From the Clinic to the Laboratory, and Back Again: Investigations on Cannabinoids and Endocannabinoid System Modulators for Treating Schizophrenia

Abstract

Introduction: From Anecdote to Clinical Trials and Animal Models

During the 1940s−50s, the principle bioactive components of cannabis–cannabinoids–were identified as delta-9 tetrahydrocannabinol (THC; the psychoactive component) and cannabidiol (CBD), through the work of Adams et al. in the USA, and Todd et al. in Great Britain [reviewed in (1)]. In the 1980s−90s, the discovery of endogenous cannabinoid receptors in the nervous system (2, 3) and their endogenous ligands anandamide and 2-arachidonylglycerol (2-AG) (4, 5) provided a physiological framework within which to investigate the psychoactive properties of THC, the possible therapeutic value of cannabinoids, and the function of the endocannabinoid system. In the context of schizophrenia pharmacotherapy, most clinical investigations have focused on CBD.

The first published report of CBD as a possible treatment for schizophrenia was a case study of a 19-year old female patient, whose symptoms were reduced by a 4-week CBD treatment [1,200 mg/day; (6)]; however, results of later case studies were equivocal (7). More recently, studies showing positive results include a double-blind randomized clinical trial comparing CBD (20 patients) to amisulpride [19 patients; (8)]. Dosing of CBD began at 200 mg/day, was raised to 800 mg/day during the first week and was maintained at that level for 3 weeks. CBD was equally as effective as amisulpride, both showing significant reductions in PANSS positive, negative, and total scores. In CBD treated patients, changes in PANSS total scores were negatively associated with increases in serum anandamide, suggesting that therapeutic effects of CBD were related to an inhibition of anandamide degredation. More recent studies indicate that CBD blocks human fatty acid amide hydrolase binding proteins (FABPs), thereby preventing transport of anandamide to fatty acid amide hydrolase (FAAH, the enzyme that degrades anandamide) (9). A second study was a randomized, double-blind, placebo controlled study comparing CBD (1,000 mg/day; 42 patients) to placebo (40 patients) as an add-on to their ongoing antipsychotic treatment (10). CBD reduced PANSS positive, but not negative or total, symptom scores. Executive function, assessed by the Brief Assessment of Cognition in Schizophrenia (BACS), showed a nearly significant improvement in the CBD group (p = 0.068). By contrast, a double-blind, placebo-controlled study of CBD (600 mg/day) as an add-on to ongoing antipsychotic found no effect of CBD on PANSS scores or on cognitive symptoms (11). Apart from CBD dosing, these two studies differed in ethnic composition, being predominantly European Caucasian in the McGuire study vs. a more mixed race population in the Boggs study. Additionally, there could have been important differences in CBD-antipsychotic interactions, if the two patient populations differed with respect to specific antipsychotics used. Finally, responsiveness of positive symptoms to CBD in the McGuire study was quite modest, and while CBD reduced positive symptoms similarly in the Boggs study, there was a significant placebo effect. Three randomized, double-blind, placebo controlled studies tested rimonabant [a cannabinoid receptor type 1 (CB1) inverse agonist]. Two of these studies, one testing rimonabant alone [20 mg/day; (12)] and the other as an add-on to ongoing antipsychotic treatment in overweight patients [20 mg/day; (13)] found no significant effects of rimonabant on positive or negative (12), or cognitive (13) symptoms. A third randomized pilot study (14) in a small group of overweight patients reported that rimonabant (20 mg/day) had no effect on negative symptoms, but improved anxiety and hostility subscales of the Brief Psychiatric Rating Scale (14).

Experimental Paradigms for Modeling Schizophrenia in Laboratory Rodents: Independent and Dependent Variables

Although the “classical” criteria of face, predictive, and construct validity are those most often applied for evaluating animal models of neuropsychiatric disorders (15), another useful context for evaluating an animal model is to consider it in terms of an experimental paradigm having a set of independent and dependent variables (16–18). Independent variables include species and sex of the subjects, their genetic characteristics, and the experimental manipulation applied. Dependent variables include quantifiable neurobiological or behavioral endpoints. An assessment of the model would first critically consider whether the model’s independent and dependent variables are homologous to known pathogenic risk factors and psychiatric symptoms, respectively. A second consideration would be whether the relationship between the independent and dependent variables corresponds to the real-life relationship between risk factor and psychiatric symptoms (“inductive validity”) (17). Regardless of the specific criteria of validity that are applied, such criteria should be considered as a means to define the strengths and limitations of the model in order to provide a context within which to critically interpret the results that the model generates.

In addition to (and often distinct from) modeling the pathophysiology of a psychiatric disorder, an important practical function of an animal model is to accurately predict pharmacological responsiveness in the clinic, or “predictive validity” (15). This mini-review will focus on pre-clinical studies that have investigated CBD or pharmacological manipulations of the endocannabinoid system in animal models of schizophrenia, with the following objectives: (1) determine how well animal models of schizophrenia predicted or corresponded to clinical responsiveness, and (2) evaluate these same animal models within the context of inductive validity, as defined above.

Experimental Modulation of the Endocannabinoid System in NMDA Receptor Antagonist and Neurodevelopmental Models of Schizophrenia

Based on reported similarities between the psychotropic effects of PCP and ketamine (glutamate NMDA receptor antagonists) and the positive, negative, and cognitive symptoms of schizophrenia, it was proposed that schizophrenia may be associated with a generalized hypofunction of NMDA receptors (51). This generalized dysfunction is believed to alter cortical information processing by altering the firing of cortical GABAergic interneurons, as well as disrupt the balance of cortical—subcortical dopamine (52). Considering this theoretical framework for schizophrenia pathophysiology, two general approaches have been taken for modeling schizophrenia in rodents: acute or subchronic/chronic challenge with NMDA receptor antagonist.

From the Clinic to the Lab and Back Again

Clinical trials suggest that CBD may be a useful pharmacotherapy for schizophrenia, but more studies are needed. Some clinical variables that are important to consider in future studies are disease chronicity (first episode vs. chronic disease), details of drug administration (alone or in combination with antipsychotics), and therapeutic objective [reducing existing symptoms or preventing disease progression; (62, 63)]. Likewise, in pre-clinical studies, these variables should be systematically considered within the experimental design. For example, acute NMDA receptor antagonist treatment can model the acute psychotic episode, while recent onset or chronic disease are best represented by persistent effects of neurodevelopmental challenges or subchronic or chronic administration of NMDA antagonists. CBD should be tested alone and in combination with distinct antipsychotics. [Indeed, at least one study indicates that CBD may reduce the effectiveness of antipsychotic treatment in an animal model (47)]. Possible preventative effects of CBD can be investigated by administering CBD before or along with the experimental challenge. More attention must be given to the comparability of the independent variables defined in animal model studies to controllable (and uncontrollable) clinical variables such as those mentioned above.

Likewise, outcome measures (dependent variables) should be standardized between pre-clinical and clinical studies. The present mini review discussed some outcome measures that are reasonably comparable between human and animal studies, and that have been applied in the laboratory (PPI and specific cognitive tests). The NOR, SI, and SR tests are highly practical for large scale use, but homologous tests for the clinic have not been developed. With respect to cognitive deficits, rodent tests that have clear human homologs, such as those identified by the CNTRICS initiative, should be more frequently applied. With respect to social deficits, new behavioral tests should be designed that more clearly capture specific domains of social dysfunction, perhaps including objective observations of human subjects in controlled social situations.

The body of studies presented here illustrates the general lack of comparability between pre-clinical and clinical studies. Of the clinical studies, only one (8) is reasonably comparable to just 4 (of 19) pre-clinical studies with respect to independent variables: CBD administered as a single drug to subjects with existing acute schizophrenia is comparable to CBD administration to rodent models that received subchronic/chronic NMDA receptor antagonist followed by drug washout (36, 38) or to neurodevelopmental models (39–41). Nevertheless, the outcome measures in these pre-clinical studies are not easily comparable to those of the Leweke et al. (8) study, except for deficits in social interaction, which are contemplated in the PANSS negative symptom subscale.

What are some other factors that may limit the translatability of pre-clinical results to the clinic? In the specific case of CBD and endocannabinoid modulators, responsiveness may be highly sensitive to dose. In the relatively few animal model studies where several different doses of CBD were tested, there was some indication that dose responsiveness to CBD might take the form of an inverted U, where lower and higher doses are ineffective or have negative effects. Given this information, it seems important that body mass be taken into consideration when analyzing clinical results, perhaps by including this factor as a covariable in the statistical analysis of the results. Secondly, clinical populations are likely heterogeneous with respect to underlying pathophysiologies, perhaps some being responsive to CBD or endocannabinoid system modulators, and others unresponsive. With this in mind, it might be useful to examine in an exploratory manner risk factor profiles of patient responders vs. non-responders, in order to define the dependent variables that are most appropriate for modeling responsiveness (and non-responsiveness) to CBD or endocannabinoid system modulators.

It is striking that the vast majority pre-clinical studies reviewed here were carried out using exclusively male animals. Human clinical populations obviously comprise both sexes, not to mention being more diverse than laboratory animals in terms of underlying pathology, age, diet, body weight, among many other characteristics. By contrast, laboratory rodent strains have been selectively bred for hundreds of generations and raised under highly controlled conditions, no doubt reducing genetic and epigenetic variability of the population. All of these factors could introduce variability into the clinical response that is not represented in pre-clinical studies, thus limiting the predictive power of the latter.

Since the results of clinical studies on a homogenous population would be almost meaningless in clinical practice, one possible way to address this limitation may be to systematically carry out pre-clinical experiments in mixed-sex animal populations that have more inherent genetic diversity, perhaps including rodent or non-rodent species that have not had an extended history of laboratory rearing, such as the prairie vole or deer mouse. Since important interspecies differences may exist with respect to pharmacological responsiveness [for example, rodent FAAH is inhibited by CBD, while the human counterpart is not (9)], it is advisable to investigate such details in a number of different species. In addition, a broader range of models should be employed: models based on different risk factors (distinct experimental manipulations) should be simultaneously tested, perhaps within the same experiment or through a coordinated multicenter effort, with the goal of approaching the diversity of the clinical population and potentially improving the predictive power of pre-clinical studies. Clearly, this strategy would require larger cohort sizes and strict multicenter coordination with regards to experimental design and outcome measures, the latter of which should be uniformly applied and readily comparable to outcome measures in clinical studies.

Animal models present an opportunity to examine in detail factors that might influence the therapeutic response to CBD in the clinic, as well as the physiological and neurobiological underpinnings of this response. In the case of cannabinoids or pharmacological manipulation of the endocannabinoid system as possible means to treat schizophrenia, studies on animal models indicate that CBD may be a useful pharmacotherapy, and suggest that pharmacotherapies that modulate anandamide and 2-AG levels should be developed and explored. Finally, studies on animal models point to interactions between cannabinoid receptors, the 5-HT1A/7 receptor, and the TRPV1 receptor that should be explored as possible targets for pharmacotherapy.

Author Contributions

KH: conceptualized, researched, and wrote the present article.

Conflict of Interest

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

References