Stress, cortisol and psychosis

Stress plays a major role in many different mental disorders, including anxiety, depression and psychotic disorders such as schizophrenia (Bebbington et al. Reference Bebbington, Wilkins, Jones, Foerster, Murray, Toone and Lewis1993; Bramon & Murray, Reference Bramon and Murray2001; Cantor-Graae & Selten, Reference Cantor-Graae and Selten2005; Gracie et al. Reference Gracie, Freeman, Green, Garety, Kuipers, Hardy, Ray, Dunn, Bebbington and Fowler2007; Aiello et al. Reference Aiello, Horowitz, Hepgul, Pariante and Mondelli2012; Belvederi Murri et al. Reference Belvederi Murri, Pariante, Dazzan, Hepgul, Papadopoulos, Zunszain, Di Forti, Murray and Mondelli2012; for a review, see Myin-Germeys & van Os, Reference Myin-Germeys and van Os2007). With particular regard to schizophrenia, a chronic disorder that typically affects young adults, early-life (Bramon & Murray, Reference Bramon and Murray2001; Beards et al. Reference Beards, Gayer-Anderson, Borges, Dewey, Fisher and Morgan2013) and adult (van Winkel et al. Reference van Winkel, Stefanis and Myin-Germeys2008) exposure to stress has been linked to an increased risk of development of the disorder. Stress in adulthood has also been associated with increased risk of relapse of pre-existing psychosis (Ventura et al. Reference Ventura, Nuechterlein, Lukoff and Hardesty1989). Consistent with this, the stress–vulnerability model implies that the interaction between vulnerability and stress increases the likelihood of psychosis, more than would be expected if the two factors were to occur independently (Zubin & Spring, Reference Zubin and Spring1977). A dose–response relationship has also been suggested, with psychotic symptoms developing when total load of stressors exceed the vulnerability threshold in a given individual (Myin-Germeys & van Os, Reference Myin-Germeys and van Os2007).

Stress has also been shown to be associated with an increase in negative affect and reduction in positive affect in psychosis patients, their relatives and healthy controls (Myin-Germeys et al. Reference Myin-Germeys, van Os, Schwartz, Stone and Delespaul2001), though the cross-sectional nature of the evidence precludes inference about a causal relationship. However, evidence that risk factors for developing psychosis and suffering relapse, including growing up in an urban area (Sundquist et al. Reference Sundquist, Frank and Sundquist2004) and/or in a highly expressed emotional household (Brent & Giuliano, Reference Brent and Giuliano2007), positively correlate with stress levels (van Winkel et al. Reference van Winkel, Stefanis and Myin-Germeys2008) further reinforces the strength of the association. Furthermore, longitudinal studies show that stressful life events often precede relapse of illness in schizophrenia (Ventura et al. Reference Ventura, Nuechterlein, Lukoff and Hardesty1989; Hirsch et al. Reference Hirsch, Bowen, Emami, Cramer, Jolley, Haw and Dickinson1996; Pallanti et al. Reference Pallanti, Quercioli and Pazzagli1997), and may alone contribute to about a quarter of the risk of relapse (Hirsch et al. Reference Hirsch, Bowen, Emami, Cramer, Jolley, Haw and Dickinson1996).

This raises the question that if stress is causally related to the onset of psychosis, exacerbation of symptoms or relapse of the illness in those with a pre-existing disorder, how may it be doing so? The HPA axis mediates the biological response to stress. When humans are faced with a stressor, corticotrophin-releasing hormone is released from the periventricular nucleus of the hypothalamus. This results in the secretion of adrenocorticotrophic hormone from the pituitary gland, which then leads to the release of glucocorticoids (cortisol in humans) from the adrenals. While the main role of cortisol is to increase blood sugar, suppress the immune response and aid metabolism, the association between the exogenous administration of glucocorticoids and the development of psychosis has been known for a long time, especially in descriptions of ‘steroid psychosis’ (Clark et al. Reference Clark, Bauer and Cobb1952; Munck et al. Reference Munck, Guyre and Holbrook1984).

Consistent with this and evidence referred to in the previous paragraph, a number of studies have reported abnormalities in cortisol levels in response to stress in patients with psychosis (Aiello et al. Reference Aiello, Horowitz, Hepgul, Pariante and Mondelli2012). These may be broadly grouped into two categories: (i) those that have investigated cortisol reactivity following awakening, exposure to stress or pharmacological challenge (Table 1); and (ii) those that have investigated baseline or diurnal abnormalities in cortisol (Table 2). Most studies have employed one or more of these strategies to investigate differences between healthy controls and patients with psychosis or those who are at high risk of developing psychosis either because they have a high genetic (e.g. first-degree relatives of patients with psychosis) or clinical (e.g. those having an ‘at-risk mental state’ for psychosis) risk. Stress and awakening cortisol reactivity studies generally show a blunted cortisol response in patients, but pharmacological challenge studies show higher cortisol levels in patients. Diurnal levels of cortisol are generally higher in patients and their relatives.

Table 1. Summary of studies investigating cortisol reactivity in patients as compared with controls

ARMS, At-risk mental state; can+, cannabis exposure; can–, no cannabis.

Table 2. Cortisol levels in patients as compared with controls

FEP, First-episode psychosis; AUC, area under the curve.

While the evidence summarized above generally tends to suggest an abnormality of the HPA axis both at baseline and following perturbation, in those with psychosis as well as in those at risk of developing psychosis, the precise mechanism through which abnormal cortisol levels may increase the risk of psychosis is unclear. A few putative mechanisms have been suggested, such as abnormal cortisol levels following recurrent stress exposure causing an activation of the dopaminergic system (Walker & Diforio, Reference Walker and Diforio1997; Walker et al. Reference Walker, Mittal and Tessner2008) or affecting neuroplasticity through an interaction with the immune system, neurotrophins and N-methyl-d-aspartate receptors (McEwen, Reference McEwen2000). Similarly, stress-induced cortisol changes affecting declarative memory (Kirschbaum et al. Reference Kirschbaum, Wolf, May, Wippich and Hellhammer1996) suggest a potential pathway to the neurocognitive changes observed in psychosis (Ivleva et al. Reference Ivleva, Shohamy, Mihalakos, Morris, Carmody and Tamminga2012). Emerging neuroimaging evidence in healthy volunteers also suggests that the effects of exposure to stress may converge on neural substrates implicated in psychosis (Akdeniz et al. Reference Akdeniz, Tost, Streit, Haddad, Wüst, Schäfer, Schneider, Rietschel, Kirsch and Meyer-Lindenberg2014). However, the precise mechanism through which exposure to stress heightens the risk of developing psychosis, the exacerbation of symptoms and an increased risk of relapse is far from clear. Therefore, the consistent nature of the evidence emerging from studies carried out by different groups suggests that there is a need to move beyond the investigation of association of stress exposure and/or HPA axis abnormalities and psychosis to studies that may suggest complementary mechanistic insight.

The endocannabinoid system

The endocannabinoid system is a lipid-signalling system involved in regulating brain development, motor control, cognition, emotional responses and homoeostasis (Monory & Lutz, Reference Monory, Lutz, Kendall and Alexander2009). The messengers of the endocannabinoid system are known as endogenous cannabinoids or endocannabinoids. So far, five different endocannabinoids have been identified. Two of the most researched of these are anandamide (AEA) and 2-arachidonoylglycerol (2-AG), which are primarily broken down by the enzymes fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase, respectively (Kozak et al. Reference Kozak, Rowlinson and Marnett2000; Kim & Alger, Reference Kim and Alger2004; Marrs et al. Reference Marrs, Blankman, Horne, Thomazeau, Lin, Coy, Bodor, Muccioli, Hu, Woodruff, Fung, Lafourcade, Alexander, Long, Li, Xu, Moller, Mackie, Manzoni, Cravatt and Stella2010). Two G-protein-coupled receptors involved in the endocannabinoid system have also been discovered: cannabinoid receptors 1 and 2 (CB1R and CB2R). Unlike CB2R, which is more common peripherally, CB1R is mostly localized in the central nervous system, particularly the cerebral cortex, basal ganglia, anterior cingulate cortex and cerebellum (for a review, see Breivogel & Sim-Selley, Reference Breivogel and Sim-Selley2009). It is present at high densities at presynaptic axon terminals (Kim & Alger, Reference Kim and Alger2010). It is thought to play a central role in homoeostasis by directly or indirectly modulating the release of neurotransmitters such as glutamate, serotonin, dopamine and noradrenaline (Melis & Pistis, Reference Melis and Pistis2007; López-Moreno et al. Reference López-Moreno, González-Cuevas, Moreno and Navarro2008; Mátyás et al. Reference Mátyás, Urbán, Watanabe, Mackie, Zimmer, Freund and Katona2008; Wang & Lupica, Reference Wang and Lupica2014). These neurotransmitters, particularly dopamine, have been implicated in the pathophysiology of psychosis (Dean et al. Reference Dean, Sundram, Bradbury, Scarr and Copolov2001; Zavitsanou et al. Reference Zavitsanou, Garrick and Huang2004; Stone et al. Reference Stone, Morrison and Pilowsky2007; Reynolds et al. Reference Reynolds, McGowan and Dalton2014). Therefore, the endocannabinoid system seems an obvious target to study, both in terms of understanding what causes the alterations in these neurotransmitters (such as hyperactive dopamine transmission) that are observed in schizophrenia, as well as in identifying potential translational entry points for correcting these abnormalities. Alterations in the endocannabinoid system have been reported in psychosis (Giuffrida et al. Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe, Faulhaber, Klosterkötter and Piomelli2004). Indeed, there has been a cannabinoid hypothesis of schizophrenia posited much like the dopamine hypothesis (Müller-Vahl & Emrich, Reference Müller-Vahl and Emrich2008).

The genesis of the cannabinoid hypothesis of schizophrenia

While it has been known for a very long time that cannabis use may induce paranoia and acute psychosis (for a review, see Murray et al. Reference Murray, Morrison, Henquet and Di Forti2007), more systematic investigation of this relationship dates back several decades (Chopra & Smith, Reference Chopra and Smith1974; Rottanburg et al. Reference Rottanburg, Ben-Arie, Robins, Teggin and Elk1982; Andréasson et al. Reference Andréasson, Engström, Allebeck and Rydberg1987). Several epidemiological studies since the study by Andréasson et al. (Reference Andréasson, Engström, Allebeck and Rydberg1987) have consistently shown an association between cannabis use and psychotic symptoms/schizophrenia (Tien & Anthony, Reference Tien and Anthony1990; Arseneault et al. Reference Arseneault, Cannon, Poulton, Murray, Caspi and Moffitt2002; Van Os et al. Reference Van Os, Bak, Hanssen, Bijl, De Graaf and Verdoux2002; Zammit et al. Reference Zammit, Allebeck, Andreasson, Lundberg and Lewis2002; Ferdinand et al. Reference Ferdinand, Sondeijker, Van Der Ende, Selten, Huizink and Verhulst2005; Fergusson et al. Reference Fergusson, Horwood and Ridder2005; Rössler et al. Reference Rössler, Hengartner, Angst and Ajdacic-Gross2012), some of which have been summarized in a number of competent and insightful reviews (Arseneault et al. Reference Arseneault, Cannon, Witton and Murray2004; Moore et al. Reference Moore, Zammit, Lingford-Hughes, Barnes, Jones, Burke and Lewis2007). Hence, we will not focus on the evidence here, but summarize the main issues. Although the association between cannabis use and psychosis is generally accepted, whether this association is causal in nature is strongly debated (Gage et al. Reference Gage, Zammit and Hickman2013). Reverse causality has been suggested as a potential explanation for the association between cannabis use and psychosis, as a result of which individuals with pre-existing psychosis are more likely to start using cannabis. However, evidence of a temporal relationship, where this is available and credible, suggests that cannabis use often predates the onset of psychosis (for a review, see Moore et al. Reference Moore, Zammit, Lingford-Hughes, Barnes, Jones, Burke and Lewis2007). Furthermore, other strategies such as the use of statistical modelling (Fergusson et al. Reference Fergusson, Horwood and Ridder2005) suggest that the direction of causality is from cannabis use to psychotic symptoms, although data using a sibling-pair design (McGrath et al. Reference McGrath, Welham, Scott, Varghese, Degenhardt, Hayatbakhsh, Alati, Williams, Bor and Najman2010) also suggest that individuals vulnerable to psychosis were at greater risk of commencing cannabis use which in turn increased their risk of subsequently developing a non-affective psychotic disorder. Similarly, another potential explanation is that cannabis use is a marker for another causative agent, such as amphetamine. Longitudinal studies that have taken such a possibility into consideration still find evidence in support of the association, albeit at a reduced strength (Zammit et al. Reference Zammit, Allebeck, Andreasson, Lundberg and Lewis2002). This does not necessarily rule out the possibility of an unknown ‘true’ causative agent driving the association between cannabis use and psychosis, though it is unclear what this agent might be (Gage et al. Reference Gage, Zammit and Hickman2013). Approaches (either statistical modelling or sibling-pair design) taken to control for unmeasured confounding (Fergusson et al. Reference Fergusson, Horwood and Ridder2005; McGrath et al. Reference McGrath, Welham, Scott, Varghese, Degenhardt, Hayatbakhsh, Alati, Williams, Bor and Najman2010) suggest that the association between cannabis use and psychosis is unlikely to be due to unmeasured residual confounding factors. Evidence summarized in a meta-analysis (Moore et al. Reference Moore, Zammit, Lingford-Hughes, Barnes, Jones, Burke and Lewis2007) as well as further new evidence (Di Forti et al. Reference Di Forti, Sallis, Allegri, Trotta, Ferraro, Stilo, Marconi, La Cascia, Marques and Pariante2014) also suggest a dose–response relationship such that heavier or more frequent use of cannabis as well as use of high-potency cannabis is associated with a greater risk. Evidence has also emerged that genetic vulnerability may influence the risk of psychotic disorder in cannabis users. While the initial studies suggested a moderating effect of functional polymorphism in the gene for catechol-O-methyltransferase (Caspi et al. Reference Caspi, Moffitt, Cannon, McClay, Murray, Harrington, Taylor, Arseneault, Williams and Braithwaite2005; Henquet et al. Reference Henquet, Rosa, Krabbendam, Papiol, Faňanás, Drukker, Ramaekers and van Os2006; Henquet et al. Reference Henquet, Rosa, Delespaul, Papiol, Faňanás, Van Os and Myin-Germeys2009), these results were not replicated in subsequent studies (Zammit et al. Reference Zammit, Spurlock, Williams, Norton, Williams, O'Donovan and Owen2007, Reference Zammit, Owen, Evans, Heron and Lewis2011; van Winkel, Reference van Winkel2011). Data from two independent cohorts instead suggest a greater risk of development of psychosis in cannabis users carrying risk variants of a polymorphism at the rs249732 locus in the gene coding for protein kinase B (AKT1). This is also consistent with independent experimental evidence (Bhattacharyya et al. Reference Bhattacharyya, Atakan, Martin-Santos, Crippa, Kambeitz, Prata, Williams, Brammer, Collier and McGuire2012a , Reference Bhattacharyya, Iyegbe, Atakan, Martin-Santos, Crippa, Xu, Williams, Brammer, Rubia, Prata, Collier and McGuire2014) that functional polymorphism at a related locus (rs1130233) in the AKT1 gene, that is in strong linkage disequilibrium (Di Forti et al. Reference Di Forti, Iyegbe, Sallis, Kolliakou, Falcone, Paparelli, Sirianni, La Cascia, Stilo and Marques2012) with the rs249732 locus mentioned earlier, moderates the sensitivity of healthy individuals to the acute psychotomimetic, cognitive and neurophysiological effects of Δ-9-tetrahydrocannabinol (THC), the main psychoactive ingredient in cannabis. Complementary experimental evidence that THC produces its behavioural effects by binding to an integral component of the endocannabinoid system, the CB1R (Huestis et al. Reference Huestis, Gorelick, Heishman, Preston, Nelson, Moolchan and Frank2001), and can induce transient psychotic symptoms similar to those seen in schizophrenia (Bhattacharyya et al. Reference Bhattacharyya, Morrison, Fusar-Poli, Martin-Santos, Borgwardt, Winton-Brown, Nosarti, MO'Carroll, Seal and Allen2010, Reference Bhattacharyya, Atakan, Martin-Santos, Crippa, Kambeitz, Prata, Williams, Brammer, Collier and McGuire2012a , Reference Bhattacharyya, Crippa, Allen, Martin-Santos, Borgwardt, Fusar-Poli, Rubia, Kambeitz, O'Carroll and Seal b ) also points toward a potential role of endocannabinoid dysfunction in schizophrenia.

The endocannabinoid system and psychosis

Independent of the evidence linking cannabis use with the development of psychotic disorders, evidence has been accumulating regarding alterations of components of the endocannabinoid system in patients with psychosis or schizophrenia, which is summarized subsequently.

Human studies

As summarized earlier, regular, frequent use of cannabis, which affects the endocannabinoid system, has been shown to be a robustly replicated environmental risk factor for the development of schizophrenia as well as for exacerbations of pre-existing disease (Hides et al. Reference Hides, Dawe, Kavanagh and Young2006). Similarly, those at risk for psychosis experience higher transition rates if they use cannabis (Kristensen & Cadenhead, Reference Kristensen and Cadenhead2007). Acute intoxication with THC may lead to transient psychotic symptoms including paranoia and hallucinations (Bhattacharyya et al. Reference Bhattacharyya, Crippa, Allen, Martin-Santos, Borgwardt, Fusar-Poli, Rubia, Kambeitz, O'Carroll and Seal2012b ) that resolve without treatment and an exacerbation of existing positive and negative symptoms in patients with schizophrenia (D'Souza et al. Reference D'Souza, Abi-Saab, Madonick, Forselius-Bielen, Doersch, Braley, Gueorguieva, Cooper and Krystal2005). Research using imaging techniques has found that these effects may be due to THC's effect on activation in striatal areas (Fusar-Poli et al. Reference Fusar-Poli, Crippa, Bhattacharyya, Borgwardt, Allen, Martin-Santos, Seal, Surguladze, O'Carrol and Atakan2009; Bhattacharyya et al. Reference Bhattacharyya, Atakan, Martin-Santos, Crippa, Kambeitz, Prata, Williams, Brammer, Collier and McGuire2012a , Reference Bhattacharyya, Crippa, Allen, Martin-Santos, Borgwardt, Fusar-Poli, Rubia, Kambeitz, O'Carroll and Seal b ). The striatum is rich in dopaminergic terminals and striatal function (Beckmann & Lauer, Reference Beckmann and Lauer1997; Lauer & Beckmann, Reference Lauer and Beckmann1997; Simpson et al. Reference Simpson, Kellendonk and Kandel2010) and the dopaminergic system (Guillin et al. Reference Guillin, Abi-Dargham and Laruelle2007) are altered in schizophrenia. While some studies (see Sami et al. Reference Sami, Rabiner and Bhattacharyya2015) have shown that acute administration of THC may increase dopamine levels (Bossong et al. Reference Bossong, van Berckel, Boellaard, Zuurman, Schuit, Windhorst, van Gerven, Ramsey, Lammertsma and Kahn2008), suggesting that the psychotomimetic effects of cannabis are mediated by dopamine, the evidence is equivocal, as others did not find this in healthy users (Stokes et al. Reference Stokes, Mehta, Curran, Breen and Grasby2009; Barkus et al. Reference Barkus, Morrison, Vuletic, Dickson, Ell, Pilowsky, Brenneisen, Holt, Powell and Kapur2011; Kuepper et al. Reference Kuepper, Ceccarini, Lataster, van Os, van Kroonenburgh, van Gerven, Marcelis, Van Laere and Henquet2013). However, Kuepper et al. (Reference Kuepper, Ceccarini, Lataster, van Os, van Kroonenburgh, van Gerven, Marcelis, Van Laere and Henquet2013) reported THC-induced dopamine release in those with increased risk of psychosis. Conversely, others have found that cannabis users have reduced dopamine synthesis capacity compared with non-users, which was not associated with cannabis-induced psychotic symptoms (Bloomfield et al. Reference Bloomfield, Morgan, Egerton, Kapur, Curran and Howes2014), suggesting perhaps that psychosis induced by cannabis use may be characterized by a reduced dopamine synthesis capacity and an increased sensitivity of the D2/D3 receptor (Murray et al. Reference Murray, Mehta and Di Forti2014).

Consistent with this, evidence has emerged of alterations in different components of the endocannabinoid system, from the target receptors to endogenous cannabinoids such as AEA, in patients with psychosis. Table 3 summarizes current evidence regarding these alterations in the different components of the endocannabinoid system in psychosis. The earliest studies investigated CB1Rs (which are the primary target for THC in the brain) in post-mortem brains of patients. Inconsistencies in the results from earlier studies reporting an increase in the density of CB1R (Dean et al. Reference Dean, Sundram, Bradbury, Scarr and Copolov2001; Zavitsanou et al. Reference Zavitsanou, Garrick and Huang2004; Newell et al. Reference Newell, Deng and Huang2006) and more recent studies reporting reduced CB1R mRNA expression levels in schizophrenia (Eggan et al. Reference Eggan, Hashimoto and Lewis2008, Reference Eggan, Stoyak, Verrico and Lewis2010) may be due to methodological reasons such as examining different brain areas. However, an important reason for difference may be that studies reporting increased CB1R density employed receptor autoradiography to estimate this, while studies that reported lower levels investigated mRNA expression or protein levels. A more recent study employing both techniques in the same post-mortem brain specimens reported reduced CB1 mRNA and higher CB1R binding levels in the prefrontal cortex (PFC) of schizophrenia and suggested that this discrepancy may be a result of higher affinity of the CB1R or its altered trafficking into the membrane (Volk et al. Reference Volk, Eggan, Horti, Wong and Lewis2014).

Table 3. Summary of human studies investigating endocannabinoid system alterations in psychosis

THC, Δ-9-Tetrahydrocannabinol; CB1R, cannabinoid receptor 1; DLPFC, dorsolateral prefrontal cortex; MDD, major depressive disorder; FAAH, fatty acid amide hydrolase; LFU, low frequency users; HFU, high frequency users.

There have also been a number of studies (Table 3) investigating endocannabinoid levels in body fluids including in peripheral blood samples, because of easy accessibility (De Marchi et al. Reference De Marchi, De Petrocellis, Orlando, Daniele, Fezza and Di Marzo2003). Use of peripheral blood samples raises the issue of whether alterations in peripheral blood are related to any changes in the brain and vice versa. However, studies investigating levels in cerebrospinal fluid (CSF) have also found similar changes. Elevated levels of AEA have been reported in early psychosis, with higher AEA levels being linked to delayed transition to psychosis in those at risk, suggestive of a protective role for AEA in psychosis (Koethe et al. Reference Koethe, Giuffrida, Schreiber, Hellmich, Schultze-Lutter, Ruhrmann, Klosterkötter, Piomelli and Leweke2009). This work would need to be replicated but if true it suggests that pharmacological treatment that increases AEA may be useful in schizophrenia. However, AEA normalized in patients prescribed typical antipsychotics but remained elevated following atypical antipsychotics (Giuffrida et al. Reference Giuffrida, Leweke, Gerth, Schreiber, Koethe, Faulhaber, Klosterkötter and Piomelli2004). In psychosis patients, regular use of cannabis led to the down-regulation of AEA compared with low-frequency users (Leweke et al. Reference Leweke, Giuffrida, Koethe, Schreiber, Nolden, Kranaster, Neatby, Schneider, Gerth and Hellmich2007). However, the cross-sectional nature of a majority of these studies makes it difficult to infer whether these alterations in different parameters of the endocannabinoid system are causally linked to psychosis or a consequence of the disease process. Nevertheless, they raise the possibility that altered peripheral endocannabinoid measures might be a marker of illness and or treatment response in psychosis. There is also a need to reconcile the nature of endocannabinoid alterations in blood/CSF with CB1R changes in the brain, perhaps using in vivo techniques such as positron emission tomography (PET) imaging in patients with psychosis to understand the extent to which peripheral measures reflect central changes.

Links between the stress response and endocannabinoid systems

Although there is considerable professional and public health concern regarding the effect of cannabis on mental health, most individuals who use it recreationally do so for its relaxing effect and a perceived beneficial effect on stress, which is consistent with the distribution of CB1Rs in brain regions associated with regulating the stress response (for example, the hypothalamus, hippocampus and amygdala: Herkenham et al. Reference Herkenham, Lynn, Johnson, Melvin, de Costa and Rice1991; Tsou et al. Reference Tsou, Nogueron, Muthian, Sañudo-Peña, Hillard, Deutsch and Walker1998). This is also consistent with human experimental evidence that cannabinoids modulate the function of limbic structures, such as the amygdala, anterior and posterior cingulate cortex (Phan et al. Reference Phan, Angstadt, Golden, Onyewuenyi, Popovska and de Wit2008; Fusar-Poli et al. Reference Fusar-Poli, Crippa, Bhattacharyya, Borgwardt, Allen, Martin-Santos, Seal, Surguladze, O'Carrol and Atakan2009; Bhattacharyya et al. Reference Bhattacharyya, Morrison, Fusar-Poli, Martin-Santos, Borgwardt, Winton-Brown, Nosarti, MO'Carroll, Seal and Allen2010) and that these neural effects may be related to an anxiogenic or anxiolytic effect (Fusar-Poli et al. Reference Fusar-Poli, Crippa, Bhattacharyya, Borgwardt, Allen, Martin-Santos, Seal, Surguladze, O'Carrol and Atakan2009; Bhattacharyya et al. Reference Bhattacharyya, Morrison, Fusar-Poli, Martin-Santos, Borgwardt, Winton-Brown, Nosarti, MO'Carroll, Seal and Allen2010). Animal (Di et al. Reference Di, Malcher-Lopes, Marcheselli, Bazan and Tasker2005; Malcher-Lopes et al. Reference Malcher-Lopes, Di, Marcheselli, Weng, Stuart, Bazan and Tasker2006) as well as human studies have shown that glucocorticoids cause elevation in endocannabinoid levels (Dlugos et al. Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012). In this section, we will focus on current evidence that points towards an inter-relationship between the stress response and endocannabinoid systems, drawing particularly upon the relevant pre-clinical literature.

Human studies

In healthy participants, acute stress leads to an increase in circulating endocannabinoids and structurally similar lipids (Hill et al. Reference Hill, Miller, Carrier, Gorzalka and Hillard2009b ; Dlugos et al. Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012). The ‘Trier Social Stress Test’ (TSST), which involves subjecting participants to social evaluative stress, induces moderate psychological stress in a laboratory setting (Kirschbaum et al. Reference Kirschbaum, Pirke and Hellhammer1993). It is an experimental paradigm that has often found favour with researchers investigating the effects of experimental stress induction in man and has proven to be a useful tool for measuring cortisol response to psychological stress (Kudielka & Kirschbaum, Reference Kudielka and Kirschbaum2005). Dlugos et al. (Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012) compared the TSST with a no-stress condition in healthy volunteers and found that compared with the no-stress condition, stress increased the concentration of AEA immediately after the stress exposure. Further, increases in cortisol were correlated with N-palmitoylethanolamine (Dlugos et al. Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012). A trend for individuals with lower levels of AEA to have higher cortisol release after stress compared with those with higher levels of AEA was also found. While pre-clinical studies have shown that peripheral concentrations of endocannabinoids correlate with levels in stress-related brain areas (Patel et al. Reference Patel, Roelke, Rademacher, Cullinan and Hillard2004), Dlugos et al. (Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012) acknowledge that the source of the circulating endocannabinoids is essentially unknown. Lack of information on previous history of stressful life events in their study participants also makes it difficult to infer the extent to which the acute effects of the TSST are moderated by previous exposure to stress, especially in light of independent evidence that previous exposure to stress can permanently alter the stress response (Heim et al. Reference Heim, Newport, Heit, Graham, Wilcox, Bonsall, Miller and Nemeroff2000). Nevertheless, this study clearly suggests a link between the stress response and endocannabinoid systems in man. In a similar study comparing the effect of the TSST in depressed relative with non-depressed women, Hill et al. (Reference Hill, Miller, Carrier, Gorzalka and Hillard2009b ) reported no change in AEA levels following acute stress but a change in 2-AG immediately after stress induction compared with baseline (Hill et al. Reference Hill, Miller, Carrier, Gorzalka and Hillard2009b ). Although, the magnitude of these changes was not different between depressed and control subjects, baseline levels of AEA and 2-AG were significantly lower in the depressed subjects compared with the controls. While apparently inconsistent, the results of the two studies may be reconciled by the fact that the Hill et al. (Reference Hill, McLaughlin, Morrish, Viau, Floresco, Hillard and Gorzalka2009a , Reference Hill, Miller, Carrier, Gorzalka and Hillard b ) study included only female participants (Hill et al. Reference Hill, Miller, Carrier, Gorzalka and Hillard2009b ) and subgroup analyses from the Dlugos et al. (Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012) study found that the results they found were only significant in males (Dlugos et al. Reference Dlugos, Childs, Stuhr, Hillard and de Wit2012). Another study which investigated the effect of parabolic flight on stress responses showed that stress-tolerant individuals demonstrate an increase in 2-AG. In contrast, highly stressed individuals do not show such an increase but have a significantly enhanced HPA axis response (Strewe et al. Reference Strewe, Feuerecker, Nichiporuk, Kaufmann, Hauer, Morukov, Schelling and Choukèr2012). Together with other evidence summarized in previous sections, this may suggest a protective role of increased endocannabinoid system signalling in maintaining homoeostasis under acutely stressful conditions.

Relevance for psychosis

From the above, it is clearly evident that the endocannabinoid system both regulates and responds to the HPA axis stress response system and habituation to stress involves alterations in the endocannabinoid system (Gorzalka et al. Reference Gorzalka, Hill and Hillard2008). However, how exposure to daily-life stress alters the HPA axis and the endocannabinoid system and whether such alterations are causally related to psychosis are unknown. Not every individual who is exposed to stress develops a psychiatric disorder, least of all psychosis. Therefore, it has been hypothesized that those who do so may have a pre-existing dysfunction in the endocannabinoid system due to its integral role in habituation to stress that makes them particularly vulnerable to adverse mental health consequences (Gorzalka et al. Reference Gorzalka, Hill and Hillard2008). Furthermore, why similar types of stress may lead to different types of psychological outcomes (e.g. affective disorder such as depression as opposed to a non-affective disorder such as schizophrenia) is unclear. Endocannabinoids have been found to generally constrain corticosterone (the primary glucocorticoid involved in the stress response in non-human species) release (Hill & Tasker, Reference Hill and Tasker2012). It has been found that administration of a CB1 antagonist leads to an increase in corticosterone secretion in animals (Wade et al. Reference Wade, Degroot and Nomikos2006). This demonstrates the endocannabinoid system's role in modulating the HPA axis and may be useful in understanding how increased stress reactivity arises and how it may ultimately lead to psychosis. Chronic restraint stress in mice leads to an increase in FAAH activity and a reduction of AEA (Hill et al. Reference Hill, Kumar, Filipski, Iverson, Stuhr, Keith, Cravatt, Hillard, Chattarji and McEwen2013). As mentioned earlier, AEA may have a protective role in stress as animal studies show that chronic restraint stress can cause a reduction of AEA (Hill et al. Reference Hill, Kumar, Filipski, Iverson, Stuhr, Keith, Cravatt, Hillard, Chattarji and McEwen2013), while increased AEA signalling due to blockade of its metabolic enzyme (FAAH) decreases anxiety-like behaviours (Moreira et al. Reference Moreira, Kaiser, Monory and Lutz2008). Similarly, a cannabinoid receptor agonist prevented endocrine alterations in a rat model of stress (Ganon-Elazar & Akirav, Reference Ganon-Elazar and Akirav2011). However, this needs to be confirmed in man. Stress-induced alterations in AEA have also been linked to alterations of the HPA axis (Rademacher et al. Reference Rademacher, Meier, Shi, Vanessa Ho, Jarrahian and Hillard2008; Hill et al. Reference Hill, McLaughlin, Morrish, Viau, Floresco, Hillard and Gorzalka2009a ) putatively via an increase in FAAH (Hill et al. Reference Hill, McLaughlin, Morrish, Viau, Floresco, Hillard and Gorzalka2009a ), the enzyme that degrades AEA. These studies suggest that enhancing the availability of certain endocannabinoids may help augment tolerance to stress. However, despite the growing body of evidence suggesting that stress-induced alterations in the HPA axis and the endocannabinoid system may have a role in psychosis, the precise nature of these relationships have not been investigated in a systematic manner in humans. Due to the abundance of evidence demonstrating the role of stress in psychosis, and the relationship between the endocannabinoid system and psychosis, a thorough understanding of the links between these two systems in man is essential.

Conclusions and future directions

Evidence summarized here generally suggests that both the baseline tone and reactivity of the HPA axis to stress, and awakening and pharmacological challenge are abnormal in those with psychosis and at risk of developing the disorder. Similarly, evidence has also emerged of alterations in components of the endocannabinoid system, from receptors (in post-mortem brain samples) to endocannabinoid levels in CSF and peripheral blood, in patients with psychosis. Evidence from basic research as well as human studies also suggests that the endocannabinoid system regulates as well as responds to the HPA axis stress response system and habituation to stress is associated with alterations in the endocannabinoid system. However, the precise nature of the inter-relationship between these two systems is unclear in man, especially in the context of psychosis. It is worth noting that the direction of alteration observed in either the HPA axis/stress response system or in components of the endocannabinoid system is not always consistent across studies, perhaps related to different methodological approaches employed. Evidence available to date, especially related to endocannabinoid alteration in man, is also limited by modest sample sizes studied. The biological relevance of endocannabinoid alteration observed peripherally in blood to effects in the brain remains unclear as well. In particular, there is a lack of evidence about how these systems interact in man and whether these are altered in those with psychosis or at risk of developing the disorder. Thus, there is a need to simultaneously investigate alteration in these two systems in the same group of subjects (healthy individuals v. patients with psychosis). Consistent evidence of alteration in these two systems emerging from studies carried out by different groups suggests that there is also a need to move beyond the investigation of statistical association of stress exposure and/or HPA axis abnormalities or endocannabinoid system abnormalities and psychosis to studies that may suggest complementary mechanistic insight, perhaps by employing a combination of experimental perturbation of either of these systems in healthy and diseased individuals. Relating peripheral endocannabinoid measures to central measures, such as CB1R density using PET, is particularly important to establish biological relevance.

As reviewed here, evidence regarding the link between abnormalities in the stress response system and the endocannabinoid system in man is limited, and this is particularly true in terms of their relationship to psychosis. It has been postulated that individuals who are susceptible to depression may possess a dysfunctional endocannabinoid system that prevents them from coping and adapting to stress adequately (Gorzalka et al. Reference Gorzalka, Hill and Hillard2008). Similarly, it may be the case that those susceptible to psychosis may possess a dysfunctional endocannabinoid system that predisposes them to altered stress sensitivity and is a mechanism for the affective pathway to psychosis (Myin-Germeys & van Os, Reference Myin-Germeys and van Os2007). This dysfunctional endocannabinoid system may, in some cases, have its genesis in the abuse of cannabis.