Developmental changes in gamma-aminobutyric acid levels in attention-deficit/hyperactivity disorder

Abstract

While the neurobiological basis and developmental course of attention-deficit/hyperactivity disorder (ADHD) have not yet been fully established, an imbalance between inhibitory/excitatory neurotransmitters is thought to have an important role in the pathophysiology of ADHD. This study examined the changes in cerebral levels of GABA+, glutamate and glutamine in children and adults with ADHD using edited magnetic resonance spectroscopy. We studied 89 participants (16 children with ADHD, 19 control children, 16 adults with ADHD and 38 control adults) in a subcortical voxel (children and adults) and a frontal voxel (adults only). ADHD adults showed increased GABA+ levels relative to controls (P=0.048), while ADHD children showed no difference in GABA+ in the subcortical voxel (P>0.1), resulting in a significant age by disorder interaction (P=0.026). Co-varying for age in an analysis of covariance model resulted in a nonsignificant age by disorder interaction (P=0.06). Glutamine levels were increased in children with ADHD (P=0.041), but there was no significant difference in adults (P>0.1). Glutamate showed no difference between controls and ADHD patients but demonstrated a strong effect of age across both groups (P<0.001). In conclusion, patients with ADHD show altered levels of GABA+ in a subcortical voxel which change with development. Further, we found increased glutamine levels in children with ADHD, but this difference normalized in adults. These observed imbalances in neurotransmitter levels are associated with ADHD symptomatology and lend new insight in the developmental trajectory and pathophysiology of ADHD.

OPEN

Citation: Transl Psychiatry (2015) 5, e589; doi:10.1038/tp.2015.79

http://www.nature.com/tp

Web End =www.nature.com/tp

S Bollmann1,2,3,4,5, C Ghisleni1,2,3, S-S Poil1,3, E Martin1,2, J Ball6, D Eich-Hchli7, RAE Edden8,9, P Klaver1,2,3,10, L Michels11,

D Brandeis2,3,6,12 and RL O'Gorman1,3,13

While the neurobiological basis and developmental course of attention-decit/hyperactivity disorder (ADHD) have not yet been fully established, an imbalance between inhibitory/excitatory neurotransmitters is thought to have an important role in the pathophysiology of ADHD. This study examined the changes in cerebral levels of GABA+, glutamate and glutamine in children and adults with ADHD using edited magnetic resonance spectroscopy. We studied 89 participants (16 children with ADHD, 19 control children, 16 adults with ADHD and 38 control adults) in a subcortical voxel (children and adults) and a frontal voxel (adults only). ADHD adults showed increased GABA+ levels relative to controls (P = 0.048), while ADHD children showed no difference in GABA+ in the subcortical voxel (P40.1), resulting in a signicant age by disorder interaction (P = 0.026). Co-varying for age in an analysis of covariance model resulted in a nonsignicant age by disorder interaction (P = 0.06). Glutamine levels were increased in children with ADHD (P = 0.041), but there was no signicant difference in adults (P40.1). Glutamate showed no difference between controls and ADHD patients but demonstrated a strong effect of age across both groups (Po0.001). In conclusion, patients with ADHD show altered levels of GABA+ in a subcortical voxel which change with development. Further, we found increased glutamine levels in children with ADHD, but this difference normalized in adults. These observed imbalances in neurotransmitter levels are

Translational Psychiatry (2015) 5, e589; doi:10.1038/tp.2015.79; published online 23 June 2015

INTRODUCTION

Attention-decit/hyperactivity disorder (ADHD) is a common developmental psychiatric disorder characterized by inattention, hyperactivity and impulsivity with a prevalence of about 5%.1 Although traditionally considered a disorder of childhood, mounting evidence suggests that ADHD often persists into adulthood.2 While the neurobiological basis and developmental course of ADHD have not yet been fully established, an imbalance between inhibitory/excitatory neurotransmitters is thought to have an important role in the pathophysiology of ADHD.3

GABA, the main inhibitory neurotransmitter in the human cerebral cortex is synthesized from neuronal glutamate (Glu), and converted back into Glu in astrocytes. The astrocytic Glu is then catalyzed into glutamine (Gln) and transported to the presynaptic neuron, where it is cycled back into Glu, some of which is then converted into GABA.4 GABA has been shown to be implicated in dopaminergic neurotransmission in the striatum5 suggesting an important role in ADHD. Further, GABA seems to be important for behavioral inhibition and self-control,6,7 as reduced GABA levels are associated with high impulsivity, less cognitive control and impaired response inhibition.

Glu, the major excitatory neurotransmitter, modulates the release of dopamine,3 and imbalances in Glu are believed to

interfere with the gating of sensory information in striato-frontal pathways in patients with ADHD.3

In children with ADHD, an increase in Glu+Gln (Glx) was observed in frontal areas,810 right prefrontal cortex10,11 and left striatum3,10,11 relative to controls, although these differences did

not reach signicance in a recent meta-analysis.12 Treatment with stimulant medication has been reported to decrease Glx in the basal ganglia in children with ADHD.13 However, there are also studies reporting no signicant Glx differences with stimulant treatment.1416 In ADHD adults, on one hand increases in Glx have been observed in the basal ganglia3 and in the left cerebellar hemisphere.17 On the other hand, studies report a reduction in Glx in the right anterior cingulate cortex,18 in a left midfrontal region19 and in the basal ganglia20 in adults with ADHD compared with controls. As the magnetic resonance spectroscopy (MRS)-visible Glx signal includes contributions from the metabolic Glx pool as well as the neurotransmitter pool, the MRS Glx signal represents only a nonspecic marker for neurotransmitter Glu. However, since neurotransmitter Glu is converted into Gln in glial cells,21

some authors have suggested that Gln may represent a more specic marker for neurotransmitter Glu.2224

In children with ADHD, only one previous study has investigated cerebral GABA+ concentrations, reporting reduced levels in the sensorimotor cortex.25 However, to date, GABA+ has not been

1Center for MR-Research, University Childrens Hospital Zurich, Zrich, Switzerland; 2Neuroscience Center Zurich, University of Zurich and ETH Zurich, Zrich, Switzerland; 3Zurich Center for Integrative Human Physiology, University of Zurich, Zrich, Switzerland; 4Institute for Biomedical Engineering, University of Zurich and ETH Zurich, Zrich, Switzerland;

5Centre for Advanced Imaging, University of Queensland, Brisbane, QLD, Australia; 6Department of Child and Adolescent Psychiatry, University of Zurich, Zrich, Switzerland;

7Psychiatric University Hospital, Zrich, Switzerland; 8Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, MD, USA; 9F. M. Kirby Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, USA; 10Department of Psychology, University of Zurich, Zrich, Switzerland; 11Institute of Neuroradiology, University Hospital of Zurich, Zrich, Switzerland; 12Department of Child and Adolescent Psychiatry and Psychotherapy, Central Institute of Mental Health Mannheim, Medical Faculty Mannheim/Heidelberg University, Mannheim, Germany and 13Pediatric Research Center, University Childrens Hospital Zurich, Zrich, Switzerland. Correspondence: Dr R O'Gorman, Center for MR-Research, University Children's Hospital Zurich, Steinwiesstrasse 75, Zrich CH 8032, Switzerland. E-mail: mailto:[email protected]

Web End [email protected] Received 4 August 2014; revised 27 April 2015; accepted 4 May 2015

ORIGINAL ARTICLE

Developmental changes in gamma-aminobutyric acid levels in attention-decit/hyperactivity disorder

associated with ADHD symptomatology and lend new insight in the developmental trajectory and pathophysiology of ADHD.

Developmental course of ADHD S Bollmann et al

2

Table 1. Demographic and clinical characteristics of the adult participants

n Mean (s.d.) Median Range P

AgeADHD (m/f) 8/8 38.4 (11.8) 37.6 23.761.1 0.04a Control (m/f) 19/19 31.6 (9.2) 27.5 21.050.6

WURS-k SumADHD 16 33.4 (11.1) 34 13.051.0 o0.001a Control 30 12.9 (10.4) 10 0.038.0

CAARS-S:L (ADHD-Index)ADHD 16 63.0 (8.4) 62 51.079.0 o0.001a Control 30 44.9 (5.7) 43 36.062.0

CAARS-O:L (ADHD-Index)ADHD 13 64.6 (11.8) 63 49.083.0 o0.001b Control 28 46.6 (6.1) 48 37.058.0

MedicationADHD Methylphenidate (n = 14); trazodone (n = 2)c; dexmethylphenidate (n = 2); bupropion (n = 1); agomelatine (n = 1); amitriptyline (n = 1)c; prednisone (n = 1); oral contraceptives (n = 3)

Control Finasteride (n = 1); uorouracil (n = 1)d; salicylic acid (n = 1); paracetamol (n = 1); minoxidil (n = 1); oral contraceptives(n = 8)

ComorbiditiesADHD Sleep problems (n = 1)

Control None

HandednessADHD (r/l) 13/3 Control (r/l) 32/6

Abbreviations: ADHD, attention-decit/hyperactivity disorder; CAARS, Conners Adult ADHD Rating Scales; f, female; l, left-handed; m, male; n, number of participants; r, right-handed; WURS, Wender Utah Rating Scale. aTwo-tailed Wilcoxon rank-sum test. bTwo-tailed two-sample t-test. cCurrent or previous reported use for sleep problems. dPrescribed for skin warts.

assessed in the basal ganglia in ADHD, although structural and functional basal ganglia abnormalities appear to represent a core nding in the ADHD literature, as highlighted in recent meta-analyses.26,27 Further, no previous studies have investigated both inhibitory and excitatory neurotransmission in both children and adults with ADHD in a single study, and the developmental trajectories of neurotransmitter levels in ADHD remain unclear, despite the known developmental abnormalities observed in the basal ganglia.28

The primary aim of this study was to investigate whether inhibitory (GABA) and excitatory (Glu/Gln) neurotransmitter levels in a subcortical voxel are altered in children and adults with ADHD. A secondary aim was to investigate the relationship between neurotransmitter levels and ADHD symptomatology and how GABA, Gln and Glu change with development. For the ADHD adults, we examined a left frontal region centered on the dorsolateral prefrontal cortex in addition to the subcortical voxel. We hypothesized that GABA+ levels would be reduced in ADHD patients in the subcortical and frontal voxels, based on a recent report of decreased GABA in the sensorimotor cortex in children with ADHD.25 We also expected to nd increased levels of Glu in children with ADHD.810

MATERIALS AND METHODSThe participant group consisted of 89 participants including 16 children with ADHD and 19 control children, 16 adults with ADHD and 38 control adults. Data were acquired as part of a large developmental study investigating ADHD-related differences in GABA, Glu and Gln in the context of the normal maturational changes in these neurotransmitter systems. For this reason, a larger control group was recruited, but the groups were matched for age, handedness and gender. One ADHD adult was excluded because of recreational drug use. All participants complied

with magnetic resonance imaging (MRI) safety standards. The control participants did not report any current or previous neurologic or psychiatric diagnoses or any current use of psychoactive medications. All spectra were inspected visually to assess t quality and the signal-to-noise ratio, leading to the exclusion of one further adult ADHD patient and one control child. One child was excluded from the control group due to elevated ADHD symptom scores to avoid a confound due to subclinical cases (Table 1 and Table 2). Because of the larger adult control group, all analyses were additionally performed with a one by one matching of the control subjects to the patients based on age and gender.

The ADHD adults were recruited from the Psychiatric University Clinic Zrich and screened for comorbidities. All the patients were diagnosed according to DSM-IV criteria, by a consultant psychiatrist specialising in ADHD. To ensure a representative ADHD sample, we included participants with current mild depressive symptoms, but excluded any patients with major depression or current severe Axis I or II disorder, substance use disorder, autism spectrum disorder, tic disorder or any other medical or neurological illness affecting brain function. The ADHD children were recruited by the Department of Child and Adolescent Psychiatry of the University of Zurich. The Kiddie-Sads-Present and Lifetime clinical interview29 was performed with all ADHD children to ensure the diagnosis of combined ADHD and to exclude participants with comorbidities. Participants taking stimulant medications were instructed to interrupt their medication at least 72 h before the MRI measurements. Antidepressant medication in six adult ADHD participants was not interrupted. A urine toxicology test was not performed before scanning.

Symptom scores were assessed with the German version of the Conners Adult ADHD Rating Scale30 and the German short form of the Wender-Utah Rating Scale.31 For children, we used the research version of the Conners-3D32 and the strengths and difculties questionnaires in their German adaptations.33

After complete description of the study to the subjects, written informed consent was obtained. For the ADHD and control children, written consent was obtained from the parents and assent was obtained from the children. The study was approved by the cantonal ethics committee of Zrich and was conducted in accordance with guidelines of the Declaration of

Translational Psychiatry (2015), 1 8

3

Table 2. Demographic and clinical characteristics of the children participants

n Mean (s.d.) Median Range P

AgeADHD (m/f) 9/7 10.8 (1.1) 10.7 8.713.1 0.89a Control (m/f) 11/8 10.8 (1.9) 11 8.114.7

Conners Parents ADHD-IndexADHD 16 69.8 (3.1) 70 61.073.0 o0.001b Control 19 46.9 (12.6) 46 20.063.0

SDQ Parents TotalADHD 15 19.1 (6.5) 18 10.034.0 o0.001a Control 19 5.1 (3.1) 5 1.014.0

MedicationADHD Methylphenidate (n = 11)

Control None

ComorbiditiesADHD F40.2 specic (isolated) phobias (n = 4); F82 specic developmental disorder of motor function (n = 2); F81.0 specic reading disorder (n = 1); F81.2 specic disorder of arithmetical skills (n = 1)

Control None

HandednessADHD (r/l) 12/4 Control (r/l) 18/1

Abbreviations: ADHD, attention-decit/hyperactivity disorder; f, female; l, left-handed; m, male; n, total number of participants; r, right-handed; SDQ, strengths and difculties questionnaire. aTwo-tailed two-sample t-test. bTwo-tailed Wilcoxon rank-sum test.

Developmental course of ADHD S Bollmann et al

Helsinki. The participants received 60 Swiss Francs in vouchers for participation in this study.

MR measurements were performed with a 3T GE HD.xt TwinSpeed MRI scanner (GE Healthcare, Milwaukee, WI, USA), using an eight-channel receive-only head coil. The MRI protocol included a T1-weighted spoiled gradient echo scan (TR = 9.94 ms; TI = 600 ms; FOV = 256 mm 192 mm; matrix = 256 192; ip angle = 8; axial plane; slice thickness = 1 mm; 172 slices) for planning the positioning of the target voxels. GABA-edited MR spectra were acquired using the MEGA-PRESS method with TE = 68 ms, TR = 2000 ms, 320 averages (160 pairs) and an eight-step phase cycle. We acquired spectra from a 28 40 25 mm3 voxel in a subcortical region centered on the left basal ganglia (also including anterior thalamus) from all the participants. An additional 25 40 30 mm3 voxel was acquired in the left frontal lobe, centered on the dorsolateral prefrontal cortex from the adult participants (see Figure 1). The assessment of GABA with MEGA-PRESS is confounded by the co-editing of macromolecules which contribute to the edited peak at 3 p.p.m., so the GABA ndings described subsequently represent GABA+ rather than pure GABA values.

The T1-weighted images were segmented in native space using SPM8 with the New Segment module (Wellcome Department of Cognitive Neurology, London, UK) running in MATLAB 2012b (The MathWorks, Natick, MA, USA) to correct the spectroscopy results for partial volume cerebrospinal uid contamination and different water compartment relaxation times.34

The spectra were coil combined with weighting factors derived from the rst point of the free induction decay signal from the unsuppressed water lines acquired with each coil. Water scaled metabolite concentrations were derived with LCModel version 6.3-1B.35 The edited spectra were analyzed with a simulated basis set including basis spectra for GABA, glutathione, NAA, Gln, Glu and N-acetyl aspartyl glutamate. GABA+/H2O, Glu/H2O and

Gln/H2O were quantied from the edited spectra, using the control parameter sptype = mega-press-2. The edited spectra provided a similar t quality compared with the edit OFF MEGA-PRESS subspectra for Glu/ H2O, but a better t quality for Gln/H2O compared with the edit OFF subspectra (Gln detected with CramerRao lower bounds (CRLB) o30% in 73/89 edited spectra and 43/89 edit OFF subspectra). The CRLB for GABA ranged from 210% for all participants. An estimate of general t quality

across each spectrum was assessed from the CRLB of the NAA t, both in the edited spectra and in the edit OFF subspectra (edited spectra: NAA CRLB = 13%, edit OFF subspectra: NAA CRLB = 15% for all cortical and subcortical spectra).

Statistical analysis was performed using R 3.1.1 (ref. 36) and ggPlot2.37

Normality of the data was assessed using QuantileQuantile plots of the analysis of variance (ANOVA) residuals. ADHD-related and developmental differences in subcortical GABA+ and Glu were assessed with a 2 2 ANOVA, including age group (adults and children) and diagnostic group (ADHD and control) as xed factors. Post hoc t-tests were performed to clarify the group differences. As the Gln levels were not normally distributed, differences between ADHD and control participants were tested using a two-tailed Wilcoxon rank-sum test. Correction for multiple comparisons was performed with Benjamini and Hochberg correction, with a false discovery rate (FDR) of 0.1.38 A multiple regression model was used to investigate the relationship between metabolite levels and ADHD symptom scores from the Conners inventory, age, methylphenidate dose and gender, using the normally distributed data. To construct the nal model, we started with the full model including gender, age, dose, inattention/memory-score, hyperactivity-score, impulsivity-score and the self-concept-score, then removed correlated regressors and nally removed nonsignicant predictors in a stepwise manner until only signicant predictors remained in the model.

RESULTS GABA+/H2Osubcortical

The gray matter content of the subcortical voxel did not differ signicantly between controls (M = 52%, s.d. = 3%) and ADHD patients (M = 50%, s.d. = 5%; P40.1). However, there was a signicant difference in gray matter content comparing adults (M = 51%, s.d. = 4%) to children (M = 55%, s.d. = 4%; Po0.001).

Comparing GABA+/H2O levels in the subcortical voxel in children and adults with and without ADHD revealed a signicant age group by diagnostic group interaction (F(1,85) = 5.11, P = 0.026, 2 = 0.04; Figure 2), which remained signicant after FDR correction for multiple comparisons. Post hoc t-tests demonstrated that ADHD adults had higher GABA+/H2O values (M = 1.98, s.d. = 0.20) compared with controls (M = 1.87, s.d. = 0.17). This difference was signicant, t(52) = 2.03, P = 0.048, d = 0.56. ADHD children showed no signicant differences in GABA+/H2O values in the subcortical voxel (M = 1.85, s.d. = 0.17) compared with controls (M = 1.91, s.d. = 0.13), t(33) = 1.25, P40.1, d = 0.43.

The same analysis was repeated with the one by one matched adult control group, resulting in similar results for the interaction (F(1,63) = 5.56, P = 0.021, 2 = 0.06), but the comparison between GABA levels in ADHD adults and controls revealed only a trend level difference (t(30) = 1.99, P = 0.055, d = 0.73), although the effect size was comparable to that observed in the full control group. Co-varying for age in an analysis of covariance model resulted in a nonsignicant age group by diagnostic group interaction in the full group (F(1,84) = 3.5, P = 0.06) but was signicant in the one by one matched group (F(1,62) = 5.4, P = 0.023).

The multiple regression in adults yielded standardized betas of0.33 for age (P = 0.017) and 0.35 (P = 0.010) for self-concept with an adjusted R2 of 0.26.

The regression analysis in children did not show any signicant effects. There were no signicant differences in estimated t error (CramerRao lower bounds) between the groups.

Glu/H2OsubcorticalComparing Glu/H2O differences in the subcortical voxel in children and adults in a 2 2 ANOVA revealed a signicant main effect of age group, F(1,85) = 40.6, Po0.001, 2 = 0.31. The post hoc t-test revealed that children have signicantly increased Glu/H2O values, t(78) = 6.59, Po0.001, d = 1.49 (see Figure 2) relative to adults. The main effect of age on Glu/H2O levels remained signicant after FDR correction for multiple comparisons. No signicant

Translational Psychiatry (2015), 1 8

Developmental course of ADHD

S Bollmann et al

4

LCModel (Version 6.3-1B) Copyright: S.W. Provencher.

Ref.: Magn. Reson. Med. 30:672-679 (1993).

LCModel (Version 6.3-1B)

Copyright: S.W. Provencher.

Ref.: Magn. Reson. Med. 30:672-679 (1993).

GABA

Gln

Glu

A

P

R

L

P

A

R

L

GABA

Gln

Glu

A

P

R

L

P

A

R

L

4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2

4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2

Chemical Shift (ppm)

Chemical Shift (ppm)

Figure 1. Left subcortical and left frontal voxel positions and representative spectra. The edited spectral ts and the position of the left subcortical and the left frontal voxel in a representative adult subject.

difference between controls and ADHD in Glu/H2O values in the subcortical voxel was observed, either for adults (t(52) = 0.07, P40.1, d = 0.02) or for children (t(33) = 0.45, P40.1, d = 0.16).

The Glu/H2O regression model in adults and children yielded no signicant effect. There were no signicant differences in t errors between ADHD and controls for Glu. However, Glu t errors were signicantly higher in adults (Mdn = 5.4%) compared with children (Mdn = 3.8%), W = 1477, Po0.001, r = 0.49.

The same analysis was repeated with the one by one matched adult control group, resulting in similar results for the age group effect (F(1,63) = 33.25, Po0.001, 2 = 0.33). The comparison between adults and children was still signicant t(64) = 5.87, Po0.001, d = 1.47.

Gln/H2OsubcorticalThe CRLB for Gln were o30% in 15/19 control children (range 13

23%), 14/16 ADHD children (range 1022%), 32/38 control adults (range 1029%) and 12/16 ADHD adults (range 1128%). Using a stricter cutoff of 20% would result in the exclusion of two additional control children, two ADHD children, 15 control adults and six ADHD adults.

ADHD adults (Mdn = 1.25) showed no signicant difference in Gln/H2O compared with control adults (Mdn = 1.15), W = 229.0,

P40.1, r = 0.14. However, ADHD children (Mdn = 1.46) showed a signicantly higher Gln/H2O level than control children (Mdn =1.01), W = 152.0, P = 0.041, r = 0.38, which remained signicant after FDR correction for multiple comparisons. There was no signicant difference in Gln levels between adults (Mdn = 1.2) and children (Mdn = 1.09), W = 726.0, P40.1, r = 0.11.

The Gln/H2O regression model was not performed due to non-normality of the data. There were no signicant differences in t errors between ADHD and controls for Gln. However, Gln t errors were signicantly higher in adults (Mdn = 19.5%) compared with children (Mdn = 16.00%) W = 918.0, P = 0.006, r = 0.32.

The same analysis was repeated with the one by one matched adult control group, where ADHD adults (Mdn = 1.25) showed no signicant difference in Gln/H2O compared with control adults (Mdn = 1.22), W = 84.0, P40.1, r = 0.06. There was no signicant

difference in Gln levels between adults (Mdn = 1.25) and children (Mdn = 1.09), W = 449.0, P40.1, r = 0.2.

GABA+/H2OfrontalThe gray matter content of the frontal voxel did not differ signicantly between controls (M = 34%, s.d. = 3%) and ADHD patients (M = 33%, s.d. = 4%; P40.1).

In the left frontal voxel, ADHD adults did not differ in GABA+/H2O

values (M = 1.47, s.d. = 0.16) compared with controls (M = 1.42, s.d. = 0.17), t(51) = 0.95, P40.1, d = 0.26 (see Figure 2). This was also the case for the one by one matched group (t(30) = 0.02, P40.1; d = 0.01).

The multiple regression in adults yielded standardized betas of0.58 for inattention/memory (P = 0.003) and 0.37 (P = 0.05) for hyperactivity with an adjusted R2 of 0.15. There were no signicant differences in GABA+ t errors between ADHD and controls (P40.1).

Glu/H2OfrontalADHD adults showed no difference in Glu/H2O values in the frontal voxel (M = 3.7, s.d. = 0.4) compared with controls (M = 3.7, s.d. = 0.4), t(51) = 0.12, P40.1, d = 0.03. This was also the case for the one by one matched group (t(30) = 1.07, P40.1; d = 0.39).

The multiple regression analysis in adults yielded no signicant effects. There were no signicant differences in Glu t errors between ADHD and controls (P40.1).

Gln/H2OfrontalFor this analysis, we excluded in total ve Gln values with an estimated t error (CramerRao lower bound) 430% (three ADHD adults, two control adults) leaving in total 48 adults (13 patients) in the analysis.

ADHD adults showed no difference in Gln/H2O values in the frontal voxel (Mdn = 0.85) compared with controls (Mdn = 0.98), W = 179.0, P40.1, r = 0.16. This was also the case for the one by one matched group (W = 65, P40.1; d = 0.24).

The multiple regression analysis in adults was not performed due to non-normality of the data. There were no signicant differences in Gln t errors between ADHD and controls (P40.1).

Translational Psychiatry (2015), 1 8

Developmental course of ADHD S Bollmann et al

5

Figure 2. GABA+, Glu and Gln levels in the left subcortical and in the left frontal voxel GABA+/H2O (top row), Glu/H2O (middle row) and Gln/ H2O (bottom row) ratios from ADHD children, control children, ADHD adults and control adults in the subcortical voxel (left column) and in the left frontal voxel (right column). ADHD, attention-decit/hyperactivity disorder; Gln, glutamine; Glu, glutamate.

DISCUSSIONThe present study examined GABA+, Glu and Gln concentrations in a subcortical voxel centered on the basal ganglia for the rst time in a developmental group of children and adults with and without ADHD,

demonstrating a developmental alteration in GABA+ levels in ADHD patients. GABA+ levels in the subcortical voxel were signicantly correlated with ADHD symptom scores in adults. In addition, Gln levels were signicantly increased in ADHD children but not in adults.

Translational Psychiatry (2015), 1 8

Developmental course of ADHD S Bollmann et al

6

The signicant interaction between age group and disorder group in GABA+/H2O in the subcortical voxel was driven by increased GABA+/H2O concentrations in ADHD adults compared with control adults, in combination with mildly reduced GABA+/H2O

concentrations in ADHD children compared with control children. Although the lower GABA+ levels in ADHD children are broadly consistent with reports in the existing ADHD MRS literature,25 the

increased GABA+ concentration in ADHD adults was unexpected. The GABA+ regression analysis revealed positive regression coefcients in adults for age and self-concept in the subcortical voxel. In the frontal voxel, GABA+ was positively correlated with inattention/memory but negatively correlated with hyperactivity in adults.

Taken together with the observed group differences in GABA+ and the signicant age by disorder interaction, these results provide evidence for a developmental alteration in subcortical GABA+ levels in ADHD patients. The neurobiological basis for this developmental change in GABA+ levels is unknown, but may result from the progression of ADHD with brain maturation. The developmental change in subcortical GABA+ levels is also broadly consistent with results from structural MRI studies, which on one hand report decreases in gray matter in the caudate and other basal ganglia structures in children which show an apparent normalization with development and treatment;26,28,3941 whereas

on the other hand, subtle decreases in caudate gray matter in adults with ADHD compared with controls can still be observed.42

In the present study, the subcortical voxel did not show any signicant differences in gray matter content between the ADHD and control groups, indicating that the apparent differences in GABA+ observed in this region are independent of structural differences which may arise from developmental or treatment effects. Therefore, the observed developmental changes in GABA+ seem to relate to functional rather than structural differences. Further, the regression results suggest that developmental changes in GABA+ levels may at least partly underlie the apparent change in ADHD symptomatology from childhood to adulthood.

It is possible that the signicant difference in GABA between ADHD adults and controls might be driven by age differences. When age was included as a covariate in the GABA+ ANOVA in adults, the age by disorder interaction altered from F(1,85) = 5.11, P = 0.026 to F(1,84) = 3.5, P = 0.06. In the one by one matched group, the interaction was signicant in the ANOVA (F(1,63) = 5.56, P = 0.021) and the analysis of covariance (F(1,62) = 5.43, P = 0.023). However, the effect of age on GABA+ is not yet clear, as some studies report a decline of GABA+ with age,43 while others observed an increase of GABA+ with age.44 The reported increase of GABA+ has also been reported to be driven by an increase of the co-edited macromolecules with age and not GABA.44

We did not nd increased subcortical Glu levels in ADHD patients compared with controls, as found in some previous studies10,11,13,45 but not in others.12,25,46 However, most previous

studies investigated the ratio of Glx (the sum of Glu+Gln) to creatine,3,11,15,16,47 rather than concentrations of Glx or Glu. Ratios

to creatine are simpler to derive since they do not require correction for partial volume effects arising from cerebrospinal uid in the MRS voxel, but the assessment of Glx/Cr ratios may be confounded by alterations in striatal Cr levels, which have been reported both in children10 and adults20 with ADHD. To our knowledge, only two previous studies investigated Glx or Glu in the left basal ganglia or striatum in ADHD, reporting an increase of Glx and Glu in children aged 8 years,10 and a decrease in Glx in both medication naive and medicated adults.20 These two ndings point towards a possible age interaction with Glu in the striatum, such that Glu may be increased in children with ADHD and decreased with age, resulting in lower Glu levels in adult ADHD patients. However, in the present study, we did not observe a signicant interaction between subcortical Glu levels and age or a signicant effect in the regression analysis, although we did

observe a signicant main effect of age for Glu/H2O in the

subcortical voxel, with children showing higher Glu concentrations than adults. Our regression analysis also suggests that Glu/ H2O decreases with age in children and adults, consistent with the age decrease of Glx/Cr reported by Gao et al.,43 although their results were derived from frontal and parietal regions.

To put our results into context with those from the literature, we identied 10 previous studies investigating Glx or Glx/Cr in the left basal ganglia/striatum in ADHD children or adults.3,10,11,13,15,16,20,45,47,48 Two of these studies did not include

a healthy comparison group13,45 and one did not report group

means or effect sizes for Glx,48 but for the remaining seven studies, we calculated the effect size (Cohens D) from the difference between group means (ADHDcontrol) and the pooled standard deviations for Glx/Cr or Glx. Where data were available for multiple cohorts (for example, medicated vs unmedicated or ADHD combined vs inattentive subtype), the effect sizes were calculated separately for each ADHD cohort relative to the control group. Effect size data are shown graphically in Supplementary Figure 1. For the Glx/Cr ratios, studies showed either small positive or negative effect sizes (both in children and adults), although two studies showed a large increase in Glx/Cr in ADHD children or young adults relative to a control group.3,11 For the Glx

concentration, data are only available for a small number of studies, but children with ADHD tended to show positive effect sizes (increased Glx) while adults with ADHD show smaller positive or negative effect sizes (reduced Glx20), but further studies would be needed to corroborate and extend this nding.

To our knowledge, the present study is the rst to examine Gln changes in a developmental cohort of adults and children with ADHD. We found increased Gln/H2O in the subcortical voxel in children with ADHD, but there were no signicant differences in Gln/H2O between ADHD and control adults. As Gln contributes to the Glx signal, the increased Gln levels are consistent with results from a previous study reporting increased Glx levels in children with ADHD10,11 (Supplementary Figure 1). Assuming that Gln represents a marker for neuronal Glu, the developmental trajectory of Gln levels observed here points towards an increase in excitatory neurotransmission in children with ADHD, which normalizes with brain maturation. While the underlying basis of this apparent increase in Gln remains unclear, the production and subsequent pruning of excitatory synapses during late childhood and adolescence could result in transiently increased Gln either from a developmental lag or from a compensatory effect in response to synaptic pruning. However, developmental changes in neurotransmitter activity are likely to result from a complex interplay between genetic and environmental factors including activity-dependent increases in synaptic strength and neuroplastic changes during adolescence. Further studies would be needed to elucidate the basis for this apparent change in Gln.

In the frontal voxel, there was no signicant difference between ADHD and control adults in GABA+/H2O, Glu/H2O or Gln/H2O.

Several methodological limitations are inherent in the detection of GABA with MEGA-PRESS MRS. Due to the low cerebral concentration of GABA, a large MRS voxel is required for sufcient sensitivity and signal-to-noise ratio, which limits the regional specicity of the measurement. In addition, the assessment of GABA with MEGA-PRESS is confounded by the co-editing of macromolecules which contribute to the edited peak at 3 p.p.m. Therefore, we cannot exclude the possibility that the differences in GABA observed in the present study are confounded by differences in the co-edited macromolecules between ADHD patients and controls, but future studies using a MEGA-PRESS protocol with reduced macromolecular contamination49 may be

able to clarify further the spectroscopic differences in GABA in ADHD.

Although the developmental effects observed in subcortical GABA+ levels and the signicant difference in Gln levels in

Translational Psychiatry (2015), 1 8

Developmental course of ADHD S Bollmann et al

7

children with ADHD relative to control children survived correction for multiple comparisons with a relatively lenient FDR threshold of 0.1, these results did not survive correction with a more stringent FDR threshold of 0.05. Given the sample size, the relatively modest signicance levels are surprising, but may result from the large variance seen both within and across age and diagnostic groups (Figure 2). Alternatively, the modest effect size may also be inuenced by variability within diagnostic groups, as evidenced by the relatively wide range of symptom scores particularly in the adult ADHD group, which included some patients with lower symptom scores indicative of a milder phenotype. Although the groups were matched for gender, the inclusion of female participants may also contribute to the large group variance, as GABA+ and Glu have been shown to vary signicantly with hormonal effects and throughout the menstrual cycle.50 However, in our sample, the variance of males and females did not differ for GABA+ in ADHD (F(1,7) = 4.7, P = 0.06) or control adults (F(1,18) = 1.5, P = 0.39). The handedness was also carefully balanced between groups but the inclusion of both left- and right-handed participants may also contribute to the group variance. Future multicenter or meta-analytic studies with larger group sizes, or groups of male, right-handed participants only may be able to elucidate further the developmental and ADHD effects observed in the present study.

In addition, medication effects could introduce variance, even though participants taking stimulants were asked to withdraw from treatment for 72 h before the MRS session. While some studies have reported a decrease in Glx levels following stimulant treatment,45 others showed no change in Glx between medicated and medication naive adults with ADHD.20 However, in addition to psychostimulant medication for ADHD, four adult participants also reported current or previous use of antidepressants within 1 month of the MRS measurement, for sleep problems. As antidepressants have been shown to alter Glx51 and GABA

levels,52 to assess the potentially confounding effects of anti-depressant treatment on the MRS results, the statistical comparisons were repeated after excluding all participants on antidepressants. In this smaller subsample, the developmental GABA+/H2O interaction was of borderline signicance with a similar effect size (F(1,80) = 3.93, P = 0.05, 2 = 0.03) compared with F(1,85) = 5.11, P = 0.026, 2 = 0.04 in the full group. The differences in GABA+/H2O between ADHD and control adults were not signicant anymore due to the smaller sample size, but showed a similar effect size (t(47) = 1.58, P = 0.1, d = 0.46 compared with t(52) = 2.03, P = 0.048, d = 0.56). Therefore, while the results do not appear to be driven by medication effects, the variability in medication status of the ADHD adults does represent a signicant source of variance, and these results should be reproduced in a sample of medication naive patients.

LCModel offers improved reproducibility for GABA detection relative to alternative spectral analysis methods,53,54 but with the

default settings the baseline t can include a large fraction of the GABA peak, introducing an error into the estimated GABA concentrations. However, with the MEGA-PRESS-2 and MEGA-PRESS-3 control parameters (available from version 6.3 onwards), the default baseline tting in LCModel is turned off and this error is reduced.

Although the CRLB cutoff used for Gln was relatively lenient (30%), using a stricter cutoff of 20% resulted in the exclusion of four additional children (two ADHD children and two control children). In this smaller subset of participants, the group difference in Gln in children was no longer signicant (W = 112.0, P = 0.137, r = 0.30, compared with W = 152.0, P = 0.041, r = 0.38 in the full group). However, as higher CRLB are expected for lower concentration metabolites, using a strict CRLB cutoff for low-concentration metabolites like Gln and GABA can introduce a systematic bias toward higher concentrations as lower concentrations are more likely to be excluded. This potential bias can

confound the analysis of group differences and should be considered carefully when selecting a CRLB cutoff.

CONCLUSIONADHD patients show altered levels of GABA+ in the subcortical voxel (centered on the basal ganglia) which change with development. Children with ADHD also show increased Gln levels in the subcortical voxel, but this difference normalized in adults. These imbalances in neurotransmitter levels are associated with symptom scores and lend new insight in the development and pathophysiology of ADHD.

CONFLICT OF INTEREST

The authors declare no conict of interest.

ACKNOWLEDGMENTS

We thank our participants and their parents for the participation in our study. We also thank Dr Hanna Christiansen for providing the unpublished CAARS questionnaires and normative data. This cooperative project was funded by the University Research Priority Program Integrative Human Physiology. SB had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

REFERENCES

1 Polanczyk G, Rohde LA. Epidemiology of attention-decit/hyperactivity disorder across the lifespan. Curr Opin Psychiatry 2007; 20: 386392.

2 American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5). American Psychiatric Association Press: Washington, DC, USA, 2013.

3 Ferreira PEMS, Palmini A, Bau CHD, Grevet EH, Hoefel JR, Rohde LA et al. Differentiating attention-decit/hyperactivity disorder inattentive and combined types: a 1H-magnetic resonance spectroscopy study of fronto-striato-thalamic regions. J Neural Transm 2009; 116: 623629.

4 Rae CD. A guide to the metabolic pathways and function of metabolites observed in human brain (1)h magnetic resonance spectra. Neurochem Res 2014; 39: 136.

5 Tritsch NX, Ding JB, Sabatini BL. Dopaminergic neurons inhibit striatal output through non-canonical release of GABA. Nature 2012; 490: 262266.

6 Boy F, Evans CJ, Edden RAE, Lawrence AD, Singh KD, Husain M et al. Dorsolateral prefrontal -amino butyric acid in men predicts individual differences in rash impulsivity. Biol Psychiatry 2011; 70: 866872.

7 Silveri MM, Sneider JT, Crowley DJ, Covell MJ, Acharya D, Rosso IM et al. Frontal lobe -aminobutyric acid levels during adolescence: associations with impulsivity and response inhibition. Biol Psychiatry 2013; 74: 296304.

8 Courvoisie H, Hooper SR, Fine C, Kwock L, Castillo M. Neurometabolic functioning and neuropsychological correlates in children with ADHD-H: preliminary ndings. J Neuropsychiatry Clin Neurosci 2004; 16: 6369.

9 Moore CM, Biederman J, Wozniak J, Mick E, Aleardi M, Wardrop M et al. Differences in brain chemistry in children and adolescents with attention decit hyperactivity disorder with and without comorbid bipolar disorder: a proton magnetic resonance spectroscopy study. Am J Psychiatry 2006; 163: 316318.

10 Carrey NJ, MacMaster FP, Gaudet L, Schmidt MH. Striatal creatine and glutamate/ glutamine in attention-decit/hyperactivity disorder. J Child Adolesc Psycho-pharmacol 2007; 17: 1117.

11 MacMaster FP, Carrey N, Sparkes S, Kusumakar V. Proton spectroscopy in medication-free pediatric attention-decit/hyperactivity disorder. Biol Psychiatry 2003; 53: 184187.

12 Perlov E, Philipsen A, Matthies S, Drieling T, Maier S, Bubl E et al. Spectroscopic ndings in attention-decit/hyperactivity disorder: review and meta-analysis. World J Biol Psychiatry 2009; 10: 355365.

13 Carrey N, MacMaster FP, Fogel J, Sparkes S, Waschbusch D, Sullivan S et al. Metabolite changes resulting from treatment in children with ADHD: a 1HMRS study. Clin Neuropharmacol 2003; 26: 218.

14 Yeo RA, Hill DE, Campbell RA, Vigil J, Petropoulos H, Hart B et al. Proton magnetic resonance spectroscopy investigation of the right frontal lobe in children with attention-decit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 2003; 42: 303310.

15 Sun L, Jin Z, Zang Y, Zeng Y, Liu G, Li Y et al. Differences between attention-decit disorder with and without hyperactivity: ao sup4 1o/sup4 H-magnetic resonance spectroscopy study. Brain Dev 2005; 27: 340344.

Translational Psychiatry (2015), 1 8

Developmental course of ADHD S Bollmann et al

8

16 Benamor L. (1)H-Magnetic resonance spectroscopy study of stimulant medication effect on brain metabolites in French Canadian children with attention decit hyperactivity disorder. Neuropsychiatr Dis Treat 2014; 10: 4754.

17 Perlov E, Tebarzt van Elst L, Buechert M, Maier S, Matthies S, Ebert D et al. H1-MR-spectroscopy of cerebellum in adult attention decit/hyperactivity disorder. J Psychiatr Res 2010; 44: 938943.

18 Perlov E, Philipsen A, Hesslinger B, Buechert M, Ahrendts J, Feige B et al. Reduced cingulate glutamate/glutamine-to-creatine ratios in adult patients with attention decit/hyperactivity disorder A magnet resonance spectroscopy study. J Psychiatr Res 2007; 41: 934941.

19 Dramsdahl M, Ersland L, Plessen KJ, Haavik J, Hugdahl K, Specht K. Adults with attention-decit/hyperactivity disordera brain magnetic resonance spectroscopy study. Front Psychiatry 2011; 2: 65.

20 Maltezos S, Horder J, Coghlan S, Skirrow C, OGorman R, Lavender TJ et al. Glutamate/glutamine and neuronal integrity in adults with ADHD: a proton MRS study. Transl Psychiatry 2014; 4: e373.

21 Bak LK, Schousboe A, Waagepetersen HS. The glutamate/GABA-glutamine cycle: aspects of transport, neurotransmitter homeostasis and ammonia transfer. J Neurochem 2006; 98: 641653.

22 Hancu I, Port J. The case of the missing glutamine. NMR Biomed 2011; 24: 529535.

23 Thberge J, Bartha R, Drost DJ, Menon RS, Malla A, Takhar J et al. Glutamate and glutamine measured with 4.0 T proton MRS in never-treated patients with schizophrenia and healthy volunteers. Am J Psychiatry 2002; 159: 19441946.

24 Rothman DL, Sibson NR, Hyder F, Shen J, Behar KL, Shulman RG. In vivo nuclear magnetic resonance spectroscopy studies of the relationship between the glutamate-glutamine neurotransmitter cycle and functional neuroenergetics. Philos Trans R Soc Lond B Biol Sci 1999; 354: 11651177.

25 Edden RAE, Crocetti D, Zhu H, Gilbert DL, Mostofsky SH. Reduced GABA concentration in attention-decit/hyperactivity disorder reduced GABA concentration in ADHD. Arch Gen Psychiatry 2012; 69: 750753.

26 Nakao T, Radua J, Rubia K, Mataix-Cols D. Gray matter volume abnormalities in ADHD: voxel-based meta-analysis exploring the effects of age and stimulant medication. Am J Psychiatry 2011; 168: 11541163.

27 Hart H, Radua J, Nakao T, Mataix-Cols D, Rubia K. Meta-analysis of functional magnetic resonance imaging studies of inhibition and attention in attention-decit/hyperactivity disorder: exploring task-specic, stimulant medication, and age effects. JAMA Psychiatry 2013; 70: 185198.

28 Castellanos FX, Lee PP, Sharp W, Jeffries NO, Greenstein DK, Clasen LS et al. Developmental trajectories of brain volume abnormalities in children and adolescents with attention-decit/hyperactivity disorder. JAMA 2002; 288: 1740.

29 Delmo C, Weiffenbach O, Stadler C, Poustka F. Diagnostisches Interview Kiddie-Sads-Present and Lifetime Version (K-SADS-PL): 5: Auage derdeutschen Forschungsversion, erweitert umICD-10-Diagnostik. Frankfurt, Germany: Klinik fr Psychiatrie und Psychotherapie des Kindes- und Jugendalters, 2001.

30 Christiansen H, Kis B, Hirsch O, Philipsen A, Henneck M, Panczuk A et al. German validation of the Conners Adult ADHD Rating Scalesself-report (CAARS-S) I: Factor structure and normative data. Eur Psychiatry 2011; 26: 100107.

31 Retz-Junginger P, Retz W, Blocher D, Stieglitz R-D, Georg T, Supprian T et al. Reliabilitt und Validitt der Wender-Utah-Rating-Scale-Kurzform. Nervenarzt 2003; 74: 987993.

32 Lidzba K Christiansen H Drechsler R. Conners Skalen zu Aufmerksamkeit und Verhalten - 3. Deutschsprachige Adaptation der Conners, 3rd edition Conners 3. von C. Keith Conners (Vol. in press): Huber: Bern, Germany, 2013.

33 Woerner W, Becker A, Rothenberger A. Normative data and scale properties of the German parent SDQ. Eur Child Adolesc Psychiatry 2004; 13: II310.

34 Gasparovic C, Song T, Devier D, Bockholt HJ, Caprihan A, Mullins PG et al. Use of tissue water as a concentration reference for proton spectroscopic imaging. Magn Reson Med 2006; 55: 12191226.

35 Provencher SW. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med 1993; 30: 672679.

36 R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing: Vienna, Austria, 2013.

37 Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer: New York, NY, USA, 2009.

38 Hochberg Y, Benjamini Y. More powerful procedures for multiple signicance testing. Stat Med 1990; 9: 811818.

39 Shaw P, Eckstrand K, Sharp W, Blumenthal J, Lerch JP, Greenstein D et al.

Attention-decit/hyperactivity disorder is characterized by a delay in cortical maturation. Proc Natl Acad Sci USA 2007; 104: 1964919654.

40 Frodl T, Skokauskas N. Meta-analysis of structural MRI studies in children and adults with attention decit hyperactivity disorder indicates treatment effects.Acta Psychiatr Scand 2012; 125: 114126.

41 Spencer TJ, Brown A, Seidman LJ, Valera EM, Makris N, Lomedico A et al. Effect of psychostimulants on brain structure and function in ADHD: a qualitative literature review of magnetic resonance imaging-based neuroimaging studies. J Clin Psychiatry 2013; 74: 902917.

42 Proal E, Reiss PT, Klein RG, Mannuzza S, Gotimer K, Ramos-Olazagasti MA et al.

Brain gray matter decits at 33-year follow-up in adults with attention-decit/ hyperactivity disorder established in childhood. Arch Gen Psychiatry 2011; 68: 11221134.

43 Gao F, Edden RAE, Li M, Puts NAJ, Wang G, Liu C et al. Edited magnetic resonance spectroscopy detects an age-related decline in brain GABA levels. Neuroimage 2013; 78: 7582.

44 Aufhaus E, Weber-Fahr W, Sack M, Tunc-Skarka N, Oberthuer G, Hoerst M et al.

Absence of changes in GABA concentrations with age and gender in the human anterior cingulate cortex: A MEGA-PRESS study with symmetric editing pulse frequencies for macromolecule suppression. Magn Reson Med 2013; 69: 317320.

45 Carrey N, MacMaster FP, Sparkes SJ, Khan SC, Kusumakar V. Glutamatergic Changes with Treatment in Attention Decit Hyperactivity Disorder: A Preliminary Case Series. J Child Adolesc Psychopharmacol 2002; 12: 331336.

46 Sparkes SJ, MacMaster FP, Carrey NC. Proton magnetic resonance spectroscopy and cognitive function in pediatric attention-decit/hyperactive disorder. Brain Cogn 2004; 54: 173175.

47 Jin Z, Zang YF, Zeng YW, Zhang L, Wang YF. Striatal neuronal loss or dys-function and choline rise in children with attention-decit hyperactivity disorder: a 1H-magnetic resonance spectroscopy study. Neurosci Lett 2001; 315: 4548.

48 Hesslinger B, Thiel T, Tebartz van Elst L, Hennig J, Ebert D. Attention-decit disorder in adults with or without hyperactivity: where is the difference? A study in humans using short echo 1H-magnetic resonance spectroscopy. Neurosci Lett 2001; 304: 117119.

49 Henry P-G, Dautry C, Hantraye P, Bloch G. Brain GABA editing without macromolecule contamination. Magn Reson Med 2001; 45: 517520.

50 Epperson CN, Haga K, Mason GF, Sellers E, Gueorguieva R, Zhang W et al. Cortical gamma-aminobutyric acid levels across the menstrual cycle in healthy women and those with premenstrual dysphoric disorder: a proton magnetic resonance spectroscopy study. Arch Gen Psychiatry 2002; 59: 851858.

51 Taylor M, Murphy SE, Selvaraj S, Wylezinkska M, Jezzard P, Cowen PJ et al. Differential effects of citalopram and reboxetine on cortical Glx measured with proton MR spectroscopy. J Psychopharmacol 2008; 22: 473476.

52 Bhagwagar Z, Wylezinska M, Taylor M, Jezzard P, Matthews PM, Cowen PJ.

Increased brain GABA concentrations following acute administration of a selective serotonin reuptake inhibitor. Am J Psychiatry 2004; 161: 368370.

53 Mullins PG, McGonigle DJ, OGorman RL, Puts NAJ, Vidyasagar R, Evans CJ et al.

Current practice in the use of MEGA-PRESS spectroscopy for the detection of GABA. Neuroimage 2014; 86: 4352.

54 OGorman RL, Michels L, Edden RAE, Murdoch JB, Martin E. In vivo detection of GABA and glutamate with MEGA-PRESS: reproducibility and gender effects.J Magn Reson Imaging 2011; 33: 12621267.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/ http://creativecommons.org/licenses/by/4.0/

Web End =by/4.0/

Supplementary Information accompanies the paper on the Translational Psychiatry website (http://www.nature.com/tp)

Translational Psychiatry (2015), 1 8