CB1 receptor-enriched functional connectivity analysis revealed target-specific modulation by cannabidiol in early psychosis

Participants

Fifteen patients with early psychosis (within 5 years of onset) were recruited from early-intervention services in the South London and Maudsley NHS Foundation Trust (London, UK). Additionally, 19 HC were recruited by local and internet-based advertisements as a comparative group. Sociodemographic and clinical measures of participants have been described previously (Gunasekera et al., 2023; O’Neill et al., 2021a). In this cohort, 13 patients with early psychosis and 14 age- and sex-matched HC completed a resting-state functional magnetic resonance imaging (rs-fMRI) acquisition and had data usable for the analysis presented in this study (see Table 1 for details of this sub-sample). The remaining participants of the original cohort were excluded due to incomplete rs-fMRI acquisition or insufficient data quality. An experienced research psychiatrist confirmed the diagnosis of psychosis using the Structured Clinical Interview for DSM-IV (Bell, 1994). Inclusion criteria required a psychotic mental illness diagnosis (meeting criteria for schizophrenia, schizophreniform, or brief psychotic disorder, but no other Axis I diagnoses) within 5 years of illness onset. Exclusion criteria included: a history of neurological disorders (other than a psychotic mental illness meeting the criteria for schizophrenia, schizophreniform, or brief psychotic disorder), current DSM-IV diagnosis of substance use disorder (except cannabis in the patient group), acute intoxication with alcohol or any other psychoactive substance on the day of experimentation, IQ of less than 70, lack of capacity to consent, urine drug screen positive for other known psychotogenic and psychedelic substances, severe intercurrent illness, pregnancy, and any contraindication to MRI. A positive urine drug screen for THC was not considered an exclusion criterion in patients with early psychosis. All patients were using antipsychotic medication (Risperidone: N = 4, Aripiprazole: N = 2, Amisulpride: N = 2, Paliperidone: N = 2, Lurasidone: N = 1, and Olanzapine: N = 1), except for one who had discontinued use of their prescribed olanzapine medication. Additional exclusion criteria for HC included diagnosis of a mental illness, current/past recipients of psychiatric treatment, a first-degree relative who had experienced psychosis, or more than 10 instances of cannabis use throughout their lifetime.

OPEN IN VIEWER

Table 1.

Sociodemographic and clinical measures of patients with early psychosis (PSY) and healthy controls (HC) that underwent resting-state fMRI acquisition and that were included in the present study. Unless otherwise reported, values are expressed as mean value ± standard deviation.

Sociodemographic and clinical measures	PSY	HC	HC vs PSY
Sample size (n)	13	14
Age (years)	27.7 ± 4.8	24.4 ± 4.7	t(25) = 1.8, p = 0.08
Sex (% males)	61.5	64.3	χ2(1) = 0.02, p = 0.88
Education (years)	14.3 ± 2.3	17.3 (2.5)	t(25) = −3.4, p = 0.003
Antipsychotic medication
Treated patients, n (%)	12 (92.3)	/
CPZ equivalent dose (mg/day)	212.6 ± 89.1
Urine drug screen results
Clean (n)	Session 1: 5	14
	Session 2: 5
THC (n)	Session 1: 7	/
	Session 2: 7
PCP (n)	Session 1: 1	/
	Session 2: 1
Morphine (n)	Session 1: 0	/
	Session 2: 0
Benzodiazepines (n)	Session 1: 0	/
	Session 2: 0

CPZ: Chlorpromazine; THC: delta-9-tetrahydrocannabinol; fMRI: functional magnetic resonance imaging; PCP: phencyclidine.

All participants provided written informed consent prior to commencing the study, approved by the National Research Ethics Service Committee of London (Camberwell, St. Giles, UK, ethics reference: 14/LO/1861).

Study design

All study visits were scheduled to start in the morning for both patients and HC. Following a standardised sequence of study procedures, MRI acquisition took place at a comparable time of day for all participants, approximately in the early afternoon.

For patients with early psychosis, a within-subject, randomised, double-blind, placebo-controlled crossover design was employed over two sessions with at least a 1-week interval to allow for the washout of CBD. On each study day, 120 minutes after a light standardised breakfast, patients received either a 600 mg CBD (approx. 99.9% pure; THC-Pharm., Frankfurt, Germany) or a visually identical gelatine placebo (PLB) capsule. Randomisation and blinding of CBD or PLB was conducted at the Maudsley Pharmacy (London, United Kingdom). fMRI acquisition took place approximately 180 minutes after drug administration.

Blood samples were obtained at three time points: (T1) 60 minutes before drug administration, (T2) 60 minutes post-drug administration, (T3) 270 minutes post-drug administration. The Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987) and the ‘state’ subscale of the State-Trait Anxiety Inventory (STAI; Spielberger, 2010) were used to assess psychopathology at timepoints T1 and T3. For the PANSS, item-level scores were collected for the Positive, Negative, and General Psychopathology subscales, and total scores were computed. For the STAI, all 20 items of the state subscale were administered, and total scores were then calculated. Changes in symptom severity over time (T3–T1) were subsequently derived for PANSS positive, PANSS negative, PANSS total, and STAI total scores (see Supplemental Table 1).

All participants were advised to avoid alcohol intake for 24 hours and caffeine intake for 12 hours before the study. Furthermore, they were asked to avoid using any recreational drugs (apart from cannabis amongst the patient group) for 2 weeks before the study day. Urine samples were obtained on each study day to screen for use of amphetamines, barbiturates, benzodiazepines, cocaine, methamphetamine, morphine, methadone, phencyclidine (PCP), tricyclic antidepressants and THC using the Alere™ Drug Screen Urine Test Cup. Carbon monoxide breath levels were also measured using the Bedfont™ Smokerlyzer. The HC group completed only a single scanning session without receiving CBD or PLB and was screened according to the same criteria, serving as a comparison group.

Image acquisition and preprocessing

MRI data were acquired on a 3T GE SIGNA HDx scanner at the Centre for Neuroimaging Sciences, King’s College London. Resting-state functional images were obtained using a gradient-echo echo-planar imaging sequence (Repetition time (TR) = 2 seconds, echo time (TE) = 30 ms, flip angle = 77°, voxel size = 3.44 × 3.44 × 5 mm³, 154 volumes). Structural images were acquired using a whole-brain sagittal T1-weighted scan based on Alzheimer’s Disease Neuroimaging Initiative parameters (TE = 2.85 ms, TR = 6.98 ms, flip angle = 11°, voxel size = 1.2 × 1.0 × 1.0 mm³).

rs-fMRI data were pre-processed with the CONN toolbox based on the SPM software implemented in MATLAB (Whitfield-Gabrieli and Nieto-Castanon, 2012) with a multistep pipeline including (i) realignment and unwarping for motion and susceptibility distortion correction, (ii) slice timing for the correction of temporal misalignment between different slices, (iii) outlier detection based on the artefact detection tool to identify potential outlier scans from the global BOLD signal and the amount of subject-motion in the scanner, (iv) normalisation into standard MNI152 space, (v) spatial smoothing by using spatial convolution with a Gaussian kernel of 6 mm full width half maximum. After preprocessing, denoising was applied to regress out motion confounds, modelled by including the six rigid-body motion parameters (three translations and three rotations) and their first-order derivatives, noise components from white matter and CSF and outlier scans as nuisance regressors. After regression, a temporal band-pass filtering was applied within the standard frequency band of 0.01–0.08 Hz, which is thought to reflect mainly neuronal fluctuations and to be less affected by physiological variables.

Receptor-enriched analysis of functional connectivity (REACT)

REACT is an analysis framework that enriches fMRI data with information on the spatial distribution of selected molecular systems and receptors derived from PET atlases (Dipasquale et al., 2019). As illustrated in Figure 1, the REACT analysis is implemented as a two-step multivariate regression analysis: in the first step, the fMRI images are restricted to grey matter voxels for which molecular information is available, using a binarised mask derived from each atlas. The PET templates are entered as spatial regressors into a first general linear model (GLM) to weight the BOLD signal across voxels and estimate, for each participant, the dominant BOLD fluctuation associated with each molecular system. In the second step, the subject-specific BOLD fluctuations obtained from the previous step are used as temporal regressors in a second GLM to estimate the corresponding molecular-enriched FC associated with each molecular system.

OPEN IN VIEWER

Figure 1.

Overview of the REACT analysis. The first GLM is used to characterise subject-specific temporal dynamics for the CB1 receptors PET atlas. The second GLM is used to estimate each subject-specific functional connectivity map that quantifies the BOLD response associated to the CB1 receptors PET atlas.

In the present study, we followed the REACT pipeline (GitHub – ottaviadipasquale/react-fmri: REACT) to rs-fMRI using molecular density distribution maps of CB1 receptors obtained from the JuSpace Toolbox (Dukart et al., 2021) to characterise CB1 receptor-enriched FC.

Statistical analysis

Group differences in sociodemographic variables (Table 1) were assessed using independent samples t-tests for continuous variables (age, education) and chi-square tests for categorical variables (sex). To assess acute changes in clinical measures, repeated-measures ANOVA was conducted for each outcome (PANSS positive, PANSS negative, PANSS total, and STAI), with Condition (CBD vs PLB) and Time (T1 vs T3) as factors. Post hoc tests were corrected for multiple comparisons using Bonferroni correction (significance threshold: p < 0.05/4, based on four planned pairwise comparisons per outcome: PLB T1 vs T3, CBD T1 vs T3, PLB vs CBD at T1, and PLB vs CBD at T3). Statistical analyses of the CB1 receptor-enriched FC maps were performed using cluster-based inference within Randomise (FSL. v 6.0; Winkler et al., 2014) with 5000 permutations per test and contrast. A cluster was considered significant if p_FWE <0.05, corrected for multiple comparisons using the threshold-free cluster enhancement option (Smith and Nichols, 2009).

Group differences between HC and patients under each condition were assessed by performing two separate two-sample t-tests (PSY-PLB vs HC and PSY-CBD vs HC).

To investigate the acute effect of CBD within the patient group, we performed a paired samples t-test comparing CBD and PLB sessions (PSY-CBD vs PSY-PLB).

Spatial correlation

To complement our REACT analysis, we computed a spatial correlation between CB1 receptor-enriched FC maps of patients under PLB and CBD and those of HC. Correlations were calculated using BrainSMASH (Burt et al., 2020), a Python-based framework that generates spatially constrained surrogate maps that preserve the spatial autocorrelation of the original FC maps. The observed correlations were then compared to the distribution obtained from these surrogates to obtain nonparametric p-values, providing a robust measure of spatial similarity between maps while accounting for their intrinsic spatial structure. This analysis returned a Spearman correlation coefficient (r) value and a p-value for PSY-PLB versus HC and PSY-CBD versus HC.

CB1 receptor-enriched FC disruptions in early psychosis

A growing body of evidence indicates that dysregulation of the endocannabinoid system may play a role in the pathophysiology of psychosis (Appiah-Kusi et al., 2020; Borgan et al., 2019; De Marchi et al., 2003; Koethe et al., 2009; Leweke et al., 2007; Newell et al., 2006; Ranganathan et al., 2016). Indeed, findings from molecular and clinical studies suggest that individuals with psychotic disorders may exhibit an altered endocannabinoid signalling tone, characterised by increased concentrations of the endogenous cannabinoid anandamide in both cerebrospinal fluid and peripheral blood, as well as elevated expression of CB1 receptors on peripheral immune cells (Minichino et al., 2019). Importantly, these alterations appear to be present across different stages of illness, ranging from the prodromal phase to established, chronic psychosis. Consistent with these observations, abnormalities in CB1 receptor expression and availability have also been documented in post-mortem brain tissue and in vivo imaging studies of patients with psychosis (Borgan et al., 2019; Ranganathan et al., 2016; Volk et al., 2016).

Our results revealed increased CB1 receptor-enriched FC in early psychosis patients under PLB compared with HC across a distributed set of regions including the anterior cingulate cortex, insula, hippocampus, amygdala, thalamus, middle temporal gyrus and frontal regions, which overlap substantially with brain networks previously reported to exhibit disrupted connectivity in psychosis. First-episode schizophrenia has been associated with heightened connectivity within and between key neurocognitive systems, including default mode, frontoparietal and limbic networks (Mei et al., 2025). Moreover, systematic reviews and meta-analyses indicate a consistent pattern of subcortical–cortical hyperconnectivity, particularly involving basal ganglia and thalamic regions and their connections with sensorimotor, temporoparietal, insular cortices, and cerebellar–cortical hyperconnectivity (Jensen et al., 2025).

CBD reduces CB1-enriched FC in the right insular cortex

Within the patient group, CBD specifically reduced CB1 receptor-enriched FC in the right insula relative to PLB. The insular cortex represents a network hub that integrates and coordinates information across multiple cognitive and affective domains (Uddin et al., 2017). Through its extensive connections with limbic, cognitive, and sensorimotor regions, the insula plays a central role in detecting behaviourally relevant stimuli, integrating internal bodily signals with external environmental information (Craig, 2009; Menon and Uddin, 2010), and facilitating dynamic switching between the default mode and central executive networks (Sridharan et al., 2008). A broad range of insular abnormalities has been consistently reported in psychotic disorders (Wylie and Tregellas, 2010), including altered activation patterns (Moran et al., 2013; Palaniyappan et al., 2013; Smieskova et al., 2015; Wilson et al., 2019) and disrupted FC (Li et al., 2019; O’Neill et al., 2019; Sheffield et al., 2020; Tian et al., 2019). Results from the present study indicate that CBD selectively attenuates CB1 receptor-enriched FC within this key hub, providing a strong support for a receptor-specific mechanism through which CBD may reduce excessive integration of aberrant interoceptive or emotional signals in psychosis. Importantly, this effect of CBD was observed in a region where CB1 receptor-enriched FC was increased in the same patients under PLB compared with HC. Although this interpretation is based on the overlap of clusters location rather than a formal three-way comparison, it is consistent with the idea that CBD engages a CB1-enriched connectivity abnormality present in psychosis and attenuates it. This is further supported by the within-subject design employed in the present study, which allows for a direct assessment of drug-related FC modulation.

CBD modulates CB1 receptor-enriched FC abnormalities

No significant differences in CB1 receptor-enriched FC were observed by comparing early psychosis under CBD and HC. Although this contrast alone does not constitute direct evidence of normalisation, when considered alongside the within-subject reduction discussed in the previous section, it suggests that acute CBD administration modulates FC within brain systems spatially enriched by CB1 receptor distribution.

Previous pharmacological neuroimaging studies have consistently shown that CBD attenuates aberrant neural responses within regions implicated in psychosis, including the medial temporal lobe, striatum, midbrain and insula across both task-based and rs-fMRI paradigms (Bhattacharyya et al., 2018; Davies et al., 2020; Gunasekera et al., 2023; O’Neill et al., 2019, 2021a; Wilson et al., 2019). Notably, these studies employed the same acute 600 mg oral CBD dose used in the present work, supporting the rationale for its selection as a dose capable of producing measurable neural effects in psychosis-related circuits. While these studies have demonstrated that CBD can partially normalise brain activation during salience, emotional and motivational processing, as well as reduce aberrant FC at rest, they did not explicitly link these effects to the molecular targets through which CBD is thought to act.

The present work extends previous findings by showing that FC changes after acute administration of CBD are expressed within networks weighted by the spatial distribution of CB1 receptors. This observation is consistent with the hypothesis that modulation of endocannabinoid signalling may contribute to CBD’s neural effects in psychosis.

Indeed, CBD has been found to inhibit endocannabinoid signalling in a dose-dependent manner, altering the potency of primary CB1 receptor ligands (Tham et al., 2019). Importantly, it may decrease CB1 receptors activity without CB1 inverse agonist-related side effects (Rohleder et al., 2016).

The present findings are consistent with the possibility that CBD may exert its therapeutic effects by modulating synaptic integration and network-level communication within CB1 receptor–rich regions, potentially contributing to its favourable tolerability profile. However, REACT does not provide a direct measure of receptor engagement, nor does it establish that the observed effects are specifically mediated by CB1 negative allosteric modulation in vivo. Given the complexity of CBD pharmacology, alternative or interactive molecular mechanisms may also contribute to the observed changes in CB1 receptor-enriched FC. Accordingly, the present results should be interpreted as providing receptor-informed evidence compatible with, but not definitive, proof of CB1 receptor-mediated effects.

Strengths and limitations

The present study has several strengths. First, the randomised, double-blind, placebo-controlled within-subject crossover design in patients with early psychosis, which increases statistical power and reduces inter-individual variability. Second, the use of the REACT analysis allowed us to investigate CB1 receptor-enriched FC without the need for PET imaging acquisition in each participant. Third, the inclusion of a comparative group of HC, matched for age and sex, undergoing the same screening procedures and scanned on the same scanner using identical acquisition protocol.

However, some important limitations should also be highlighted. Firstly, the present findings must be considered in light of the modest sizes of samples studied. In particular, it remains unclear whether the lack of significant differences between HC and psychosis patients under CBD condition is merely an effect of an underpowered sample. Therefore, these results need independent confirmation in larger cohorts of patients. Moreover, HC had a significantly higher level of education than patients, which may represent a potential confound, and they were scanned only once. Future studies may consider administering both CBD and placebo to this control group in parallel with the design used in patients with early psychosis.

Approximately half of the patients tested positive for THC, and therefore a potential interaction between residual THC exposure and acute CBD effects cannot be ruled out. Moreover, most patients were receiving antipsychotic medication at the time of scanning, making it challenging to disentangle whether the observed differences between patients under PLB and HC reflect effects of psychosis, medication or their interaction. Although the within-subject crossover design ensures stable medication exposure across conditions, a potential pharmacodynamic interaction between CBD and ongoing antipsychotic treatment cannot be excluded. This is supported by evidence that CBD inhibits cytochrome P450 (CYP) enzymes, particularly CYP3A4 and CYP2D6, which are responsible for the metabolism of many antipsychotics, raising the risk of drug–drug interactions (Stout and Cimino, 2014; Woo et al., 2022; Yamaori et al., 2011).

In addition, as the PANSS is primarily designed to capture symptoms severity over longer temporal windows, the assessment within few hours after pharmacological administration may have limited sensitivity to detect subtle clinical changes.

Several factors should also be considered when interpreting the mechanistic implications of the present findings. As discussed above, previous neuroimaging studies support the use of an acute CBD dose of 600 mg to produce measurable neural effects in psychosis-related brain areas. However, recent evidence suggests that acute administration of CBD may exert a dose-dependent effects on circulating endocannabinoid markers, with significant increase of anandamide and related N-acylethanolamines observed at 800 mg but not consistently at 600 mg (Couttas et al., 2024). Accordingly, the complexity of CBD pharmacology should be acknowledged, as the neural effects observed in the present study may not necessarily reflect measurable changes in peripheral endocannabinoid signalling and could involve additional molecular mechanisms. The present study focused specifically on CB1 receptor-enriched FC and therefore does not capture the full range of potential systems through which CBD may influence brain function. While REACT provides a receptor-informed functional framework based on the spatial distribution of CB1 receptors, it does not demonstrate direct receptor occupancy or isolate CB1-specific pharmacodynamic effects in vivo. Consequently, the observed changes should be interpreted as reflecting one candidate pathway within a broader and still incompletely understood pharmacological profile. Future dose-response studies directly comparing multiple CBD doses and integrating neuroimaging, pharmacokinetic, biological markers and receptor-informed measures across additional receptor systems, will be important to better characterise the relationship between CBD dose, molecular target engagement and clinical effects.

Finally, it remains unclear whether the CB1 receptor-enriched FC reduction observed in the right insula under acute administration of CBD would be also observed with a sustained CBD administration. Demonstrating longitudinal changes in CB1 receptor-enriched FC will be essential to establish the clinical potential of CBD as a therapeutic intervention for psychosis.