Glutamatergic Response to Heat Pain Stress in Schizophrenia

Abstract

Regulation of stress response involves top-down mechanisms of the frontal-limbic glutamatergic system. As schizophrenia is associated with glutamatergic abnormalities, we hypothesized that schizophrenia patients may have abnormal glutamatergic reactivity within the dorsal anterior cingulate cortex (dACC), a key region involved in perception of and reaction to stress. To test this, we developed a somatic stress paradigm involving pseudorandom application of safe but painfully hot stimuli to the forearm of participants while they were undergoing serial proton magnetic resonance spectroscopy to measure changes in glutamate and glutamine levels in the dACC. This paradigm was tested in a sample of 21 healthy controls and 23 patients with schizophrenia. Across groups, glutamate levels significantly decreased following exposure to thermal pain, while ratio of glutamine to glutamate significantly increased. However, schizophrenia patients exhibited an initial increase in glutamate levels during challenge that was significantly different from controls, after controlling for heat pain tolerance. Furthermore, in patients, the acute glutamate response was positively correlated with childhood trauma (r = .41, P = .050) and inversely correlated with working memory (r = −.49, P = .023). These results provide preliminary evidence for abnormal glutamatergic response to stress in schizophrenia patients, which may point toward novel approaches to understanding how stress contributes to the illness.

Introduction

Stress, broadly defined as actual or perceived threat to well-being and demands a homeostatic response,¹ is implicated as an etiological factor in multiple psychiatric disorders, including schizophrenia. Prenatal exposure to stress experienced by mothers² and early life childhood trauma^3,4 may be determinants of vulnerability to schizophrenia later in life. Onset, exacerbation, and relapse of psychosis in schizophrenia patients have all been linked to multiple somatic and environmental stressors.^5–9 However, the biological mechanisms by which stress exposure can increase vulnerability to schizophrenia are unclear.

The initiation and resolution of a stress response require top-down signaling from the central nervous system. Of particular interest in schizophrenia research is the role of glutamatergic signaling in regulating hypothalamic-pituitary-adrenal (HPA) axis functioning^10,11 and the glucocorticoid-mediated effects of stress on glutamatergic signaling.¹² Frontal and limbic regions, particularly the medial frontal and dorsal anterior cingulate cortex (dACC), exert top-down regulation of HPA stress responses, in part mediated by glutamatergic signaling to the hypothalamus.¹⁰ Glutamatergic dysfunction also represents a major mechanistic theory in the pathophysiology of schizophrenia.^13–16 Abnormal glutamine and glutamate levels as measured with proton magnetic resonance spectroscopy (¹H-MRS) have been found in several schizophrenia studies including in the dACC, although the findings were not consistent across studies.^17–20 Excitatory glutamatergic neurotransmission in the ACC plays a critical role in detection, monitoring, and emotional response to stressful and threatening stimuli.^21,22 We hypothesized that glutamatergic dysfunction may be involved in the mechanistic pathways by which stress contributes to schizophrenia pathophysiology.

Direct testing of this hypothesis in patients requires laboratory-based stress challenges with concurrent assessment of their frontal glutamate response. Several previous studies employing functional spectroscopy have found that heat pain can induce transient elevations in glutamate metabolites in the ACC and insula of healthy volunteers^23,24; however, no study to date has employed this technique in schizophrenia research. Although there is individual variability in sensitivity to pain, and schizophrenia is associated with altered pain perception,^25,26 pain thresholds can be objectively measured, and challenge stimuli can be titrated to achieve a more uniform magnitude of subjective aversive response. Furthermore, this challenge can be done concurrently with MRS, allowing measurement of glutamate levels. We chose to study response to pain instead of a social stressor in order to minimize potential confounding factors introduced by schizophrenia psychopathology, with the goal of developing a more reliable model system for studying stress pathophysiology. We hypothesized that schizophrenia would be associated with altered glutamatergic response to pain stress, and that glutamatergic reactivity to stress would be related to clinical characteristics of the illness. We also examine changes in the ratio of glutamine to glutamate, which is emerging as a marker that may be more specific to glutamatergic neurotransmission than glutamate level alone.^19,27

Methods

Participants

Individuals with schizophrenia spectrum disorder were recruited from the outpatient clinics at the Maryland Psychiatric Research Center and neighboring mental health clinics. Healthy controls were recruited through media advertisements. Diagnoses were confirmed with the Structured Clinical Interview (SCID) for DSM-IV in all participants. Exclusion criteria included all major medical conditions, history of epilepsy, cerebrovascular accident, head injury with cognitive sequelae, and mental retardation. Individuals with certain common medical conditions such as diabetes and hypertension were not necessarily excluded as long as the condition was stable and the individuals were under proper treatment. Patients and controls with substance dependence within the past 6 months, or current substance abuse (except nicotine or marijuana) were excluded. Controls had no current DSM-IV Axis I diagnoses and no family history of psychosis in the prior 2 generations. After excluding participants for whom MRS data failed to meet quality standards (see below), the final sample size was 23 patients (n = 21 schizophrenia and n = 2 schizoaffective disorder, mean duration of illness 12.7 ± 9.7 years) and 21 controls. Except for 3 medication-free participants, all schizophrenia patients were on antipsychotic medications, including 16 taking atypical antipsychotics, 2 taking typical antipsychotics, and 2 taking a combination of antipsychotic types. Participants gave written informed consent. Individuals with schizophrenia were evaluated for understanding of study procedures using the Evaluation of Signed Consent.²⁸ Patients did not undergo any study procedures if they did not score at least 10 points out of a possible maximum of 12. This study was approved by the University of Maryland Baltimore IRB.

Clinical Assessments

Overall psychiatric symptoms were assessed with the mean of the 20-item Brief Psychiatric Rating Scale (BPRS).²⁹ Negative symptoms were assessed using the Brief Negative Symptom Scale (BNSS).³⁰ To assess cognitive deficits, participants were tested with the Digit Symbol Coding task of the WAIS-3³¹ and the Digit Sequencing task from the Brief Assessment of Cognition in Schizophrenia,³² to assess processing speed and working memory, respectively. Deficits in these domains are among the most robust cognitive impairments in schizophrenia.^33,34 Major early life stress was measured with the Childhood Trauma Questionnaire (CTQ).³⁵

Thermal Pain Tolerance Testing and Pain Challenge

Thermal stimuli were delivered using the Medoc Pathways system (Medoc) via a 30 mm × 30 mm thermode attached to the forearm of participants. This is a FDA approved device commonly used in pain research²⁴ using a Peltier element thermode with safety features that prevent stimuli greater than 52°C. On a separate day before the pain challenge, all participants underwent testing of heat pain threshold and tolerance. This consisted of a gradually increasing temperature, with participants instructed to click a button to stop the stimulus when they first detect pain (threshold) or at the point when they are longer willing to tolerate the pain (tolerance). Threshold and tolerance were each measured 3 times. Subsequently, participants rated the painfulness of a series of stimuli within the range between threshold and tolerance on a scale of 0–10 based on a visual analogue scale. Temperatures that participants rated as moderately painful (6–8 on a 0–10 scale) were then used as the target temperature for the heat pain challenge during MRS. This was done to individualize stimulation intensity to control for variability in pain thresholds, as schizophrenia patients may have higher thresholds.^25,26

Subsequently, the pain challenge during the MRS study consisted of a preprogrammed series of 10 trials of stimuli around (±0.5°C) the target temperature in one block. For each trial, the temperature ramped up from body temperature to the target temperature, which was maintained for a brief duration before ramping down back to body temperature. The rate of temperature ramping up and down (0.7–5.25°C/s) in each trial and the length of time at target temperature for each stimulus (3–10 s) were pseudorandomly varied. The variations from trial to trial in the duration and characteristics of the stimuli were designed to maximize the unpredictability of the painful stimuli. The interstimulus intervals were also pseudorandomly varied (5–24 s). The total duration of the pain challenge block was 5 min. To confirm the challenge was aversive, participants rated how painful the experience was on a scale of 0–10 following the challenge.

Magnetic Resonance Spectroscopy

All imaging data were acquired using a 3-T Siemens TIM Trio MRI scanner equipped with a 32 channel head coil. A T₁-weighted structural image was acquired for spectroscopic voxel prescription and anatomical reference. Spectra were acquired from a 4.0 × 2.0 × 1.5 cm³ dACC voxel using a very short TE phase rotation stimulated echo acquisition mode localization technique (PR-STEAM: TR/TM/TE = 2000/10/6.5 ms, 2500-Hz spectral width, 2048 complex points, φ1 = 135°, φ2 = 22.5°, φ3 = 112.5°, and φADC = 135°, NEX = 128) (figure 1). A water reference (NEX = 16) was also acquired for phase and eddy current correction as well as quantification. This technique has been demonstrated to be able to detect and distinguish between glutamate and glutamine.^36,37 A basis set of 19 metabolites was simulated using the GAVA software package.³⁸ The basis set was imported into LCModel (6.3-0I) and used for quantification.³⁹ Correction for the proportion of the gray matter, white matter, and cerebrospinal fluid (CSF) within each spectroscopic voxel was performed using SPM8 and in-house Matlab code.⁴⁰ Only glutamate values with Cramer Rao lower bounds (CRLB) <20%, and Gln values with CRLB <30%, were included in statistical analyses. Spectra with LCModel reported linewidths (LW) greater than 0.1 Hz and signal-to-noise ratio (SNR) less than 10 were excluded. Metabolite levels are reported in institutional units (i.u.).

Fig. 1.

(A) Voxel placement in the dorsal anterior cingulate cortex, (B) representative spectrum (gray) with fit (red) and residual (gray line at top), with locations of major peaks for glutamate (Glu) and glutamine (Gln) identified by arrows.

Spectra were acquired in 5 blocks, with each acquisition taking about 5 min: 2 blocks before the painful heat stimuli (prechallenge), one during challenge (challenge) and 2 blocks immediately following the heat pain challenge (postchallenge). By obtaining 2 baseline blocks and averaging them, the repeated baseline block should reduce variability in potential anticipatory responses to MRI and/or the pain block, providing a more stable estimate for the prechallenge glutamate level. The primary assessment was acute glutamate response, calculated by the difference between the challenge—averaged prechallenge values. This was repeated for the acute glutamine/glutamate ratio response. Similarly, resolution of the glutamatergic response was evaluated using the difference between averaged postchallenge assessment—challenge values.

Data from at least one time-point of MRS acquisition were discarded for 3 participants due to poor quality. Furthermore, glutamine data for one or more time-points were discarded for 5 participants (8 time-points total) due to CRLB values >30%. For data retained in analyses, average spectra LWs did not significantly differ between patients (0.038 ± 0.016) and controls (0.033 ± 0.007; F(1, 41) = 1.80, P = .19). Average SNR did not significantly differ between patients (28.6 ± 8.7) and controls (30.9 ± 6.5; F(1, 41) = 0.98, P = .33). Average CRLBs for glutamate did not differ between patients (4.7 ± 1.20) and controls (4.1 ± 0.74; F(1, 41) = 2.93, P = .10). Average CRLBs for glutamine did not differ between patients (16.8 ± 4.67) and controls (15.5 ± 3.39; F(1, 41) = 1.22, P = .28).

Cortisol Response to the Pain Stressor

Saliva samples were collected using oral swabs (Salimetrics) at 5 time-points during the heat challenge, with 2 time-points collected while the participant was in the MRI (figure 2). For estimating potential anticipatory effects on cortisol response during the experiment day, for a subset of participants, saliva samples were also acquired on a nontesting day at about the same time of day as the heat challenge. Saliva swabs were kept at 4°C during the experiment, processed per manufacturer recommendation, and were then stored at −80°C until time of assay. Cortisol was measured using a commercial enzyme-linked immuosorbent assay kit (Salimetrics), using the protocol recommended by the manufacturer.

Fig. 2.

Overview of integrated magnetic resonance spectroscopy (MRS)—thermal pain stress paradigm. Arrows indicate the 5 time-points of saliva collection. Bars between the dotted lines reflect the periods of MRS acquisition. Bar at the top indicates the period of heat stimulation. Dotted line indicates entry into and exit from the MRI scanner.

Statistical Analyses

Group differences in demographic and behavioral variables were examined with t-test or chi-square test as indicated. Cortisol, glutamate, and glutamine/glutamate ratio responses to heat challenge were examined using repeated-measures ANOVAs. Greenhouse-Geisser corrections were applied when sphericity was violated per Mauchly’s test. Linear regression analyses were performed where acute glutamate (and then glutamine/glutamate) response was the dependent variable, with diagnosis, age, sex, and heat tolerance entered as independent variables. Sex was included as females tend to have lower tolerance for heat pain than males⁴¹ and heat pain tolerance was included as it was used to determine the individualized temperature used as the challenge stimuli. Correlations between variables were examined with Pearson’s correlation coefficients, except for variables that deviated from a normal distribution as determined by Kolmogorov–Smirnov tests, in which case Spearman’s rank correlation coefficient was used. All tests were 2-tailed, with alpha set at 0.05.

Results

Heat Threshold and Tolerance

Age and sex ratio were matched between groups per study design (table 1). There were no significant differences between patients and controls in heat sensitivity (t(42) = 1.09, P = .28) or heat tolerance (t(42) = 0.50, P = .62), though the calculated effect size (Hedge’s g) was 0.32, comparable to an effect size of 0.38 found in a previous quantitative review.⁴² There was a nonsignificant trend for heat pain threshold in individuals with schizophrenia to be associated with chlorpromazine-equivalent dose (CPZ) of antipsychotic medication (rho = 0.39, P = .069). Target temperatures used for the heat challenge were not significantly different between patients and controls (t(42) = 0.79, P = .44). Post-MRS ratings of how painful the heat challenge was inside the scanner also did not differ between patients (5.52 ± 2.4) and controls (5.45 ± 1.7; t(42) = 0.11, P = .91).

Table 1.

Demographic, Clinical, and Behavioral Characteristics

	Healthy Control (n = 21)	Schizophrenia (n = 23)	Test-Statistic	P-Value
Age (years)	37.6 ± 15.2	38.0 ± 13.8	t = 0.092	.93
Smoker/nonsmoker	4/17	9/14	χ2 = 2.13	.15
Male/female	12/9	14/9	χ2 = 0.063	.80
Cannabis use within past month/no recent use	3/18	1/22	χ2 = 1.31	.25
Working memory	20.7 ± 3.6	17.5 ± 5.3	t = 2.26	.03
Processing speed	75.9 ± 18.8	61.4 ± 21.3	t = 2.27	.03
CTQ total	37.4 ± 11.5	44.1 ± 12.0	t = 1.87	.07
Heat threshold (°C)	42.6 ± 5.6	44.3 ± 4.8	t = 1.09	.28
Heat tolerance (°C)	48.6 ± 3.6	49.1 ± 2.8	t = 0.50	.62
Challenge stimulus temperature (°C)	46.1 ± 2.0	45.7 ± 2.1	t = 0.79	.44

Note: Variance reported as ± standard deviation. Values for working memory and processing speed are the raw scores on the digit sequencing task and digit symbol coding task, respectively. All variables examined with t-tests were normally distributed as determined by Kolmogorov–Smirnov tests. CTQ, Childhood Trauma Questionnaire.

Acute Glutamatergic Responses

A repeated-measures ANOVA of glutamate at 3 time-points (prechallenge average, challenge, and postchallenge average) with diagnosis and gender as group variables revealed a significant effect of time [F(df = 2) = 4.97, P = .009], with contrasts revealing that glutamate was significantly lower postchallenge compared to previous time-points (figure 3A). There were no significant main effects of diagnosis or time × diagnosis interactions. A similar analysis examining glutamine/glutamate ratios revealed a significant effect of time [F(df = 2) = 3.53, P = .035], with glutamine/glutamate significantly increasing over the course of the experiment (figure 3B). Using a more stringent criteria for glutamine (only including values with CRLB < 20%), the trend for increase in gln/glu ratio over the course of the experiment is still present, but attenuated [F(1.55, 43.5) = 1.60, P = .22]; a paired t-test comparing prechallenge gln/glu to postchallenge gln/glu shows a significant increase [t(33) = 2.19, P = .036).

Fig. 3.

Levels of dACC glutamate (A) and glutamine/glutamate (B) levels during and following pain challenge. The raw values for the acute glutamate response (difference between glutamate levels during challenge and prechallenge level) by diagnosis is displayed in C whereas the group difference in acute glutamate response values predicted by a linear regression model correcting for age, gender, and heat pain tolerance is displayed in D. Similarly, raw values and predicted values from linear regression model for glutamine/glutamate response are shown in E and F, respectively. Error bars represent standard error.

Linear regression analysis for glutamate response showed that there was a significant effect of diagnosis (β = 0.286, P = .042; partial η² = 0.102) and sex (β = −0.324, P = .028; partial η² = 0.118), but no significant effect of age (P = .12) and heat pain tolerance (P = .06), indicating that patients had greater glutamate response to painful heat stimulation compared with controls, after correcting for age, sex, and individual variability in tolerance to heat pain. Females had lower glutamate response compared with males. Linear regression analysis on glutamine/glutamate response showed that there were no significant effects of diagnosis (P = .72), gender (P = .22), age (P = .46), or heat pain tolerance (P = .80). Diagnosis effects on glutamate and glutamine/glutamate responses (raw and normalized to age, sex, and pain tolerance) were plotted in figure 3.

Resolution of Glutamatergic Responses

Linear regression analysis showed that there was no significant effect of diagnosis or sex for glutamate resolution. There was also no significant effect of diagnosis or sex for glutamine/glutamate resolution, suggesting a similar glutamatergic resolution after acute pain challenge in patients and controls. The time course of prechallenge, challenge, and postchallenge glutamate and glutamine/glutamate are plotted in figure 3.

Cortisol Response to Heat Pain Challenge

A repeated-measures ANOVA found a significant effect of time on salivary cortisol levels [F(2.34, 80.0) = 4.95, P = .006], but no main effect of diagnosis (P = .16) or time × diagnosis interaction (P = .84). The pattern of cortisol response suggests that cortisol levels were elevated before the heat challenge, possibly indicating an anticipatory stress effect. This possibility was further supported by cortisol levels in a subset of participants (n = 20) obtained at a similar time of day on a separate day when they were not expecting any MRI or pain testing (figure 4). Patients had similar pattern of response and resolution of the cortisol response compared with the controls under this paradigm.

Fig. 4.

Salivary cortisol responses to heat pain challenge in patients (n = 19; triangles) and controls (n = 17; circles). Bar between Pre-0 and Post 0 min indicates the period when heat pain stimuli were applied. The “resting baseline” was a saliva sample collected in a subset of participants (10 patients and 10 controls) at approximately the same time of day as the “prestress baseline,” but on a separate day when participants were not anticipating any stress challenge. Error bars represent standard error mean.

Association of Cortisol and Glutamatergic Response

Cortisol levels obtained when participants were anticipating the heat pain challenge were correlated with the prechallenge glutamate levels (r = .38, P = .011), but not prechallenge glutamine/glutamate ratio (r = −.16, P = .35). There were no significant correlations between subsequent cortisol levels and subsequent glutamate or glutamine/glutamate ratio levels (all P > .09). Change in cortisol over the course of the study (calculated as area under the curve with respect to increase) did not significantly correlate with the acute glutamate response to heat pain (r = .18, P = .29) nor with glutamine/glutamate ratio response (r = .11, P = .54).

Influence of Childhood Trauma on Glutamate Response

Patients reported insignificantly higher level of childhood trauma as assessed by CTQ total score compared to controls (t = 1.87, P = .07). CTQ score was positively associated with the acute glutamate response (r = .29, P = .05) and also the acute glutamine/glutamate ratio response to heat pain (r = .36, P = .03) in the whole sample. Examining these relationships by diagnosis, in patients, the CTQ summary score was correlated with both acute glutamate (r = .41, P = .05) and also glutamine/glutamate ratio (r = .45, P = .05) responses, but these correlations were not significant in controls (P > .17). However, none of these findings was statistically significant after applying correction for multiple comparisons.

Relationship to Antipsychotic Medication and Symptoms

Among patients, acute glutamate response, but not acute glutamine/glutamate ratio response, was significantly positively correlated with CPZ (r = .51, P = .01). Further investigation of this trend revealed that patients who are taking an atypical antipsychotic (n = 18) had a greater glutamate response (M = 0.56 ± 0.23) than patients who are not on an atypical [n = 5, M = −0.70 ± 0.41; t(21) = 2.6, P = .017]. Cortisol response was not significantly associated with CPZ (r = .18, P = .46). Among patients, neither acute glutamate response nor glutamine/glutamate ratio response were significantly correlated with BPRS or BNSS scores. Cortisol response was also not associated with BPRS or BNSS.

Association of Glutamatergic Response and Cognition

In the entire sample, the acute glutamate response to stress was negatively correlated with working memory (r = −.51, P = .001) and processing speed scores (r = −.36, P = .03), while glutamine/glutamate ratio response was not significantly correlated with either cognitive measure. To examine these relationships in more detail, linear regression analyses were examined in each diagnostic group separately to examine if working memory and processing speed were correlated with acute glutamate response after controlling for age, sex, pain tolerance, and CPZ (in patients only). In patients, working memory was negatively correlated with acute glutamate response (β = −0.69, P = .02) while in controls this was insignificant but in the same direction (β = −0.39, P = .15). Processing speed was also still significantly correlated with acute glutamate response in patients (β = −0.63, P = .04) but not in controls (β = 0.01, P = .96).

Discussion

Glutamate levels in the dACC decreased and glutamine/glutamate ratios increased following exposure to heat pain, possibly consistent with increased glutamate release and conversion to glutamine in response to the pain stimuli. In contrast to controls, schizophrenia patients exhibit an initial increase in glutamate during stress response. The magnitude of this acute glutamate response was associated with more severe working memory and processing speed deficits in patients. These initial results support the use of this paradigm to probe glutamatergic reactivity in schizophrenia research.

MRS studies on glutamate levels in schizophrenia have yielded variable findings, likely due to methodological factors, as well as variability due to age and stage of illness.¹⁸ Some studies suggest elevated levels of glutamate in first episode psychosis, while other evidence indicates that glutamate levels are lower in chronic, older patients. One hypothesis accounting for this apparent aging effect is that abnormalities in glutamatergic metabolism leave patients vulnerable to excitotoxicity, leading to degenerative processes that result in lower cortical thickness and lower glutamate levels in advanced stages of the illness.^43,44 A mechanism preventing glutamate excitotoxicity is the rapid clearance of glutamate from the synapse by transporters and by the conversion of glutamate to glutamine by glutamine synthetase. Expression of glutamine synthetase is upregulated by glucocorticoids in the brain,⁴⁵ possibly a mechanism of buffering against the excitotoxic consequences of glucocorticoid-enhanced glutamate release.^46,47 However, expression of glutamine synthetase has been found to be decreased in the ACC of schizophrenia patients,⁴⁸ which in combination with evidence of altered expression of glutamate transporters,⁴⁹ suggests schizophrenia patients may be more vulnerable to excitotoxicity. In this context, it is worth considering the observations in this study of a rise in glutamine/glutamate ratio in response to heat pain in both patients and controls, while glutamate acutely rose only in patients. It must be noted that glutamate measured with MRS cannot distinguish between the neurotransmitter pool and the metabolic pool of glutamate; however, animal studies and recent clinical work has suggested that the glutamine/glutamate ratio measured with MRS may track glutamatergic neurotransmission and/or abnormalities in the glutamate–glutamine cycle.^19,27 Further studies with parallel animal experiments will be necessary to test the hypothesis that stress is a precipitant contributing to elevated glutamate levels and possibly subsequent excitotoxicity in vulnerable individuals.

Animal models have demonstrated that corticosteroids can rapidly induce a transient increase in glutamate concentration in the brain.^47,50 In this study, baseline cortisol levels were positively correlated with glutamate levels before the pain challenge, but cortisol and glutamate levels were not correlated during or following the pain challenge. However, the pattern of cortisol levels over the course of this experiment suggests that the participants experienced stress from the anticipation of pain; thus, it is possible the glutamate changes in response to the pain were partly masked by anticipatory effects. This, in addition to differences in study design, may explain the discrepancy between our findings and those of earlier studies on pain-induced changes in neurochemistry.^23,24 Further refinement of the paradigm will be necessary to clarify if this paradigm can capture the relationship between glucocorticoid and glutamatergic responses to stress.

The observed relationship between childhood trauma and glutamate response may be consistent with previous evidence linking early life stress to stress reactivity in schizophrenia.⁵¹ The “sensitization” hypothesis predicts that environmental risk factors, including childhood trauma, cumulatively produce a liability for psychosis via persistent and dysfunctional neurobiological responses to stress.^52,53 Evidence from animal models suggests that the limbic and frontal glutamatergic system could be the substrate for this sensitization: epigenetic modulation of mGlu2 receptors by glucocorticoids was found to contribute to individual differences in stress sensitivity in mice⁵⁴; and repeated stress in juvenile rats caused downregulation of glutamate receptors with concomitant cognitive impairment.⁵⁵ In humans, administration of the NMDA antagonist ketamine acutely increases glutamate or glutamine levels^56,57 and can mimic symptoms of schizophrenia.⁵⁸ The neurochemical effects of NMDA antagonists are hypothesized to be due to suppression of GABAergic interneurons, in turn, disinhibiting glutamatergic neurons^56,59; however, acute stress may directly enhance the pool of glutamate available for neurotransmission.⁶⁰ Though we cannot speculate which of these mechanisms underlie our results, our finding here of enhanced glutamate response to heat pain in schizophrenia patients, which was correlated both with childhood trauma and impaired working memory, supports abnormal glutamatergic response to stress as a potential mechanism linking environmental risk factors to schizophrenia.

In interpreting the results of this study, there are several limitations that must be considered. First is the nature of the stressor. We chose to use thermal pain as a stressor because pain is universally aversive, and variability in pain tolerance can be controlled with good precision with the Medoc device used. In contrast, uniform stress burden may be more difficult to achieve with stressors such as social judgment and/or cognitive challenges, especially in a clinical sample of individuals with variable levels of paranoia, negative symptoms, and cognitive deficits. Most research on stress diathesis for schizophrenia has focused on social stressors; hence it remains to be determined if the current results are generalizable to the types of stress patients are more likely to encounter in everyday life. A second limitation is the cross-sectional approach. To determine if abnormal glutamatergic responses to stress are causally related to cognitive deficits will require study of patients in the very early course of illness with longitudinal follow-up. Most of the patients in this study were taking antipsychotic medications, which represents another potential confound, especially as we did find a positive correlation between acute glutamate response and CPZ. Of note, we found modest evidence that this correlation was driven by use of atypical antipsychotics. Animal studies have found that atypical antipsychotics can prevent increases in extracellular glutamate levels induced by NMDA receptor antagonists, and may normalize elevated basal glutamate levels.^61,62 Additionally, there is preclinical evidence that atypical, but not typical, antipsychotics can downregulate the expression of mGlu2⁶³; this receptor is often located presynaptically and helps fine-tune signaling at glutamatergic synapses. However, given the small sample size, discussion of the mechanism underlying the relationship of medications to glutamate change remains speculative.

Overall, this initial application of a MRS—pain stress paradigm in schizophrenia research found that brief exposure to moderate pain induces a rise in glutamine/glutamate ratio in dACC that is similar in schizophrenia patients and healthy controls. In addition, schizophrenia patients exhibit an initial rise in glutamate levels relative to controls. Although this difference was of modest effect size, this effect was significantly associated with basal cognitive functions and may indicate altered glutamatergic response to stress in schizophrenia. Further improvement of methods to assess glutamate responses to stress may facilitate direct testing on the interplay of stress and the glutamatergic system in the development of schizophrenia.

Funding

Support was received from National Institutes of Health (K23MH112010, R01MH085646, R01DA027680, R01MH094520, R01MH096263, T32MH067533, P50 MH103222, and U01MH108148), and a NARSAD Young Investigator Award from the Brain and Behavior Foundation.