The endocannabinoid system is emerging as crucial to many regulatory functions in the brain. The known endocannabinoids are anandamide, 2-arachidonoylglycerol (2-AG), oleoylethanolamide (OEA) and palmitoylethanolamide (PEA). Anandamide levels have been found to be elevated in the cerebrospinal fluid (CSF) of patients with antipsychotic-naive first-break schizophrenic psychosis Reference Leweke, Giuffrida, Wurster, Emrich and Piomelli1,Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 and in the initial prodromal states of psychosis. Reference Koethe, Giuffrida, Schreiber, Hellmich, Schultze-Lutter and Ruhrmann3 This elevation is reversed by treatment with typical, but not atypical, antipsychotics, Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 leading to the suggestion that anandamide has an adaptive, if not protective, role in psychosis and has a homeostatic role in controlling, for example, excessive dopamine release. Reference Leweke4
Repeated treatment with Δ9-tetrahydrocannabinol (THC), the major psychotomimetic compound of cannabis, has been shown to downregulate anandamide signalling in the central nervous system (CNS) of rats (see for example Di Marzo et al Reference Di Marzo, Berrendero, Bisogno, Gonzalez, Cavaliere and Romero5 ). Additionally, in patients with schizophrenia, cannabis use was shown to reduce the elevation in anandamide levels in CSF, when compared with patients who had never or only occasionally used cannabis. Reference Leweke, Giuffrida, Koethe, Schreiber, Nolden and Neatby6 However, among healthy people who had not used cannabis significantly in their lives (<5 occasions in their lifetime) and those who were low-frequency users (<50 occasions in their lifetime) there were no differences in anandamide, suggesting that at least low-level cannabis use does not downregulate endocannabinoid signalling. No previous study had investigated levels of CSF endocannabinoids in heavy cannabis users and how these might relate to differences in the response to cannabis and associated risk of schizophrenia.
Thirty-three individuals were recruited from a database created during previous studies. Reference Morgan, Schafer, Freeman and Curran7 Individuals who used cannabis >10 times in a month were assigned to the ‘heavy’ use condition (n = 10), whereas individuals who used cannabis <10 times per month (n = 10) were assigned to the ‘light’ use condition. Controls were a group of 13 non-cannabis-using individuals. All controls had used cannabis less than eight times in their lifetime; two individuals had never used cannabis. All participants were free from any diagnosis of psychopathology or current medical problems according to DSM-IV criteria. 8 All participants provided written, witnessed, informed consent. This study was approved by the National Research Ethics Committee for National Health Service research. Participants were compensated £10 per hour (up to £50) for participation in the lumbar puncture procedure.
Participants were asked to abstain from recreational drugs and alcohol for 24 h before testing commenced. Each participant was tested at a single session lasting around 2 h. Urine tests were also administered where available to assess the presence of THC and other substances. Drug use histories were taken and demographic data collected from participants. Lumbar punctures were conducted, with the participants choosing between the lying and sitting position. The needle was inserted in the lumbar 4-5 interspace. An atraumatic Sprotte needle was used to collect CSF. A sample of blood was taken from each participant. Blood samples were centrifuged for 10 min at 500 rpm. Samples were stored at −80°C and then all samples were analysed at the Central Institute of Mental Health, Mannheim, Germany. To quantify fatty acid ethanolamides, 1 ml aliquots from CSF and serum samples were spiked with 25 pmol of [ Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 H4]anandamide, [ Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 H4]2-AG, [ Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 H4]PEA and [ Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 H4]OEA to resemble the analytes and analyses were conducted by standard procedures reported elsewhere. Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2,Reference Leweke4
Data from a previous study reported elsewhere Reference Morgan, Schafer, Freeman and Curran7 indexed individual's acute psychotic response to naturalistically smoked cannabis on the Psychotomimetic States Inventory (PSI). Reference Mason, Morgan, Stefanovic and Curran9 These data were used for correlational analyses for all but three cannabis users who had not participated in the previous study. Non-parametric tests were used throughout because of non-normality of the data.
The whole sample comprised 15 males and 18 females. One-way analysis of variance (ANOVA) showed no significant group differences in age (22.1 years, s.d. = 2.45) or years of education (16.5 years, s.d. = 1.99). There were no significant group differences in the number of regular alcohol users or level of alcohol use. Urine tests revealed the presence of THC in urine of all heavy cannabis users (n = 10/10), all but one light cannabis users (n = 9/10) and no controls (n = 0/13). One heavy cannabis user and one light cannabis user tested positive for MDMA (3,4-methylenedioxymethamphetamine).
The number of regular cannabis users and the number of years of cannabis use was not significantly different between the light (4.65 years, s.d. = 3.28) and heavy (6.5 years, s.d. = 2.92) groups. As expected, heavy users used cannabis on significantly more days per month (22.6 days, s.d. = 7.24) than light users (3.85 days, s.d. = 3.12) (F(1,16) = 43.02, P<0.001).
A Kruskal-Wallis test results showed a trend-level group difference for anandamide levels in CSF (χ2 = 4.62, P= 0.096) (Fig. 1). Orthogonal contrasts showed that heavy users had significantly lower CSF anandamide than light users (Mann-Whitney U-test = 18.5, P = 0.015). There were no significant group differences for OEA, PEA or 2-AG in the CSF. A Kruskal-Wallis test showed a significant group difference (χ2 = 13.669, P = 0.001) in serum 2-AG levels (online Fig. DS1). A subsequent U-test indicated that heavy users had significantly higher serum 2-AG levels than controls (U = 3.000, P<0.001). There were no significant group differences for OEA, PEA or anandamide in serum. Cerebrospinal fluid anandamide and 2-AG did not correlate with serum. Among cannabis users, CSF anandamide correlated negatively with state psychotic symptoms when non-intoxicated (r = −0.568, P= 0.017).
This was the first study to examine CSF eicosanoid levels in cannabis users. The study found lower CSF anandamide in heavy cannabis users compared with light cannabis users. Serum 2-AG levels were significantly higher in cannabis users than in controls. Cerebrospinal fluid anandamide correlated negatively with drug-free psychotic symptoms.
Preclinical work has suggested a downregulation of the endocannabinoid system Reference Di Marzo, Berrendero, Bisogno, Gonzalez, Cavaliere and Romero5 following chronic THC administration in rats, which is partially supported by these findings. However, the current study extends this to suggest a possible upregulation of endocannabinoid signalling following moderate cannabis use, evident here in the elevated levels of CSF anandamide in light cannabis users but downregulation following heavy use. Control group anandamide levels were intermediate between the two groups, supporting this assertion.
We also observed a negative relationship between CSF anandamide levels and persistent psychotic symptoms. This is consistent with the findings of previous studies Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2 and is suggestive of a protective role of anandamide against psychotic-like symptoms, Reference Leweke4 in particular negative symptoms. Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe and Faulhaber2,Reference Leweke, Giuffrida, Koethe, Schreiber, Nolden and Neatby6 This may relate to the putative role of anandamide in stabilising dopamine release, given suggestions of increased dopamine release following smoking cannabis. Reference Bossong, van Berckel, Boellaard, Zuurman, Schuit and Windhorst10 Individuals with lower anandamide may be less able to modulate dopamine release, which increases psychotic-like symptoms both acutely and chronically.
Serum levels of 2-AG were found to differ between heavy cannabis users and controls, with greater levels seen in cannabis users. This was an unexpected finding and, given the absence of correlation between serum and CSF levels, one that is difficult to interpret. Indeed, 2-AG is thought to be neuroprotective against excessive glutamate release and previous work has suggested a link between serum 2-AG and depression. Reference Hill, Miller, Ho, Gorzalka and Hillard11 However, the link between peripheral endocannabinoids and the CNS remains conjectural at present and warrants further research.
Our findings suggest that anandamide signalling is altered following cannabis use differentially in light and heavy users and psychotic-like symptoms are related to levels of this endocannabinoid in the brain. These findings suggest that healthy individuals who smoke cannabis and have lower levels of anandamide - either endogenously or through their cannabis use - may be more vulnerable to the acute and chronic psychotomimetic effects of cannabis, and therefore potentially more vulnerable to psychosis.
This work was supported by an ECNP Research Grant for Young Scientists to C.J.A.M.