The antipsychotic drugs remain the most effective form of pharmacotherapy for schizophrenia spectrum disorders. These compounds, which were initially developed in the 1950s (Shen, 1999; Seeman, 2021), are used widely in the field of psychiatry to treat a range of psychotic disorders, and have expanded in use beyond their original indications to include other psychiatric conditions and patient populations (Hershenberg et al., 2014; Vasudev et al., 2018; Del Casale et al., 2019; Mela et al., 2020; Ameer and Saadabadi, 2021; Lian et al., 2021). The antipsychotics are traditionally classified into at least two categories, which include the earlier, first-generation drugs, and the more recent second-generation compounds [some consider the dopamine D2 receptor partial agonists to represent a third-generation (Mailman and Murthy, 2010)]. While the second-generation drugs have significantly displaced the first-generation ones in clinical use, due to their lower incidence of neurological side-effects (Rummel-Kluge et al., 2010), the second-generation antipsychotics are associated with their own side-effect profiles, which commonly include metabolic dysregulation. These metabolic side-effects result in increased risk for developing cardiometabolic disorders with chronic use (Newcomer, 2007; Kessing et al., 2010; Deng, 2013; Tse et al., 2014; Bozymski et al., 2018; Kim et al., 2018; Yuen et al., 2018; Tumiel et al., 2019), which is a particular concern as many idiopathic psychotic disorders require lifelong treatment with these medications (Lindenmayer and Khan, 2004).

At present, the biochemical pathways that underlie the metabolic side-effects of the second-generation antipsychotics are not fully established. It appears likely that the metabolic side-effects are unrelated to the primary mechanism of clinical efficacy of these drugs (Reynolds and Kirk, 2010), which is antagonism of mesolimbic dopamine D2 receptors. Instead, a wide number of non-dopaminergic receptors and other physiological substrates are posited to play key roles in the metabolic side-effects of the second-generation antipsychotics, including both central and peripheral mechanisms (De Hert et al., 2011; Reynolds and McGowan, 2017; Bush et al., 2018; Kowalchuk et al., 2019; Marteene et al., 2019). In this regard, preclinical rodent models are especially helpful in advancing our understanding of the biochemical pathways involved, as they provide the opportunity for invasive experimental procedures, direct experimental control, and are not prone to the multifactorial causes of metabolic dysregulation observed in antipsychotic-treated patients, which include diet, exercise and non-prescribed substance use (Laursen et al., 2012; Marteene et al., 2019).

Rodent models of antipsychotic-induced side-effects reliably demonstrate metabolic changes that parallel those observed in humans, including glucose dysregulation, insulin resistance and weight gain (Baptista et al., 1999; Baptista et al., 2004; Girault et al., 2012; Weston-Green et al., 2013; Ersland et al., 2015; Babic et al., 2018; Shamshoum et al., 2019; Lian and Deng, 2020; Ashraf et al., 2021), using homologous techniques such as the glucose tolerance test (GTT) (Eyth et al., 2021). A number of these preclinical studies provide strong evidence that the peripheral catecholamines norepinephrine and epinephrine play an important role in regulating these metabolic side-effects (Savoy et al., 2010; Nagata et al., 2016; Nagata et al., 2018), and are consistent with clinical studies in antipsychotic-treated patients (Brown et al., 1997; Spivak et al., 1998; Elman et al., 1999). Given the potent control that peripheral catecholamines exert over glucose regulation and insulin release (Cryer, 1993; Barth et al., 2007), they represent a viable substrate for further, translational study. Recently, we demonstrated that acute treatment with different antipsychotic drugs caused substantial increases in levels of peripheral norepinephrine, epinephrine and dopamine, and these effects were greatest by the drugs with the highest metabolic liability (Boyda et al., 2020). However, metabolic indices were not assessed, and so the link between elevated catecholamines and metabolic dysregulation was not confirmed. Therefore, the goal of the present study was to use a preclinical model to treat rats with multiple different antipsychotic drugs, across a range of doses, and concurrently measure the effects on both peripheral catecholamine levels and glucose dysregulation, using the GTT. This would potentially allow us to confirm that glucose intolerance was associated with elevated catecholamine levels, as well as determine the strength of that association for each of the individual catecholamines involved.

The goal of the current study was to determine if treatment with antipsychotic drugs resulted in elevation of peripheral catecholamines concurrently with metabolic dysregulation, measured by glucose intolerance. Our overall findings strongly confirmed this hypothesis. Consistent with prior preclinical observations, the four different antipsychotic drugs used in the present study all caused potent glucose intolerance, and this was largely dose-dependent. Reflecting the pharmacological validity of the model, the effects on glucose intolerance paralleled those seen in the clinic, whereby the smallest metabolic effects were noted with haloperidol, and the largest with clozapine. In the same animals, blood that was collected at three separate timepoints during the GTT confirmed large increases in peripheral catecholamines, which was statistically significant for the three second-generation drugs at the higher doses. Importantly, there were strong, highly significant associations between levels of norepinephrine/epinephrine and elevated plasma glucose levels for most of the antipsychotics, as well as for dopamine with clozapine. Mediation analyses indicated that most of these correlations represented intervening effects between the drug treatment and glucose intolerance, whereby a large proportion of the effects of drug dose on glucose intolerance were directly mediated through the increases in catecholamine concentrations.

These findings therefore provide additional evidence that peripheral catecholamines may play a key role in the acute metabolic effects of antipsychotic drugs. The metabolic side-effects of antipsychotic drugs in clinical populations are now well established (Allison et al., 1999; McEvoy et al., 2005; Newcomer, 2007; Kessing et al., 2010; Mitchell et al., 2013; Kim et al., 2017; Tumiel et al., 2019; Yuen et al., 2021a), and animal studies have faithfully modelled these many of these effects, using comparable techniques that are commonly used in humans, including the GTT (Monzillo and Hamdy, 2003). We and many other research groups (Smith et al., 2008; Muller et al., 2010; Boyda et al., 2013a; Boyda et al., 2014; Evers et al., 2017; Lord et al., 2017; Stefanidis et al., 2017; Oxenkrug and Summergrad, 2020; Sylvester et al., 2020; Shamshoum et al., 2021) routinely observe that antipsychotics generate acute glucose intolerance of an equivalent magnitude to that noted in human studies (Albaugh et al., 2011) without a change in factors such as food intake or obesity, indicating a direct effect of the drugs on glucose dysregulation. Furthermore, previous experiments in our laboratory have reported that these effects on glucose intolerance remained stable with repeated, chronic treatment over months (Boyda et al., 2012; Boyda et al., 2014), comparable to the human condition. The drug-specific degree of glucose intolerance exhibited by the current four antipsychotic drugs is homologous with the degree of metabolic dysregulation noted in patients, which typically follows the sequence: clozapine > olanzapine ≥ risperidone > haloperidol (Tandon and Halbreich, 2003; Newcomer, 2005). Our observations noted increased AUC glucose levels of +130%, +96%, +42% and +18% with the highest doses of each of these drugs, respectively. The present findings therefore suggest that these acute metabolic effects may provide a meaningfully predictive model for further understanding the underlying physiology.

In parallel to glucose intolerance, most doses of drugs also caused large increases in the peripheral catecholamines, with the greatest increases (both in absolute and relative terms) in norepinephrine, followed by epinephrine and then dopamine. The pattern of elevated catecholamines was slightly different to that of the IGTT, with the largest changes caused by clozapine, followed by risperidone (rather than olanzapine), olanzapine and then haloperidol. However, it should be noted that the increases in norepinephrine were large even for haloperidol, whereby all doses approximately doubled norepinephrine levels. These effects of the antipsychotics on peripheral catecholamine levels recapitulate the findings from our recent study (Boyda et al., 2020), where we measured catecholamine levels with the same drugs, but with fewer doses and without any measure of metabolic dysregulation. This suggests that these effects are stable and reliable, and also consistent with a limited number of other studies in the preclinical literature which have observed the effects of individual antipsychotics on specific catecholamines, including one which noted that intravenous treatment with the first-generation drug chlorpromazine caused an increase of both norepinephrine and epinephrine (Sato et al., 1988). Meanwhile, indirect evidence for increased catecholamines was evident in a study where hyperglycemia, caused by treatment with clozapine and chlorpromazine, was mitigated by co-treatment with the ganglionic blocker hexamethonium, which presumably worked by preventing increases in peripheral norepinephrine and epinephrine (Savoy et al., 2010) - although these were not actually measured in the study. Human clinical findings also indicate that both norepinephrine and epinephrine are elevated by antipsychotic drugs with higher metabolic liability (Green et al., 1993; Schulz et al., 1996; Brown et al., 1997; Elman et al., 1999). For instance, two well-controlled studies observed that levels of peripheral norepinephrine were three times greater in patients treated with clozapine than either of the first-generation drugs fluphenazine or haloperidol (Brown et al., 1997; Elman et al., 1999). However, to our knowledge, clinical studies have yet to determine a link between peripheral catecholamine levels and metabolic dysregulation.

It is therefore apposite that in the present study, there were strong and highly significant correlations between glucose levels and norepinephrine/epinephrine concentrations for all four antipsychotics. While associations clearly do not determine causality, the consistent and robust correlations for all four drugs between glucose intolerance and norepinephrine/epinephrine levels—as well as the results of the mediation analyses—provide data to suggest that further study of the link between these two physiological pathways is warranted. Both norepinephrine and epinephrine serve as hormones that are highly active in the regulation of many metabolic processes, and are therefore strong candidates—a priori—to examine with regards to the metabolic side-effects of antipsychotics. Pertaining to glycemic control, both catecholamines play an essential role in glycogenolysis, hepatic gluconeogenesis, glucagon secretion and insulin release (Barth et al., 2007; Boyda et al., 2013b). Elevated concentrations of both norepinephrine and epinephrine might therefore be expected to contribute in manifold ways to excessively elevated glucose levels, and will require future study—although some potential candidate substrates of these catecholamines, such as glucagon, are already being confirmed as presumed downstream mediators of antipsychotic-induced glucose intolerance (Medak et al., 2019; Shamshoum et al., 2021). Physiologically, the relative importance of each of the individual catecholamines towards the drug-induced glucose dysregulation will require more targeted evaluation, involving both pharmacological and anatomical techniques [such as denervation of the pancreas (Nilsson et al., 2001)]. With regards to the former, in a recent study we sought to understand which of the adrenoceptors may be responsible for mediating the effects of putative increases in peripheral catecholamines on whole-body insulin resistance (Yuen et al., 2021b) following treatment with clozapine. Using selective antagonists at α1 (prazosin), α2 (idazoxan), β1 (atenolol) and β2 (butoxamine) adrenoceptors, we observed that both atenolol and butoxamine could significantly reverse clozapine-induced insulin resistance. Additionally, the ganglionic blocker mecamylamine was able to partially reverse the insulin resistance caused by treatment with clozapine. Mecamylamine is a potent non-selective antagonist at nicotinic receptors, which blocks the effects of acetylcholine released by preganglionic nerve fibers, thereby preventing downstream release of norepinephrine from post-ganglionic fibers and epinephrine from the adrenal medulla. Less well understood is the role of peripheral dopamine as a potential mediator of metabolic side-effects. In the body’s periphery, dopamine can be released from three main sources, including neuronal fibers, the adrenal medulla, and neuroendocrine cells (Pearse, 1969). It exerts its hormonal effects across a diverse range of substrates. There is considerable evidence to show that peripheral dopamine regulates body weight and glucose homeostasis via insulin release (Rubí and Maechler, 2010). Studies have demonstrated that dopamine directly inhibits the secretory response of β-cells in the pancreas, and dopamine D2 receptors are expressed in both rodent and human islets, where the receptors regulate inhibition of the secretory response (Rubí et al., 2005).

While the present study is hopefully informative, and provides additional novel evidence to support a key role for the peripheral catecholamines on antipsychotic metabolic side-effects, there are a number of limitations. Firstly, we studied four antipsychotic drugs, but there are a considerably greater number of antipsychotic drugs than this in clinical use. While the full range of metabolic liability was represented with the current drugs, spanning from haloperidol to clozapine (Newcomer, 2005), it will be important to determine how other antipsychotics with known metabolic side-effects (Barr et al., 2006; Houseknecht et al., 2007) work in this model. It will be of particular interest to assess the impact of third generation antipsychotic drugs, which exert partial agonist activity at the dopamine D2 receptor—but also serve as functional antagonists when the endogenous ligand is at a higher concentration (Mailman and Murthy, 2010). Given that peripheral dopamine can regulate insulin release in the pancreas (Rubí and Maechler, 2010), the impact of this partial agonism/antagonism under basal and stimulated levels of dopamine before and following antipsychotic treatment will need further study. It will also be interesting to study a drug such as aripiprazole in combination with clozapine, which has been shown to reduce the metabolic side-effects of the latter (Fan et al., 2013). Whether aripiprazole’s partial agonist effect at D2 receptors at peripheral sites, such as the pancreas, might serve to counter significant elevations in catecholamines (including dopamine) caused by clozapine will be feasible to study. A second major limitation of the present study is, the doses of drugs that were used in the present study may be considered high, in relation to doses needed to achieve occupancy of mesolimbic dopamine receptors at the same degree as observed with antipsychotic effects in humans (Kapur et al., 2003). This could theoretically call into question the external validity and translational value of the model. While this remains uncertain at present, there is considerable evidence that the physiological substrates and pathways that mediate the therapeutic and metabolic side-effects of antipsychotic drugs are different (Reynolds, 2012), and so may not be equally as sensitive to drug effects in different species. It may be that in rats, comparably higher doses are needed to observe metabolic effects of an equivalent magnitude to those seen in humans. For example, Albaugh et al. (2011) acutely administered a therapeutic dose of olanzapine to healthy human volunteers and observed an increase in glucose intolerance (measured by the AUC) of 42%. In our laboratory, acute treatment with olanzapine in rats requires a dose of at least 10 mg/kg to observe a comparable increase in glucose intolerance of 39% (Wu et al., 2014); this dose is notably greater than that needed to achieve species-equivalent “therapeutic” D2 receptor occupancy. Further study will therefore be necessary to determine the relationship between drug dosing for side-effects vs efficacy in rodent models to better understand the relevance to the human condition. A third important limitation of our study is that we only used female rats. The preclinical literature on antipsychotic drug-induced metabolic side-effects reveals a complex pattern when comparing between males and females, in which effects vary depending on both the specific drug and metabolic parameter being studied (see Castellani et al. (2019) for a comprehensive review). With regards to glucose intolerance, it has been claimed that female rats exhibit less reliable acute glucose dysregulation than males (Castellani et al., 2019). For the specific purposes of studying acute mechanisms associated with drug effects, the use of one sex alone may be justified in the earlier stages of exploratory research. However, it will be important in the future to compare the effects of antipsychotic drugs on peripheral catecholamine levels and both males and females to increase the generalizability to the clinical population. A fourth limitation of the present study is the administration of antipsychotic drugs acutely rather than chronically, as per the patient population. It is entirely possible that the current increases in peripheral catecholamines may diminish or disappear with repeated treatment. In our laboratory, we have observed consistent effects of olanzapine with chronic treatment over 9–10 weeks (Boyda et al., 2012; Boyda et al., 2014), in a manner whereby the magnitude of glucose tolerance remains stable and does not increase or diminish after weeks or months. It will therefore be important to determine if peripheral catecholamines follow this same trajectory. Interestingly, we did observe that intermittent treatment (vs continuous treatment) was associated with a “sensitization” of metabolic effects (Boyda et al., 2012), which is consistent with catecholamines’ role in central sensitization (Featherstone et al., 2007). Clinical studies that have noted elevated peripheral epinephrine and norepinephrine in patients treated with second generation antipsychotics did so in individuals who had received chronic treatment (Brown et al., 1997; Spivak et al., 1998; Elman et al., 2002).

In conclusion, the current experiments provide further evidence for the potential role of peripheral catecholamines in the metabolic side-effects of antipsychotic drugs. Acute treatment with four different antipsychotics of varying metabolic liability in humans induced similar changes in glucose intolerance, and these metabolic changes were strongly associated with increases in peripheral concentrations of norepinephrine and epinephrine. These findings indicate that further research into the links between glucose dysregulation and elevated catecholamines may provide a productive avenue for future research, while acute measurement of peripheral catecholamines may have predictive potential for antipsychotic-induced metabolic dysregulation, when assessed using procedures such as the GTT.