New Drug Treatments for Schizophrenia: A Review of Approaches to Target Circuit Dysfunction

Schizophrenia is a leading cause of global disease burden. Current drug treatments are associated with signi ﬁ cant side effects and have limited ef ﬁ cacy for many patients, highlighting the need to develop new approaches that target other aspects of the neurobiology of schizophrenia. Preclinical, in vivo imaging, postmortem, genetic, and pharmacological studies have highlighted the key role of cortical GABAergic (gamma-aminobutyric acidergic)-glutamatergic microcircuits and their projections to subcortical dopaminergic circuits in the pathoetiology of negative, cognitive, and psychotic symptoms. Antipsychotics primarily act downstream of the dopaminergic component of this circuit. However, multiple drugs are currently in development that could target other elements of this circuit to treat schizophrenia. These include drugs for GABAergic or glutamatergic targets, including glycine transporters, D-amino acid oxidase, sodium channels, or potassium channels. Other drugs in development are likely to primarily act on pathways that regulate the dopaminergic system, such as muscarinic or trace amine receptors or 5-HT 2A receptors, while PDE10A inhibitors are being developed to modulate the downstream consequences of dopaminergic dysfunction. Our review considers where new drugs may act on this circuit and their latest clinical trial evidence in terms of indication, ef ﬁ cacy, and side effects. Limitations of the circuit model, including whether there are neurobiologically distinct subgroups of patients, and future directions are also considered. Several drugs based on the mechanisms reviewed have promising clinical data, with the muscarinic agonist KarXT most advanced. If these drugs are approved for clinical use, they have the potential to revolutionize understanding of the pathophysiology and treatment of schizophrenia.

https://doi.org/10.1016/j.biopsych.2024.05.014Schizophrenia is characterized by psychotic symptoms, negative symptoms, and cognitive impairments (1).It is a leading cause of global disease burden (2,3).All currently licensed antipsychotic drugs are D 2 /D 3 receptor blockers (4).However, they have limitations in terms of efficacy and tolerability (4)(5)(6), highlighting the need for new approaches.In this review, we outline the circuits implicated in schizophrenia, focusing on the neurochemical systems that could be targeted by drugs, and then consider new treatment approaches that potentially target these abnormalities.

DOPAMINERGIC CIRCUITS INVOLVED IN SCHIZOPHRENIA
Dysfunction in projections from the midbrain to the striatum is central to many models of schizophrenia (7,8).Supporting this, molecular imaging studies provide evidence of higher striatal dopamine synthesis and release capacity and higher synaptic dopamine levels in patients (7) and people at risk of schizophrenia (9-12) compared with control participants and that these elevations are related to symptoms (Supplement) (13).Alterations in the mesostriatal dopaminergic circuit seen in schizophrenia are predominantly presynaptic in nature (7,8) and most prominent in the dorsal striatum, an area that receives projections from the frontal cortex as well as dopaminergic projections from the substantia nigra (Supplement) (7,8).
Mesostriatal dopamine neurons in the striatum play a key role in information processing (14)(15)(16), and this is thought to underlie the development of psychotic symptoms (Supplement) (17)(18)(19).Dopaminergic dysfunction in this circuit is thought to generate false associations and misperceptions, which, interacting with dysfunctional cognitive schema, result in psychotic symptoms (8,20,21).
In contrast to the overactivity in mesostriatal projections, it is theorized that there is hypofunction in dopaminergic projections from the midbrain to the frontal cortex (22,23).Supporting this, studies have found evidence for lower dopamine release in participants with schizophrenia relative to control participants in the frontal cortex in response to challenge with amphetamine (24) or a cognitive task (25).It is thought that hypodopaminergia in the frontal cortex could contribute to negative and cognitive symptoms, although this remains to be established (24).

GAMMA-AMINOBUTYRIC ACIDERGIC AND GLUTAMATERGIC CIRCUITS
Several lines of evidence indicate that there is dysfunction in cortical microcircuits involving GABAergic (gamma-aminobutyric acidergic) interneurons and glutamatergic pyramidal neurons in schizophrenia (26,27).For example, postmortem studies have found lower levels of GABAergic markers in the frontal cortex in subjects with schizophrenia relative to control subjects (Supplement) (28)(29)(30)(31)(32).A meta-analysis of proton magnetic resonance spectroscopy studies found that GABA levels were significantly lower in midcingulate cortex in patients with first-episode psychosis relative to control participants (33).These findings suggest that there may be deficient GABAergic neurotransmission in frontal cortical areas in schizophrenia, including in parvalbumin-positive GABAergic interneurons.
Studies provide evidence of lower brain levels of NMDA receptors (NMDARs) (34)(35)(36) and lower blood and cerebrospinal fluid levels of NMDAR co-agonists (e.g., D-serine and/or glycine) in patients compared with control participants (37)(38)(39)(40).Blocking NMDARs with ketamine induces short-lived schizophrenia-like symptoms in healthy people and worsens symptoms in people with schizophrenia (41,42).Ketamine is also a treatment for major depression, but emerging evidence indicates that its clinical effect in depression is likely to be downstream of NMDAR antagonism and different from its acute psychotogenic effects (43).
Numerous proton magnetic resonance spectroscopy studies have investigated glutamatergic markers in vivo in patients with schizophrenia.A recent meta-analysis found lower levels of glutamate in the medial frontal cortex (Supplement) (44).Taken together, these findings of lower cortical glutamate levels and NMDARs suggest that there is a loss of glutamatergic inputs in frontal cortical regions in people with schizophrenia relative to control participants.
Combined, these findings suggest that cortical abnormalities in both glutamatergic and GABAergic neurotransmission exist in people with schizophrenia (45).Given the opposing roles of these 2 neurotransmitters in neuronal firing, it has been hypothesized that such deficiencies lead to an excitation/inhibition imbalance in the disorder (26,46).This is also supported by electrophysiological studies in schizophrenia (Supplement) (46)(47)(48)(49)(50)(51).
These lines of evidence, primarily from the frontal cortex, provide evidence for a circuit in which the loss of glutamatergic inputs to GABAergic interneurons reduces the tonic inhibition of glutamatergic pyramidal cells (26,27).This disruption of normal excitation/inhibition balance in cortical microcircuits is thought to impair cortical function and result in the cognitive impairments and negative symptoms seen in the disorder (52).While we have concentrated on the frontal cortex as a site of these alterations, there is evidence for similar abnormalities in other regions, including the hippocampus, which may also give rise to disinhibited subcortical circuits and potentially contribute to cognitive and negative symptoms (27,53,54).

COMBINED MODEL OF CIRCUIT DYSFUNCTION IN SCHIZOPHRENIA
The midbrain and dorsal striatum receive glutamatergic projections from the frontal cortex (Figure 1) (55).Reduced cortical glutamatergic input onto GABA neurons could lower inhibitory inputs to the glutamatergic neurons projecting from the frontal cortex to the basal ganglia.Increased firing in these projections to the midbrain is thought to lead to overactivity in the mesostriatal dopaminergic circuit discussed in Dopaminergic Circuits Involved in Schizophrenia (Figure 1; Supplement) (22,(56)(57)(58)(59)(60)(61)(62).This circuit model provides a useful framework to consider where future treatments might act to treat the illness.

CURRENT DRUG TREATMENT
Dopamine D 2 /D 3 receptor blockade is central to the action of antipsychotic drugs (4,(63)(64)(65).This fits with the evidence for striatal dopaminergic overactivity discussed in Dopaminergic Circuits Involved in Schizophrenia, as it suggests that antipsychotics block the consequences of dysregulated dopamine release in mesostriatal circuits.However, there are limitations to this approach.First, substantial striatal D 2 /D 3 receptor blockade is required for clinical response (64).This likely blunts the capacity for adaptive striatal dopaminergic signaling and results in motor and other side effects (64,66).Second, antipsychotics block D 2 /D 3 receptors throughout the brain and outside of it, leading to side effects (67,68).Finally, schizophrenia resistant to antipsychotic treatment (69) is seen in around 30% of patients (70)(71)(72) and could develop due to the consequences of long-term D 2 /D 3 receptor blockade in some patients (73).These limitations have motivated the development of new treatment approaches.

DRUGS IN DEVELOPMENT TO TARGET CIRCUIT DYSFUNCTION IN SCHIZOPHRENIA
The following sections consider drug targets within the circuit, which is summarized in Figure 2.For each target we first consider how this may address the circuit abnormalities linked to schizophrenia and then summarize clinical development.In each case, further details are provided in the Supplement.Table 1 summarizes the affinities of current and novel drugs to relevant receptors (search method is described in the Supplement).

Muscarinic Agonists
There are 5 subtypes of muscarinic acetylcholine receptors (M 1 to M 5 ).Of these, M 1 and M 4 are highly expressed in the brain ( 74), with M 1 the most common subtype in the cortex and hippocampus and M 4 highly expressed in the midbrain and striatum in humans (75).In rodents, agonism of muscarinic acetylcholine receptors on the subpopulations of cortical GABAergic interneurons that express somatostatin or vasoactive intestinal polypeptide increases GABAergic inhibition of cortical pyramidal cells (76,77).M 1 agonism therefore has the potential to restore excitation/inhibition balance in schizophrenia and decrease activity in glutamatergic projections (78).
In support of effects on excitation/inhibition balance, activation of the M 1 receptor in the hippocampus modulates gamma oscillation frequency (79).These effects of M 1 agonism on cortical GABAergic/glutamatergic function suggest potential to treat cognitive impairments in schizophrenia (52).
In contrast to the M 1 receptor, M 4 is an autoreceptor that reduces cholinergic activity (80).Cholinergic projections stimulate dopamine neurons in the midbrain (81).Consistent with the autoreceptor function of M 4 , M 4 agonists reduce acetylcholine release from these neurons to reduce dopamine neuron activity and, consequently, striatal dopamine release (80,82).M 4 agonists may have additional effects via activation of M 4 receptors expressed on inhibitory interneurons in the striatum to reduce dopamine release there (83).Thus, M 4 agonism may have multiple effects on the dopamine system that are relevant for the treatment of psychosis (Figure 2).In contrast, anticholinergics increase dopamine release (84), presumably because they lack selectivity at specific muscarinic receptors.
Xanomeline is an M 1 /M 4 -preferring muscarinic receptor agonist (85).Compared with placebo, this drug reduced symptoms in patients with schizophrenia but was associated with high rates of gastrointestinal side effects, likely due to peripheral M 1 agonism (86).To improve tolerability, xanomeline was combined with an anticholinergic drug with low brain penetrance, trospium, forming KarXT (87,88).One phase 2 and two phase 3 double-blind trials of KarXT relative to placebo have been completed.All trials found improvements in total Positive and Negative Syndrome Scale (PANSS) scores in patients with acute schizophrenia (88)(89)(90).An alternative approach to KarXT could be M 4 selective drugs (91).NBI-1117568 is currently being tested in a phase 2 placebocontrolled study that is due to be completed in December 2024 (92).
A related approach is to use M 4 positive allosteric modulators (PAMs) to enhance receptor response to endogenous ligands (93).When emraclidine (CVL-231), a highly selective M 4 PAM ( 94), was evaluated in a phase 1b trial, there was improvement in total PANSS in patients with acute schizophrenia (95).Two phase 2 trials are now underway, with data expected in 2024 (91,96).
While preclinical evidence suggests that these M 4 agonists/ PAMs could act on circuits proximal to mesostriatal dopamine neurons to treat psychotic symptoms and on cortical circuits to address cognitive symptoms, this needs to be investigated in humans and linked to clinical response in patients.One potential issue is that postmortem and imaging studies have found evidence for lower levels of M 1 and M 4 receptors in the cortex of subjects with schizophrenia relative to control subjects (97-103), which could complicate dose selection of muscarinic agonists and PAMs.
Overall, the preclinical data indicate that muscarinic agonists could target components of the circuits thought to underlie schizophrenia, and results of clinical trials so far are promising.KarXT, a mixed M 1 /M 4 agonist, is most advanced and has a date for review by the U.S. Food and Drug Administration in fall 2024 (104).

PDE10A Inhibitors
PDE10A is an intracellular enzyme that regulates the levels of neuronal second messengers (105,106).It is highly expressed in medium spiny neurons in the striatum (107,108).Inhibition of PDE10A increases the levels of second messengers in medium spiny neurons in both direct and indirect pathways to activate them (106).This contrasts with D 2 blockers, which block D 2mediated inhibition of the indirect pathway, but do not directly modulate the D 1 pathway (8,109).PDE10A inhibition may result in predominant activation of the D 2 pathway in basal dopaminergic conditions but activate both pathways in conditions of elevated dopaminergic transmission (110).Either way, on the circuit that is thought to be dysfunctional in schizophrenia.GlyT1 and DAA oxidase inhibitors increase levels of NMDA receptor coagonists to potentially restore GABAergic interneuron function.Potassium channel modulators on subtypes of GABAergic interneurons potentiate their action.5 PDE10A inhibitors act downstream of the mesostriatal dopaminergic dysfunction implicated in schizophrenia (Figure 2).Four different PDE10A inhibitors (PF-02545920, TAK-063, Lu AF1167, MK-8189) have been studied to date (106,(111)(112)(113)(114). Compared with placebo, these drugs failed to show improvement in PANSS scores, although in some cases a high placebo rate was noted (Supplement).The most common side effects reported with these PDE10A inhibitors were somnolence, akathisia, acute dystonia, extrapyramidal disorder, and headache (106,(112)(113)(114).
Overall, there are no clearly positive phase 2 trials showing efficacy of PDE10A inhibitors in schizophrenia, though the results of 2 studies are awaited (115,116).One issue is whether the doses of the drugs tried to date are inhibiting PDE10A to the right degree.Dosing so far has been based on occupancy studies in healthy volunteers, but it is not clear if PDE10A availability is altered in patients with schizophrenia, with studies finding no difference in (117), lower (118), and higher (119) striatal PDE10A availability relative to healthy control participants.Resolving this would help inform dosing in schizophrenia and help ensure that future trials use the optimal dose.

TAAR1 Agonists
Endogenous amine neuromodulators that are present at trace levels, such as tyramine and b-phenylethylamine, act on traceamine associated receptors in the brain (120,121).One subtype of these receptors, TAAR1, is expressed in the ventral tegmental area, substantia nigra, and dorsal raphe nucleus (122,123).In preclinical models, TAAR1 agonists reduce dopamine neuron activity (124) and dopamine release (124,125).Together, this suggests that TAAR1 agonists have an inhibitory effect on dopamine neuron firing that could target the dopaminergic mesostriatal circuit implicated in schizophrenia (Figure 2).Several TAAR1 agonists have been developed for the treatment of schizophrenia (85).Ralmitaront, a partial TAAR1 agonist with no appreciable effect on 5-HT 1A receptors, and ulotaront, a full TAAR1 agonist that has affinity for 5-HT 1A receptors, are the most clinically advanced (126).
Ralmitaront (RO6889450) has been investigated in two phase 2 randomized placebo-controlled trials (127,128).One trial did not find a significant difference between placebo and active drug (127), and the other was terminated early based on an interim futility analysis (128).Ulotaront (SEP-363856) (129) has been tested in a translational mouse model ( 58) designed to replicate the elevated striatal dopamine synthesis capacity found in schizophrenia, where it reduced striatal dopamine synthesis capacity (58).This suggests that ulotaront may target this element of the pathophysiology (4).A subsequent study in people with schizophrenia showed that 2 weeks of treatment with ulotaront resulted in a reduction in striatal dopamine synthesis capacity and that greater reduction was correlated with larger decreases in psychotic symptoms (130).
In a randomized double-blind phase 2 trial in patients with acute schizophrenia, patients who received ulotaront showed greater improvement on PANSS total scores relative to patients who received placebo (131).An open-label extension study of ulotaront indicated that it was generally well tolerated (132).Thus, these early clinical studies suggest that ulotaront has promise.However, two subsequent phase 3 trials in people with acute psychosis found no significant difference in PANSS total scores compared with placebo (133,134).Both trials showed improvements in the groups receiving ulotaront but also large placebo responses (135), limiting the conclusions that can be drawn.Further clinical trials currently underway will help determine if ulotaront is effective [e.g., (136)].

Glutamate Modulators
NMDAR hypofunction leading to reduced activation of inhibitory interneurons is thought to be a key alteration leading to the circuit dysfunction (Figure 1).In addition to glutamate, the NMDAR requires co-agonists, such as glycine ( 137) and Dserine, for it to be activated (138).Thus, one approach to combat NMDAR hypofunction could be to increase the synaptic levels of glycine.To do this, companies have developed drugs that inhibit GlyT1, which is responsible for the removal of glycine from the synapse (139,140).
Bitopertin, a high-affinity, noncompetitive GlyT1 inhibitor, showed improvement in negative symptoms relative to placebo in a phase 2 double-blind randomized controlled study in schizophrenia (140).However, a subsequent phase 3 trial showed no significant separation from placebo (141).Another GlyT1 inhibitor, PF-03463275, was initially abandoned in phase 2 clinical trials (142) but has been re-examined in recent years using positron emission tomography to determine the doseoccupancy relationship (143), as it was unclear if the dose used in the phase 2 trial was adequate.However, a subsequent double-blind placebo-controlled randomized trial of the drug in addition to cognitive training found that it did not enhance the benefits of computerized training despite dose optimization (144).
Iclepertin (BI 425809) is a selective, potent GlyT1 inhibitor (145) in development for the treatment of cognitive impairment in schizophrenia.A large phase 2 randomized placebocontrolled study found a significant improvement in cognitive function with iclepertin relative to placebo across several doses (146), although the effect sizes were modest.The Food and Drug Administration granted iclepertin breakthrough therapy status (147), and further studies, including in combination with cognitive training (148) and a phase 3 study (149), are underway.
Another potential approach to augment NMDAR function is to increase synaptic levels of D-serine.DAAO (D-amino acid oxidase) is an enzyme that degrades D-serine (150).Inhibition of DAAO function has been shown to increase levels of D-serine in rodents (151) and enhance NMDAR function (152).
Luvadaxistat (TAK-831), a selective DAAO inhibitor, was tested in a phase 2 study for negative symptoms, but showed no difference relative to placebo (153).However, there was a signal for cognitive improvements at a dose of 50 mg/day (153,154).Based on this, a placebo-controlled phase 2 study testing its effects on cognitive function in schizophrenia has started (155).
Overall, there have been mixed results so far for drugs that aim to increase the levels of co-agonists at the NMDAR, although iclepertin and luvadaxistat show promise for improving cognitive impairment.If this promise is confirmed in Drugs to Target Circuit Abnormalities in Schizophrenia

Potassium Channel Modulators
As previously discussed in Gamma-Aminobutyric Acidergic and Glutamatergic Circuits, postmortem studies show lower levels of parvalbumin, a calcium binding protein expressed by some GABAergic interneurons, in cortical brain regions in subjects with schizophrenia relative to control subjects, and this is thought to contribute to impairments in high-frequency neuronal firing associated with the disorder (30,31).Potassium Kv3.1 and Kv3.2 channels are highly expressed on parvalbumin-positive interneurons (156)(157)(158), where they facilitate high-rate action potential propagation and neurotransmitter release (159).Modulation of Kv3.1 and Kv3.2 channels increases GABAergic interneuron firing frequency and improves gamma oscillation regularity (160,161).These actions should help restore excitation/inhibition balance (Figure 1).Therefore, this mechanism has been identified as a potential treatment target in schizophrenia (162,163), and novel selective Kv3.1/Kv3.2channel modulators (AUT1, AUT6, AUT9) were developed (164)(165)(166).AUT1 rescued the fastspiking phenotype of parvalbumin-positive interneurons in a pharmacological model (164), and AUT6 reversed subchronic phencyclidine-induced learning deficits in rats (165), supporting the potential of this approach for schizophrenia.These preclinical studies led to a small phase 1b study (167) in people with schizophrenia using a similar Kv3.1/Kv3.2modulating compound, AUT00206 (168).It was well tolerated (168)(169)(170) and was found to decrease frontal resting gamma power (169) and increase striatal neural activation during reward anticipation measured using functional magnetic resonance imaging (170).These biomarker findings suggest that the compound engages with the circuits relevant to schizophrenia (Figure 2).However, clinical efficacy is yet to be tested in trials.
5-HT 2A Antagonists/Inverse Agonists 5-HT 2A receptors are present on cell bodies of dopaminergic neurons in the midbrain, including the substantia nigra and ventral tegmentum (171).Blockade of 5-HT 2A receptors attenuates stimulated release of dopamine (172).5-HT 2A receptor inverse agonists bind to the receptor and induce a pharmacological action opposite to its activation by serotonin (173).These lines of evidence, therefore, indicate that 5-HT 2A receptor antagonists or inverse agonists should reduce dopamine neurotransmission (Figure 1).Theoretically, as this mechanism reduces but still permits dopamine release, it could treat psychotic symptoms without inducing the secondary negative symptoms often associated with D 2 /D 3 receptor blockade (174).In addition, it should be recognized that 5-HT 2A receptors are expressed on cortical glutamate neurons, so drugs that block them may also modulate cortical glutamatergic function (173).A number of licensed antipsychotics are also potent 5-HT 2A receptor antagonists (171).However, there are several drugs in development that specifically target the 5-HT 2A receptors without appreciable affinity for the D 2 /D 3 receptor.Two of these, pimavanserin and roluperidone, have reached phase 3 testing.
Pimavanserin is a 5-HT 2A receptor inverse agonist that has been approved by the Food and Drug Administration to treat Parkinson's disease psychosis (175).There have been 2 placebo-controlled randomized clinical trials of pimavanserin in schizophrenia (176,177).One trial found no significant difference in PANSS total change relative to placebo but found some improvement in secondary outcomes (including negative symptoms); however these analyses were not corrected for multiple comparisons (Supplement).The other trial found improvement in negative, but not total, PANSS scores compared with placebo, and it is now being tested further for negative symptoms in an open-label study (178,179).Reports so far indicate that pimavanserin is well tolerated, with a low incidence of extrapyramidal side effects in schizophrenia and no evidence of effects on weight or plasma glucose or lipid levels (177).However, increases in QTc interval have been reported (177).
Roluperidone is a 5-HT 2A receptor antagonist that has been developed for the treatment of negative symptoms (180).There have been 3 phase 2/3 trials of this compound in schizophrenia.One study in patients (N = 244) with negative symptoms found a significant improvement in PANSS negative symptom factor score with roluperidone relative to placebo as well as an overall improvement in PANSS and cognitive scores (181).Another trial found no significant improvement in PANSS total score with roluperidone relative to placebo (182).A larger study (N = 514) failed to show significant improvement in negative symptom factor score with roluperidone over placebo, although there was a large placebo effect (183).Roluperidone was generally well tolerated in these studies, with no evidence of extrapyramidal side effects, hyperprolactinemia, weight increase, or hyperlipidemia (181).
Trials of other selective serotonin antagonists/inverse agonists, such as ritanserin (a 5-HT 2A/2C antagonist), SR46349B (a 5-HT 2A/2C antagonist), fananserin (a 5-HT 2A/ D 4 antagonist), and mianserin (a 5-HT 2A/2B/2C and a 2 antagonist) have so far had mixed results, although these trials have generally not been performed specifically for negative symptoms (184).In summary, some trials of selective 5-HT 2A antagonists/inverse agonists have suggested potential efficacy for negative symptoms, but there have been a number of inconsistencies.The results of further clinical trials are required to determine if this approach will be effective in schizophrenia.

OUTSTANDING ISSUES AND FUTURE DIRECTIONS
While there is converging evidence to support the involvement of the GABA-glutamate-dopamine circuit in schizophrenia, there are areas of uncertainty with the circuit model (Figure 1).In particular, the link between the cortical GABAergicglutamatergic and dopaminergic circuits has not been tested fully in patients, and abnormalities in other areas such as the hippocampus and thalamus may also contribute to circuit dysfunction in schizophrenia (53,185,186).The ultimate test will be establishing that a drug or other manipulation of elements of the circuit effectively treats symptoms in patients with schizophrenia.The only element of the circuit where this is well established is for the dopamine system in the striatum, where a  (187).Future studies will also need to determine whether drugs acting to correct the excitation/inhibition balance act too far upstream of the dopamine dysfunction to really become effective antipsychotic agents.
Another key issue is the degree to which the same circuit abnormality underlies all cases of schizophrenia (44,188).Meta-analyses of variance have found greater variability in aspects of the dopaminergic (188) and glutamatergic ( 44) system in people with schizophrenia relative to control participants, indicating that there is heterogeneity in these systems in the disorder.One potential source of heterogeneity is the inclusion of people with treatment-resistant schizophrenia (TRS).Studies indicate that there are differences in the nature of the glutamatergic and dopaminergic changes seen in this subgroup of patients relative to the majority of patients (189)(190)(191)(192).For example, higher glutamatergic measures in the anterior cingulate cortex have been found in people with TRS relative to antipsychotic-responsive schizophrenia (189,191,192).Two drugs, evenamide and riluzole, thought to reduce glutamatergic activity are being tested for TRS (193)(194)(195).Consistent with this, evidence suggests that riluzole modulates glutamatergic measures in people with TRS (196).A phase 2 open-label study of evenamide found that nearly 50% of patients with TRS showed at least a 20% improvement from baseline total PANSS (194).However, this requires testing in double-blind, placebo-controlled studies.
One way to reduce heterogeneity in future trials could be to stratify patients to treatment options based on neurobiology.Embedding biomarker assessment within clinical studies will also allow evaluation of drug and disease mechanisms [see (197) and reviews on other measures (198)(199)(200) for further details].
However, it is worth considering that while selecting a subgroup of patients on the basis of biomarkers or clinical characteristics may be more specific, these markers may not be common, limiting the generalizability of findings (201).Clinical trials already have a tendency to select only partially representative samples, and additional selection could exacerbate this (202).Identification of specific characteristics/biomarkers during post hoc analysis of clinical trials that are representative of the whole schizophrenia population may be the best compromise (201).
Recent clinical trials have shown large placebo responses, which affect the assay sensitivity of the trials.To address this, larger samples have been recruited, requiring a greater number of sites.However, the use of more trial sites is associated with greater placebo response (201,203), potentially exacerbating the problem.Another way to tackle placebo response might be to use a sequential parallel comparison trial design (201,204,205).Other factors that may increase the drugplacebo difference include smaller samples with better selected patients with stricter inclusion criteria, choice of rating scales, inclusion of patients earlier in the course of illness, and biomarker outcome measures (6,201,206).
Another avenue of future research will be to determine if the drugs described can be used for related psychotic conditions (e.g., bipolar affective disorder), given that there is evidence for similar neurobiological alterations in these conditions (207)(208)(209).Finally, while all current clinical trials in schizophrenia focus on symptom reduction, moving to disease modification approaches, for example, to correct the circuit dysfunction outlined here (210), could result in a step change improvement in outcomes for patients.

CONCLUSIONS
The evidence implicates abnormalities in cortical GABAergicglutamatergic circuits that regulate subcortical dopaminergic circuits in the pathoetiology of schizophrenia.This circuit model allows the identification of multiple targets to treat schizophrenia.Drugs acting at these targets are in varying stages of clinical development, with muscarinic agonists closest to regulatory approval.If approved, these drugs would revolutionize the treatment of schizophrenia by offering non-D 2 /D 3 receptor blocking approaches.They would also advance understanding of the pathophysiology underlying the illness, particularly if complemented by mechanistic studies in patients.Establishing that the mechanisms we have discussed underlie the clinical efficacy of these new drugs would be an important test of the model and help in the development of related approaches.Conversely, if drugs we have discussed are clinically ineffective, this may require the circuit model to be revised, particularly if the drug has been shown to engage the relevant mechanism in patients.This highlights the value of including biomarker studies in drug development and the potential of these novel approaches to advance understanding of schizophrenia as well as its treatment.

Figure 1 .
Figure1.Illustration of a key circuit thought to underlie symptoms in schizophrenia.In the cortex, lower GABAergic activity due to lower excitatory inputs is thought to disinhibit glutamatergic neurons, leading to an E/I imbalance, impairing cortical function.In addition, the disinhibition of glutamatergic neurons projecting to the midbrain is thought to activate dopaminergic projections to the striatum, leading to increased dopaminergic activity there.E/I, excitation/inhibition; GABA, gamma-aminobutyric acid.(Created with BioRender.com.)

Figure 2 .
Figure 2. Illustration of putative drug targets

Table 1 .
-HT 2A receptor antagonists/ inverse agonists are thought to reduce dopaminergic and glutamatergic neurotransmission.Evenamide, a sodium channel modulator, and riluzole reduce glutamatergic activity.TAAR1 agonists and M 4 receptor agonists both have the potential to modulate mesostriatal dopaminergic activity.PDE10A inhibitors increase the levels of second messengers in medium spiny neurons to potentially counter the effects of excess dopaminergic activity at dopamine D 2 receptors in the striatum.DAA, D-amino acid; GABA, gamma-aminobutyric acid.(Createdwith BioRender.com.)Drugs to Target Circuit Abnormalities in Schizophrenia Median K i Values of New and Established Antipsychotics Showing Varying Affinities to Dopamine, Serotonin, a-Adrenergic, Histamine, Muscarinic, and Trace-Amine Associated Receptors 640Biological Psychiatry October 15, 2024; 96:638-650 www.sobp.org/journal Drugs to Target Circuit Abnormalities in Schizophrenia Biological Psychiatry October 15, 2024; 96:638-650 www.sobp.org/journal643 number of interventions that reduce dopamine neurotransmission, not just D 2 /D 3 blockers, have been shown to reduce psychotic symptoms