News The latest developments in autism research.

Imaging shows loss of brain chemical in autism

by  /  1 July 2013

Big boxes: A downside of magnetic resonance spectroscopy is that it can’t measure brain regions smaller than 27 cubic centimeters, at least an order of magnitude larger than for functional magnetic resonance imaging.

Using a little-known brain-imaging technique, a new study shows that children with autism have low levels of gamma-aminobutyric acid (GABA), a chemical that keeps brain signals in check.

The report, published 24 May in NeuroImage, is the third imaging study in two years to find that children with the disorder have less GABA than controls do.

Other studies, using mutant mice, genetic screens or postmortem brain tissue, have also linked autism and GABA. The imaging studies are the only ones to confirm low GABA levels in the brains of living people with autism.

In the new study, William Gaetz‘s group showed that children with autism have less GABA in brain regions related to sound processing and motor control1. This may partly explain why many people with autism are hypersensitive to loud noises or have motor impairments.

Motor ability and sensory responses depend on the coordinated activity of thousands of neurons — and of neurotransmitters or chemical messengers such as GABA — so it’s important to study these chemicals, says Gaetz, assistant professor of radiology at the Children’s Hospital of Philadelphia.

The studies use a technique called in vivo magnetic resonance spectroscopy (MRS), which measures the concentrations of brain chemicals. More well-known methods — such as functional magnetic resonance imaging, electroencephalography and magnetoencephalography  (MEG) — instead gauge the brain’s structure or neuronal firing patterns.

MRS has been around for decades, but researchers only learned to use it to track GABA a few years ago, notes Donald Rojas, associate professor of psychiatry at the University of Colorado in Denver. In January, his team published an MRS study in NeuroImage showing that children with autism and their unaffected siblings have less GABA in the auditory cortex, where sound is processed, than controls do2.

But it’s unclear what these results mean, Rojas adds.

“MRS is, ultimately, a somewhat crude technique that cannot tell us the mechanism by which GABA is reduced,” Rojas says. For example, it may be that people with autism have fewer GABA-producing cells, or that they lack an enzyme needed to make GABA. “At this point we do not know the answer.”

Resonance renaissance:

Like its more famous counterparts, MRS relies on a magnetic resonance imaging scanner. The strong magnetic field and radio-wave pulses the machine produces cause the hydrogen atoms in different chemicals, including GABA, to absorb energy at varying frequencies. By charting these frequencies, MRS can estimate the chemical makeup of a particular brain region.

Gaetz and his colleague Timothy Roberts had previously used MEG, which uses magnetic fields to detect split-second changes in brain activity, to look for biomarkers of autism.

Their studies found that children with autism have delayed brain responses to sounds, as well as abnormal brain waves, or oscillations, in the cortex. “Those cortical oscillations are thought to be dependent on neurotransmitters like GABA,” Gaetz says. “So that’s why we’re getting into magnetic spectroscopy.”

The researchers scanned the brains of 17 children with autism and 17 controls. They zeroed in on the chemical composition of three brain areas concerned with movement, hearing and sight, respectively: the motor cortex of the frontal lobe, the auditory cortex of the temporal lobe and the visual cortex at the back of the brain.

Compared with controls, children with autism have less GABA in the motor cortex and the auditory cortex, but show no differences in the visual cortex, the study found.

The researchers plan to analyze the same group of children with autism using MEG, Gaetz says. The GABA data would predict that these participants will show abnormal brain waves in sound and motor areas, but not in the visual region, he says.

The new findings echo the auditory cortex results from the Rojas study. They also replicate the first study of this kind, published in 2011, which found GABA loss in the frontal lobe of children with autism3.

Other researchers, though, say it’s premature to make region-specific predictions based on MRS data.

The analyses for visual areas relied on data from just eight children with autism, notes Jeffrey Stanley, associate professor of psychiatry and behavioral neurosciences at Wayne State University in Detroit, Michigan, who was not involved in the work. “The negative finding from the visual area may have been due to poor [statistical] power,” Stanley says.

Still, the motor and auditory findings are intriguing, he says, and complement reports of sensory abnormalities in autism.

The main downside of MRS is that it takes time to obtain a strong signal. But a child — and especially a child with autism — can only stand to be in the uncomfortable scanner for about an hour.

That means there’s only enough time to measure the chemical composition of fairly large brain regions: roughly 27 cubic centimeters. “About the size of a shot glass,” Gaetz says.

With a territory that big, it’s difficult to discriminate between different functional areas of the brain. “We’re excited, but we’re also cautious,” Gaetz says. “We know that we’re covering an awful lot of brain with these measures.”

Correction: This article has been modified from the original. The previous version estimated MRS’ minimum volume as 3 cubic centimeters, instead of 27 cubic centimeters.


1. Gaetz W. et al. Neuroimage Epub ahead of print (2013) PubMed

2. Rojas D.C. et al. Neuroimage Epub ahead of print (2013) PubMed

3. Harada M. et al. J. Autism Dev. Disord. 41, 447-454 (2011) PubMed

  • Steve White

    As a parent this finding has a lot of interest for me. I would like to throw out a few speculations for comment by scientists.
    My son’s autism is almost certainly Maternal Antibody related, and I don’t know if that is tied to any difference in GABA levels or not, but as a young child, he had extreme hearing abnormalities, you could scream at him and he did not seem to hear it, but an ABA therapist found she could whisper to him and he would hear that.
    If I remember right, the researchers in Autism Phenome Project, a large longitudinal study we put him in, which among many other things measured brain response to sounds, call this a paradoxical response. They found four different types of responses to sound in autism. I hope I can explain then correctly, I am not using the terms they used: Normal. Decreased, Increased, and Paradoxical, So, some kids have normal response, (the testing is done by EEG in a Faraday cage with a cap with about 60 electrodes on it) some kids have decreased for all sounds but louder still gets more response than softer, the same graph of loudness/response, some kids increased for all sounds, louder still has a bigger response than softer, and some paradoxically have more response to soft sounds than loud ones.
    As he’s grown older, (6 now) all the abnormalities seem to have disappeared, he hears all sounds, responds more to loud than soft, looks when his name is called, and so forth. Well, maybe he’s got some abnormal sensitivities but nothing extreme.
    As a layman I don’t really know this at all, but seems like many neurotransmitters can be regulated by drugs, so I am hoping maybe GABA can be and maybe that will lead to improvements in brain function in some cases.
    As a parent I am very much hoping that President Obama’s program to map the brain will include funding for turning the research findings in autism into clinical medicine sooner than that would otherwise have happened.
    Thank you for the article. One question I had, do you have a percentage, and I know this is a small sample, but is there any indication what percentage of autistic people have this condition? In general as you know, new research findings are often reported without giving the percentage of patients effected and it’s kind of frustrating to not know.
    One more suggestion to anyone reading and thinking about these things.
    It would be helpful to understanding for researchers to tie known traits of autism to others.
    Maybe it would make sense where possible for the researchers doing the MRS, or any other kind of study but particularly something new, to draw their participants from large longitudinal studies where there is already a great deal of data to correlate with the new findings.
    In this case for example, if the folks in this study had been drawn from a previous study where sound response was already known, they could try to tie the new findings with the old, and further both lines of research.
    I know there are efforts to share databases and maybe someone has another paper already written on this, but as a layman it’s hard to tell if this is done routinely or not.
    Thanks again for the article and any responses.

  • ASDDad

    Inhibitory role for GABA in autoimmune inflammation

    There is some research from memory that would suggest 18% of ASD children may fall into an immune system phenotype.

    • Steve White

      The 2nd link you gave was to a non autism related paper. Maybe one of digits at the end is wrong?
      The other paper is very interesting, thanks for the link.
      If you search PubMed for “autism” “autoantibodies” and “GABA” I think you will find the papers about autistic people having antibodies to their own GABAergic neurons which I mention in a comment below.

  • Payman

    This study has a huge shortcoming. To which autism does it apply? there are aspergers, PDD, PDD-NOS, classic autiss. For what its worth, they may very well have different causes both genetic and environmental. All these studies are wide too general to pull out any conclusion. For examples, Aspergers in family roots, others aspergers can be traced…which is genetic application. But in regressive autism, kids losing abilities after normal growth, often no genetics roots to autism can be established, there could be other factors such as immune system. The study is too general to pull any conclusions…next

    • Steve White

      I think your comment is a bit too much. If the MRS works it could be huge, for one thing, even if GABA is not really all that important in autism.
      Also, we can’t be sure from this article that the original paper does not answer some of your questions.

  • Steve White

    Actually, I double-checked PubMed and found that many of the same folks doing the Maternal Antibody at UC Davis and the MIND Institute up there have found antibodies to GABAergic neurons in a large percentage of autistic people, (not sure what age but if you search on PubMed you can find the studies).
    Those studies were published two years back, I don’t where this research has gone recently.

    • Payman

      have found antibodies to GABAergic neurons in a large percentage of autistic people…

      What is a large % of autistic people?. In autism, they may have similar symptomes to some extent but their cause or biological process could be far too different. Its just too nonspecific.
      The only way is to cluster subpopultaion in autism and apply the study to pull out real fruitfull data.

      In any process, you have to know what is input and what is output, take into account the variables of the process…inorder to understand it and control it.

      • Steve White

        Payman — I don’t know if your question is rhetorical or not, as I said you could find it on PubMed, but it was around 20% who had antibodies to some of the GABAergic cells in their own brains.
        In autism, when you have some biomarker very rare in non-autistic people, (I think they have 1 control out of over 100 controls who also had the antibody) with a clear reason to think it would be harmful to the brain, (there is no doubt at all GABA is very important to normal brain function) then 20% is a VERY large percentage.
        If it’s all bearing fruit. Frankly I am puzzled why we don’t hear about a follow up study of some sort. In the case of maternal antibodies the same folks did monkey studies to confirm the pathogenicity of the antibodies, I would think they would want to do that with these antibodies.

        • Payman

          Hi Steve,
          Do you find that with your son on autism spectrum that there are periodic changes in his awareness, alertness and his stimming (if he has some.

          I have observed some of these kids very closely…what is driving me crazy is that for example: one kid stimms everyday for the whole week, which within this episode total lack of awareness like brain is too foggy followed by period of total lack of stims and complete awareness as if autism has just dispapeared…have you ever noticed anything like that with your kid…

          well with these kids depending on the subset of autism…in some it looks as if there are period of regressions constantly but the brain wiring seems there since when the brain is not too foggy the behavior is almost neurotypical…

  • ASDDad

    Steve White

    Cluster analysis of autistic patients based on principal pathogenetic components.

    We have recently described four principal pathogenetic components in autism:

    (I) circadian and sensory dysfunction,

    (II) immune abnormalities,

    (III) neurodevelopmental delay, and

    (IV) stereotypic behaviors. Using hierarchical and k-means clustering, the same 245 patients assessed in our principal component analysis can be partitioned into four clusters:

    (a) 43 (17.6%) have prominent immune abnormalities accompanied by some circadian and sensory issues;

    (b) 44 (18.0%) display major circadian and sensory dysfunction, with little or no immune symptoms;

    (c) stereotypies predominate in 75 (31.0%); and

    (d) 83 (33.9%) show a mixture of all four components, with greater disruptive behaviors and mental retardation. The “immune” component provides the largest contributions to phenotypic variance (P = 2.7 x 10(-45)), followed by “stereotypic behaviors.” These patient clusters may likely differ in genetic and immune underpinnings, developmental trajectories, and response to treatment.”

    You should also read the very enlightening article by Isaac Kohane Harvard just placed here on SFARI

  • Jarod

    Picamilon (niacin+GABA) and Phenibut(GABA+phenyl ring) are dietary supplements that could cross the blodd brain barrier.

    • Grizzle

      You are correct. I have damage to the cerebellum which can be attributed to a course of treatment with Dilantin (produces atrophy of the cerebellum in some people). I was born in 1967, so, not a lot of help or awareness then. I have an odd presentation of autism. I became functional because there was no other choice, but life has been a terror since I was five. School I endured purely by force of will, and by imitating others. I recently tried Phenibut, and although it develops tolerance in many after only a week, with horrific withdrawals if one quits,I am continuing, as it is as if very dark gray lenses seem to have been lifted from my eyes. I suggest you also look into Inosine, an “endogenous GABA receptor ligand”. I used that to overcome the symptoms of Protracted Benzodiazepine Withdrawal Syndrome, which involves one’s GABA system being shut off almost entirely, which was very traumatic for me, as you can imagine, as I already had a deficiency of GABA receptor function.
      I also used something I call “fronto-lumbar electrical stimulation”, developed by a scientist for the revivification of those who had died, Dr. Louise G. Robinovitch. An article is available on the NYT site, in pdf format, and her book, Sommeil Electrique is available for download in various places, for free (it is very old). She used DC, low level, I used AC, to try to work awareness back and forth, to get some GABA receptors working again, it worked, but it is dangerous if used at too high a level, I sensed that. But Phenibut is like enlightenment in a pill, for me. It is an enactogen. I never thought life could be this good. I wish I’d heard of it when I was going through benzo withdrawals (which can last forever, and can kill, and can completely render one bedridden). I surmise that much could be learned about autism from scientific studies of people with this syndrome (there are none). I will be returning to this page if there are any questions, or if anyone wants to contact me.You can read an article I wrote on the use of mid-infrared light to stimulate the cerebellum by this link: and one on using complex, time-varying magnetic fields for the same by this link: Also, look up the terms ledoux + sesquiterpenes , schmahmann + cerebellum , heath + cerebellar vermis , James w. Prescott + origins of human love and violence , basically look up the links in the article on the insomniaandstress site. If they no longer work, try wayback machine. You will know so much more about how autism works. It is all about the amygdala / cerebellar axis, and GABA and Glutamate. This same type of damage can either produce people who are autistic and passive, or violently psychotic and criminally insane. After studying my condition all these years, I do not know what makes the difference.

      • James

        By the way, although I did when I wrote the articles, I no longer sell anything. Not right now, anyway. Also, I was five when I got the Dilantin, not a neonate. Just after the course of treatment was over, as both parents told me much later, I changed from happy normal child, to dark, brooding, over-intelligent child that wanted to stay indoors and read books all the time. When I was 12, I tested as having a 2nd year college reading level, when I was in kindergarten, I tested in the top three percentile on whatever intellectual ability tests they did them. A side effect of cerebellar dysfunction that allows the amygdala to put more information into long term memory storage in the hippocampus. Not bragging, I definitely would have traded that for a normal life.

  • Jarod

    J Altern Complement Med. 2007 May;13(4):419-26.
    Yoga Asana sessions increase brain GABA levels: a pilot study.
    Streeter CC, Jensen JE, Perlmutter RM, Cabral HJ, Tian H, Terhune DB, Ciraulo DA, Renshaw PF.

    Division of Psychiatry, Boston University School of Medicine, Boston, MA 02118, USA. [email protected]

    The aim of this study was to compare changes in brain gamma-aminobutyric (GABA) levels associated with an acute yoga session versus a reading session. It was hypothesized that an individual yoga session would be associated with an increase in brain GABA levels.

    This is a parallel-groups design.

    Screenings, scan acquisitions, and interventions took place at medical school-affiliated centers.

    The sample comprised 8 yoga practitioners and 11 comparison subjects.

    Yoga practitioners completed a 60-minute yoga session and comparison subjects completed a 60-minute reading session.

    GABA-to-creatine ratios were measured in a 2-cm axial slab using magnetic resonance spectroscopic imaging immediately prior to and immediately after interventions.

    THERE WAS A 27% INCREASE IN GABA LEVELS in the yoga practitioner group after the yoga session (0.20 mmol/kg) but no change in the comparison subject group after the reading session ( -0.001 mmol/kg) (t = -2.99, df = 7.87, p = 0.018).

    These findings demonstrate that in experienced yoga practitioners, brain GABA levels increase after a session of yoga. This suggests that the practice of yoga should be explored as a treatment for disorders with low GABA levels such as depression and anxiety disorders. Future studies should compare yoga to other forms of exercise to help determine whether yoga or exercise alone can alter GABA levels.

    [PubMed – indexed for MEDLINE]

  • Unistasis

    Grizzle, I’d like to hear about Your experiences With Phenibut. Do you use it long term, have you experienced peak tolerance With it?

    Ive read so many stories of tragedies regarding pheni, but ive also seen several People being able to take it for years without neccecarily having to increase dosage or get terrible side effect.
    I know it can be Dangerous to mess around too much With gaba-regulation and it can lead to horrible withdrawal symptoms, but then again, many of us are suffering from gaba/glutamate/dopamin etc abnormalities, but there must be a reason some People tend to be able to normalize its use, function normally without it stopping to work.

    i suppose that, as With any other supplement, or drug, you are dependent on cofactors, and when you stimulate or overstimulate certain parts of Your physiology, you tend to start lacking perhaps certain minerals and vitamins. i have experienced the co-factor factor many times in my years of experiementing With vits and mins.

    so besides downregulating or abnormally regulating receptors, the psychological trauma of hitting baseline again, if Your baseline was at the starting point very anxious and depressed, i think co-factors and getting the right supportive nutrients to continue using phenibut efficiently.
    there are many discussions going on online about this, but id like as much input as possible, since if im going to go into this phenibut use, i need to have as much information. i deeply respect its ability to fuck People over.

    take a look at this thread by the way:


Log in to your Spectrum Wiki account

Email Address:



Request your Spectrum Wiki account

Spectrum Wiki is a community of researchers affiliated with an academic or research institutions. To be considered for participation, please fill out this form and a member of our team will respond to your request.


Email Address:

Title and Lab:

Area of Expertise: