied to between-group studies. Consequently, there may be a number of neuroimaging-related variables that are important sources of noise or diminished signal, leading to false-negative findings of group differences, or bias, leading to false-positive findings of group differences. An example of the former relates to the phased array head coils frequently used in GABA studies. With these receive-only coils, signal strength decreases linearly as a function of the distance between the coil element that detects the spectroscopy signal and the brain region being sampled. Thus, the signal from brain regions farther away from these elements, for example, deep midline and subcortical regions, will be much lower than regions that are adjacent to these elements, for example, the occipital pole. Consequently, our ability to detect true differences between groups in these low-signal regions will be constrained. Another important variable may be in-scanner head movement. From our own work, we are coming to believe that in-scanner head movement may produce significant over- or underestimation of true GABA concentration, depending on the type of movement. The effect of head movement may be particularly important in between-group studies in which one group may exhibit a significantly different amount of movement compared to the other group. Even a few patients with excessive movement during a prolonged MRS acquisition could generate outlying and erroneous GABA values and lead to false-positive group differences.

In summary, we are in the very early stages of MRS studies of GABA in schizophrenia. There are many unanswered questions regarding the meaning of this signal and how it relates to GABA physiology, function, and their impairment in schizophrenia. The answers to these questions will require intense efforts relying on animal and human models to unravel the complex relationships between bulk GABA measurements and GABA signaling. As a methodology, much more work needs to be done to validly and reliably translate this method to clinical studies. In the immediate future, it will be critical to identify the important sources of noise and bias, and to develop methods controlling for these variables in clinical studies so that the true nature of GABA levels in schizophrenia may be established.


Asada H, Kawamura Y, Maruyama K, Kume H, Ding RG, Kanbara N, et al (1997): Cleft Palate and Decreased Brain Gamma-aminobutyric Acid in Mice Lacking the 67-kDa Isoform of Glutamic Acid Decarboxylase. Proc Natl Acad Sci U S A 94:6496-6499. Abstract

Edden RAE, Muthukumaraswamy SD, Freeman TCA, Singh KD. (2009) Orientation Discrimination Performance Is Predicted by GABA Concentration and Gamma Oscillation Frequency in Human Primary Visual Cortex. Journal of Neuroscience 29(50):15721-15726. Abstract

Sumner P, Edden RAE, Bompas A, Evans JC, Singh KD (2010) More GABA, Less Distraction: a Neurochemical Predictor of Motor Decision Speed. Nature Neuroscience 13:825-827. Abstract

Waagepetersen HS, Sonnewald U, Larsson OM, Schousboe A. (1999) Synthesis of Vesicular GABA From Glutamine Involves TCA Cycle Metabolism in Neocortical Neurons. Journal of Neuroscience Research 57:342-349. Abstract

Waagepetersen HS, Sonnewald U, Schousboe A (2007) Glutamine, Glutamate, and GABA: Metabolic Aspects. In: Lajtha A, Oja S, Schousboe A, Saransaari P (eds) Handbook of Neurochemistry and Molecular Neurobiology: Amino Acids and Peptides in the Nervous System. Springer, New York, pp 1-21.

Wu Y, Wang W, Diez-Sampedro A, Richerson GB (2007) Nonvesicular Inhibitory Neurotransmission Via Reversal of the GABA Transporter GAT-1. Neuron 56:851-865. Abstract

Yoon JH, Maddock RJ, Rokem AS, Silver MA, Minzenberg MJ, Ragland JD, Carter CS. (2010) Gamma-aminobutyric Acid Concentration is Reduced in Visual Cortex in Schizophrenia and Correlates with Orientation-Specific Surround Suppression. Journal of Neuroscience 10;30(10):3777-81. Abstract

View all comments by Jong H. Yoon
View all comments by Richard J. MaddockComment by:  Robert McCarleyMargaret NiznikiewiczMartina M. VoglmaierKevin Spencer (Disclosure), Nick BoloAlexander P. LinYouji HiranoElisabetta del ReIsrael MolinaVicky LiaoSai Merugumala
Submitted 13 February 2012
Posted 14 February 2012
  I recommend the Primary Papers

The important and elegantly controlled work by Kegeles et al., and the informed comments of Ongur, Yoshimura, and Yoon and Maddock, on GABA in schizophrenia raise a series of potentially key factors about the sources of variability of MRS findings in this disorder (medication, stage of illness, and region of interest [ROI]). They also point out the need for association of MRS GABA findings with physiologic measures such as γ oscillations (40 Hz), a functional measure particularly relevant because of the involvement of GABA interneurons interacting with pyramidal neurons in generating this oscillation.

We would like to call the reader's attention to a potentially informative schizophrenia spectrum disorder, schizotypal personality disorder (SPD), that may help shed light on and respond to these issues. As has been documented by Kendler (Kendler et al., 1993; Fanous et al., 2007), SPD shares a genetic relationship with schizophrenia. Although sharing the symptoms of schizophrenia, SPD individuals have an attenuated version of the symptoms and are not psychotic. One thus can recruit SPD individuals who are living in the community, have never been neuroleptic medicated, who have no current medication, and who do not show the profound lifestyle disturbance of individuals with schizophrenia.

We have begun MRS evaluations on SPD subjects with these characteristics, choosing ROI in the superior temporal gyrus (STG) because of the strong evidence of the association of this region with the auditory steady-state (ASSR) γ oscillation response, as well as structural MRI evidence for left STG reduced gray matter volume. Our still quite preliminary data showed, compared with matched healthy controls, a mean reduction in GABA levels and an increase in glutamate. Although the levels were not yet statistically significantly different in our preliminary data, what was notable, and statistically significant, was the very high correlation of the left STG glutamate and GABA levels with the levels of the ASSR γ oscillation, measured as the strength of the phase locking factor (PLF) over left-sided electrodes. As predicted, GABA levels were positively correlated with the PLF, while glutamate levels were inversely (negatively) correlated with the PLF. Obviously, more data are needed, but these initial findings suggest the promise of using SPD subjects with both MRS and γ oscillation measurements in the STG.


Preliminary results to be presented at the 3rd Biennial Schizophrenia International Research Society Conference 14-18 April 2012, Florence, Italy, as a poster and an oral presentation, and at the 20th Annual Meeting of the International Society of Magnetic Resonance in Medicine 5-11 May 2012, Melbourne, Australia.

Kendler KS, McGuire M, Gruenberg AM, O'Hare A, Spellman M, Walsh D. (1993). The Roscommon Family Study. III. Schizophrenia-related personality disorders in relatives. Arch Gen Psychiatry, 50(10):781-788. Abstract

Fanous AH, Neale MC, Gardner CO, Webb BT, Straub RE, O'Neill FA, Walsh D, Riley BP, Kendler KS. Significant correlation in linkage signals from genome-wide scans of schizophrenia and schizotypy. Mol Psychiatry. 2007 Oct;12(10):958-65. Abstract

View all comments by Robert McCarley
View all comments by Margaret Niznikiewicz
View all comments by Martina M. Voglmaier
View all comments by Kevin Spencer
View all comments by Nick Bolo
View all comments by Alexander P. Lin
View all comments by Youji Hirano
View all comments by Elisabetta del Re
View all comments by Israel Molina
View all comments by Vicky Liao
View all comments by Sai MerugumalaComment by:  Lawrence KegelesDikoma C. Shungu
Submitted 4 April 2012
Posted 5 April 2012

The news story by Allison Curley on our recent paper gives a concise and insightful overview of in-vivo studies of GABA levels in schizophrenia. As the story notes, for those keeping score, studies measuring GABA in schizophrenia are evenly split in that two showed increases, two found decreases, and one reported no change. A major theme running through the thoughtful commentaries by Ongur, Yoshimura, Yoon and Maddock, and McCarley and colleagues is how to understand the variability across studies.

Some regularities can already be found in these and similar studies of the glutamate system. If we confine the scorekeeping to GABA in the prefrontal cortex (PFC), the studies are more uniform: two showed increases (Ongur et al., 2010; Kegeles et al., 2012) and two showed no change (Goto et al., 2009; Tayoshi et al., 2010). If we further limit attention to unmedicated patients, but broaden the review to include the glutamatergic system as well as GABA in the PFC, the studies all agree: glutamine, glutamate-glutamine (Glx), or GABA is increased in the medial PFC (Bartha et al., 1997; Théberge et al., 2002; Théberge et al., 2007; Kegeles et al., 2012), but unchanged in the dorsolateral PFC (Stanley et al., 1996; Ohrmann et al., 2007; Kegeles et al., 2012).

It is encouraging to find patterns where we can, but so far these are limited. We still need (and have begun) to investigate other important brain regions, and it is essential to understand the effects of antipsychotic medication. The commentary by Yoshimura describes the subjects studied by Goto et al. (2009) as medicated and unmedicated, and we wonder if a comparison between those subsamples, as we did in our study, might be informative.

Besides brain region and medication status, the commentaries suggest other patient, treatment, or technical measurement factors contributing to the variability. These include chronicity or duration of illness, medications other than antipsychotics such as benzodiazepines and anticonvulsants, and specifics of MRS methodology. We share these views and encourage any efforts to find systematic impacts of these variables.

Yoon and Maddock raise technical cautions: use of phased-array head coils can limit signal detection in deeper brain regions, and movement artifacts might introduce spurious group differences. As their commentary notes, regions adjacent to the coil elements, such as the occipital lobe (or the dorsolateral PFC) will yield greater signal than deeper structures. In our study, it was the surface region, the dorsolateral PFC, where no group difference was detected, and the slightly deeper medial PFC that showed differences, suggesting adequate sensitivity in the deeper region. Acquisition parameters can be used to offset the coil depth effect. In our study, we enhanced the medial PFC signal by doubling the volume, tending to offset the greater distance from the coil array. Head movement might raise special concerns as a source of artifact in a technique such as MEGA PRESS that relies on subtraction of sequentially acquired spectra, and Yoon and Maddock raise the possibility of resulting over- or underestimation of GABA concentration. Evidence that this may not have occurred in our study is the agreement of our Glx data with prior studies in both medial (Bartha et al., 1997; Théberge et al., 2002; Théberge et al., 2007) and dorsolateral PFC (Stanley et al., 1996; Ohrmann et al., 2007) that did not use MEGA PRESS. Our Glx and GABA measurements that did use MEGA PRESS were correlated and were both elevated in medial PFC, so the agreement with prior methodologies seems to lessen the likelihood of artifacts specific to subtraction methodology. Also, the deeper region (medial PFC) would be expected to undergo less movement than the surface region, yet showed the elevations. Additional evidence that movement artifact may not be a confounder in MEGA PRESS measurements is a recent study by Hasler et al., (2007) in major depression, where a very different pattern of abnormalities was seen in medial PFC (decreased Glx and unchanged GABA). It seems unlikely that patients with depression and schizophrenia would exhibit movement patterns systematically different from controls, yet so different from each other as to have generally opposite impacts on the outcome measures. However, these are all indirect considerations. Systematic characterization of movement effects in MEGA PRESS and other acquisition sequences could add important specific data on potential artifacts, and these issues deserve further study.

Another theme of the commentary is the apparent discrepancy between postmortem markers of GABA function and parenchymal GABA measured in vivo with MRS. There is a clear indication of diminished GABA function associated with fast-spiking, parvalbumin-positive GABA interneurons in the postmortem findings, yet we reported an elevation of parenchymal GABA concentration in vivo in the medial PFC. Ongur’s commentary raises the interesting possibility of abnormally increased storage in synaptic vesicles, while Yoon and Maddock cite evidence from animal studies of equilibrium between vesicular and non-vesicular GABA pools. Possibilities are a disruption of this normal equilibrium in schizophrenia and, alternatively, a compensatory increase in GABA signaling from the non-parvalbumin interneurons. These speculative possibilities raise the questions of detectable postmortem markers of abnormal vesicular function or heightened signaling by the non-fast-spiking interneurons.

Finally, the commentaries offered important observations on the functional role of total tissue GABA levels. Since neurotransmission is only one of several compartments contributing to parenchymal GABA, it is reasonable to wonder whether this MRS measurement has any detectable functional significance at all. Our study found no relation between elevated parenchymal GABA and working memory performance. We did find a relationship to positive symptoms that did not survive multiple comparisons correction, but suggests a focus for future testing. Yoon and Maddock cite several studies documenting functional importance of total GABA (Edden et al., 2009; Sumner et al., 2010; Yoon et al., 2010). McCarley and colleagues note in their commentary that relationships to physiological measures such as gamma oscillations suggest that bulk GABA is functionally meaningful (see also Muthukumaraswamy et al., 2009).

In the end, if we can develop a consistent picture of GABA abnormalities in schizophrenia, the primary motivation for all of these studies is to establish their functional relevance, and to raise the possibility of interventions designed to restore not only normal levels, but also, more importantly, normal function.


Bartha R, Williamson PC, Drost DJ, Malla A, Carr TJ, Cortese L, Canaran G, Rylett RJ, Neufeld RWJ (1997) Measurement of glutamate and glutamine in the medial prefrontal cortex of never-treated schizophrenic patients and healthy controls by proton magnetic resonance spectroscopy. Arch Gen Psychiatry 54:959-65. Abstract

Edden RA, Muthukumaraswamy SD, Freeman TC, Singh KD (2009) Orientation discrimination performance is predicted by GABA concentration and gamma oscillation frequency in human primary visual cortex. J Neurosci 29:15721-6. Abstract

Goto N, Yoshimura R, Moriya J, Kakeda S, Ueda N, Ikenouchi-Sugita A, Umene-Nakano W, Hayashi K, Oonari N, Korogi Y, Nakamura J (2009) Reduction of brain gamma-aminobutyric acid (GABA) concentrations in early-stage schizophrenia patients: 3T Proton MRS study. Schizophr Res 112:192-3. Abstract

Hasler G, van der Veen JW, Tumonis T, Meyers N, Shen J, Drevets WC (2007) Reduced prefrontal glutamate/glutamine and gamma-aminobutyric acid levels in major depression determined using proton magnetic resonance spectroscopy. Arch Gen Psychiatry 64:193-200. Abstract

Muthukumaraswamy SD, Edden RA, Jones DK, Swettenham JB, Singh KD (2009) Resting GABA concentration predicts peak gamma frequency and fMRI amplitude in response to visual stimulation in humans. Proc Natl Acad Sci U S A 106:8356-61. Abstract

Ohrmann P, Siegmund A, Suslow T, Pedersen A, Spitzberg K, Kersting A, Rothermundt M, Arolt V, Heindel W, Pfleiderer B (2007) Cognitive impairment and in vivo metabolites in first-episode neuroleptic-naive and chronic medicated schizophrenic patients: a proton magnetic resonance spectroscopy study. J Psychiatr Res 41:625-34. Abstract

Ongur D, Prescot AP, McCarthy J, Cohen BM, Renshaw PF (2010) Elevated gamma-aminobutyric acid levels in chronic schizophrenia. Biol Psychiatry 68:667-70. Abstract

Stanley JA, Williamson PC, Drost DJ, Rylett RJ, Carr TJ, Malla A, Thompson RT (1996) An in vivo proton magnetic resonance spectroscopy study of schizophrenia patients. Schizophr Bull 22:597-609. Abstract

Sumner P, Edden RA, Bompas A, Evans CJ, Singh KD (2010) More GABA, less distraction: a neurochemical predictor of motor decision speed. Nat Neurosci 13:825-7. Abstract

Tayoshi S, Nakataki M, Sumitani S, Taniguchi K, Shibuya-Tayoshi S, Numata S, Iga J, Ueno S, Harada M, Ohmori T (2010) GABA concentration in schizophrenia patients and the effects of antipsychotic medication: a proton magnetic resonance spectroscopy study. Schizophr Res 117:83-91. Abstract

Théberge J, Williamson KE, Aoyama N, Drost DJ, Manchanda R, Malla AK, Northcott S, Menon RS, Neufeld RW, Rajakumar N, Pavlosky W, Densmore M, Schaefer B, Williamson PC (2007) Longitudinal grey-matter and glutamatergic losses in first-episode schizophrenia. Br J Psychiatry 191:325-34. Abstract

Théberge J, Bartha R, Drost DJ, Menon RS, Malla A, Takhar J, Neufeld RW, Rogers J, Pavlosky W, Schaefer B, Densmore M, Al-Semaan Y, Williamson PC (2002) Glutamate and glutamine measured with 4.0 T proton MRS in never-treated patients with schizophrenia and healthy volunteers. Am J Psychiatry 159:1944-6. Abstract

Yoon JH, Maddock RJ, Rokem A, Silver MA, Minzenberg MJ, Ragland JD, Carter CS (2010) GABA concentration is reduced in visual cortex in schizophrenia and correlates with orientation-specific surround suppression. J Neurosci 30:3777-81. Abstract

View all comments by Lawrence Kegeles
View all comments by Dikoma C. Shungu