The dopamine hypothesis

Perhaps the most well-known model, the dopamine (or monoamine) hypothesis, is considered a key configuration for understanding schizophrenia. This hypothesis suggests that an aberration of dopamine transmission within some of the key primary dopaminergic pathways contributes to the emergence of psychotic features. In the mesolimbic pathway, there is evidence of dopaminergic overactivity – due to dysregulation of striatal dopamine synthesis, release or receptor sensitivity. Overactivity within this pathway leads to the emergence of the ‘classic’ positive features of psychosis, such as hallucinations and delusions. Underpinning this are atypical salience mechanisms, which appear to place undue emphasis on irrelevant stimuli, bringing new meaning and contributing to the alternative, subjective reality of those experiencing psychosis (Howes Reference Howes and Kapur2009). In contrast to the hyperactivity of the mesolimbic pathway, the mesocortical pathway is subject to an underactivity of dopamine. This contributes to the development of negative and cognitive features of schizophrenia.

The widespread functions facilitated by these two dysfunctional dopaminergic pathways is one of the possible reasons for the heterogeneous presentation of psychosis and schizophrenia seen in clinical practice (Howes Reference Howes and Kapur2009). It is noteworthy that evidence of dopaminergic dysfunction, as proven via neuroimaging using ¹⁸F-fluorodopa (F-DOPA) positron emission tomography (PET), is seen in both established illness and first-episode psychosis (Jauhar Reference Jauhar, Veronese and Nour2019). However, although there is much focus on dopaminergic activity, there is evidence that dopaminergic dysfunction may be more characteristic of those who are treatment responsive, compared with those who are resistant to conventional treatment approaches (Vanes Reference Vanes, Mouchlianitis and Collier2018a). There is accumulating evidence indicating that, in TRS, striatal dopaminergic activity appears unremarkable, whereas glutamate levels in the anterior cingulate cortex (ACC) are elevated (Demjaha Reference Demjaha, Egerton and Murray2014). Increased glutamate metabolites in the ACC have been linked to poor response to dopaminergic agents (Egerton Reference Egerton, Griffiths and Casetta2023), suggesting that glutamatergic dysfunction may underlie resistance to conventional dopaminergic therapies. This observation challenges the classic dopamine hypothesis in TRS, instead implicating broader neurotransmitter imbalances – particularly involving glutamate and gamma-aminobutyric acid (GABA) – in its pathophysiology.

Such findings regarding neurotransmitter imbalances align with the developmental risk factor model, which posits that cumulative early life insults, including prenatal stress, aberrant synaptic pruning, and genetic or environmental perturbations, disrupt neurotransmitter homeostasis beyond dopaminergic systems, thereby contributing to the emergence of treatment-resistant symptomatology (Murray Reference Murray, Bhavsar and Tripoli2017). These early neurodevelopmental disturbances, which impair the maturation of the prefrontal cortex (PFC) and hippocampi, contribute to deficits in cognitive and emotion regulation. Supporting this, neuroimaging studies consistently report structural abnormalities, such as enlargement of the lateral ventricles and pallidum when compared to healthy controls (van Erp Reference van Erp, Walton and Hibar2018). These components of the corticolimbic circuit can be examined at structural, functional and molecular levels to yield insight into schizophrenia’s neurobiology and inform novel experimental interventions (Gee Reference Gee, Dazzan and Grace2025). Overall, converging evidence supports conceptualising TRS as a mechanistically distinct subtype, warranting the exploration of glutamatergic and other non-dopaminergic treatment strategies (Table 1).

TABLE 1

Dopaminergic pathways – origin, terminus and functions

The glutamatergic–GABAergic hypothesis

A less widely known, but increasingly studied, framework for understanding schizophrenia is the glutamate hypothesis, which posits that dysfunction in the glutamatergic system plays a central role in the pathophysiology of psychosis, particularly in the context of TRS. This hypothesis highlights an imbalance between excitatory glutamatergic and inhibitory GABAergic neurotransmission as a potential mechanism underlying the disorder. At the core of this model is the proposed hypofunction of N-methyl-d-aspartate (NMDA) receptors, which are essential for synaptic plasticity, learning and neurodevelopment. It is theorised that impaired NMDA receptor activity, particularly due to deficient glutamatergic signalling, results in an excitation/inhibition (E/I) imbalance, leading to widespread downstream neural disruptions. A key consequence of this disruption involves the dysregulation of GABAergic parvalbumin-expressing interneurons, which are critical for maintaining cortical E/I balance. Their dysfunction has been linked to abnormal activity in the hippocampus and PFC, regions strongly implicated in the manifestation of the positive, negative and cognitive symptoms of psychosis, including TRS (Gee Reference Gee, Dazzan and Grace2025).

Empirical support for the glutamate hypothesis includes observational studies in which administration of NMDA receptor antagonists (such as phencyclidine or ketamine) induce psychosis-like symptoms in healthy individuals. Rather than questioning the dopamine hypothesis, the glutamate model may complement it, suggesting an integrated neurochemical framework in which glutamatergic dysfunction contributes to dopaminergic dysregulation, which may represent a final common pathway for psychosis (Howes Reference Howes and Kapur2009). Despite the growing body of evidence, pharmacological interventions targeting the glutamatergic system have thus far yielded inconsistent clinical efficacy (Begni Reference Begni, Marchesin and Riva2025). Nonetheless, glutamate-modulating agents, such as D-serine, may be a promising area of therapeutic development. These compounds aim to restore NMDA receptor functionality and potentially alleviate the negative and cognitive symptoms of schizophrenia.

Neuroimaging research has contributed valuable insights to this hypothesis. Early PET studies reported reduced NMDA receptor binding in individuals with schizophrenia, although subsequent work using alternative glutamatergic tracers has produced mixed results. More recently, proton magnetic resonance spectroscopy (¹H-MRS), a neuroimaging technique that quantifies in vivo brain metabolites using 3T magnetic resonance imaging (MRI), found no significant difference in occipital cortex glutamate levels between individuals with ultra-TRS and healthy controls (Lopes Reference Lopes, Rae and Meyer2025), although further ¹H-MRS studies have identified reduced PFC GABA and glutamate levels in those with schizophrenia (Gee Reference Gee, Dazzan and Grace2025). However, Egerton et al (Reference Egerton, Griffiths and Casetta2023) have demonstrated that TRS, identifiable via poor response to conventional antipsychotic drugs, is associated with elevated glutamate metabolites in the ACC.

It may be argued that targeted glutamatergic treatments are beneficial for those with TRS by engaging non-dopaminergic pathways, especially given that TRS is defined by a lack of response to conventional dopaminergic treatments. However, optimism has been tempered by recent findings from Boehringer Ingelheim’s Phase III CONNEX trial, which investigated iclepertin, a selective α₇ nicotinic acetylcholine receptor agonist, for cognitive enhancement in schizophrenia. The trial was terminated prematurely after primary and secondary trial end-points were not met (Boehringer Ingelheim 2025).

Cholinergic hypothesis

The cholinergic hypothesis centres on dysregulation of the cholinergic system being integral to the development and maintenance of psychosis. This is reportedly driven by dysregulation of both muscarinic acetylcholine receptors (mAChRs) and nicotinic acetylcholine receptors. The hypothesis, which is not new, has come to the fore over recent decades owing to the evidence that mAChR agonists such as xanomeline appear to reduce the clinical features of psychosis. This is considered promising, particularly given the impact of xanomeline on negative and cognitive features, which are often difficult to treat with standard dopaminergic agents (Begni Reference Begni, Marchesin and Riva2025). This has led to the development of a xanomeline–trospium chloride combination, licensed under the trade name Conbenfy by the Food and Drug Administration (FDA) in the USA. Xanomeline–trospium chloride has a novel mechanism as a combined agent. Xanomeline offers central M₁ and M₄ muscarinic receptor agonism, whereas trospium chloride provides muscarinic antagonism, primarily in peripheral tissues. Trospium chloride therefore prevents the peripheral adverse effects of xanomeline. This is possible owing to its hydrophilicity, which prevents penetration across the blood–brain barrier while permitting activity in peripheral tissues (Smith Reference Smith, Augustine and Dorrough2024).

This drug represents enhanced clinical options for the 20–30% of people with schizophrenia who are resistant to the mainstay options of dopamine-based medications (Nair Reference Nair, Shergill and Whiskey2025). Its approval by the FDA is a significant milestone as it introduces a non-dopaminergic pharmacological treatment for psychosis and schizophrenia. In addition, given its non-dopaminergic mechanism, xanomeline–trospium chloride is not associated with the archetypical adverse drug reactions often associated with treating psychosis and schizophrenia, such as extrapyramidal side-effects, metabolic syndrome and hormonal abnormalities (Smith Reference Smith, Augustine and Dorrough2024). The promise of xanomeline–trospium is leading the way for other non-dopaminergic medications, which are currently in development (Begni Reference Begni, Marchesin and Riva2025; McCutcheon Reference McCutcheon, Pillinger and Varvari2025). This may be of particular benefit for those with TRS, who, as a criterion for the diagnosis, have not responded adequately to dopaminergic interventions previously.

Overview of neuroimaging findings

Advanced imaging techniques, including MRI and PET, provide valuable insights into the brain changes that underlie schizophrenia. Structural MRI studies have shown widespread grey matter reductions in schizophrenia, particularly in the prefrontal cortex, temporal lobes and hippocampi – areas implicated in memory, emotion regulation and cognitive processes such as language. Although grey matter is affected, white matter appears more resilient to these changes, with some evidence of plasticity that may mitigate cognitive decline to a degree (Emsley Reference Emsley, du Plessis and Phahladira2023). Longitudinal meta-analyses indicate that grey matter atrophy is progressive, but particularly noticeable in the early stages of illness, emphasising the importance of reducing the duration of untreated psychosis (Yang Reference Yang, Jing and Naici2021). The left hemispheric structures are primarily affected by this atrophy. It has been argued that many of the abnormalities identified in chronic illness are also present in those experiencing a first episode of psychosis; this could indicate that this neurobiological basis is a primary pathological feature rather than a consequence secondary to chronic illness or the impact of treatment. However, exposure to antipsychotic drugs is an important consideration, with higher cumulative exposure linked to greater grey matter atrophy (Yang Reference Yang, Jing and Naici2021). Clozapine therapy, employed in the management of TRS, is demonstrated to be implicated in decreases of grey matter volume in a multitude of areas, including frontotemporal regions, parietal and occipital cortices, the striatum and cerebellum (Sun Reference Sun, Zelaya and Sendt2024).

Furthermore, there is evidence that the detection of lesions in the hippocampal areas may precipitate psychotic features that would not suitably respond to dopaminergic antipsychotic medications (Pines Reference Pines, Frandsen and Drew2025). Identification of these lesions (which affect functional circuitry on the posterior hippocampal–subicular axis) via neuroimaging can offer potential for non-dopaminergic therapeutics by way of neuromodulatory interventions such as transcranial magnetic stimulation (Pines Reference Pines, Frandsen and Drew2025).

Functional neuroimaging has advanced our understanding of the neural mechanisms underlying schizophrenia. There is considerable evidence of hypofrontality – reduced perfusion of the parenchyma – in the PFC, as well as altered metabolism in the frontal cortex, with deficits of cognitive control and executive functioning (Townsend Reference Townsend, Pillinger and Selvaggi2023). As mentioned above, F-DOPA is a radiotracer utilised in PET imaging; F-DOPA studies have revealed that striatal dopamine is higher in those with psychosis, even during the first episode of illness, compared with controls; moreover, imaging has also shown that elevated dopamine synthesis is associated with responsiveness to standard antipsychotic drugs (Jauhar Reference Jauhar, Veronese and Nour2019). This may be a potential biomarker in future practice; however, the aberration of dopamine activity is heterogeneous, outlining the need for a bespoke, personalised approach (Giacomel Reference Giacomel, Martins and Nordio2025).

Neuroimaging in TRS

Neuroimaging has emerged as a valuable tool in the investigation and diagnosis of TRS, offering the ability to identify both structural and functional abnormalities through advanced neuroradiological techniques (Vanes Reference Vanes, Mouchlianitis and Collier2018a, Reference Vanes, Mouchlianitis and Wood2018b; Mouchlianitis Reference Mouchlianitis, Vanes and Tracy2023). Modalities such as functional magnetic resonance imaging (fMRI) and PET have provided critical insights into dopaminergic and glutamatergic dysfunctions, thereby informing targeted therapeutic strategies and enhancing our understanding of the neurobiological mechanisms underlying resistance to conventional antipsychotic treatments.

PET and single-photon emission computed tomography (SPECT) studies have demonstrated reduced cerebral blood flow (CBF) in individuals with TRS relative to healthy controls. This reduction in CBF extends beyond the traditional concept of hypofrontality, encompassing not only the PFC but also parietal and subcortical regions (Sun Reference Sun, Zelaya and Sendt2024). Furthermore, clozapine therapy appears to exacerbate this phenomenon, with additional reductions in CBF observed specifically in the ACC (Sun Reference Sun, Zelaya and Sendt2024). Intriguingly, elevated baseline CBF in subcortical structures – particularly the striatum and hippocampi – has been associated with a more favourable response to clozapine treatment (Sun Reference Sun, Zelaya and Sendt2024). Resting-state fMRI studies have also revealed disrupted connectivity patterns in TRS, including reduced functional connectivity between the ventral striatum and the substantia nigra (White Reference White, Wigton and Joyce2016), and cognitive activation studies have shown differential cognitive control in TRS (Horne Reference Horne, Vanes and Verneuil2021). Structural neuroimaging findings further support this altered neurobiology (Vanes Reference Vanes, Mouchlianitis and Wood2018b), with evidence of reduced cortical mantle thickness in several regions, notably the left hemispheric temporal cortex and bilateral cingulate cortices, when comparing individuals with TRS with those who have adequately responded to standard antipsychotic therapies (Barry Reference Barry, Vanes and Andrews2019).

Collectively, these findings underscore the potential of neuroimaging to inform a more individualised, precision medicine approach to the diagnosis and management of TRS, with potential core pathognomic features observed in TRS but not in typical treatment-responsive schizophrenia (Sun Reference Sun, Zelaya and Sendt2024; Giacomel Reference Giacomel, Martins and Nordio2025).