What Do Thirty Years of Neuroimaging Research Tell Us About Recreational Cannabis Use and Brain Integrity? A Narrative Review of the Multimodal Neuroimaging Evidence to Date

Introduction

Over 228 million individuals report consuming cannabinoids worldwide, making cannabis the most commonly used regulated psychoactive substance in the world.¹ As liberal cannabis policies have spread internationally, cannabinoid-based products have become increasingly more accessible and diverse.² Meanwhile, the amount of Δ-9-tetrahydrocannabinol (THC), the primary psychoactive cannabinoid of cannabis, has increased over time in different world regions, including from 4% in 1995 to 14% in 2019 in the United States^3,4 and from 7% in 2010 to 11% in 2019 in Europe.⁵ These statistics are worrying, as THC has psychoactive, anxiogenic, psychotogenic and addictive properties.^6–8

Relatedly, the consumption of cannabis can come with negative physical and psychosocial outcomes. For example, intoxication with cannabinoids has been associated with increased anxiety and psychotic-like symptoms,^9,10 motor vehicle crashes resulting in injury or death,^10–12 and reduced cognitive function, including working memory, verbal fluency, executive functioning, and time estimation.^13,14 Regular cannabis consumption has also been linked to respiratory issues,¹⁵ cannabis craving and withdrawal,¹⁶ and greater symptoms and presence of psychiatric disorders, including cannabis use disorder (CUD),^14,17 and psychotic, mood, and anxiety disorders.^14,18

The adverse outcomes associated with acute intoxication with cannabinoids and the regular consumption of cannabinoids have been partly attributed to the impact of THC on the brain.^8,19 Indeed, in vitro animal evidence shows that THC exerts its psychoactive effects via binding to cannabinoid type 1 (CB1) G protein-coupled receptors.⁷ In vitro human evidence demonstrates that CB1 receptors are densely innervated in multiple brain regions implicated in cognitive processes reportedly altered during cannabinoid intoxication and in regular cannabis users.²⁰ The regions include: the prefrontal cortex (PFC), including the orbitofrontal cortex (OFC) and the anterior cingulate cortex (ACC), which are implicated in executive function, cognitive control, and disinhibition;²¹ the amygdala, which is implicated in craving;²² the hippocampus, which is involved in stress, learning, and memory;²³ the cerebellum, which is implicated in salience, motivation, and executive function;²⁴ and the basal ganglia, including the striatum and the globus pallidus, which are involved in reward processing.²⁵ According to prominent neuroscientific theories of addiction, the chronic consumption of cannabinoids can be ascribed to neuroadaptations²² in pathways high in CB1 receptors outlined above, and implicated in addiction-related cognitive processes such as interoception (eg, insula).²²

In the last three decades, in vivo neuroimaging studies in humans have begun to uncover the neurobiology of cannabinoid intoxication and regular cannabis use compared to non-use. While several high-quality reviews of the neuroimaging evidence in cannabis users to date have been published,^26,27 the reviews focus either on outdated evidence (eg, up to year 2019)²⁸ and therefore lack an up-to-date summary; or on individual neuroimaging techniques or two neuroimaging techniques, such as structural T1-weighted magnetic resonance imaging (sMRI)^29,30 and functional MRI (fMRI).^26,27 The neurobiological mechanisms of cannabis use and related problems are likely complex and may not be fully understood with an individual neuroimaging modality. A comprehensive synthesis of the evidence regarding the impact of cannabis exposure on brain integrity is warranted, especially in light of global decriminalization trends and the subsequent rise in cannabis availability² and potency.^3–5 Such an overview is essential to provide evidence-based guidance on cannabis use and brain health for the general community, vulnerable people who consume cannabinoids, clinical practitioners, and policy-makers. However, the current evidence should be considered in light of several methodological limitations, including but not limited to cross-sectional study designs, heterogeneous metrics and reporting of cannabis use patterns and related problems, variable cannabinoid dosing and product heterogeneity (eg, THC:CBD ratios, routes of administration), under-representation of females, and inconsistent collection and adjusting for confounding factors, such as polysubstance use, via means of design (eg, matching between groups) or statistical analyses (eg, covariate).

This narrative review aims to synthesize the evidence from multimodal human neuroimaging studies to date. The synthesis was achieved via examining the neural correlates of acute cannabinoid intoxication and of regular cannabis consumption in non-intoxicated participants (ie, henceforth termed cannabis users) compared to controls. The multimodal neuroimaging techniques included in the review are briefly described in Supplementary Materials Section 1.1, and illustrated in Figure 1, and include positron emission tomography (PET);³¹ fMRI;³² sMRI;³² diffusion-weighted imaging (DWI);³³ and magnetic resonance spectroscopy (MRS).³⁴ To that end, a series of searches of the electronic database Google Scholar were conducted to identify relevant studies between any time and October 2024. The following search terms were used: “cannabis”; OR “THC”; “tetrahydrocannabinol” OR “marijuana”; AND “neuroimaging”; OR “brain imaging”; OR “fMRI”; OR “functional magnetic resonance imaging”; OR “BOLD”; OR “brain function”; OR “magnetic resonance imaging”; OR “sMRI”; OR “structural MRI”; OR “structural magnetic resonance imaging”; OR “brain volume”; OR “brain structure”; OR “PET”; OR “positron emission tomography”; OR “MRS”; OR “magnetic resonance spectroscopy”. Studies were included if they: (i) examined human participants; (ii) were written in the English language; (iii) were full-text peer-reviewed articles; (iv) assessed a sample of people who consume cannabis as defined by each study, compared to non-using controls, or a sample who were acutely administered THC, compared to placebo. We excluded studies that (i) examined non-human participants; (ii) were published in languages other than English; (iii) were non-peer-reviewed (eg, conference abstracts); (iv) the sample selected, endorsed at a group level dependence/severe substance use disorders on substances other than cannabis or tobacco; (v) the sample selected, endorsed major psychiatric (eg, schizophrenia) or neurological (eg, epilepsy) or health disorders (eg, cancer) at a group level. Full-text articles were cross-referenced for additional relevant studies.

Figure 1 An Overview of the Non-invasive Neuroimaging Tools used to Assess Brain Integrity and Cannabis Use. Abbreviations: ACC, anterior cingulate cortex; BOLD, blood oxygen level-dependent; CB1, cannabinoid type 1 receptor; CBD, cannabidiol; CUD, cannabis use disorder; CUDIT, cannabis use disorder identification test; dlPFC, dorsolateral prefrontal cortex; DWI, diffusion-weighted imaging; fMRI, functional magnetic resonance imaging; MPS, marijuana problem scale; MRS, magnetic resonance spectroscopy; NAA, N-acetyl aspartate; OFC, orbitofrontal cortex; PCC, posterior cingulate cortex; PET, positron emission tomography; PFC, prefrontal cortex; ROI, region-of-interest; sMRI, structural T1-weighted magnetic resonance imaging; THC, delta-9-tetrahydrocannabinol; VBM, voxel-based morphometry.

Neuroimaging Evidence on Brain Integrity During THC Intoxication

As overviewed in Table 1, a substantial number of PET and fMRI studies have examined brain metabolism, blood flow, neurotransmission, and function during acute intoxication with THC. Overall, the studies provided consistent evidence that THC intoxication affects distinct metrics of brain integrity, most commonly brain function within fronto-striatal regions.

Table 1 Overview of Demographics, THC Administration, and Neuroimaging Results from THC Intoxication Studies

PET Studies Investigating Changes in Brain Metabolism, Blood Flow, and Neurotransmission During THC Intoxication

Results from PET studies consistently demonstrated that acute THC intoxication changes brain metabolism and blood flow, and reduces dopamine release in frontal-insular regions in cannabis users compared to controls. A series of four studies employing H₂[¹⁵O]-PET, administered 20 mg of inhaled THC vs placebo to 76 cannabis users (M_age = 29).^37–40 Cannabis users intoxicated with THC vs placebo showed increased cerebral blood flow in the insula and several frontal regions, including the OFC and the ACC.^37–40 Second, two PET studies examined how THC intoxication affects dopamine release using D2/D3 receptor tracer [11C]raclopride.^35,36 Administration of 8 mg of inhaled THC vs placebo increased dopamine release in controls (n = 31; M_age = 26).^35,36 In contrast, intravenous injection of 0.5 mg THC/kg vs placebo did not change dopamine release in cannabis users (n = 24; M_age = 27).^35,36 Third, an individual 2-deoxy-2-(¹⁸F)fluoro-D-glucose PET study administered 2 mg of THC intravenously to 8 cannabis users (M_age = 31) and 8 controls (M_age = 35).⁴¹ THC intoxication increased cerebellar glucose metabolism in both cannabis users and controls.⁴¹ However, only the cannabis group demonstrated increased metabolism in the OFC, PFC, and basal ganglia.⁴¹ Overall, PET evidence shows THC intoxication-induced changes in brain metabolism, blood flow, and dopamine release.

fMRI Studies: Brain Functional Changes During THC Intoxication

The fMRI evidence to date examining brain function during acute intoxication is growing and shows consistent changes in brain function, particularly within fronto-striatal pathways.

Resting State Functional Connectivity fMRI During THC Intoxication

There was consistent evidence that acute intoxication with cannabis alters brain function at rest, particularly between striatal and frontal regions involved in reward processing and disinhibition, respectively.⁵³ One review of 13 studies in young cannabis users (n = 318; M_age = 25) showed that the type of cannabinoid administered (eg, THC, cannabidiol [CBD] or cannabis plant matter) differentially changed resting state functional connectivity.⁴² For example, inhalation of 12.5 mg THC vs placebo reduced frontal-striatal connectivity (eg, ACC, frontal cortex, nucleus accumbens).⁴² Similarly, lower resting state functional connectivity emerged during intoxication with cannabis plant matter vs placebo, between fronto-insular, fronto-temporal and fronto-striatal pairs,⁴² and in correlation with greater self-report subjective intoxication.^44,54,55 In young cannabis users (n = 19), oral administration of 10 mg of THC vs 600 mg of CBD increased fronto-striatal connectivity.⁴⁵ Additionally, emerging evidence from two studies in occasional cannabis users (n = 17; M_age = 26) suggests that co-administering THC and CBD reduces the effect of THC on resting state functional connectivity. Indeed, inhaling 8 mg of THC reduced resting state functional connectivity between striatal-frontal and fronto-insular pairs, compared to inhaling 8 mg of THC concurrently with 10 mg of CBD.^43,44 Overall, THC and cannabis intoxication were associated with functional changes in fronto-striatal pathways at rest.

Task-Based fMRI Studies: Brain Function During THC Intoxication

There were inconsistent results from a growing number of task-based fMRI studies that acute THC intoxication affects brain activity across reward, learning and memory, and additional fMRI tasks.

Reward Processing Tasks During THC Intoxication

During fMRI tasks of reward processing, there was emerging, inconsistent, and mixed evidence that acute THC intoxication changes brain activity, particularly in striatal regions during the receipt and anticipation of rewards.⁵⁶ For example, a group of cannabis users (n = 11; M_age = 22) administered 10 mg of inhaled THC vs placebo, demonstrated reduced frontal-striatal activity when receiving monetary rewards.⁴⁸ In separate samples of youth (n = 24; M_age = 17) and adult cannabis users (n = 23; M_age = 27), intoxication with 8 mg inhaled THC vs placebo increased ventral striatal and insular activity while anticipating monetary rewards.⁴⁶ Further, in cannabis users (n = 16; M_age = 26), intoxication with 10 mg inhaled THC vs placebo lowered striatal, hippocampal, amygdalar, and auditory activity while listening to music.⁴⁷ Overall, the direction and location of THC-induced changes were inconsistent during reward processing fMRI tasks. The inconsistent findings might be explained by the use of heterogeneous fMRI tasks.

Verbal Learning and Memory fMRI Tasks During THC Intoxication

Only two studies in occasional cannabis users examined the relationship between brain activity during verbal pair associate fMRI tasks following the oral administration of 10 mg THC vs placebo.^49,50 Both studies provided mixed results for THC vs placebo; one study (n = 15; M_age = 27) demonstrated increased activity in the ventral striatum during memory retrieval and parahippocampus during memory encoding,⁵⁰ while a second study (n = 13; M_age = 22) found lower insular and inferior frontal/occipital activity during memory encoding.⁴⁹

Inhibitory and Working Memory fMRI Tasks During THC Intoxication

Only individual studies have examined how THC intoxication affects brain function during additional fMRI tasks. One study examined brain activity during a go/no-go inhibitory fMRI task in cannabis users (n = 15; M_age = 27) who ingested 10 mg of THC.⁵¹ Cannabis users showed decreased inhibitory-related ACC and inferior frontal activity during THC intoxication vs placebo.⁵¹ A separate study examined brain activity during a working memory fMRI task in young cannabis users (n = 17; M_age = 21) who inhaled 9 mg THC vs placebo.⁵² Young cannabis users showed increased working memory-related activity in dorsolateral PFC (dlPFC) and inferior frontal/parietal during THC vs placebo.⁵²

Brain Integrity from Case Control Studies: Non-Intoxicated Cannabis Users Compared to Controls

As summarized in Table 2, this section overviews the multimodal neuroimaging evidence to date from PET, sMRI, DWI, MRS, and fMRI studies, investigating differences in brain integrity between non-intoxicated cannabis users and controls.

Table 2 Overview of Demographics, Cannabis Use, and Neuroimaging Results from Case-Control Studies

PET Studies Investigating Group Differences in Brain Metabolism, Cannabinoid Receptor Density, and Neuroimmune Activation in Cannabis Users Compared to Controls

Emerging evidence from PET studies has examined group differences between cannabis users and controls in brain metabolism,^59,62 CB1 receptor density,⁶¹ and neuroimmune activation⁵⁸ using different techniques, such as 2-deoxy-2-(¹⁸F)fluoro-D-glucose (FDG), [¹⁸F]FMPEP-d2, and [¹⁸F]FEPPA PET. A seminal FDG PET study showed lower cerebellar glucose metabolism in cannabis users (n = 8; M_age = 31) compared to controls (n = 8; M_age = 35).⁴¹ Similarly, lower glucose metabolism was reported in two [¹⁸F]-FDG PET studies in cannabis users (n = 30; M_age = 26) compared to controls (n = 30; M_age = 26).^59,62 Further, lower CB1 receptor density was demonstrated in the neocortex and limbic regions using [¹⁸F]FMPEP-d2 PET in cannabis users (n = 30; M_age = 30) compared to controls (n = 28; M_age = 30), in correlation with the duration of cannabis use;⁶¹ in the ACC using [¹⁸F]MK-9470 PET in cannabis users (n = 10; M_age = 26) compared to controls (n = 10; M_age = 23);^57,60 and in the temporal lobe using [¹¹C]OMAR PET in cannabis users (n = 10; M_age = 23) compared to controls (n = 10; M_age = 26).^57,60 Notably, an individual study reported that CB1 receptor density normalized after a 4-week monitored abstinence.⁶¹ Additional [¹⁸F]FEPPA PET work demonstrated higher whole-brain neuroimmune activation (ie, translocator protein levels) in cannabis users, particularly those with a CUD (n = 23; M_age = 23), compared to controls (n = 27; M_age = 24).⁵⁸ Overall, PET studies of cannabis users, compared to controls, demonstrated lower cerebellar metabolism, reduced CB1 receptor density in cortical and limbic regions that may be restored following a period of abstinence, and greater whole-brain neuroimmune activation.

T1-Weighted Studies in Cannabis Users Compared to Controls: Volumetry

A substantial number of T1-weighted MRI studies to date have compared volumetry between cannabis users and controls.^29,63 There were no significant differences between cannabis users (n = 778; M_age = 26) and controls (n = 710; M_age = 25) for global brain metrics, including total brain volume, intracranial volume, total grey matter volume, and total white matter volume.⁶³ Lower volumes in the OFC and hippocampus emerged in a review of 31 studies in cannabis users (n = 778; M_age = 26) compared to controls (n = 710; M_age = 25).⁶³ Lower OFC and hippocampal volumes were confirmed in a meta-analysis of cannabis users (n = 717; M_age = 26) compared to controls (n = 778; M_age = 25).³⁰ Age may moderate the relationship between cannabis use and volumetry. In younger samples aged up to 26 years, no volumetric differences emerged between cannabis users (n = 386; M_age = 23) and controls (n = 444; M_age = 23).²⁹

Specific cannabinoids may moderate the association between cannabis use and volumetry. One study compared hippocampal NAA between 4 groups, including 3 groups of cannabis users and a control group (n = 37; M_age = 30).⁶⁴ The cannabis groups were separated depending on cannabinoid profile based on toxicology analyses from urine and hair samples, and were: (i) presence of THC metabolites in urine, but unknown cannabinoids in hair analysis (CBDx, n = 19, M_age = 34); (ii) confirmed THC but no CBD from hair analysis (CBD−, n = 30; M_age = 32); (iii) presence of both THC and CBD in hair analysis (CBD+, n = 12; M_age = 32); and (iv) regular cannabis users who reported abstinence for an average of 2.4 years, corroborated by absence of cannabinoids in urine and hair analysis (n = 13; M_age = 39).⁶⁴ Volumes were significantly lower in the cannabis groups: without CBD (CBD-) than controls (11% reduction), and with unknown hair cannabinoids (CBDx) than controls (7% reduction). There were no significant hippocampal volume differences between former regular users and controls.⁶⁴ Therefore, CBD might mitigate the impact of cannabis use on hippocampal volumetry, and abstinence may recover volume reductions. Finally, the evidence to date provided inconsistent evidence for an association between specific cannabis use patterns and volumetry, and only a few studies reported significant correlations between lower volumes and greater cannabis dosage.⁶³ However, no significant associations emerged in a meta-analytical evaluation of the evidence.³⁰ Overall, T1-weighted evidence showed that reduced volumes were confined to specific brain regions, including the OFC and the hippocampus.

DWI Studies in Cannabis Users Compared to Controls: White Matter Microstructure

DWI evidence consistently demonstrated different white matter microstructure in cannabis users compared to controls, particularly within the corpus callosum and superior longitudinal fasciculus.^65–67 A review of 30 DWI studies in cannabis users (n = 1457; M_age = 24) compared to controls (n = 1441; M_age = 24) consistently reported lower white matter microstructural integrity (ie, lower fractional anisotropy and greater mean diffusivity) in cannabis users.⁶⁷ The more consistently altered white matter fiber tracts included the corpus callosum, which connects the right and left cerebral hemispheres; and the superior longitudinal fasciculus, which connects the frontal, parietal, and temporal lobes.⁶⁷ A few DWI studies showed consistent associations between earlier age of cannabis use onset and a longer duration of use and lower white matter integrity in cannabis users.^86–89 Two recent DWI studies also confirm lower white matter microstructural integrity. A study from the UK Biobank (https://www.ukbiobank.ac.uk/) reported worse corpus callosum integrity (eg, lower fractional anisotropy and higher mean diffusivity) in older participants who reported consuming cannabis in their lifetime (n = 3641; M_age = 61) compared to controls (n = 12255; M_age = 65).⁶⁵ Notably, Mendelian randomization demonstrated no causal relationship between cannabis use in the same sample of older adults and white matter microstructure.⁶⁵ Finally, a recent study combined connectome and Fixel-Based Analysis methodologies to compare CUD (n = 56; M_age = 25) and control groups (n = 38; M_age = 32).⁶⁶ CUD individuals exhibited higher white matter density in the corpus callosum and lower white matter density in the cingulum and right cerebellum.⁶⁶ Overall, DWI evidence points to consistently different white matter integrity in the corpus callosum in cannabis users compared to controls.

MRS Studies in Cannabis Users Compared to Controls: Neurochemistry

There was mixed MRS evidence that cannabis users demonstrated alterations in markers of neuronal viability (NAA) and glial cell metabolite markers (myoinositol), glutamate and other neurochemicals (eg, adenosine triphosphate, phosphocreatine).^64,71 A review of eight MRS studies reported emerging evidence that cannabis users (n = 124; M_age = 25) compared to controls (n = 154; M_age = 25) had reduced NAA levels, most consistently in the ACC and dorsolateral PFC.⁷¹ There was also emerging evidence of associations between frontal NAA and duration of cannabis use.^90–93 However, a more recent study found no differences in NAA levels within the ACC in cannabis users with adolescent onset (n = 22; M_age = 30) compared to controls (n = 21; M_age = 24).⁶⁹ A further study found that hippocampal NAA was significantly lower in the cannabis groups: without CBD in hair analysis (CBD-), compared to controls (14% lower), and to former regular users (16% lower); with unknown hair cannabinoids (CBDx) than former regular users.⁶⁴ There were no significant NAA differences between former regular users and controls.⁶⁴ Thus, CBD might minimize the effect of cannabis use on hippocampal NAA, and abstinence may recover hippocampal NAA levels. Additional MRS research on brain metabolites other than NAA comprises largely individual studies. For example, myoinositol or glutamate markers in the ACC did not significantly differ between cannabis users (n = 26; M_age = 24) and controls (n = 47; M_age = 24).⁷⁰ Similarly, levels of adenosine triphosphate and inorganic phosphate across five brain regions – frontal grey and white matter, thalamus, basal ganglia, and temporal lobe – were comparable between cannabis users (n = 45; M_age = 24) and controls (n = 47; M_age = 24).⁶⁸ Yet, in the same sample, phosphocreatine was lower in the cannabis group in the left basal ganglia and left frontal white matter.⁶⁸ Further, in a separate study, myoinositol but not glutamate levels in the hippocampus were lower in cannabis users with adolescent use onset (n = 22; M_age = 30) compared to controls (n = 21; M_age = 24).⁶⁹ In sum, MRS studies showed that NAA levels in the hippocampus and ACC were inconsistently altered in cannabis users.

Cue-Reactivity fMRI Studies Investigating Brain Function in Cannabis Users Compared to Controls

Mounting fMRI cue-reactivity studies have investigated brain function in cannabis users and demonstrated differences while watching cannabis cues compared to neutral cues in several brain pathways implicated in decision-making and reward processing.^72,73 A review of 18 fMRI studies comparing brain function between regular cannabis users (n = 603; M_age = 23) and controls (n = 315; M_age = 23) demonstrated a pattern of greater activity in frontal and striatal regions: ACC, posterior cingulate cortex [PCC], and nucleus accumbens.⁷³ In two studies, cue-reactivity-related brain function in frontal regions (eg, OFC, ACC) and PCC correlated with cannabis use problems (eg, scores on the Marijuana Problem Scale [MPS] and Cannabis Use Disorder Identification Test [CUDIT])^94,95 However, no association emerged between cue-reactivity-related brain function and cannabis frequency or dosage.⁷³ A more recent multi-site study recruited 104 participants from Texas (USA) and 83 from the Netherlands.⁷² The CUD group (n = 104; M_age = 23) compared to controls (n = 83; M_age = 23) showed greater activity in the frontal medial cortex, striatum (eg, nucleus accumbens), precuneus, lateral occipital cortex, and thalamus.⁷² There were no correlations with cannabis dosage, but there were significant correlations between precuneus/occipital activity and CUDIT scores and cue-induced craving, and the direction of the correlations was positive for the Texan CUD group, and negative for participants from the Netherlands.⁷² Overall, cannabis cue reactivity was associated with greater brain activity and connectivity in prefrontal-striatal pathways, in some instances exacerbated with more problematic cannabis use and moderated by world region/culture.

Reward Processing fMRI Studies Investigating Differences in Brain Function in Cannabis Users Compared to Controls

The fMRI evidence comparing brain reward function between cannabis users and controls was emerging and largely non-significant. A recent review summarized the evidence from 13 fMRI studies comparing cannabis users (n = 379; M_age = 21) and controls (n = 333; M_age = 21),⁷⁵ during varied fMRI tasks such as card-guessing and monetary incentive delay.^74,75 About half of the fMRI evidence pointed to altered reward anticipatory activity in prefrontal-striatal regions and altered reward receipt activity in prefrontal, limbic, and sensorimotor regions. The remaining findings were non-significant.⁷⁵ A separate review synthesized findings comparing brain activity between cannabis users (n = 255; M_age = 22) and controls (n = 279; M_age = 22) during the monetary incentive delay fMRI task.⁷⁴ The review found emerging but inconsistent evidence of altered brain reward function.⁷⁴ A more recent fMRI study found no significant differences between cannabis users (n = 63; M_age = 22) and controls (n = 62; M_age = 22) during the anticipation or receipt of monetary rewards.⁵⁶ Overall, the findings on brain reward function in cannabis users were largely non-significant.

Impulsivity fMRI Studies Investigating Differences in Brain Function in Cannabis Users Compared to Controls

A handful of fMRI studies examined brain function during impulsivity-related fMRI tasks (eg, decision-making, response inhibition). In one review of 4 fMRI studies comparing cannabis users (n = 64; M_age = 24) and controls (n = 58; M_age = 22),⁷⁷ about half of the studies showed group differences in prefrontal activity (eg, ACC, dlPFC and left/right middle frontal-parietal lobe).⁷⁷ There was emerging evidence of significant correlations between brain activity during inhibitory processing and cannabis use patterns, such as duration of use.⁹⁶ A recent study demonstrated no group differences between cannabis users (n = 17; M_age = 21) and controls (n = 14; M_age = 24) during the Balloon Analogue Risk fMRI task measuring risky behaviour.⁷⁶ Overall, there is insufficient and mixed fMRI evidence on brain functional differences during impulsivity-related tasks between cannabis users and controls (eg, decision-making, response inhibition).⁷⁷

Memory and Working Memory fMRI Studies Investigating Differences in Brain Function in Cannabis Users Compared to Controls

There was consistent evidence of altered brain activity during memory-related and working-memory tasks in cannabis users compared to controls.^78–80 One review in cannabis users (n = 267; M_age = 24) and controls (n = 261; M_age = 24) consistently observed group differences in memory-related brain activity, most frequently in frontal and temporal/hippocampal regions.⁷⁹ There was sparse evidence demonstrating correlations between memory-related brain activity and the age of cannabis use onset.⁷⁹ A more recent study in cannabis users (n = 100; M_age = 23) and controls (n = 84; M_age = 23) investigated brain activity during a working memory task (N-back).⁷⁸ The cannabis group displayed reduced working memory-load brain activity in the lateral occipital cortex, right precuneus and left and right medial frontal gyrus.⁷⁸ An additional review examined the correlates of cannabis abstinence on brain activity in adolescents during various fMRI tasks, including spatial working memory.⁸⁰ Three studies investigated group differences between cannabis users (n = 69; M_age = 18), cannabis users who were abstinent for at least 25 days (n = 98; M_age = 18) and controls (n = 106; M_age = 18).⁸⁰ Both abstinent and current cannabis users consistently displayed differences compared to controls in activity in occipital areas and higher-order frontal-parietal regions. No significant differences were found between the abstinent and current cannabis-using groups.⁸⁰ Overall, the fMRI evidence demonstrated altered brain activity during memory-related and working-memory tasks, particularly in the higher-order cognitive regions and the temporal lobe.^78–80 In some cases, these differences in brain activity were associated with cannabis use patterns, including age of onset.⁷⁹ The emerging evidence from abstinence studies suggests that memory-related differences in brain activity persist beyond at least 25 days of abstinence.⁸⁰

Resting State Functional Connectivity Studies Investigating Differences in Brain Function in Cannabis Users Compared to Controls

There was consistent evidence that cannabis users, compared to controls, demonstrated altered patterns of resting state functional connectivity between frontal and other cortical regions, in correlation with cannabis use patterns. For instance, one review summarized the evidence from 21 studies in cannabis users (n = 737; M_age = 23) and controls (n = 659; M_age = 23).⁸⁴ Studies consistently demonstrated greater connectivity between fronto-frontal, fronto-striatal, and fronto-temporal region pairings, and in correlation with cannabis use frequency and cannabis-related problems (eg, MPS and CUDIT scores).⁸⁴ Additional evidence from three studies revealed that similar patterns of altered resting state functional connectivity within the dorsal salience network and temporal regions persist across cannabis using samples across different ages including youth (n = 70, M_age = 17), adults (n = 180, M_age = 25), and older adults (n = 43, M_age = 68).^81–83 Resting state functional connectivity differences in cannabis users correlated with cannabis use patterns, including heavier cannabis use and participants’ perceived benefits and harms of cannabis.^81–83 Finally, one individual study showed that young cannabis users (n = 36; M_age = 21) compared to controls (n = 39; M_age = 21), displayed lower resting state functional connectivity within the dorsal attention network after two weeks of monitored abstinence.⁸⁵ In sum, different frontal-cortical resting state functional connectivity consistently emerged in cannabis users compared to controls and across distinct age groups.

Discussion

This review summarizes the multimodal neuroimaging evidence published over the past 3 decades, on the neurobiology of cannabinoid consumption, during cannabinoid intoxication and in non-intoxicated cannabis consumers. PET and fMRI evidence consistently showed that acute cannabis THC intoxication altered brain fronto-striatal function. Case-control PET and MRS studies consistently showed differences between cannabis users and controls, in glucose metabolism and CB1 receptor density, and the latter reversed after a 4-week abstinence.⁶¹ Case-control fMRI studies consistently showed that cannabis users, compared to controls, had different brain function in frontal pathways implicated in higher-order cognitive control.²¹ There was largely inconsistent MRS evidence on altered neurochemistry (eg, NAA level in frontal regions) in cannabis users. Consistent findings from sMRI and DTI studies demonstrated that cannabis users compared to controls showed lower volumetry in selected brain regions (eg, OFC/hippocampus), and altered white matter microstructure in specific tracts (eg, superior longitudinal fasciculus). There was inconsistent MRS evidence of altered brain metabolism. The evidence from abstinent samples was preliminary and suggests recovery of brain integrity, albeit inconsistently. Emerging evidence suggests that brain changes associated with cannabinoid consumption may be moderated by age, cannabis dependence/CUD, heavier cannabis use, amount of THC and other cannabinoids (eg, CBD) consumed. Overall, the most consistently affected brain metrics across both the evidence on THC intoxication and case–control studies were activity, connectivity and structure of selected brain pathways (eg, OFC, ACC) and lower volumes of selected regions (eg, OFC, hippocampus). Additionally, case–control studies demonstrated altered activity, connectivity, and metabolism of striatal regions implicated in the experience of reward.²²

Early findings suggest that some variables may moderate the association between cannabis use and brain integrity. First, age may moderate the relationship between cannabis and brain anatomy but not brain function. Brain functional differences were consistently reported across adult and youth samples,⁹⁷ but neuroanatomical differences were consistent in adult samples only.²⁹ Perhaps, adult age or prolonged cannabis exposure or both might be required for volumetric differences to occur. Second, it remains unclear if neurobiological changes resolve with abstinence. There was early evidence from individual studies that CB1 receptor density normalized after 4 weeks of monitored abstinence,⁶¹ and hippocampal NAA and volumes recovered after periods of abstinence from regular use (eg, 24 years).^61,64 However, individual fMRI studies indicated that differences in brain function between cannabis users and controls persist after periods of abstinence (range: 14-to-25 days).^80,85 Multimodal neuroimaging research tracking participants who are abstinent from cannabis multiple times and over time is required to confirm if differences in brain integrity between cannabis users and controls dissipate over time and their time course. Third, chronicity of cannabis use patterns and the severity of cannabis use-related problems may moderate alterations in the integrity of prefronto-temporal-striatal pathways. Indeed, group differences emerged more consistently in samples with more chronic exposure to cannabis (eg, greater frequency, longer duration and greater dosage) and with greater cannabis use-related problems (eg, cannabis dependence or CUD). Consistently, prominent neuroscientific theories of addiction propose that neuroadaptations within the addiction neurocircuitry occur with chronic and problematic consumption of substances.²²

This review is characterized by methodological limitations, including a focus on published studies in English that may be biased towards positive findings, and the lack of a systematic and meta-analytical approach. The multimodal neuroimaging literature reviewed herein also needs to be considered in light of several methodological limitations. First, most case-control studies were cross-sectional, and the temporality or causality of the observed neurobiological correlates cannot be established. Longitudinal neuroimaging cohort studies represent valuable tools to determine whether brain differences predate or confer risk for cannabis use or if they dissipate with abstinence. Several longitudinal cohort studies could prove useful to this end, including the Connectome Coordination Facility (https://www.humanconnectome.org) and the ABCD study (https://abcdstudy.org/). Second, cannabis use patterns (eg, frequency, occasions, dosage, age of onset, duration), cannabis use-related problems (eg, dependence/CUD), and cannabinoid dosing and products used (eg, THC:CBD ratios, routes of administration) were measured rarely and inconsistently. Thus, it cannot be established how specific patterns of use can be harmful to brain integrity.^98,99 Future research should use consistent metrics for cannabis consumption (eg, iCannToolkit)¹⁰⁰ and the Standard THC Unit¹⁰¹ to quantify cannabis use patterns and their link to neurobiological indices. Third, females were largely under-represented,¹⁰² and the role of sex was largely not accounted for in analyses.^102,103 Importantly, males and females show different cannabinoid metabolism, CB1 receptors and endocannabinoids.¹⁰⁴ Therefore, the findings to date may not generalize to female participants. Further, males consume greater quantities of cannabis and on more days than females.^105–107 In case-control studies, sex was also often statistically unaccounted for. Therefore, it is unresolved whether findings were driven by cannabis use or by heavier consumption in males. Future research should include diverse samples and statistically account for the effect of sex (eg, sex as an interaction term with cannabis use).¹⁰³ Fourth, the literature failed to consistently collect data, report, and statistically examine the contribution of mental health and the use of other substances to the observed neurobiological correlates. The presence and severity of psychopathology (eg, anxiety, mood, and psychotic disorders) and substances other than cannabis (eg, tobacco and alcohol) are entrenched with cannabis use and also impact brain integrity independently, and in interaction with cannabis use.^108–110 Therefore, future work is required to confirm which of the emerging effects are specific to cannabinoids or to confounders. Outstanding gaps also exist in MRS studies during THC intoxication, which limit the understanding of the neurochemical correlates of cannabinoid consumption. Further, there was little evidence on the neurobiological correlates of cannabis withdrawal, and future work is required to confirm if the theorized mechanisms underlying drug withdrawal or problematic substance use apply to cannabinoid consumption, eg, amygdala, habenula involvement.²² New neuroimaging work is required to address the limitations described above and shed some light on the neurobiological correlates of cannabis use to inform the general community, vulnerable people who consume cannabinoids, clinical practitioners, and policy-makers. The reviewed evidence suggests that cannabis use is associated with changes in various metrics of brain integrity, particularly in acute THC intoxication studies and emerging case-control studies where changes in brain integrity correlated with heavier cannabis use and amount of THC use. As such, the early evidence implies that policy-makers could consider the current trend towards decrimininalization and increased availability of cannabis products² and potency^3–5 in light of the relationship between cannabis use and brain integrity. However, more work that addresses the limitations described above is required to fully understand the neurobiological mechanisms of cannabis use to confidently inform public health and policy-making. Further, characterizing the neurobiological correlates of CUD will inform targets for neuromodulation interventions that have been gathering interest in other substance use disorders. These interventions can include fMRI-neurofeedback and repetitive transcranial magnetic stimulation, which have shown promise to modulate brain activity in other substance-using cohorts (eg, alcohol, cocaine, tobacco).^111,112 Indeed, novel work has begun to investigate the use of fMRI-neurofeedback to mitigate craving-related brain activity in individuals with CUD.¹¹³

Conclusion

We summarized the last thirty years of multimodal neuroimaging evidence on the neural correlates of cannabinoid use. During THC intoxication, consistent evidence showed changes in the function of frontal-striatal regions, including increased cerebral blood flow as measured with PET, as well as decreased rsFC and activity, as per fMRI. There were consistent findings that non-intoxicated cannabis users, compared to controls, showed: changes in glucose metabolism and CB1 receptor density (which reversed with a 4-week abstinence), as measured with PET; changes in brain function as per fMRI, most consistently increases during cue-reactivity; lower OFC/hippocampal volumetry measured by T1-weighted imaging; and lower superior longitudinal fasciculus white matter microstructure. The reviewed evidence came with several methodological limitations. Indeed, the evidence was largely cross-sectional and did not account for potentially confounders entrenched with the consumption of cannabis (eg, mental ill health, polysubstance use). Further, many studies were constrained by small sample sizes, particularly those investigating the brain integrity and cannabis abstinence. As such, the reviewed evidence should be considered as hypothesis-generating rather than hypothesis-testing. More longitudinal multimodal neuroimaging work that controls for confounds is required to confirm the time course and causal mechanisms of cannabinoid consumption. The lack of brain-behavior correlations restricts the behavioural and clinical relevance of the findings; however, alterations were most commonly reported in brain pathways implicated in disinhibition, stress and interoception pathways and in samples with more severe problems with cannabis use and chronicity of use. Therefore, the findings support the notion that severe problems with/chronicity of cannabis use might require the targeting of neurocognitive systems linked to inhibition, stress and affect, via individually tailored psychological approaches, exercise-based interventions, or non-invasive neuromodulation techniques (eg, neurofeedback, TMS).