Effects of N-Methyl-d-Aspartate Receptor Antagonists on Gamma-Band Activity During Auditory Stimulation Compared With Electro/Magneto-encephalographic Data in Schizophrenia and Early-Stage Psychosis: A Systematic Review and Perspective

Abstract

Background and Hypothesis

N-Methyl-d-aspartate receptor (NMDA-R) hypofunctioning has been hypothesized to be involved in circuit dysfunctions in schizophrenia (ScZ). Yet, it remains to be determined whether the physiological changes observed following NMDA-R antagonist administration are consistent with auditory gamma-band activity in ScZ which is dependent on NMDA-R activity.

Study Design

This systematic review investigated the effects of NMDA-R antagonists on auditory gamma-band activity in preclinical (n = 15) and human (n = 3) studies and compared these data to electro/magneto-encephalographic measurements in ScZ patients (n = 37) and 9 studies in early-stage psychosis. The following gamma-band parameters were examined: (1) evoked spectral power, (2) intertrial phase coherence (ITPC), (3) induced spectral power, and (4) baseline power.

Study Results

Animal and human pharmacological data reported a reduction, especially for evoked gamma-band power and ITPC, as well as an increase and biphasic effects of gamma-band activity following NMDA-R antagonist administration. In addition, NMDA-R antagonists increased baseline gamma-band activity in preclinical studies. Reductions in ITPC and evoked gamma-band power were broadly compatible with findings observed in ScZ and early-stage psychosis patients where the majority of studies observed decreased gamma-band spectral power and ITPC. In regard to baseline gamma-band power, there were inconsistent findings. Finally, a publication bias was observed in studies investigating auditory gamma-band activity in ScZ patients.

Conclusions

Our systematic review indicates that NMDA-R antagonists may partially recreate reductions in gamma-band spectral power and ITPC during auditory stimulation in ScZ. These findings are discussed in the context of current theories involving alteration in E/I balance and the role of NMDA hypofunction in the pathophysiology of ScZ.

Introduction

The pathophysiological processes underlying schizophrenia (ScZ), a severe psychiatric syndrome, remain to be determined. Dopaminergic neurotransmission has been implicated in the generation of positive symptoms, such as delusions and hallucinations, and is the main target for antipsychotic medications (APMs).¹ However, APMs have only limited efficacy for negative symptoms and cognitive impairments.^2,3 More recently, a glutamatergic dysfunction has gained attention as a plausible mechanism, which could potentially account for the developmental profile and cognitive dysfunctions associated with the disorder.^4–7

In particular, aberrant N-methyl-d-aspartate receptor (NMDA-R) functioning has been implicated in circuit dysfunctions in ScZ.⁴ Expression of the NR1 subunit in the prefrontal cortex (PFC) is decreased in postmortem studies,⁸ but also other NMDA-R subunits.^9,10 Importantly, there are findings to suggest NMDA-Rs may be reduced on gamma-aminobutyric acid (GABA) interneurons.¹¹ Finally, genetic evidence has also shown that Copy Number Variants in ScZ patients are enriched for NMDA-Rs.¹²

Further evidence for the role of NMDA-Rs in the pathophysiology of ScZ comes from studies that investigated the effects of NMDA-Rs antagonists, such as ketamine and phencyclidine (PCP), in healthy volunteers.^13,14 NMDA-R antagonists elicit transient positive, negative, and cognitive symptoms in healthy volunteers.^14,15 In addition, acute and chronic administration of NMDA-R antagonists in animal models recreate several core circuit dysfunctions observed in ScZ,^16,17 such as deficits in parvalbumin-expressing GABAergic interneurons.¹⁸

GABAergic interneurons are also critically involved in the generation of neural oscillations as the precise inhibition of pyramidal cell activity regulates the output of neuronal assemblies, leading to rhythmic fluctuations in neural excitability.^19–21 Specifically, parvalbumin-expressing (PV+) GABAergic interneurons have been involved in the generation of high-frequency oscillations¹⁹ as well α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)- and NMDA-R-mediated activation of PV+ interneurons.²²

In ScZ, there is consistent evidence for alterations in both resting-state^23–25 and task-related gamma-band (30–120 Hz) activity.^26–28 These data in turn have been linked to changes in excitation and inhibition balance (E/I-balance) parameters in ScZ, such as the reduced levels of PV+ interneuron activity.^29,30 Moreover, NMDA-R hypofunctioning has also been proposed to be compatible with alterations in both resting-state³¹ and task-related gamma-band parameters in ScZ.³²

In the current study, we carried out a systematic review to test the hypothesis that NMDA-R antagonists, such as ketamine, MK-801, and PCP, in animal models and human participants recreate the pattern of changes observed in electro/magneto-encephalographic (EEG/MEG) data during auditory stimulation in ScZ. A large body of work has investigated gamma-band activity during auditory tasks, such as oddball paradigms^33,34 as well as auditory steady-state responses (ASSR).^35,36 There is evidence for reductions in gamma-band power in chronic ScZ patients,³⁷ first-episode psychosis (FEP) and in high risk for psychosis (CHR-P) participants³⁸ as well as in first-degree relatives.³⁹ However, not all studies have confirmed these findings.^40–42

To link these data with the preclinical and pharmacological literature, we systematically examined studies that investigated the effects of NMDA-R antagonists on gamma-band activity during auditory stimulation in rodent and human electrophysiological recordings with the aim to compare these changes with EEG/MEG data in ScZ and early-stage psychosis patients. Moreover, we examined the potential impact of alterations in baseline gamma-band power and its relationship to NMDA-R antagonists. Accordingly, we hypothesized that auditory gamma-band activity during NMDA-R hypofunctioning in preclinical and human EEG/MEG recordings would be compatible with the findings observed in ScZ patients as well as during early-stage psychosis.^43–45

Method

Publication searches were conducted in Google Scholar and PubMed using a combination of the following search terms for the preclinical literature: “Rodents”/“Monkey,” “Ketamine”/“PCP”/“MK801”/“NMDA-R,” “Gamma oscillations,” “Auditory Task” until January 2022. For human studies investigating the effects of ketamine on gamma-band oscillations, “Human,” “Ketamine,” “NMDA-R,” “Gamma,” and “Auditory Task” were entered. Additionally, studies that examined gamma-band oscillations EEG/MEG data in healthy humans as well as EEG, local field potentials (LFPs), and electrocorticography (ECoG) for preclinical studies were identified with the search terms: “EEG/MEG” and “EEG/LFP/ECoG.” Furthermore, reference lists of appropriate publications were searched to identify studies matching our search criteria. To exclude duplicates from the study search, results from all search terms were combined and PMIDs (unique identifier numbers used in PubMed) were used to exclude potential duplicates. The title and abstract of publications were carefully inspected and where necessary studies were removed that contained data reported in previous publications.

Inclusion criteria for preclinical studies on NMDA-R antagonists were as follows: (1) ketamine, PCP, MK-801, and/or memantine administration, (2) in vivo studies, (3) subanesthetic dosage, and (4) EEG/LFP/EcoG recordings. For human ketamine studies, inclusion criteria were: (1) EEG/MEG recordings, (2) subanesthetic dosage, (3) no previous or current history of neuropsychological disorder/substance abuse, and (4) sample size of n = 10 or more participants.

Finally, inclusion criteria for EEG/MEG studies in ScZ patients were as follows: (1) EEG/MEG and (2) sample size of n = 10 or more patients. In addition, publications that consisted of reviews, meta-analyses, case studies, and case reports were also excluded from further analysis. Preclinical and human studies were excluded where additional drugs in addition to NMDA-R antagonists were administered, such as psychotropic drugs. Finally, genetically modified animals were excluded.

The following parameters were retrieved: (1) preclinical studies: number, type, gender and age of animals, auditory task type, recording and location, frequency range, and type of NMDA-R antagonist drug as well as dosage; (2) healthy human studies: number, gender and age of participants, task type, imaging technique, frequency range, and pharmacological administration; (3) patient studies: patient type (CHR-P, FEP vs chronic ScZ), age and gender of participants, mean illness duration, medication status, task type, imaging technique, and frequency range.

Statistical Analysis

For the estimation of effect sizes, the Comprehensive Meta-Analysis (CMA) software version 3.3.070 was used. Hedges’ g was calculated based on mean scores of gamma-band power values. When these values were not reported, exact F, t, or P values were employed. When effect sizes or appropriate statistics were missing from the reviewed articles, the authors were contacted to request the relevant data.

Effect size analysis was conducted on evoked gamma-band activity in ScZ patients and early-stage psychosis studies and data were available in 39 studies. We used R to plot the standardized mean differences with 95% confidence intervals. Funnel plots were visually inspected for symmetry to assess publication bias and outliers. Egger’s regression test, performed in R was used to assess potential asymmetry and interpreted.^46,47

Gamma-Band Measures

We examined the following measures of gamma-band (30–120 Hz) activity across studies: (1) Evoked power: Evoked activity occurs at a consistent time lag after the onset of an external event and can be identified by averaging the responses over several trials. Because of the close relationship with stimulus-onset, evoked oscillatory activity may reflect the encoding of stimulus information. (2) Intertrial phase coherence (ITPC): Assesses the phase modulations across trials.⁴⁸ An ITPC value close to 0 reflects high variability of phase angles across epochs, whereas an ITPC value of 1 reflects all epochs having the same phase angle. (3) Induced oscillatory power: Induced oscillations are not locked to the onset of a stimulus. Analysis of induced oscillations must therefore be performed on a single-trial basis because averaging across trials would cancel out oscillations owing to random phase shifts.⁴⁸

In addition, we included the analysis of baseline power where such parameters were available based on findings that NMDA-R antagonists increase baseline gamma-band power which in turn could impact task-related gamma-band oscillations.^44,49

Experimental Auditory Tasks

We included auditory tasks that are commonly employed in both preclinical and human studies, such as ASSRs, oddball- or pre-pulse inhibition (PPI) paradigms. However, only reports were considered that analyzed gamma-band responses to tones or stimulus trains. Studies that reported relative measures, such as the gamma-band band response as an expression of PPI or Mismatch Negativity (MMN), were not considered. Moreover, studies that used speech stimuli were not included.

Risk of Bias

The risk of bias and quality of studies were assessed following Cochrane risk of bias guidelines⁴⁷ for human randomized controlled studies, SYRCLE’s risk of bias tool for animal studies⁵⁰ and the ROBINS-I assessment tool⁵¹ for matched cohort studies in ScZ patients (see supplementary tables S1–S3).

Results

Study Selection

1130 preclinical studies were identified. After duplicate removal (n = 551), 468 reports were excluded after title or abstract screening. The remaining studies (n = 111) were assessed for eligibility and 96 were excluded, leading to 15 studies (see figure 1, supplementary table S4).

Fig. 1.

PRISMA chart of preclinical studies.

For studies in human participants with ketamine, n = 701 records were identified and n = 189 duplicates were removed. Following title and/or abstract screening n = 459 records were further excluded. Of the remaining 53 studies, 50 studies were excluded after full-text inspection, leading to a final sample consisting of 3 studies (see figure 2, supplementary table S5).

Fig. 2.

PRISMA chart of human pharmacological studies.

ScZ studies included 661 records, of which 166 were excluded after duplicate removal. Of the identified studies, 21 were obtained from a meta-analysis on ASSRs³⁵ and titles and/or abstracts were screened for 387 studies. Full-text inspection of the reports assessed for eligibility (n = 105 studies) resulted in the further exclusion of 52 studies. Moreover, the papers by Maharajh et al,⁵² Roach et al,⁵³ and Mulert et al⁵⁴ were excluded due to the fact that these were either reanalyses of datasets from previously published studies or that clinical samples were partially overlapping with previous publications.⁵⁵ The final sample consisted of 46 studies (see figure 3, supplementary table S6).

Fig. 3.

PRISMA Chart of EEG/MEG-Studies in Schizophrenia and Early-Stage Psychosis.

Study Characteristics: Preclinical Studies

Twelve studies examined NMDA-R antagonists in rats and 3 studies in mice (supplementary material S4). A variety of auditory tasks were employed: 8 studies used ASSR paradigms, 4 studies employed an auditory oddball paradigm, 2 studies used a PPI, and 2 studies employed a paired-click paradigm.⁵⁶ Pharmacological administration included ketamine (n = 9), 3 studies used MK-801 and 4 studies PCP. Two studies included both ketamine and MK-801.^57,58 Twelve studies examined acute continuous dosages of NMDA-R antagonists, while n = 5 studies used chronic NMDA-R administration. Four studies examined both acute and chronic doses.

The majority of studies employed EEG recordings (13 studies) and n = 1 study utilized EcoG.⁵⁷ EEG recordings included placement of electrodes over different cortical areas, including frontal regions (n = 3 studies), parietal lobe (n = 3 studies), temporal regions (n = 6 studies), occipital areas (n = 2 studies), and motor cortex (n = 1 study).⁵⁹ Subcortical recordings included the hippocampus (n = 2 studies) and the cerebellum (n = 1 study).

All studies examined evoked spectral power (n = 14), while n = 5 also examined ITPC. We differentiated between low (30–60 Hz) (n = 14) and high gamma-band activity (60–120 Hz) (n = 10 studies). The majority of studies investigated evoked activity (n = 10). Six studies investigated induced gamma-band activity, with 2 studies examining both evoked and induced power.^60,61 There were 6 studies that investigated baseline spectral power.

Study Characteristics: Healthy Human Studies

N = 3 studies administered a bolus of ketamine followed by continuous, subanesthetic infusion (0.006 mg/kg/min) (see supplementary table S4). Auditory paradigms included auditory reaction tasks, (n = 2) and a paired-click paradigm (n = 1). All studies used EEG. Spectral power analysis was conducted by Hong et al⁶² while 2 studies investigated PLF and 1 investigated both power and PLF.⁶³ All studies measured evoked activity and baseline spectral power.⁶⁴

Study Characteristics: ScZ Studies

Thirty studies examined patients with ScZ, while n = 4 studies focused on patients with FEP. One study⁶⁵ investigated only CHR-P individuals (supplementary table S6). The remaining studies (n = 11) examined patient cohorts at different illness stages (CHR-P, FEP, and ScZ), including n = 5 studies with CHR-Ps. Only 1 study included antipsychotic naive FEP patients.³⁸

The majority of studies employed ASSR paradigms (n = 29), while 10 studies used an auditory oddball task. Three studies investigated gamma-band activity during an auditory reaction task, and 4 studies during a paired-click paradigm. Thirty-four studies employed EEG while MEG was used in n = 11 studies. ITPC was examined in 32 studies, while evoked power was investigated in 40 studies and induced power in 8 studies. Of the 40 studies investigating evoked power, n = 2 studies^66,67 reported total power and n = 5 reported event-related spectral perturbation. Finally, baseline activity was investigated in 6 studies.

Risk of Bias Within and Across Studies

Preclinical NMDA-R Antagonist Studies

Only 1 study had a high overall risk of bias,⁶⁸ while 4 studies presented unclear overall risk of bias^69–72 and the remaining 10 studies had low overall risk. Random sequence allocation was high in 1 study⁶⁸ and 3 studies were at unclear risk of bias.^68,70,71 Random housing was considered at unclear risk of bias in 2 studies.^56,72 Random outcome assessment was unclear in 3 studies and at high risk in 1 study⁶⁸ (see supplementary table S1).

Human Ketamine Studies

Three studies were considered to have a low risk of bias. The blinding of outcome assessment was unclear in 1 study⁶² (see supplementary table S2).

ScZ Auditory Task Studies

Seven studies were considered to have a high overall risk of bias, 3 studies were at unclear overall risk and 42 studies presented with a low risk of bias (see supplementary table S3). High risk was detected in 6 studies concerning confounders and in 3 studies concerning selection bias,^73–75 1 study for attrition bias.⁵⁵ Unclear risk was observed in 4 studies for confounding,^55,74,76 4 studies concerning selection bias,^66,74,77 1 study for misclassification bias³⁹ and 1 study for measurement bias.⁷⁸

Risk of Publication Bias

Egger’s regression test was significant (t = 3.65, df = 37, P ≤ .0001), suggesting the presence of publication bias. Moreover, a funnel plot indicated some asymmetry, suggesting that the studies investigated in the meta-analysis could be affected by reporting bias (supplementary figure S1).

Additionally, we used the “trim and fill” method to estimate the number of studies needed for the funnel plot to be symmetrical⁷⁹ which indicates that the original results are overestimated due to publication bias (see supplementary figure S2).

Preclinical Studies: Effects of NMDA-R Antagonists on Gamma-Band Activity

Evoked power: Eight out of 14 studies reported reductions in evoked gamma-band power following the administration of NMDA-R antagonists. The remaining studies report enhanced (n = 3),^49,60,69 no change⁸⁰ as well as bidirectional findings (n = 2). Ahnaou et al⁸¹ showed that acute ketamine increased gamma-band activity to standard and deviant stimuli during an oddball paradigm, while chronic ketamine administration decreased high gamma-band activity.

ITPC: Five out of 9 studies showed reduced gamma-band ITPC following NMDA-R antagonists. In contrast, 3 studies reported elevated gamma-band ITPC. Sivarao et al⁸² demonstrated a dose-dependent effect of ketamine: gamma-band ITPC was initially increased at 1 mg/kg while 30 mg/kg resulted initially in a reduction of gamma-band ITPC, followed by an increase after 30 min intravenous administration.

Induced power: Only the study by Lazarewicz et al⁴⁹ examined the effects of ketamine on induced power, reporting a reduction of 30–80 Hz activity.

Baseline activity: Changes in baseline activity following the administration of NMDA-R antagonists were examined in 6 studies. Five studies^{57,59,70,80,83} reported an increase while only the study by Lee et al⁸⁴ showed no effect of PCP on baseline power.

Human Studies: Effects of NMDA-R Antagonists on Gamma-Band Oscillations

Evoked power: Two studies exhibited a reduction in evoked gamma activity following ketamine administration,^63,85 while 1 study⁶² observed an increase in evoked gamma-band power (see supplementary table S6).

ITPC: Two studies^63,85 observed reduced gamma-band ITPC following ketamine administration.

Correlations with psychopathology: There was an association between auditory gamma-band activity and negative symptoms, but the direction of the effect differed across studies. Thus, there was evidence for a positive correlation between negative symptoms and reduced gamma-band oscillations⁶² while Haaf et al⁸⁵ and Curic et al.⁶³ reported a negative relationship between gamma-band power and negative symptoms.

Gamma-Band Activity in ScZ and Early-Stage Psychosis

Gamma-Band Activity in ScZ

Evoked power: Out of 37 studies investigating evoked gamma-band power, 17 studies reported significant reductions. Conversely, 4 studies^86–88 documented an increase in evoked gamma-band activity, all of which were conducted in ScZ patients. Moreover, 7 studies^{32,40,41,77,89–91} did not identify significant differences between ScZ patients and healthy controls, although Rass et al⁹² discerned a trend level. The remaining studies presented mixed results concerning specific frequencies: Brenner et al,⁶⁶ Puvvada et al,⁹³ and Vierling-Claassen et al⁶⁷ detected reductions toward 40 Hz ASSRs but not at other gamma-band frequencies.

ITPC: Twenty-four studies investigated ITPC of which 16 reported decreased gamma-band activity. Two studies^87,88 reported increased ITPC whereas 3 did not observe group differences relative to healthy controls.^41,91,92 Hirano et al⁹⁴ reported no differences at 30 Hz but a significant reduction during 40 Hz ASSRs. Similar results were obtained by Tsuchimoto et al.⁹⁴

Induced power: Seven studies compared induced power between ScZ-patient groups and healthy controls. One study detected a decrease⁹⁵ while 3 reported an increase in induced power.^32,87,96 The studies by Gallinat et al,³³ Nguyen et al,⁹⁷ and Popov et al⁹⁸ found no differences in induced gamma-band power.

Baseline activity: Three studies^32,75,99 reported an increase in baseline activity in ScZ patients vs healthy controls while 3 studies^39,73,88 found no differences.

Effect Size Evoked Gamma-Band Power

The average Hedges’ g random-effect size was −0.63 for evoked gamma-band power (see supplementary figure 3).

Gamma-Band Activity in Early-Stage Psychosis

Evoked power: Out of 8 studies that examined evoked power in CHR-P and FEP groups, 5 reported reduced gamma-band power. Two studies indicated no significant differences,^74,100 although Perez et al⁷⁴ reported a trend-level reduction.⁷⁴ Tada et al¹⁰¹ showed a reduction at 40 Hz ASSRs in both FEP and CHR groups but not at 30 Hz stimulation frequencies.

ITPC: Out of 9 studies, 6 detected reductions in ITPC while 2 studies^100,102 found no difference albeit Leicht et al¹⁰² reported a trend-level reduction in CHR-Ps. Tada et al¹⁰¹ reported reduced ITPC at 40 Hz but not 30 Hz ASSRs in both FEP and CHR-Ps.

Induced power: Wang et al¹⁰³ detected a reduction at 40 Hz power in FEPs.

Gamma-Band Activity and Symptom Correlations

The relationship between clinical symptoms and gamma-band oscillations in ScZ and early-stage psychosis was investigated in 26 studies. Fourteen reported no association between auditory gamma-band activity and psychopathology, while 9 studies observed positive and negative correlations with positive symptoms. Five studies observed negative correlations between reduced gamma-band activity and negative symptoms.^{37,39,95,104,105} In addition, 2 studies^101,106 found a correlation with the Positive and Negative Syndrome Scale (PANSS) total scores while Fujimoto et al¹⁰⁷ reported a correlation between gamma-band activity and disorganized symptoms.

Discussion

The systematic review investigated the effects of NMDA-R antagonists on gamma-band activity during auditory stimulation in animal models and human electrophysiological recordings with the aim to compare these changes with the pattern of gamma-band deficits observed in ScZ and early-stage psychosis patients. It was hypothesized that NMDA-R hypofunction would recreate the pattern of gamma-band activity observed in clinical populations, given the extensive evidence for similarities in psychopathology between the effects of NMDA-R hypofunctioning in healthy volunteers and symptoms of ScZ¹⁴ as well as the potential overlap in circuit dysfunctions elicited by NMDA-R antagonists and observations in ScZ and early-stage psychosis.^44,108

Specifically, we expected NMDA-R antagonists would lead to a deficit in the generation of gamma-band activity characterized by changes either to the signal-to-noise ratio⁴⁴ or impaired generation of high-frequency activity.³² Crucially, we expected that NMDA-R administration would result in changes compatible with task-related deficits observed in ScZ patients and circuit deficits that may differ across illness stages.^38,91,109

Overall, we could partially support these hypotheses. The majority of preclinical studies involving NMDA-R antagonists as well as findings in ScZ and early-stage psychosis support a reduction in evoked and ITPC gamma-band activity. Evoked gamma-band power and ITPC occur typically in a latency range of 50–150 ms, suggesting a role in the initial sensory registration and feedforward propagation of stimulus information.¹¹⁰ However, there was also evidence for increased gamma-band activity and biphasic effects in both patient^86,111 and preclinical studies.^69,112

In regard to induced gamma-band activity, the pattern of deficits was more variable, especially in patient populations where there was evidence for both up- and downregulated gamma-band activity. In contrast to evoked oscillatory, induced gamma-band oscillations are non-phased locked and tend to reflect cognitive processes, such as attention and memory.¹¹³

Finally, while NMDA-R antagonists were consistently associated with an increase in baseline activity in preclinical studies, the evidence from studies in early-stage psychosis and ScZ was inconclusive, suggesting that the mechanisms through which NMDA-R hypofunctioning impacts on aberrant gamma-band oscillations and circuit deficits may be different from those observed in clinical populations. However, further studies are required in ScZ and/or early-stage psychosis populations to substantiate this conclusion.

Previous studies^32,44 have suggested that impairments in sensory-related gamma-band activity in ScZ are the result of an SNR deficit whereby elevated, ongoing gamma-band activity interferes with the generation of precisely timed neuronal oscillations. These data are consistent with findings from a previous systematic review of our group²⁵ that examined the effects of NMDA-R antagonist administration on resting-state gamma-band activity. NMDA-R antagonists in preclinical and human studies were associated with an upregulation of high-frequency power while this observation was not consistent, however, with evidence from ScZ and early-stage psychosis cohorts.

There was considerable heterogeneity in pharmacological agents, recording and analysis techniques as well as experimental tasks that could have contributed toward differences in the direction of effects across studies. Of relevance is the contribution of different dosages of NMDA-R antagonists. While in human EEG/MEG data, studies used consistently subanesthetic dosages,^62,63,85 only preclinical studies were able to examine the effects of different administration regimes and dosages of NMDA-R antagonists.

Previous studies had indicated that acute vs chronic administration of NMDA-R antagonists was associated with differences in both E/I-balance parameters¹¹⁴ and behavioral deficits.¹⁷ Consistent with these findings, Ahnaou et al⁸¹ found that acute ketamine administration increased gamma-band activity while chronic application reduced gamma-band oscillations. However, this was not confirmed by Martin et al.⁶¹ In addition, Sivarao et al⁸² showed a transient augmentation of gamma-band activity through low doses of ketamine (1 mg/kg) in mice while a high dose of ketamine (30 mg/kg) produced a sustained suppression of gamma-band activity in a 40-Hz ASSRs paradigm.

In ScZ patients, there was consistent evidence for a reduction in ITPC and evoked gamma-band activity although this was not confirmed in all studies, with evidence for both an absence of group differences⁹⁰ as well as for increased gamma-band activity.⁸⁸ Notably, a similar pattern was observed in FEP and CHR-P populations, suggesting that deficits in the generation of sensory-related gamma-band oscillations are present across illness stages and may have prognostic potential, for example, in predicting clinical outcomes in CHR-P populations.³⁸ However, the number of studies in early-stage psychosis is still small, and further studies are needed to confirm these findings.

Across EEG/MEG studies in ScZ- and early-stage psychosis groups, there was also no consistent association with clinical symptoms. Changes in E/I balance that are mediated by gamma-band oscillations and glutamate levels have been proposed as a possible mechanism involved in the generation of auditory hallucinations^115,116 and more recent evidence also suggests that ketamine induces both negative and positive symptoms in healthy volunteers.¹¹⁷

Limitations

There are several limitations of the current systematic review. First, we compared several auditory paradigms across preclinical and human studies that may be associated with different corresponding gamma-band signatures. However, our review of ASSR studies in ScZ patients and early-stage psychosis (see supplementary analysis 1) suggests a similar pattern of deficits between ASSR studies and those employing other auditory paradigms. Moreover, both preclinical and studies in ScZ patients and early-stage psychosis measured gamma-band activity from different cortical and sometimes subcortical areas.

In addition, the presence of a publication bias in studies investigating auditory gamma-band activity in ScZ patients indicates that the reduction in gamma-band activity may have to be interpreted with caution. Furthermore, there is a limited number of studies that examined the differences and similarities between the acute and chronic effects of NMDA-R antagonists on auditory gamma-band oscillations. Current studies point to potential differences between acute and chronic dosages.^81,114

Importantly, other neurotransmitters and receptor systems, such as dopamine and serotonin, are affected by NMDA-R antagonists, such as PCP and ketamine.^118,119 Hence, aberrant auditory gamma-band activity may not be strictly generated from NMDA-Rs hypofunction, which could explain the mixed patterns emerging from the animal and human studies examined in the present systematic review.

The number of studies of human EEG/MEG recordings following ketamine administration was unfortunately small (n = 3) so no robust conclusions could be drawn. However, there was evidence for both increases⁶² as well as decreases^63,85 of gamma-band activity across studies.

Summary and Perspective

Our systematic review has highlighted that the administration of NMDA-R antagonists may have similar effects on auditory gamma-band activity as observed in ScZ and early-stage psychosis. However, the underlying mechanisms through which these changes occur may be different. While NMDA-R antagonists in preclinical studies lead to disinhibition of neural circuits, the associated increase of baseline gamma-band activity is not consistently observed in ScZ and early-stage psychosis patients. This conclusion is in line with the outcome of a recent systematic review,²⁵ which compared changes in resting-state gamma-band activity following NMDA-R antagonists.

Thus, further research is required to identify the origin of circuit deficits in ScZ as well as the precise actions of NMDA-R antagonists on neural circuits and gamma-band activity. Recent work has, for example, indicated that ketamine causes spontaneously active neurons to become suppressed while previously silent neurons become spontaneously activated. This mechanism of action is mediated by the suppression of PV+ and somatostatin interneuron activity and inhibition of NMDA receptors and hyperpolarization-activated cyclic-nucleotide-gated channels.¹²⁰ Accordingly, it is conceivable that a more complex pattern of GABAergic interneurons contributes toward the dysregulation of gamma-band oscillations following NMDA-R hypofunctioning which may be distinct from circuit deficits observed in ScZ.

Supplementary Material

Supplementary material is available at https://academic.oup.com/schizophreniabulletin/.

Funding

P.J.U. was supported by project MR/L011689/1 from the Medical Research Council (MRC), the ERA-NET project 01EW2007A, and through the Einstein Stiftung Berlin (A-2020-613).