INTRODUCTION

Schizophrenia is a chronic mental illness characterized by psychotic or positive symptoms (hallucinations and delusions), and negative and cognitive symptoms (apathy, social withdrawal, decreased attention, decreased executive function, abnormal psychomotor speed of processing and impairment of verbal memory) (APA, 2000; Rajji et al, 2009). Despite treatment advances, schizophrenia remains as a seriously disabling, lifelong illness, that is among the world's top ten causes of long-term disability (Saraceno, 2002). The onset of schizophrenia is usually preceded by a ‘prodromal phase’, characterized by subthreshold psychotic symptoms, a high likelihood of a family history of schizophrenia, and a decline in everyday functioning (Yung and McGorry, 1996). Longitudinal studies, in developed countries, have found that between 19% and 35% of individuals with prodromal symptoms experience conversion to a primary psychotic illness across 1- to 2.5-year follow-up intervals (Yung et al, 2003; Cannon et al, 2008; Ruhrmann et al, 2010), which provides an ideal window of opportunity to assess biomarkers associated with progression of full-blown psychotic illness.

At the present time, pharmacological management of schizophrenia is based on antagonists or partial agonists of the dopamine D2 receptors (Kapur et al, 2000; Seeman and Kapur, 2000; Mamo et al, 2007). Although pharmacotherapy improves psychotic symptoms in most patients, improvement in negative and cognitive symptoms is at best minimal (Lieberman et al, 2005), and specific therapeutic strategies have yet to be tested in adequately powered samples to fully assess their impact on the management and conversion rate of the prodromal phase (Cadenhead et al, 2010).

The clinical effect of dopamine antagonists has been the basis for ‘The dopamine hypothesis of schizophrenia’ (Howes and Kapur, 2009a), which posits an aberrant function of the dopaminergic system in patients with schizophrenia (Sato et al, 1992; Laruelle et al, 1996; Abi-Dargham et al, 1998; Hietala et al, 1999; Kapur, 2003). The hypothesis has been bolstered by the recent observation of an elevated striatal (18)F-dopa uptake in subjects with prodromal symptoms (Howes et al, 2009b).

Although the dopamine hypothesis has been a useful model in our understanding and study of the psychotic state, it does not explain the deteriorating course in terms of cognition and function seen in the first few years of schizophrenia. Glutamate antagonists are well known to induce positive and negative psychotic symptoms more akin to schizophrenia than the positive symptoms induced by dopamine agonists alone (Javitt, 2007). It has been proposed that this deterioration course may be partially explained by cortical neuronal toxicity secondary to enhanced glutamate exposure (Sharp et al, 2001), which in turn is thought to reflect a compensatory increase in cortical glutamatergic activity due to hypofunction of the N-methyl-D-aspartate (NMDA) receptor (Olney and Farber, 1995).

The interaction between glutamate and dopamine is widely documented (Cepeda and Levine, 1998; Levine and Cepeda, 1998; Kulagina et al, 2001; West et al, 2003; David et al, 2005). In schizophrenia, dopaminergic dysregulation is thought to be the final common pathway resulting from an altered glutamatergic neurotransmission (Carlsson and Carlsson, 1990; Javitt and Zukin, 1991; Olney and Farber, 1995). Disruption of the cortical glutamatergic afferents induce decreased tonic dopamine release with a subsequent disinhibition of phasic dopamine release, causing abnormal responses to insignificant stimuli (Grace, 1991, 1993, 2000).

Single photon emission computed tomography (SPECT) and positron emission tomography (PET) studies in humans have provided evidence that non-competitive glutamate NMDA receptor antagonists, such as ketamine, increase amphetamine-induced dopamine release (Kegeles et al, 2000), and decrease D2/3 binding in the posterior cingulate cortex (Aalto et al, 2005) and striatum (Breier et al, 1998; Smith et al, 1998; Vollenweider et al, 2000). Other studies though have not supported this finding in the striatum (Aalto et al, 2002; Kegeles et al, 2002).

Recent developments in proton magnetic resonance spectroscopy (1H-MRS) permit the in vivo study of the glutamatergic system (Di Costanzo et al, 2003; Abbott and Bustillo, 2006; Di Costanzo et al, 2007).

We therefore used 1H-MRS to compare the glutamate levels in antipsychotic-naïve subjects with prodromal symptoms, antipsychotic-naïve patients with a first episode of psychosis and sex-and-age similar healthy controls. We selected two regions of interest for our analysis: the precommissural dorsal caudate, which is characterized by a high density of dopamine receptors and dopamine afferents, and the cerebellum, a brain region with negligible quantity of dopamine receptors and absence of dopamine afferents. It was recently shown that both patients with schizophrenia and prodromal subjects have higher dopamine levels in the associative striatum (Howes et al, 2009b; Kegeles et al, 2010), of which the precommissural dorsal caudate (henceforth, dorsal caudate) is the main component (Mawlawi et al, 2001). Our hypothesis was that subjects with prodromal symptoms and patients with a first episode of psychosis would show higher glutamate than controls in the dorsal caudate as measured by 1H-MRS, and no differences of glutamate between groups in the cerebellum.

MATERIALS AND METHODS

Clinical Sample

The study was approved by the Ethics and Scientific Committees of the National Institute of Neurology and Neurosurgery of Mexico (INNN). All the subjects were included following successful completion of an informed consent procedure, and with written consent of both parents for subjects under 18 years old.

Eighteen subjects with ultra high-risk for schizophrenia–or prodromal symptoms (UHR), and 18 patients during their first non-affective psychotic episode (FEP) were recruited from the inpatient psychiatric service, first psychotic episode clinic, and the Adolescent Program of Neuropsychiatric and Imaging Study (PIENSA) of the INNN. All subjects were interviewed using the structured clinical interview for DSM-IV (First et al, 1997) and the UHR group met Structured Interview for Prodromal Syndromes (SIPS) criteria (Miller et al, 2003) for study entry. Both groups were antipsychotic naïve, and were capable to grant informed consent. Patients were excluded if they had the following: any concomitant medical or neurological illness, current substance abuse or history of substance dependence (excluding nicotine), comorbidity of any other axis I disorders, were considered to be at high risk for suicide, or show psychomotor agitation. Use of psychotropic medications was not permitted for the duration of the study. Forty right-handed age- and gender-matched healthy controls were also recruited. The control subjects were assessed in the same manner as the patients and any subject with a history of psychiatric illness or positive familiar history for schizophrenia was excluded. All participants were screened for drugs of abuse (eg, cannabis, cocaine, heroin, opioids and benzodiazepines) before the 1H-MRS studies.

Magnetic Resonance Studies

The 1H-MRS studies were performed in a 3T GE (GE Healthcare, Milwaukee, WI) whole-body scanner with a high-resolution 8-channel head coil (Invivo, Orlando, FL). The participant’s head was positioned along the cantho-meatal line and immobilized by means of a forehead strap. 1H-MRS spectra were obtained using point-resolved spectroscopy (PRESS; TE=35 ms, TR=2000 ms, spectral width=5000 Hz, 4096 data points used, 128 water-suppressed, and 16 water-unsuppressed averages) in volume elements (voxels) of 8 ml (2 × 2 × 2 cm) centered in the right dorsal caudate nucleus and right cerebellar cortex in all studied subjects. This acquisition allowed for the quantification of glutamate (Glu), N-acetyl-aspartate (NAA), creatine (Cr), phosphocreatine (PCr), glycerophosphocholine (GPC), phosphocholine (PCh), and myo-inositol (mI).

The spectra were shimmed to achieve full width at half maximum (FWHM) ⩽12 Hz (measured on the unsuppressed water signal from the voxel), and spectra with larger FWHM were excluded. The voxels were defined in a T1 weighed volumetric image in axial projection (SPGR, TE=5.7 ms, TR=13.4 ms, TI=450 ms, flip angle=20°, FOV=25.6 cm, 256 × 256 matrix, slice thickness=1.2 mm) and oriented above and parallel to the anterior (AC) to the posterior commissure (PC).

The lower end of the dorsal caudate (or associative-striatum) voxel was located 3 mm dorsal to the AC, including the maximum amount of gray matter and with a dorsal extension (thickness) of 2 cm. The cerebellar voxel was located in the cerebellar cortex below the inferior cerebellar peduncle avoiding the midline (Figure 1). All spectra were analyzed with the Linear Combination Model included in the LCModel program, version 6.2-1T (Provencher, 1993). The metabolites included in the basis set were as follows: L-alanine (Ala), aspartate (Asp), Cr, γ-aminobutyric acid (GABA), glucose (Glc), glutamine (Gln), Glu, GPC, PCh, L-lactate (Lac), mI, NAA, N- acetylaspartylglutamate acid (NAAG), Scyllo-inositol (Scyllo), taurine (Tau), Cr methylene group (–CrCH2), guanidinoacetate (Gua), GPC+PCh, NAA+NAAG, Glu+Gln, lipids (Lip), and macromolecules (MM): Lip13a, Lip13b, Lip09, MM09, Lip20, MM20, MM12, MM14, MM17, Lip13a+Lip13b, MM14+Lip13a+Lip13b+MM12, MM09+Lip09, and MM20+Lip20. All metabolites with Cramer-Rao lower-bound >20% reported by LCModel were excluded.

Figure 1
figure 1

Spectroscopic voxel placement in right dorsal caudate and right cerebellum and representative spectra for each region. Glu, glutamate; Gln, glutamine; NAA, N-acetylaspartate.

PowerPoint slide

Full size image

SPGR scans used for localization of the voxels were subsequently segmented into gray matter, white matter, and cerebrospinal fluid (CSF) using Statistical Parametric Mapping 8 (Friston et al, 1995) (SPM8, Wellcome Department of Cognitive Neurology, London; http://www.fil.ion.ucl.ac.uk/spm). Then, the size and location of each area from the spectra file headers were extracted to calculate the percentage of gray, white, and CSF content within the voxel, allowing for correction of the CSF fraction of the spectroscopic values. CSF correction assumed zero metabolite concentration in the CSF using the equation SVcorr=SV/(1−CSF), where SV are the spectroscopic values, CSF is the CSF fraction within the voxel and SVcorr are the corrected spectroscopic values.

Statistical Analysis

The results are presented in means and standard deviations (±SD). Statistical analyses were performed using SPSS v16.0 software (SPSS, Chicago, IL). Demographic and clinical characteristics of the sample were compared between controls, FEP, and UHR groups with analysis of variance (ANOVA), with the exception of frequency data. Frequency data were analyzed using χ2 or Fisher's exact tests. Metabolite measures between groups were compared using a general linear model (GLM). The percentages of gray matter in the dorsal caudate and cerebellum, as well as age were included as covariates in the GLM. On the basis of significant main effects per metabolite and region, post hoc comparisons with Bonferroni correction were performed. The statistical comparisons were carried out at a significance level set at p<0.05.

Partial Spearman correlations controlling for the effect of gray matter were conducted to examine the relationship between clinical scales and Glu concentration for each region. Spearman correlations (non-parametric) rather than Pearson coefficients were used due to the relatively small sample size. The statistical threshold was established with p⩽ 0.05 to control for multiple comparisons, p=0.05/4 clinical scales for UHR group (SIPS positive, negative, disorganization, general), and p=0.05/3 clinical scales for FEP (PANSS positive, negative, and general).

RESULTS

Demographic and Clinical Characteristics

The UHR group was younger than the FEP (F[2,73]=3.60, p=0.03), but both groups did not differ from controls. Education was higher in controls compared with FEP and UHR groups (F[2,73]=9.59, p<0.001). One subject in the UHR and two in the FEP groups previously used cannabis (Fisher's exact test=6.12, p=0.03). Four subjects in the UHR group were taking a selective serotonin reuptake inhibitor (two fluoxetine, one paroxetine, one sertraline) at the time of the study (Fisher's exact test=9.26, p=0.005). UHR and FEP groups did not differ in gender, handedness, or tobacco use compared with controls (see Table 1).

Table 1 Demographic and Clinical Characteristics of the Sample
Full size table

Cerebral Metabolites

There was a significant difference in Glu levels in the dorsal caudate between groups (F[2,73]=5.99, p=0.004). Post hoc tests revealed higher Glu in both UHR and FEP groups than controls (p=0.04 and p=0.01, respectively), with no statistical difference between FEP and UHR groups (p=0.64). On the contrary, cerebellar Glu concentrations did not differ significantly between the three groups (F[2,73]=1.41, p=0.25) (Figure 2). Glutamine (Gln) was not analyzed due to poor spectra fitting in the caudate (number of spectra rejected in controls=37, UHR=18, and FEP=17) and the cerebellum (number of spectra rejected in controls=40, UHR=18, and FEP=18).

Figure 2
figure 2

Glutamate (Glu) levels of each participant in the dorsal caudate and cerebellum of controls, ultra high-risk for schizophrenia (UHR), and first episode-psychosis patients (FEP). Bars represent the mean for that group. *vs control, p<0.05.

PowerPoint slide

Full size image

NAA levels in the dorsal caudate were higher among patient groups (F[2,73]=6.72, p=0.002), with higher NAA in the UHR group compared with controls (p=0.002). Although NAA was also higher in the FEP group, this finding did not survive correction for multiple comparisons (Table 2). We also found a difference in NAA concentration between groups in the cerebellum (F[2,73]=7.65, p=0.001), with higher NAA in the FEP group vs both controls (p=0.003) and the UHR group (p=0.001).

Table 2 Means (±SD) for Each Metabolite in Dorsal Caudate and Cerebellum
Full size table

GPC+PCh concentrations, for which we had no a priori hypotheses, in the dorsal caudate and cerebellum were different between groups (F[2,73]=7.57, p=0.001 and F[2,73]=4.03, p=0.02, respectively), with higher GPC+PCh in the FEP group compared with controls in the dorsal caudate (p=0.002), as well as the cerebellum (p=0.03).

No differences in Glu+Gln were found between groups in dorsal caudate (F[2,73]=2.11, p=0.13). Although an effect was found in Glu+Gln in the cerebellum (F[2,73]=3.39, p=0.04), post hoc analysis did not show any differences between groups.

The results of metabolite comparisons did not change when age was included as a covariate (data not shown). No significant correlations were found in the dorsal caudate or cerebellum between any clinical measure in UHR or FEP groups and all the metabolites concentrations.

The percentages of gray matter in the dorsal caudate and cerebellum were different between the three groups (F[2,73]=3.44, p=0.04, F[2,73]=9.38, p=0.001, respectively). The post hoc analysis showed that the gray matter in the dorsal caudate was lower in the UHR group than in FEP (p=0.03). The gray matter in the cerebellum was lower in the FEP in comparison with controls (p=0.002) and with UHR (p=0.001). Besides the differences of the percentage of gray matter found in dorsal caudate and cerebellum between the groups, the results of metabolites comparisons did not change whether or not the percentages of gray matter were included as a covariate (data not shown).

Glu and NAA levels were correlated in the caudate and cerebellum of all participants (r=0.77, p⩽0.001 and r=0.71, p⩽0.001, respectively). Moreover, Glu and NAA levels were correlated in the caudate of controls (r=0.70, p⩽0.001), UHR (r=0.74, p⩽0.001), and FEP groups (r=0.78, p⩽0.001), and the cerebellum of controls (r=0.81, p⩽0.001), UHR (r=0.56, p=0.01), and FEP groups (r=0.64, p=0.002) (Figure 3).

Figure 3
figure 3

Correlation between glutamate (Glu) and N-acetylaspartate (NAA) in the dorsal caudate and cerebellum of the participants. Dorsal caudate: Pearson's r=0.70, p⩽0.001 in controls, Pearson's r=0.74, p⩽0.001 in ultra high-risk for schizophrenia (UHR), Pearson's r=0.78, p⩽0.001 in first episode-psychosis patients (FEP). Cerebellum: Pearson's r=0.81, p⩽0.001 in controls, Pearson's r=0.56, p=0.01 in UHR, Pearson's r=0.64, p=0.002 in FEP.

PowerPoint slide

Full size image

DISCUSSION

Summary of Main Results

Our results confirmed our hypothesis that subjects with prodromal symptoms of schizophrenia and unmedicated subjects experiencing a first episode of psychosis, have increased Glu in the dorsal caudate but not in the cerebellum. The absence of differences in the cerebellum suggests that the alterations in Glu in subjects with psychotic and pre-psychotic symptoms are not ubiquitous within the brain, and that the differences may be restricted to brain dopamine-rich regions such as the associative striatum, a region thought to be involved in the pathophysiology of schizophrenia (Howes et al, 2009b; Kegeles et al, 2010).

The associative striatum (or cognitive striatum) (Mawlawi et al, 2001) includes the rostral and dorsal part of the caudate nuclei or precommissural dorsal section, also known as neostriatum (Brodal, 2004). This structure is rich in dopamine afferents and D2 receptors, and is frequently included in the quantification of in vivo occupancy of antipsychotics (Farde et al, 1988; Graff-Guerrero et al, 2008). In addition, the associative striatum establishes major connections with the frontal lobe (Lehericy et al, 2004). The frontal cortex has been implicated in the neurocognitive deficits seen in schizophrenia (Villalta-Gil et al, 2006), deficits which are also present in subjects with prodromal symptoms and FEP patients (Cadenhead, 2002; Jahshan et al, 2010).

The associative striatum, and especially the precommissural dorsal caudate, has shown the highest D2 receptor availability after acute pharmacologically induced dopamine depletion in antipsychotic-free patients with schizophrenia (Kegeles et al, 2010). In this same study, no differences in receptor availability were seen in the other functional subdivisions of the striatum (limbic and sensorimotor striatum), suggesting that schizophrenia is associated with elevated dopamine function in associative regions of the striatum. Moreover, UHR subjects had elevated DOPA decarboxylase activity in this same functional region, suggesting that the subcortical dopamine synthesis is enhanced before the expression of psychosis (Howes et al, 2009b).

Although our main interest was to explore the differences in Glu associated with a dopamine rich region, we included the cerebellar cortex for comparison. The cerebellar cortex has a negligible amount of dopamine receptors and has no dopamine afferents (De Keyser et al, 1988; Camps et al, 1989). On the other hand, both the dorsal caudate and the cerebellar cortex are abundant in glutamatergic cells (Brodal, 2004) and cortical afferents from the frontal cortex (Schmahmann and Pandya, 1995; Middleton and Strick, 2001; Dum and Strick, 2003; Kelly and Strick, 2003). In this sense, one of the differences between the dorsal caudate and the cerebellum are the dopaminergic afferents, which are restricted to the dorsal caudate. Although it is tempting to speculate that some differences observed in the Glu in the dorsal caudate may be related to the dopaminergic tone, our study was not designed to address this question. Although, in agreement with our results, preclinical studies have shown that elevation in striatal endogenous Glu induces an increase in striatal dopamine release (Segovia et al, 1997). Preliminary results from our group (de la Fuente-Sandoval et al, 2009) have shown that patients with schizophrenia during an acute psychotic episode and after 6 weeks of antipsychotic treatment presented with higher Glu levels in the dorsal caudate compared with controls, and with no differences in the cerebellum. These data suggest that higher Glu levels are a stable finding in the dorsal caudate regardless of treatment. However, we acknowledge that to confirm the association between higher Glu with dopamine, a direct measure of both Glu and dopamine should be performed in the same subject.

Glutamate-Dopamine Interaction and Schizophrenia

Abnormal interaction between Glu and dopamine neurotransmission systems has been shown in preclinical model of schizophrenia (Grace, 2000). In addition, using SPECT in normal controls, Kegeles et al. (Kegeles et al, 2000) demonstrated that ketamine administration enhanced an increase in amphetamine-induced dopamine release. In a PET study, ketamine administration resulted in a decrease in striatal [11C]-raclopride binding, reflecting an increase on striatal synaptic dopamine. Moreover, this decrease in [11C]-raclopride binding was similar to the decrease shown after amphetamine administration (Breier et al, 1998). Others (Smith et al, 1998; Vollenweider et al, 2000; Aalto et al, 2005) have also documented this interaction, concluding that glutamatergic antagonism is associated to an increase of dopamine release in the cerebral cortex and the striatum, and inducing positive and negative symptoms similar to those observed in schizophrenia. In addition, NMDA-receptor blockade in healthy humans has been shown to increase Gln levels in the anterior cingulate as measured by 1H-MRS (Rowland et al, 2005).

Glu system dysfunction has been suggested to have a role in schizophrenia (Carlsson and Carlsson, 1990; Javitt and Zukin, 1991; Olney and Farber, 1995). Uncompetitive NMDA receptor antagonists, such as phencyclidine (PCP) and ketamine, induce the full range of positive and negative psychotic symptoms observed in schizophrenia patients (Javitt, 2007). In rats, ketamine administration produced an increase in glutamate release in prefrontal cortex (Moghaddam et al, 1997).

The results of this study, which show higher Glu levels in subjects with pre-psychotic and psychotic symptoms in the dorsal caudate, a brain region previously associated with enhanced dopaminergic transmission in schizophrenia, are consistent with our hypothesis. However, it is important to note that direct measurement of Glu neurotransmission is not possible with 1H-MRS. This technique measures both metabolic and vesicular Glu. A summary of findings from previous studies that measured Glu are presented in Table 3. The reasons for the discrepant findings between studies are unclear; nevertheless, our study illustrates, using a within subject design, that Glu levels are elevated in the dorsal caudate in comparison with the cerebellum.

Table 3 Summary of Glutamate 1H-MRS Studies
Full size table

Glutamate and Neuronal Degeneration

Excessive glutamatergic activity is also associated with neuronal degeneration (Olney et al, 1989). Our study revealed a lower percentage of gray matter in caudate voxels of UHR participants, which may be secondary to excessive glutamate levels. Other studies have also reported a decrease in gray matter in UHR subjects in a variety of other brain regions, including the right medial temporal, lateral temporal, orbitofrontal and cingulate cortex, insula, and inferior frontal and superior temporal gyrus (Pantelis et al, 2003; Borgwardt et al, 2007; Stone et al, 2009). Other factors, however, may have also contributed to our finding of reduced gray matter as a lower percentage of gray matter was not observed in FEP patients. Moreover, our study was not intended to compare gray matter differences between regions, which were principally included in this study as a covariate. As noted above, no differences were seen in the metabolites results whether or not we included this measure as a covariate. Interestingly, we found a decrease in cerebellar gray matter of the FEP group compared with UHR subjects and controls. This finding agrees with previous studies of FEP, drug-naïve patients (Jayakumar et al, 2005; Chua et al, 2007).

Other Metabolites

Our results indicate that Glu levels are significantly correlated with NAA in the caudate and cerebellum in all comparison groups. NAA, one of the prominent peaks consistently shown in 1H-MRS, is present almost exclusively in neurons and is thought to be a marker of neuronal functional integrity (Barker, 2001) and axonal mitochondrial metabolism (Bates et al, 1996). Higher NAA levels may be driven by increased axonal mitochondrial metabolism to maintain axonal conduction (Ariyannur et al, 2008). It is not surprising then that an increase in Glu, an excitatory neurotransmitter, would be associated with an increase in local neuronal metabolism.

We report higher NAA in the cerebellum of FEP patients compared with both controls and UHR subjects. This is inconsistent with the findings of a recent meta-analysis, (Steen et al, 2005) which found no difference in NAA levels between patients with schizophrenia and controls. Another study of early schizophrenia patients with minimal antipsychotic exposure also found no differences in the cerebellum compared with controls (Bustillo et al, 2008). The reasons for these discrepant findings are unclear, but may be related to the clinical status of patients, previous exposure to antipsychotic medication, spectroscopic acquisition (single-voxel vs chemical shift imaging and varying echo time), and the use of ratios (NAA to Cre or Cho) vs metabolite concentrations. Future replication studies are necessary to shed light on these discrepant findings.

Elevated choline containing compounds levels (represented collectively as GPC+PCh) have been interpreted as supportive of the ‘membrane hypothesis’ of schizophrenia (Horrobin et al, 1994), suggesting that phospholipid disturbances and an increased myelin degradation supports a generalized membrane disorder in patients with schizophrenia (Auer et al, 2001). Our results in elevated choline containing compounds (GPC+PCh) found in the dorsal caudate of FEP patients agrees with previous reports in patients with schizophrenia (Bustillo et al, 2002; Lutkenhoff et al, 2010), and in childhood-onset schizophrenia (O’Neill et al, 2004). Three previous reports have failed to find differences in choline in the cerebellum of patients with schizophrenia (Eluri et al, 1998; Tibbo et al, 2000; Bustillo et al, 2008).

Glycerophosphocholine is one the main products of membrane phospholipid breakdown. The observed increase in the FEP group but not in the UHR suggests increased membrane turnover that occurs after a switch to FEP, and may result from changes in membrane mass or proliferation of dendrites and synaptic connections in these regions (Stanley et al, 2006).

Limitations

Limitations of this study need to be considered. First, we did not include cognitive evaluations: we therefore could not address the possibility of an effect of cognition associated with Glu levels. Second, the groups could not be matched for education; education was greater in controls than in FEP and UHR. Parental education and socioeconomic status were not collected but would be better variables for subject matching given that the subjects groups were young and became ill before reaching full educational potential. Third, the spectroscopic voxels involved functionally dissimilar areas and included different proportions of CSF, gray and white matter, making interpretations regarding tissue specificity problematic. We tried to minimize this limitation by correcting for CSF proportion and using the gray matter as a covariate. Although we found differences in dorsal caudate and cerebellum gray matter percentage between the groups, this did not change the results. Fourth, altered NAAG has been described in schizophrenia. Although NAAG could contribute to the quantification of NAA, it could not be differentiated in a reliable way with the methodology employed in our study (Lutkenhoff et al, 2010). Specific techniques to measure NAAG (Edden et al, 2007) could be used in future studies. Fifth, Glu-Gln contamination of the NAA peak has been observed at short echo times, including the one used in this study (TE=35 ms) (Clementi et al, 2005). Thus, this could be reflected as an artificial increase of NAA when Glu is increased. However, this is unlikely as at least the dorsal caudate of FEP showed higher Glu with no difference in NAA levels. Further studies using specific techniques to measure Glu, such as multiple echo times (Zhang et al, 2007), TE-averaged PRESS (Hurd et al, 2004), or constant time PRESS (Mayer and Spielman, 2005) are needed.

Summary

Our results indicate an increase in Glu in the dorsal caudate of antipsychotic naïve subjects with prodromal symptoms of schizophrenia and at a FEP, but not in the cerebellum. Due to the heterogeneity of outcomes in the UHR population, further longitudinal studies are needed to determine the true difference of Glu of those subjects that will convert to psychosis. Replication of our results and the results of longitudinal studies could help to develop a better prediction algorithm for those subjects that will develop a primary psychotic disorder.