This is the first human study to investigate the acute effect of intravenous Δ9-THC administration on brain glutamate levels, and whether glutamate level alterations underlie the acute psychotomimetic effects of Δ9-THC. Consistent with our first hypothesis and with previous evidence from animal studies [28–31] we found that acute administration of Δ9-THC significantly increased Glx levels in the left caudate head compared to the placebo. As predicted, we also found that this was associated with transient psychotomimetic effects induced by Δ9-THC. Our most novel finding is that individuals who experienced transient psychotomimetic effects following Δ9-THC (~70% of the sample) had significantly lower baseline Glx (as under placebo) and an almost two-and-a-half-fold higher increase in Glx following Δ9-THC administration compared to individuals who were not sensitive to the psychotomimetic effects of Δ9-THC. Finally, consistent with our prediction, previous cannabis exposure was positively associated with Δ9-THC-induced Glx increase. Exploratory analyses also suggested that the acute effects of Δ9-THC on human brain Glx levels are region-specific, as Δ9-THC administration increased Glx levels in the striatum, but not in the hippocampus or the ACC.

In addition to the brain stem projections, the key inputs to the striatum are mesolimbic dopaminergic (DA) projections from the ventral tegmental area (VTA) as well as cortical and thalamic glutamatergic projections [55, 56]. The activity of VTA DA neurons in vivo is dominated by pacemaker-like tonic firing interrupted by phasic bursts leading to striatal dopamine release [57], and this shift from tonic pacemaker firing to bursting is strongly controlled by synaptic input from glutamatergic and GABAergic afferents to the dopamine neurons [58]. Upon acute exposure, endocannabinoids regulate synaptic strength by acting on glutamatergic afferents to VTA dopamine neurons via activation of CB1 receptors [59]. Preclinical evidence also indicates that the intravenous administration of Δ9-THC can increase VTA dopamine neuronal activity, being ultimately responsible for an increase in striatal dopamine levels through the mesolimbic pathway [19]. However, in vitro studies have demonstrated that cannabinoids do not affect dopamine concentrations when locally applied in the striatum [60], while increasing dopaminergic firing when administered in the VTA [61]. Since VTA dopamine neurons do not express CB1 receptor protein nor mRNA [62], this argues against a direct effect of cannabinoids on dopamine neuron activity, also potentially accounting for the inconsistent evidence on effects of Δ9-THC administration on striatal dopamine release in man [20–25]. In contrast, the glutamatergic inputs to the striatum are especially relevant as they are involved in the processing of different stimuli, such as rewarding and stressful information, and the selection of related behavioral responses [63]. In line with this, acute Δ9-THC administration in animal models has been shown to consistently increase glutamate levels in a number of brain regions including the striatum [28–31]. Therefore, glutamate rather than dopamine may play a more important role in the neurochemical underpinnings of the acute psychotomimetic effects of cannabis [31]. However, no acute challenge study had investigated the effect of acute administration of Δ9-THC on glutamate metabolism in humans. In sum, our study confirms preclinical evidence [28–31] that a single dose of Δ9-THC may increase striatal glutamate levels and suggests this as a potential mechanism underlying the acute psychotomimetic effects of cannabis.

Previous preclinical research indicates that a history of psychostimulant self-administration leads to decreased basal glutamate in both the striatum and the primary neuronal source of striatal glutamate, the prefrontal cortex. Instead, acute psychostimulant drug administration in abstinent animal models previously exposed to the drug induces an enhancement of cortical and striatal glutamate release not seen in drug-naive subjects [64, 65]. Drug-induced heightened activation of cortical glutamatergic afferents to the VTA has been proposed to modulate behavioral sensitization and addiction [66]. In line with this evidence, we found that the lower the striatal Glx levels at baseline, the higher was the increase after Δ9-THC administration. Furthermore, we found that the higher the previous cannabis exposure, the higher was the Δ9-THC-induced striatal increase, potentially suggesting sensitization due to the effects of previous cannabis exposure. However, an important caveat to such an interpretation is the relatively modest levels of previous cannabis use (1 ≤ previous use ≤60 times, 10.4 ± 14.4 times on average). Nevertheless, sensitization (if it is present) to such low levels of exposure is not surprising as preclinical evidence suggests that even a single exposure to psychostimulant drugs can be sufficient to induce long-lasting behavioral and cellular sensitization [66]. One may therefore speculate that the glutamatergic system may have a different steady-state homeostasis as a function of previous exposure, from where the system is particularly susceptible to destabilizing influences that may affect it, such as the acute administration of Δ9-THC. As lower baseline Glx levels and related higher Δ9-THC-induced increase were evident in individuals who developed psychotomimetic symptoms under Δ9-THC compared to individuals who did not, our findings suggest that Δ9-THC-induced psychotomimetic symptoms could be explained by Δ9-THC effects on tonic (basal) versus phasic (burst) glutamate system function, which in turn are modulated by previous cannabis use.

ACC and hippocampal Glx were not significantly increased by acute Δ9-THC administration, suggesting that glutamate alteration in the striatum may represent a specific locus of abnormality underlying sensitivity to the acute and transient psychotomimetic effects of Δ9-THC. However, this needs to be confirmed in larger samples.

Glutamate steady-state homeostasis and sensitization may also account for the apparent discrepancy between the reduction in glutamate observed in studies of chronic cannabis use in man [32–35, 37] and the increase in glutamate observed in preclinical acute challenge studies [28–31] as well as in the present acute challenge study in humans. Animal studies of chronic psychostimulant use clearly indicate drug-induced changes in glutamate regulation, such that basal glutamate levels are decreased while glutamate release is enhanced during drug exposure [64, 65]. Cannabis use may involve progressive neurochemical adaptations in glutamate function, which need to be further investigated. Determining the regional changes in glutamate function that may result from repeated cannabis exposure is also imperative to understanding their relevance to the acute and chronic psychoactive effects of cannabis use. To date, there is robust evidence for altered glutamate steady-state homeostasis in animal models of addiction [64, 65]. However, aberrant glutamate function has also been suggested in psychosis and related disorders. In particular, a systematic review of 63 studies investigating metabolite biomarkers of schizophrenia has indicated glutamate increase as one of the most consistent potential metabolite signatures of the disorder [67]. Increased striatal glutamate levels have also been described in subjects at ultra-high risk for psychosis, and they are also associated with conversion to psychosis [68, 69].

The major strength of this study is its design. Study subjects were recruited if they had a minimal history of cannabis use, had been abstinent from cannabis for at least 6 months, and had negligible use of other substances (alcohol, tobacco, and other illicit drugs). Therefore, we can reasonably rule out the possibility that some of the results observed could be attributed to the effects of other substance use or cannabis withdrawal, dependence, or intoxication. Moreover, for each study participant, there was an interval of at least 14 days between the two study visits. This helps exclude the possibility of any carryover effects as Δ9-THC has been shown to have an elimination half-life of 18 h to 4.3 days [70]. Also, all the participants’ urine samples collected at each study visit at baseline were negative for the presence of Δ9-THC. However, these strict inclusion criteria, while offering advantages in terms of a controlled sample, may at the same time limit the generalizability of the present results to the wider population of cannabis users. Also, the intravenous route for Δ9-THC administration allowed much more consistent Δ9-THC blood levels across study subjects [14], but might have similarly affected the generalizability of the results to the effects of recreational cannabis use. Another limitation of the present study is that, due to its design, it was not possible to examine the test–retest reliability of the MRS Glx measure for the regions investigated. However, evidence indicates that GM Glx in healthy subjects has relatively high reproducibility and test–retest reliability at 3 Tesla [71]. Furthermore, the within-subject design helped avoid the confounding effect of between-subject differences in the outcome variable [71].

It is worth noting that 1H-MRS does not allow us to disentangle whether measured glutamate is from the neurotransmitter or the metabolic pool. Nevertheless, research evidence indicates that majority of the brain glutamate is cycled through the neurotransmitter pool [53] and 1H-MRS-related glutamate measures are likely to be related to glutamatergic neurotransmission [54].

It is worth considering a few other potential alternative explanations for the results presented here, such as effect of spectral quality differences, effect of T2 relaxation and test–retest effect. Cramér–Rao lower-bound values were considerably below the 20% threshold under both drug conditions. Although full width at half maximum (FWHM) values showed a trend toward difference between the two drug conditions, they were also within the spectral quality recommended by Kreis (FWHM of metabolites < 0.07–0.1 ppm) [72]. Collectively, they suggest good quality data, and are more informative than a comparison of quality measures across the drug conditions (∆9-THC, placebo) [72]. Results presented here also point toward regional specificity of the acute effects of THC, as no effects on any metabolite were observed in the hippocampus and anterior cingulate, arguing against these effects being a result of spectral quality differences.

With reference to the possibility that a T2 relaxation effect might have occurred, decreasing the signal, this is expected to happen when a longer echo time (TE) is used [73]. Instead, the combination of a short TE and a long repetition time (TR) used in this study is considered to allow the acquisition of signals with minimal signal loss due to T2- and T1-weighting [73]. Nevertheless, the T2 relaxation of water, which differs between white matter, grey matter, and CSF may be different between individuals because of the presence of different fractions of these components in the MRS voxel, and arguably may introduce a systematic bias in metabolite quantification due to group difference in T2 relaxation. However, this is unlikely to have systematically affected the results of the present study as we employed a within-subject repeated measures design, thereby mitigating the effect of group difference in the fraction of grey matter, white matter, and CSF in the MRS voxel contributing to difference in T2 relaxation under the two drug conditions (∆9-THC versus Placebo). Furthermore, the randomized crossover design with counterbalanced order of drug administration employed in this study also helped mitigate a possible test–retest effect.

In summary, this study suggests that striatal glutamate levels are increased following a single dose of ∆9-THC in healthy individuals. This ∆9-THC-induced glutamate increase likely underlies the acute cannabis-induced psychotomimetic effects, as it seems to be specific to subjects experiencing psychotomimetic effects. These results also suggest that lower baseline levels of striatal glutamate may be a marker of sensitivity to the acute psychotomimetic effects of cannabis and potential sensitization to the modulation of striatal glutamate levels by ∆9-THC as a function of previous cannabis exposure may develop early. Collectively, these results provide novel insight into the neurochemical underpinnings of the effects of cannabis in man and may point towards potential approaches towards mitigating the adverse effects of cannabis, which may have important public health implications.