Converging lines of evidence, including postmortem (Benes, 2000; Benes and Berretta, 2001; Benes et al, 1996; Hashimoto et al, 2003; Lewis et al, 2005; Ohnuma et al, 1999; Volk et al, 2000; Volk and Lewis, 2002; Volk et al, 2002; Woo et al, 1998), genetic (reviewed by Charych et al (2009), and brain imaging studies (Ball et al, 1998; Busatto et al, 1997; Schroder et al, 1997; Verhoeff et al, 1999; Yoon et al, 2010), suggest that dysfunction of the gamma-amino butyric acid (GABA) system contributes to the pathophysiology of schizophrenia. In postmortem studies of schizophrenic patients, alterations in GABAergic transmission have been shown in many ways, including (1) reduced mRNA levels for the GABA-synthesizing enzyme glutamic acid decarboxylase-67 (Impagnatiello et al, 1998; Volk et al, 2000, 2) decreased density of axon cartridges of chandelier neurons (Woo et al, 1998, 3) decreased gene expression of the GABA membrane transporter-1 (Ohnuma et al, 1999; Volk and Lewis, 2002, 4) increased density of GABA-A receptors (Benes et al, 1996, 5) elevated density of α1- (Impagnatiello et al, 1998; Ohnuma et al, 1999) and α2-subunit-containing GABA-A receptors at pyramidal neuron axon segments (Volk et al, 2002, 6) decreased Reelin mRNA, which is preferentially expressed in GABAergic interneurons (Impagnatiello et al, 1998), and (7) decreased levels of ankyrin-G, a membrane protein that anchors the GABA receptor complex onto initial axonal segments of pyramidal cells in the area of chandelier cell synapses in superficial cortical area (Cruz et al, 2009). Many of these findings appear to be specific to schizophrenia (Volk and Lewis, 2002). Besides post-mortem data, some in vivo brain imaging (SPECT) studies suggest reduced benzodiazepine (BZ) receptor binding in schizophrenia (Ball et al, 1998; Busatto et al, 1997; Schroder et al, 1997).

GABAergic deficits described above may contribute to the pathophysiology of psychosis by several mechanisms. While it is out of the scope of this paper to discuss all the possible mechanism, a few putative mechanism are discussed below. One mechanism may involve the critical role that GABA interneurons have in synchronizing neural activity and, consequently, in information processing (reviewed in Uhlhaas and Singer, 2010). Sensory perception, emotion, attention, and memory, which are characteristically disturbed in schizophrenia, are based on distributed processes among multiple cortical and subcortical regions. It has been proposed that the neural assemblies representing these processes are functionally ‘bound’ together by synchronous high-frequency oscillatory activity to create a coherent cortical representation (Engel and Singer, 2001; Singer, 1999; Singer and Gray, 1995; Tallon-Baudry, 2003; Varela, 2001; Varela, 1995). The spike timing of pyramidal cells and the tuning of neuronal rhythms are largely governed by GABAergic interneurons. Loss of the inhibitory influence of GABAergic interneurons from cell loss or reduced function could lead to a loss of synchronization of pyramidal cell activity, resulting in the loss of associative functions, disruption of normal gating mechanisms, and eventually psychotic symptoms. Consistent with this hypothesis, several lines of evidence suggest that dysfunctional synchronized oscillatory neuronal activity may contribute to the pathophysiology of the perceptual and cognitive abnormalities in schizophrenia (Lewis et al, 2005; Uhlhaas and Singer, 2010).

Another potential mechanism by which GABA contribute to the pathophysiology of psychosis may involve the interactions between the GABA and dopamine (DA) systems. Converging lines of evidence suggest that the activity of DA neurons in the VTA is under tonic inhibitory control by GABA-A receptors (Fritschy and Mohler, 1995; Pirker et al, 2000; Waldvogel et al, 2008). Specifically relevant to this study, systemic administration of GABA-A receptor inverse agonist, FG7142 has been shown to activate VTA neurons (Murphy et al, 1996). Interestingly, FG7142-induced VTA activation was reversed by DA receptor antagonists. It should be noted that FG7142 and iomazenil are both inverse agonists. Therefore, a reduction in GABAergic transmission by GABA-A receptors, as would be the case with iomazenil, in the presence of pre-existing dysregulation of DA function, as is the case in schizophrenia, would be expected to further disinhibit DA systems, leading to a worsening of the DA-related symptoms in schizophrenia (Abi-Dargham et al, 1998; Laruelle et al, 1996). We acknowledge that this study does not directly test whether iomazenil increases neural synchrony deficits or DA dysregulation. Rather, the study measures a very distal outcome—psychotic symptoms that could result from an exacerbation of DA dysregulation or altered neural synchrony. The current report describes how pharmacological induction of GABAergic deficits, with iomazenil, increases psychosis.

Pharmacological induction of GABAergic deficits increases vulnerability to psychosis. Iomazenil (Ro 16–0154) is an iodine analog of the BZ receptor competitive antagonist flumazenil. Iomazenil has high affinity and selectivity for BZ receptors (Kd=0.5 nM) (Johnson et al, 1990). Some of its pharmacological properties are comparable with those of flumazenil (Beer et al, 1990). However, unlike the competitive antagonist flumazenil, which blocks the effects of BZ agonists but lacks intrinsic pharmacological effects (Hunkeler et al, 1981), inverse agonists have intrinsic pharmacological effects opposite to those of BZs (Tallman and Gallager, 1985). In preclinical studies, iomazenil has been shown to behave as a BZ receptor competitive antagonist with inverse agonist effects (Beer et al, 1990; Roche; Schubiger and Hasler, 1989). Similarly, clinical studies demonstrate that iomazenil has anxiogenic effects and at higher doses has proconvulsant effects (Randall, personal communication) that are consistent with inverse agonist activity at BZ receptors. Iomazenil produces a net deficit in GABA function.

Iomazenil has been shown to increase the psychotomimetic effects of the 5-HT2 partial agonist 1-(m-chlorophenyl)piperazine (m-CPP) in healthy subjects (D’Souza et al, 2006). Thus, although neither iomazenil nor m-CPP alone induces psychosis, the combination of iomazenil followed by m-CPP causes measurable psychotic symptoms in healthy subjects. Similarly, unpublished preliminary data from our laboratory suggest that GABA deficits induced by iomazenil pretreatment may also increase the psychotomimetic effects of low-dose amphetamine (0.1 mg/kg, intravenous (IV) infusion over 1 min), which by itself does not induce psychotic symptoms, and delta-9-tetrahydrocannabinol (Δ9-THC) in healthy subjects. Collectively, these studies suggest that the known vulnerability of schizophrenia patients to the psychotomimetic effects of amphetamine (Laruelle et al, 1999; Laruelle et al, 1996) and Δ9-THC (D’Souza et al, 2005) might result from pre-existing GABA deficits.

Although there is strong support for the existence of a GABA deficit in schizophrenia, the proportion of schizophrenia patients with this deficit is not known. The limited data available suggest that only 50% of schizophrenia patients have lower GABA levels than the lowest level found in healthy normal controls (Yoon et al, 2010). Furthermore, BZ augmentation reduces psychosis in only 30–50% of schizophrenia patients (Wolkowitz and Pickar, 1991).

We hypothesized that if GABA deficits contribute to an increased propensity toward psychosis in schizophrenia, then enhancement of these deficits should exacerbate psychotic symptoms in some schizophrenia patients, but not in healthy normal controls.



Thirteen chronic, stable schizophrenia patients were recruited and compared with data from 20 healthy normal controls from a previous study (D’Souza et al, 2006). Diagnosis was confirmed by SCID-III-R or SCID-IV. Inclusion and exclusion criteria for both studies were identical other than diagnosis. Included were men, 18–70 years old, who were able to provide informed consent; excluded were those with other Axis-I disorders; substance abuse (by history or urine toxicology); treatment with BZs in the previous week, clozapine, or low-potency antipsychotics (thorazine, thioridazine, mesoridazine); unstable medical conditions; active neurological illness requiring treatment; seizure history; abnormal baseline EKG; suboptimally controlled psychosis (defined as >16 on the four-key positive symptom subscale of the Brief Psychiatric Rating Scale (BPRS)); high risk for violence or suicide; and women, as the teratogenic potential of iomazenil is unknown. Potential subjects underwent a thorough medical and psychiatric history, complete physical examination, and a battery of laboratory tests including EEG, EKG, blood chemistry (CBC, BUN, creatinine, fasting blood glucose, electrolytes, liver and thyroid function tests), and urinalysis.

The study was conducted at the Neurobiological Studies Unit (VA Connecticut Healthcare System, West Haven, CT) with the approval of the Institutional Review Boards of VA Connecticut Healthcare System and Yale University School of Medicine. Subjects were recruited by word of mouth, clinician referral, and public advertisement. Subjects were compensated $100 per test day for participating. Confidentiality of study data was assured. Data on the effects of iomazenil in healthy subjects were taken from the iomazenil-only and placebo-only test days of a published study of the interactions of iomazenil and m-CPP (D’Souza et al, 2006). These studies were carried out in the same setting, by the same investigators and raters, during the same time period, using almost identical protocols.

Experimental Design

Subjects completed two test days, during which they received placebo or active (3.7 μg/kg) iomazenil over 10 min in random, counterbalanced order under double-blind conditions. Test days were separated by at least 72 h in order to minimize any carryover effects.

Schedule of testing

The detailed schedule of test procedures is described elsewhere (D’Souza et al, 2006). Subjects were required to refrain from using alcohol, street drugs, psychotropic medications, or caffeinated beverages for 2 weeks before testing and throughout the entire study. Urine toxicology was conducted on each test day to rule out recent drug use; a positive screen resulted in exclusion from the study.

After obtaining IV access and two sets of baseline assessments, subjects were administered active IV iomazenil or placebo (saline over a 10-min period). Behavioral ratings were conducted before (−55 to −30 min) and after (+10 to +30 min) the administration of iomazenil.


Psychotic symptoms were measured using the four-key positive symptom subscale of the BPRS, which has items for hallucinatory behavior, conceptual disorganization, unusual thought content, and suspiciousness (Overall and Gorham, 1962; Woerner et al, 1988). Perceptual alterations were measured using the Clinician Administered Dissociative Symptoms Scale (CADSS) (Bremner et al, 1998), a scale consisting of 19 self-report items and 8 clinician-rated items (0, not at all; 4, extremely), that evaluates aspects of altered environmental perception, time perception, body perception, feelings of unreality, and memory impairment. The scale has been shown to be sensitive to the effects of other psychoactive drugs, including ketamine and THC (D’Souza et al, 2004; Krystal et al, 1994). Anxiety was measured using the clinician-rated anxiety item of the BPRS, which is sensitive to iomazenil effects (D’Souza et al, 2006). The same research assistant rated both days of a subject and the same group of staff rated both schizophrenia patients and healthy controls. Interrater reliability was assessed every 1–2 months, and intraclass coefficients for the BPRS and CADSS were consistently >0.85.

Statistical Analysis

All statistical analyses were performed using SPSS Version 17. Demographic data were compared using independent t-tests. As the BPRS and CADSS data were not normally distributed in controls, and as there were obvious baseline differences between the two groups, peak change from baseline was used in the analysis of these behavioral outcome variables. These variables were analyzed using repeated analysis of variance (ANOVA) with iomazenil (active vs placebo) as within-subject factor and group (schizophrenia vs control) as between-subject factors.


Schizophrenia patients were significantly older, less educated, and less employed than control subjects (Table 1).

Table 1 Demographic Characteristics

Positive Symptoms

There was a significant main effect of drug (iomazenil vs placebo, F(1,31)=7.26, p=0.011) and drug-by-group interaction (F(1,31)=5.84, p=0.022), but no main effect of group (schizophrenia vs control, F(1,31)=0.20, p=0.66) on peak change in four-key BPRS scores. Although iomazenil did not have any significant effect on four-key BPRS scores in controls (0.1±0.31 for placebo and 0.15±0.67 for iomazenil, F(1,12)=0.09, p=0.77), it produced significant increases in schizophrenia subjects (−0.46±0.97 for placebo and 0.46±1.66 for iomazenil, F(1,12)=6.35, p=0.027, P2=0.35; Figure 1).

Figure 1
figure 1

Peak change in 4-key BPRS.

PowerPoint slide

Overall Symptoms

There were no significant effects of group (F(1,31)=3.35, p=0.08), drug (F(1,31)=0.87, p=0.36), or drug × group interaction (F(1,31)=0.29, p=0.60) in total BPRS score peak change.

Perceptual Alterations

There was a significant main effect of drug (iomazenil vs placebo, F(1,31)=10.08, p=0.003) and a drug-by-group interaction (F(1,31)=5.05, p=0.032) but no main effect of group (schizophrenia vs control, F(1,31)=0.004, p=0.95) on peak change in CADSS total scores. Although iomazenil did not have any significant effect on peak change in CADSS total scores in controls (0.35±1.04 for placebo and 0.60±1.76 for iomazenil, F(1,12)=0.80, p=0.38), it produced significant increases in schizophrenia subjects (−0.23±1.17 for placebo and 1.23±1.42 for iomazenil, F(1,12)=8.08, p=0.015, ηP2=0.40; Figure 2). There were no significant correlations between the peak change in CADSS and BPRS four-key positive symptom scores for each drug condition.

Figure 2
figure 2

Peak change in CADSS total score.

PowerPoint slide


Baseline BPRS anxiety scores were comparable between placebo day and iomazenil day in both schizophrenia (Table 1). There were no significant effects of group (F(1,31)=0.03, p=0.87), drug (F(1,31)=0.73, p=0.40), or drug × group interaction (F(1,31)=0.73, p=0.40) on the peak change of BPRS anxiety subscale.


Consistent with the GABA-deficit hypothesis of schizophrenia, iomazenil increased psychotic symptoms and perceptual alterations in schizophrenia patients, but not in controls. The effects of iomazenil were modest; effect size was small (ηP2=0.35). It should be noted that the schizophrenia patients were symptomatically stable and taking D2-receptor antagonists, which could have blunted their response to iomazenil. Furthermore, as GABA deficits likely contribute to a vulnerability to psychosis rather than being intrinsically pro-psychotic, the small effect of iomazenil was not unexpected, and is noteworthy especially as iomazenil did not have any effects in healthy normal subjects. The notion that GABA deficits increase the vulnerability to psychosis is illustrated by the observation that although m-CPP, a partial serotonin 2C, 1A, 1B, and 1D receptor agonist, did not produce psychosis in healthy subjects when administered alone, it did so when administered after iomazenil (D’Souza et al, 2006).

Iomazenil increased psychosis and perceptual alterations in about 50% of schizophrenia patients (Figures 1 and 2), supporting the hypothesis that GABA deficits may be present only in a subgroup. As discussed elsewhere, GABA deficits may increase vulnerability to pro-psychotic drugs. Thus, the hypothesis that only a subgroup of schizophrenia patients have GABA deficits might explain why only a subgroup of schizophrenia patients experience worsened psychosis in response to m-CPP (Iqbal et al, 1991; Krystal et al, 1993), amphetamine (Lieberman et al, 1987), ketamine (Lahti et al, 1995; Malhotra et al, 1997), or Δ9-THC (D’Souza et al, 2005).

The use of iomazenil to interrogate GABA function in schizophrenia is novel. This study has a number of strengths, including a double-blind, placebo-controlled design, and use of well-validated measures. It also has some limitations. First, the sample size was relatively small, thus confirmatory study with a larger sample size may be necessary. The groups were not matched for treatment with antipsychotic drugs. However, treatment with antipsychotic medications (DA D2 receptor antagonists) would be expected to blunt the response of schizophrenia patients to iomazenil, obscuring group differences rather than enhancing them. Further studies comparing schizophrenia patients on and off antipsychotic medications will need to be conducted. The groups were also not matched by age. However, there is no reason to presume that older subjects show increased response to iomazenil, and a significant response to iomazenil persisted within the schizophrenic group even when age was used as a covariate in the analysis. As only men were studied, because the teratogenic potential of iomazenil is unknown, the results may not generalize to women. Finally, on the basis of other evidence suggesting GABA deficits in schizophrenia (Lewis and Hashimoto, 2007), the study presumes that any differential responses to iomazenil reflect a trait rather than a state. However, this will need to be confirmed by studying the effects of iomazenil in different stages of schizophrenia.

In conclusion, iomazenil exacerbated psychosis in some schizophrenia patients. Identifying patients with GABA deficits is of therapeutic relevance, as a number of drugs can be used to enhance GABA function, such as MK-077, BZs, tiagabine, and vigabatrin. The findings of this small study justify future studies with iomazenil in a larger sample with more proximal measures of GABA function (such as electrophysiological indices of neural synchrony) that might be more sensitive in assessing GABA deficits in schizophrenia. Further studies should also compare schizophrenia patients on and off antipsychotic medications, and compare different stages of schizophrenia to verify that any observed GABA deficits are stable over time.