Review
Molecular pathophysiology of metabolic effects of antipsychotic medications Jacob S. Ballon1,2*, Utpal Pajvani3,4*, Zachary Freyberg1,5*, Rudolph L. Leibel4,6, and Jeffrey A. Lieberman1,2 1
Department of Psychiatry, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA Division of Experimental Therapeutics, New York State Psychiatric Institute, New York, NY 10032, USA 3 Department of Medicine, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA 4 Naomi Berrie Diabetes Institute, New York, NY 10032, USA 5 Division of Molecular Therapeutics, New York State Psychiatric Institute, New York, NY 10032, USA 6 Department of Pediatrics, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA 2
Antipsychotic medications are associated with major metabolic changes that contribute to medical morbidity and a significantly shortened life span. The mechanisms for these changes provide us with a broader understanding of central nervous and peripheral organ-mediated metabolic regulation. This paper reviews an extensive literature regarding putative mechanisms for effects of antipsychotic medications on weight regulation and glucose homeostasis as well as potential inherent metabolic risks of schizophrenia itself. We present a model suggesting that peripheral antipsychotic targets play a critical role in drug-induced weight gain and diabetes. We propose that a better understanding of these mechanisms will be crucial to developing improved treatments for serious mental illnesses as well as providing potentially novel therapeutic targets of metabolic disorders including diabetes. Introduction Schizophrenia affects 1% of the world’s population. It is manifest by positive, negative, and cognitive symptoms that typically emerge in adolescence and early adulthood. People with schizophrenia have a 20% shorter life expectancy than the general population, with cardiovascular disease as the leading cause of death [1]. Schizophreniainduced cardiovascular disease is in part attributable to higher rates of smoking but is likely to be further exacerbated by glucose and lipid abnormalities inherent to the pathophysiology of schizophrenia [2]. Antipsychotic drugs (APDs) are the primary medications used to treat schizophrenia and are increasingly used as adjunctive treatments for mood disorders. Second-generation APDs, often termed atypical antipsychotics (AAPs), are the current standard-of-care for treatment of schizophrenia Corresponding author: Lieberman, J.A. ([email protected]). Keywords: schizophrenia; antipsychotic drugs; insulin resistance; weight regulation; diabetes. * These authors are joint first authors. 1043-2760/ ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tem.2014.07.004
because they are generally better tolerated than earlier medications (Box 1) [3]. Improved tolerability, however, comes with significant metabolic side effects, including obesity, type 2 diabetes (T2D), and dyslipidemia, that contribute to overall morbidity and mortality [4]. Nevertheless, no specific mechanisms have yet been identified to account for the relationship between schizophrenia and its metabolic comorbidities, nor the effects/interactions of antipsychotic medications in this regard. Although APDinduced adiposity may lead to insulin resistance, there is emerging evidence that weight gain and T2D may also be independent consequences of APD therapy [5]. Schizophrenia and abnormal glucose metabolism For decades, there has been literature documenting an association between metabolic disease and schizophrenia [6] suggesting a metabolic phenotype intrinsic to schizophrenia. Before the APD era, cohort studies noted increased incidence of abnormal glucose metabolism in people with schizophrenia [7]. These observations corroborated cross-sectional results demonstrating that the prevalence of diabetes was greater in schizophrenic patients compared to the general population [8]. More recent studies confirmed these findings, and have shown impaired glucose tolerance in drug-naı¨ve schizophrenic patients, as compared to healthy controls [9]. Impaired glucose tolerance has also been demonstrated in nonpsychotic, firstdegree relatives of schizophrenic patients, further indicating a heritable phenotype that tracks with the risk of psychosis but is independent of the actual development of a psychotic disorder [10]. The risk of metabolic abnormalities further increases significantly with duration of illness with those who have chronic illness showing increased rates of metabolic dysfunction compared to firstepisode and drug-naı¨ve patients [11]. Recent genome-wide association studies (GWASs) have provided candidates for the association between schizophrenia and T2D [12]. AKT1, the gene encoding for the serine–threonine protein kinase AKT1, has been identified as a candidate gene for schizophrenia susceptibility in several global communities, across multiple studies [13]. Trends in Endocrinology and Metabolism, November 2014, Vol. 25, No. 11
593
Review Box 1. Antipsychotic drugs (APDs) Antipsychotic medications have been available for use in the United States since the FDA approval of chlorpromazine in 1954. As these new medications came to the market, they were initially categorized based on the potency of antagonism of the D2 dopamine receptor. High-potency agents, such as haloperidol and fluphenazine, and low-potency agents, like chlorpromazine and thioridazine, were in wide usage until the release of the second-generation drugs, or AAPs, which became available in the 1990s. The neuropsychiatric mechanism of action of AAPs is also through antagonism of D2 receptors, although adjunctive neurotransmitter effects, including serotonergic and histaminergic, have been related to differentiation in side effect profiles and efficacy between the two classes [67]. The primary side effect concern for the first-generation or conventional antipsychotics focused on neurological problems. Blocking D2 receptors in the nigrostriatal pathway of the brain is associated with Parkinsonian movements (pill-rolling tremor, bradykinesia, masked facies). Tardive dyskinesia, a permanent tremor which may present after several months of antipsychotic exposure, is also seen in greater numbers in high-potency conventional agents [68]. The higher the potency of the D2 blockade, the greater the likelihood of these neurological side effects [69]. Like the newer medications, low-potency conventional antipsychotics, particularly chlorpromazine, were associated with weight gain and insulin resistance [70]. The risk for weight gain from antipsychotic medications is inversely related to the potency at D2 [71]. Despite the purported differences between conventional and atypical antipsychotics, all antipsychotic medications have been associated with both neurological and metabolic side effects [72].
AKT1 is also recognized as a key mediator of insulin signaling and glucose metabolism, making it an attractive molecular interface for neuropsychiatric and metabolic consequences of schizophrenia. Loss-of-function AKT1 mutations result in glucose intolerance, probably as a result of reduced insulin secretion and/or impaired action in liver, muscle, and hypothalamus [14]. In schizophrenia, lower levels of AKT1 expression have been found in lymphocytes and frontal cortex of brain [13]. This reduction was associated with decreased phosphorylation of AKT1 substrates, including glycogen synthase kinase 3 beta (GSK3b), suggesting a functional defect in insulin signaling in the brain [15]. Conversely, treatment of mice with the APD haloperidol increased AKT signaling in the CNS – whether this holds true for peripheral organs, or in patients with schizophrenia, remains to be determined [15]. It is also unclear whether in schizophrenia, insulin/ AKT signaling is reduced in hypothalamic regions that mediate satiety and systemic insulin sensitivity. APDs induce metabolic dysfunction APDs are associated with substantial risk for adverse metabolic conditions. The high prevalence and poor tolerability of these metabolic side effects frequently leads to suboptimal medication compliance and high rates of APD discontinuation, resulting in symptomatic relapse and poor long-term patient outcomes. Concerns regarding the metabolic side effects of APDs were greatly increased following the introduction of the ‘second generation’ or ‘atypical’ drugs, which have come to dominate the APD market. Interestingly, among this group of medications, clozapine and olanzapine simultaneously cause the most metabolic dysfunction and weight 594
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
gain of all the APDs, while demonstrating the greatest clinical efficacy for core psychotic symptoms [3]. Because APDs have relatively well defined nominal molecular targets through which they convey their therapeutic effects, a framework within which to study cellular mechanisms of APD action centrally and peripherally has emerged (Figure 1). In fact, the same neurotransmitter signaling networks targeted by APDs have also been implicated in metabolic dysregulation and obesity, highlighting potential shared molecular mechanisms of schizophrenia and obesity/T2D. Antipsychotic-induced obesity is primarily the result of altered energy intake Weight gain only occurs when energy intake exceeds expenditure, both of which can be affected by pathologic states and pharmacotherapy. When access to food is not limited, excess caloric consumption is the principal driver of positive energy balance and consequent deposition of lean tissue and fat. Additive or compensatory effects on energy expenditure may follow and affect rates of weight gain. APDs have been hypothesized to affect both energy intake and expenditure (Figure 2); however, in humans, weight gain with APDs has been more consistently associated with increased food consumption [16]. A well-designed crossover study by Fountaine et al. showed that olanzapine treatment resulted in an estimated 345 kcal/day (18%) excess energy intake in 30 healthy male volunteers and 2.62 kg increased body weight (over 15 days), with greater food intake seen at all meals [17]. This acute perturbation may be an underestimate of what has long been recognized in clinical practice [16] – increased caloric intake and weight gain effects may in fact be higher in APD-treated schizophrenia patients than in healthy controls, and may be associated with increased food craving and binge eating [18]. For example, in olanzapine-naı¨ve adolescents with schizophrenia, 4 weeks of olanzapine treatment was associated with an estimated increase of energy intake of 589 kcal/day (28%), leading to significant increases in abdominal circumference and body weight [19]. In the study by Fountaine et al. cited above, an excess of 345 kcal/day for 15 days would equate to approximately 5175 total excess kcal consumed over the trial length. Thus, the net weight gain cannot be accounted for solely by caloric excess, because 2.62 kg of increased body weight would equate to approximately 13 100 additional kcal (assuming 5000 kcal/kg mixed tissue). This ‘back-of-the-envelope’ calculation suggests two possible hypotheses – either APDinduced weight gain is a combination of increased food intake as well as decreased energy expenditure, and/or the caloric excess is accompanied by a large increase in body water, which may account for a substantial fraction (>50%) of the acute weight gain in short-term studies. The hypothesis that APDs reduce resting energy expenditure (REE) has been tested with mixed results. Although multiple groups have reported minimal to no change in energy expenditure [20,21], Fountaine et al. actually found that olanzapine increased REE (+113 kcal/24 h [17]. This is not a wholly unexpected result – energy intake and expenditure are not independent variables. In weight-reduced humans, energy
Review
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
Brain Global behavioral control • Hedonic response • Decision making
Hypothalamus Appete regulaon/saety (arcuate nucleus) • • • • •
Lepn/ghrelin Insulin/IRS signaling Histamine (H1R antagonism) -> AMPK Dopamine (D2R antagonism) -> AKT NPY/POMC
Fat Glucose uptake • Insulin-responsive (Glut4-mediated)
Muscle
Adipokine acon
Glucose uptake
• Lepn • Adiponecn
• Insulin-responsive (Glut4-mediated) • Contracon-dependent (AMPK-dependent)
Inflammatory state • TNF-alpha/IL-6 and others
Liver Hepac glucose producon • Hormone (insulin/glucagon)-regulated • Serotonin/SERT
Pancreas Insulin secreon • Dopamine co-secreon • Serotonin/SERT-regulated
Glucagon secreon
Hepac de novo lipogenesis • Insulin-regulated • Nutrient (mTorc1)-regulated
Gut Insulin/glucose regulaon • • • •
GLP-1 GIP Serotonin Dopamine TRENDS in Endocrinology & Metabolism
Figure 1. Pharmacological targets for metabolic dysfunction throughout the body. Antipsychotic medications are active throughout the body, including in pathways involved in metabolic processes. Based on the current body of evidence, we hypothesize that antipsychotic drug (APD)-induced metabolic disturbances arise from a combination of drug actions on targets in both the central nervous system (CNS) and peripheral organs. Within the CNS an important series of targets include hypothalamic nuclei. Primary peripheral targets of APD-induced metabolic dysfunction include pancreas, liver, muscle, and adipose tissue. Both central and peripheral targets are likely to be linked through a series of convergent molecular pathways including dopaminergic, serotoninergic, histaminergic, and adipokine signaling. Through action at receptors for these transmitters, APDs have synergistic effects to amplify metabolic risk. Therefore, based on this framework, we suggest a model whereby (i) APDs act directly on insulin-sensitive peripheral organs including pancreas, liver, muscle, and adipose tissue as well as the hypothalamic–liver circuit, which leads to peripheral insulin resistance; (ii) APDs act directly upon hypothalamic nuclei responsible for central regulation of food intake and metabolism; and (iii) these first two processes increase adiposity, which further exacerbates overall insulin resistance. Abbreviations: AKT, a serine–threonine protein kinase; AMPK, AMP-activated protein kinase; D2R, dopamine D2-like receptor; GIP, gastric inhibitory polypeptide; GLP-1, glucagon-like peptide-1; Glut4, glucose transporter type 4; H1R, histamine receptor 1; IL-6, interleukin-6; mTorc1, mammalian target of rapamycin complex 1; NPY, neuropeptide Y; POMC, proopiomelanocortin; SERT, serotonin transporter; TNF, tumor necrosis factor.
expenditure is reduced more than predicted by the decrease in somatic mass [22]. Conversely, short-term overfeeding results in proportionately higher energy expenditure. This may account for the greater energy expenditure seen in the Fountaine study. The range of acute alterations in energy expenditure appears to be 10–15% in either direction [23], suggesting that prolonged or excessive energy intake can overcome this physiologic compensation, leading to weight gain. In summary, APD-induced weight gain results from a combination of excess caloric intake, coupled with increased water weight, but is unlikely to result from a significant decrease in metabolic rate. Molecular mechanisms of antipsychotic-induced weight gain Despite years of study, the molecular mechanisms underlying APD-induced weight gain remain unclear, in part because animal models have only partially recapitulated results seen in human studies and in clinical practice. For
example, most [24–26] but not all [27] studies have reported increased food intake and body weight in APDtreated mice or rats. Outlying studies may result from the known sedative properties of APDs, which leads to issues with dose selection. Upon titration to doses that result in plasma drug concentrations comparable to those at steady state in schizophrenic patients [28], rodents are less active [29], which may artifactually reduce food intake and body weight gain as compared to animals treated with lower APD doses [30]. In addition, food intake and body weight gain in response to APDs are likely to be gender-, strain-, and diet-specific [31], possibly leading to divergent results. Nonetheless, rodent studies have suggested plausible hypotheses for the molecular mechanisms by which these drugs achieve their metabolic effects. For instance, one well-conducted study revealed that olanzapine-induced weight gain was mitigated by pair-feeding to placebo-treated rats, indicating that APD-induced weight gain is primarily the result of increased food intake [32]. In this 595
Review
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
Anpsychoc exposure
Current
Revised R
Increased food intake
Increased food intake
Obesity
Obesity
Insulin resistance Beta cell dysfuncon
Insulin nsulin istance resistance
Beta cell dysfuncon
T2D
T2D
TRENDS in Endocrinology & Metabolism
Figure 2. Proposed pathways for antipsychotic drug (APD)-induced metabolic dysfunction. Multiple converging pathways in the periphery mediate APD-induced metabolic dysregulation. Our current understanding of this phenomenon is based on the assumption that the increased food intake resulting from APDs is the principal driver of obesity and insulin resistance. However, recent evidence suggests that more than one pathway may be responsible, and that the obesity and insulin resistance may develop from different sources. This is consistent with mounting data showing that not all people who are obese are prone to develop type 2 diabetes (T2D), and vice versa.
study, the authors demonstrated increased orexigenic [neuropeptide Y (NPY), agouti related protein (AgRP)] decreased anorexigenic [proopiomelanocortin and (POMC)] patterns of neuropeptide expression in the hypothalami of rats treated with olanzapine, as compared to placebo, regardless of whether the animals were pair-fed or not [32]. This result is consistent with the primary effect of APDs on hypothalamic gene expression, with secondary effects on food intake [32]. Energy intake is homeostatically regulated by neuropeptide expression in the arcuate and other hypothalamic nuclei [33], as well as ‘higher’ brain regions mediating hedonic responses and decision-making [34]. An important candidate molecule for conferring APD-induced weight gain is the melanocortin receptor MC4R4. MC4R4 plays an important role in hypothalamic control of both food intake and energy expenditure; mutations reducing the functional activity of this receptor result in severe obesity in mice and humans [35]. An association has been shown between common DNA sequence variants in the vicinity of MC4R4 with APD-induced weight gain in a recent trial combining a GWAS and a prospective cohort analysis [36]. In recent years, multiple groups have postulated that APDs exert their orexigenic effects through effects on neurotransmitter signaling. Because this has been thoroughly reviewed by others [37], we briefly highlight the most compelling evidence for this hypothesis here. 596
Histamine hypothesis Histamine (HA) is a monoamine that binds four HA receptors (H1R–H4R), three of which (H1R–H3R) are expressed in the brain. Histaminergic neurons are abundant in the posterior hypothalamus and project throughout the brain, leading to speculation that HA may mediate homeostatic and hedonic aspects of feeding. Intracerebroventricular (ICV) HA infusion in rodents reduces food intake, probably mediated by H1R, as specific H1R agonists recapitulate the anorectic effect of HA, and H1R knockout mice consume more food after stimulation with the orexigenic neuropeptide NPY [38]. Consistent with these observations, APDs with the greatest antagonist H1R affinity, clozapine and olanzapine, preferentially stimulate hypothalamic AMPactivated protein kinase (AMPK) activity in proportion to their H1R antagonism [39]. AMPK activity is negatively regulated by the anorexigenic hormone leptin and leptin receptor signaling [40]; APD antagonism of this effect could contribute to APD-induced weight gain. In support of this theory, clozapine does not stimulate hypothalamic AMPK phosphorylation and activity in H1R-knockout mice [39]; effects on food intake or weight gain in these mice were not reported – this was likely to be related to the difficulty in modeling APD action in rodent models, as discussed above. Serotonin hypothesis Serotonergic (5-HT) neurotransmission is a common target of psychotropic drugs, particularly for antidepressants and atypical APDs. Notably, 5-HT agents have a long history as anorexiants, depending on their receptor specificity and affinity. The serotonin secretagogue and re-uptake inhibitor fenfluramine was widely prescribed for obesity treatment before being withdrawn in 1997 as a result of the risk of pulmonary hypertension. Lorcaserin, a selective 5HT2c receptor agonist, was recently approved for short-term treatment of obesity [41]. Conversely, mice with a hypomorphic 5HT2c receptor mutation are hyperphagic and insulin-resistant [42]. Second-generation APDs have variable activity at presynaptic 5-HT receptors in nigrostriatal neurons. 5HT2a and 5HT2c receptor single-nucleotide polymorphisms (SNPs) have also been associated with obesity, glucose intolerance, and susceptibility to APDinduced weight gain [43]. These observations led to the suggestion that APD blockade of hypothalamic 5HT2a and 5HT2c receptors causes drug-induced weight gain, but this hypothesis has not been rigorously tested. Obesity-independent effects of APDs on glucose metabolism Temporal relationships of hyperglycemia and new-onset diabetes to initiation of APD treatment and their reversibility on drug withdrawal suggest a causal relationship. Most new-onset diabetes occurs within the first 6 months after drug initiation, but increased risk of hyperglycemia persists with extended treatment. Hyperglycemia may be quite severe – presentation in diabetic ketoacidosis, although rare, has been observed with most APDs and may represent the first clinical manifestation of APD-induced T2D or even incipient type 1 diabetes (T1D). The most parsimonious explanation for this increased diabetes risk would be the APD-associated weight gain;
Review
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
Table 1. Weight gain results from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) [3] Weight change from baseline (kg) Weight gain >7% from baseline Hemoglobin A1c % change from baseline
Olanzapine 4.3 (0.4) 92/307 (30%) 0.41 (0.09)
Quetiapine 0.5 (0.4) 49/305 (16%) 0.05 (0.05)
however, the mechanisms driving the obesogenic and diabetogenic effects of APD appear to be partially independent. Clozapine has been shown to cause increased insulin secretion independent of body mass [44]. Even relatively weight-neutral APDs show higher than expected rates of hyperglycemia (Table 1), although the risk increased with the more obesogenic drugs [3]. This association does not preclude the possibility that increased adiposity with chronic treatment exacerbates early APD-induced insulin resistance and/or beta cell dysfunction, leading to greater hyperglycemia and worsening diabetes. Nevertheless, there is clear evidence that early metabolic dysfunction precedes significant weight gain. In a well-conducted study, 12 healthy volunteers (18–30 years) were administered olanzapine 10 mg/day or placebo for 3 days, evaluated for multiple metabolic parameters, then crossed-over to the other treatment after a sufficient drug washout period [45]. No significant differences were detected between placebo and olanzapine treatments in terms of body weight, but olanzapine treatment caused a marked deterioration (42% increase of glucose area under the curve) of oral glucose tolerance (OGTT). Interestingly, neither fasting insulin levels nor insulin concentration during the OGTT was increased in olanzapine-treated subjects, suggesting that a compensatory insulin response to hyperglycemia was absent in this short-term exposure experiment [45]. Plasma leptin levels were significantly increased with olanzapine treatment [45], which is suggestive of an increase in body fat and/or increased leptin production as a result of the anabolic effects on adipocytes. Many of these findings have been corroborated in a recently published inpatient study with healthy volunteers, where slightly longer olanzapine treatment (9 days) led to weight-independent impaired glucose homeostasis [46]. In this study, olanzapine raised fasting insulin levels, suggesting increased insulin resistance, but both olanzapine and aripiprazole (which is considered weight-neutral) increased post-prandial insulin levels and reduced glucose disposal in hyperinsulinemic–euglycemic clamp studies independently of body weight effects when compared to placebo-treated volunteers [46]. These complementary studies confirm the weight-independent adverse effects on glucose homeostasis seen in animal models. However, the molecular mechanisms of APD-associated hyperglycemia remain as unclear as those underlying APD-associated weight gain. Further, it remains uncertain whether schizophrenia per se predisposes patients to respond differently to the metabolic effects of these agents. Molecular mechanisms of antipsychotic-induced hyperglycemia The hypothesis that APDs provoke weight gain-independent disturbances in glucose/insulin homeostasis has been tested in animals. In a seminal study, healthy rats treated
Risperidone 0.4 (0.4) 42/300 (14%) 0.08 (0.04)
Perphenazine 0.9 (0.5) 29/243 (12%) 0.10 (0.06)
Ziprasidone 0.7 (0.5) 12/161 (7%) 0.10 (0.14)
P
Molecular pathophysiology of metabolic effects of antipsychotic medications Jacob S. Ballon1,2*, Utpal Pajvani3,4*, Zachary Freyberg1,5*, Rudolph L. Leibel4,6, and Jeffrey A. Lieberman1,2 1
Department of Psychiatry, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA Division of Experimental Therapeutics, New York State Psychiatric Institute, New York, NY 10032, USA 3 Department of Medicine, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA 4 Naomi Berrie Diabetes Institute, New York, NY 10032, USA 5 Division of Molecular Therapeutics, New York State Psychiatric Institute, New York, NY 10032, USA 6 Department of Pediatrics, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA 2
Antipsychotic medications are associated with major metabolic changes that contribute to medical morbidity and a significantly shortened life span. The mechanisms for these changes provide us with a broader understanding of central nervous and peripheral organ-mediated metabolic regulation. This paper reviews an extensive literature regarding putative mechanisms for effects of antipsychotic medications on weight regulation and glucose homeostasis as well as potential inherent metabolic risks of schizophrenia itself. We present a model suggesting that peripheral antipsychotic targets play a critical role in drug-induced weight gain and diabetes. We propose that a better understanding of these mechanisms will be crucial to developing improved treatments for serious mental illnesses as well as providing potentially novel therapeutic targets of metabolic disorders including diabetes. Introduction Schizophrenia affects 1% of the world’s population. It is manifest by positive, negative, and cognitive symptoms that typically emerge in adolescence and early adulthood. People with schizophrenia have a 20% shorter life expectancy than the general population, with cardiovascular disease as the leading cause of death [1]. Schizophreniainduced cardiovascular disease is in part attributable to higher rates of smoking but is likely to be further exacerbated by glucose and lipid abnormalities inherent to the pathophysiology of schizophrenia [2]. Antipsychotic drugs (APDs) are the primary medications used to treat schizophrenia and are increasingly used as adjunctive treatments for mood disorders. Second-generation APDs, often termed atypical antipsychotics (AAPs), are the current standard-of-care for treatment of schizophrenia Corresponding author: Lieberman, J.A. ([email protected]). Keywords: schizophrenia; antipsychotic drugs; insulin resistance; weight regulation; diabetes. * These authors are joint first authors. 1043-2760/ ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tem.2014.07.004
because they are generally better tolerated than earlier medications (Box 1) [3]. Improved tolerability, however, comes with significant metabolic side effects, including obesity, type 2 diabetes (T2D), and dyslipidemia, that contribute to overall morbidity and mortality [4]. Nevertheless, no specific mechanisms have yet been identified to account for the relationship between schizophrenia and its metabolic comorbidities, nor the effects/interactions of antipsychotic medications in this regard. Although APDinduced adiposity may lead to insulin resistance, there is emerging evidence that weight gain and T2D may also be independent consequences of APD therapy [5]. Schizophrenia and abnormal glucose metabolism For decades, there has been literature documenting an association between metabolic disease and schizophrenia [6] suggesting a metabolic phenotype intrinsic to schizophrenia. Before the APD era, cohort studies noted increased incidence of abnormal glucose metabolism in people with schizophrenia [7]. These observations corroborated cross-sectional results demonstrating that the prevalence of diabetes was greater in schizophrenic patients compared to the general population [8]. More recent studies confirmed these findings, and have shown impaired glucose tolerance in drug-naı¨ve schizophrenic patients, as compared to healthy controls [9]. Impaired glucose tolerance has also been demonstrated in nonpsychotic, firstdegree relatives of schizophrenic patients, further indicating a heritable phenotype that tracks with the risk of psychosis but is independent of the actual development of a psychotic disorder [10]. The risk of metabolic abnormalities further increases significantly with duration of illness with those who have chronic illness showing increased rates of metabolic dysfunction compared to firstepisode and drug-naı¨ve patients [11]. Recent genome-wide association studies (GWASs) have provided candidates for the association between schizophrenia and T2D [12]. AKT1, the gene encoding for the serine–threonine protein kinase AKT1, has been identified as a candidate gene for schizophrenia susceptibility in several global communities, across multiple studies [13]. Trends in Endocrinology and Metabolism, November 2014, Vol. 25, No. 11
593
Review Box 1. Antipsychotic drugs (APDs) Antipsychotic medications have been available for use in the United States since the FDA approval of chlorpromazine in 1954. As these new medications came to the market, they were initially categorized based on the potency of antagonism of the D2 dopamine receptor. High-potency agents, such as haloperidol and fluphenazine, and low-potency agents, like chlorpromazine and thioridazine, were in wide usage until the release of the second-generation drugs, or AAPs, which became available in the 1990s. The neuropsychiatric mechanism of action of AAPs is also through antagonism of D2 receptors, although adjunctive neurotransmitter effects, including serotonergic and histaminergic, have been related to differentiation in side effect profiles and efficacy between the two classes [67]. The primary side effect concern for the first-generation or conventional antipsychotics focused on neurological problems. Blocking D2 receptors in the nigrostriatal pathway of the brain is associated with Parkinsonian movements (pill-rolling tremor, bradykinesia, masked facies). Tardive dyskinesia, a permanent tremor which may present after several months of antipsychotic exposure, is also seen in greater numbers in high-potency conventional agents [68]. The higher the potency of the D2 blockade, the greater the likelihood of these neurological side effects [69]. Like the newer medications, low-potency conventional antipsychotics, particularly chlorpromazine, were associated with weight gain and insulin resistance [70]. The risk for weight gain from antipsychotic medications is inversely related to the potency at D2 [71]. Despite the purported differences between conventional and atypical antipsychotics, all antipsychotic medications have been associated with both neurological and metabolic side effects [72].
AKT1 is also recognized as a key mediator of insulin signaling and glucose metabolism, making it an attractive molecular interface for neuropsychiatric and metabolic consequences of schizophrenia. Loss-of-function AKT1 mutations result in glucose intolerance, probably as a result of reduced insulin secretion and/or impaired action in liver, muscle, and hypothalamus [14]. In schizophrenia, lower levels of AKT1 expression have been found in lymphocytes and frontal cortex of brain [13]. This reduction was associated with decreased phosphorylation of AKT1 substrates, including glycogen synthase kinase 3 beta (GSK3b), suggesting a functional defect in insulin signaling in the brain [15]. Conversely, treatment of mice with the APD haloperidol increased AKT signaling in the CNS – whether this holds true for peripheral organs, or in patients with schizophrenia, remains to be determined [15]. It is also unclear whether in schizophrenia, insulin/ AKT signaling is reduced in hypothalamic regions that mediate satiety and systemic insulin sensitivity. APDs induce metabolic dysfunction APDs are associated with substantial risk for adverse metabolic conditions. The high prevalence and poor tolerability of these metabolic side effects frequently leads to suboptimal medication compliance and high rates of APD discontinuation, resulting in symptomatic relapse and poor long-term patient outcomes. Concerns regarding the metabolic side effects of APDs were greatly increased following the introduction of the ‘second generation’ or ‘atypical’ drugs, which have come to dominate the APD market. Interestingly, among this group of medications, clozapine and olanzapine simultaneously cause the most metabolic dysfunction and weight 594
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
gain of all the APDs, while demonstrating the greatest clinical efficacy for core psychotic symptoms [3]. Because APDs have relatively well defined nominal molecular targets through which they convey their therapeutic effects, a framework within which to study cellular mechanisms of APD action centrally and peripherally has emerged (Figure 1). In fact, the same neurotransmitter signaling networks targeted by APDs have also been implicated in metabolic dysregulation and obesity, highlighting potential shared molecular mechanisms of schizophrenia and obesity/T2D. Antipsychotic-induced obesity is primarily the result of altered energy intake Weight gain only occurs when energy intake exceeds expenditure, both of which can be affected by pathologic states and pharmacotherapy. When access to food is not limited, excess caloric consumption is the principal driver of positive energy balance and consequent deposition of lean tissue and fat. Additive or compensatory effects on energy expenditure may follow and affect rates of weight gain. APDs have been hypothesized to affect both energy intake and expenditure (Figure 2); however, in humans, weight gain with APDs has been more consistently associated with increased food consumption [16]. A well-designed crossover study by Fountaine et al. showed that olanzapine treatment resulted in an estimated 345 kcal/day (18%) excess energy intake in 30 healthy male volunteers and 2.62 kg increased body weight (over 15 days), with greater food intake seen at all meals [17]. This acute perturbation may be an underestimate of what has long been recognized in clinical practice [16] – increased caloric intake and weight gain effects may in fact be higher in APD-treated schizophrenia patients than in healthy controls, and may be associated with increased food craving and binge eating [18]. For example, in olanzapine-naı¨ve adolescents with schizophrenia, 4 weeks of olanzapine treatment was associated with an estimated increase of energy intake of 589 kcal/day (28%), leading to significant increases in abdominal circumference and body weight [19]. In the study by Fountaine et al. cited above, an excess of 345 kcal/day for 15 days would equate to approximately 5175 total excess kcal consumed over the trial length. Thus, the net weight gain cannot be accounted for solely by caloric excess, because 2.62 kg of increased body weight would equate to approximately 13 100 additional kcal (assuming 5000 kcal/kg mixed tissue). This ‘back-of-the-envelope’ calculation suggests two possible hypotheses – either APDinduced weight gain is a combination of increased food intake as well as decreased energy expenditure, and/or the caloric excess is accompanied by a large increase in body water, which may account for a substantial fraction (>50%) of the acute weight gain in short-term studies. The hypothesis that APDs reduce resting energy expenditure (REE) has been tested with mixed results. Although multiple groups have reported minimal to no change in energy expenditure [20,21], Fountaine et al. actually found that olanzapine increased REE (+113 kcal/24 h [17]. This is not a wholly unexpected result – energy intake and expenditure are not independent variables. In weight-reduced humans, energy
Review
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
Brain Global behavioral control • Hedonic response • Decision making
Hypothalamus Appete regulaon/saety (arcuate nucleus) • • • • •
Lepn/ghrelin Insulin/IRS signaling Histamine (H1R antagonism) -> AMPK Dopamine (D2R antagonism) -> AKT NPY/POMC
Fat Glucose uptake • Insulin-responsive (Glut4-mediated)
Muscle
Adipokine acon
Glucose uptake
• Lepn • Adiponecn
• Insulin-responsive (Glut4-mediated) • Contracon-dependent (AMPK-dependent)
Inflammatory state • TNF-alpha/IL-6 and others
Liver Hepac glucose producon • Hormone (insulin/glucagon)-regulated • Serotonin/SERT
Pancreas Insulin secreon • Dopamine co-secreon • Serotonin/SERT-regulated
Glucagon secreon
Hepac de novo lipogenesis • Insulin-regulated • Nutrient (mTorc1)-regulated
Gut Insulin/glucose regulaon • • • •
GLP-1 GIP Serotonin Dopamine TRENDS in Endocrinology & Metabolism
Figure 1. Pharmacological targets for metabolic dysfunction throughout the body. Antipsychotic medications are active throughout the body, including in pathways involved in metabolic processes. Based on the current body of evidence, we hypothesize that antipsychotic drug (APD)-induced metabolic disturbances arise from a combination of drug actions on targets in both the central nervous system (CNS) and peripheral organs. Within the CNS an important series of targets include hypothalamic nuclei. Primary peripheral targets of APD-induced metabolic dysfunction include pancreas, liver, muscle, and adipose tissue. Both central and peripheral targets are likely to be linked through a series of convergent molecular pathways including dopaminergic, serotoninergic, histaminergic, and adipokine signaling. Through action at receptors for these transmitters, APDs have synergistic effects to amplify metabolic risk. Therefore, based on this framework, we suggest a model whereby (i) APDs act directly on insulin-sensitive peripheral organs including pancreas, liver, muscle, and adipose tissue as well as the hypothalamic–liver circuit, which leads to peripheral insulin resistance; (ii) APDs act directly upon hypothalamic nuclei responsible for central regulation of food intake and metabolism; and (iii) these first two processes increase adiposity, which further exacerbates overall insulin resistance. Abbreviations: AKT, a serine–threonine protein kinase; AMPK, AMP-activated protein kinase; D2R, dopamine D2-like receptor; GIP, gastric inhibitory polypeptide; GLP-1, glucagon-like peptide-1; Glut4, glucose transporter type 4; H1R, histamine receptor 1; IL-6, interleukin-6; mTorc1, mammalian target of rapamycin complex 1; NPY, neuropeptide Y; POMC, proopiomelanocortin; SERT, serotonin transporter; TNF, tumor necrosis factor.
expenditure is reduced more than predicted by the decrease in somatic mass [22]. Conversely, short-term overfeeding results in proportionately higher energy expenditure. This may account for the greater energy expenditure seen in the Fountaine study. The range of acute alterations in energy expenditure appears to be 10–15% in either direction [23], suggesting that prolonged or excessive energy intake can overcome this physiologic compensation, leading to weight gain. In summary, APD-induced weight gain results from a combination of excess caloric intake, coupled with increased water weight, but is unlikely to result from a significant decrease in metabolic rate. Molecular mechanisms of antipsychotic-induced weight gain Despite years of study, the molecular mechanisms underlying APD-induced weight gain remain unclear, in part because animal models have only partially recapitulated results seen in human studies and in clinical practice. For
example, most [24–26] but not all [27] studies have reported increased food intake and body weight in APDtreated mice or rats. Outlying studies may result from the known sedative properties of APDs, which leads to issues with dose selection. Upon titration to doses that result in plasma drug concentrations comparable to those at steady state in schizophrenic patients [28], rodents are less active [29], which may artifactually reduce food intake and body weight gain as compared to animals treated with lower APD doses [30]. In addition, food intake and body weight gain in response to APDs are likely to be gender-, strain-, and diet-specific [31], possibly leading to divergent results. Nonetheless, rodent studies have suggested plausible hypotheses for the molecular mechanisms by which these drugs achieve their metabolic effects. For instance, one well-conducted study revealed that olanzapine-induced weight gain was mitigated by pair-feeding to placebo-treated rats, indicating that APD-induced weight gain is primarily the result of increased food intake [32]. In this 595
Review
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
Anpsychoc exposure
Current
Revised R
Increased food intake
Increased food intake
Obesity
Obesity
Insulin resistance Beta cell dysfuncon
Insulin nsulin istance resistance
Beta cell dysfuncon
T2D
T2D
TRENDS in Endocrinology & Metabolism
Figure 2. Proposed pathways for antipsychotic drug (APD)-induced metabolic dysfunction. Multiple converging pathways in the periphery mediate APD-induced metabolic dysregulation. Our current understanding of this phenomenon is based on the assumption that the increased food intake resulting from APDs is the principal driver of obesity and insulin resistance. However, recent evidence suggests that more than one pathway may be responsible, and that the obesity and insulin resistance may develop from different sources. This is consistent with mounting data showing that not all people who are obese are prone to develop type 2 diabetes (T2D), and vice versa.
study, the authors demonstrated increased orexigenic [neuropeptide Y (NPY), agouti related protein (AgRP)] decreased anorexigenic [proopiomelanocortin and (POMC)] patterns of neuropeptide expression in the hypothalami of rats treated with olanzapine, as compared to placebo, regardless of whether the animals were pair-fed or not [32]. This result is consistent with the primary effect of APDs on hypothalamic gene expression, with secondary effects on food intake [32]. Energy intake is homeostatically regulated by neuropeptide expression in the arcuate and other hypothalamic nuclei [33], as well as ‘higher’ brain regions mediating hedonic responses and decision-making [34]. An important candidate molecule for conferring APD-induced weight gain is the melanocortin receptor MC4R4. MC4R4 plays an important role in hypothalamic control of both food intake and energy expenditure; mutations reducing the functional activity of this receptor result in severe obesity in mice and humans [35]. An association has been shown between common DNA sequence variants in the vicinity of MC4R4 with APD-induced weight gain in a recent trial combining a GWAS and a prospective cohort analysis [36]. In recent years, multiple groups have postulated that APDs exert their orexigenic effects through effects on neurotransmitter signaling. Because this has been thoroughly reviewed by others [37], we briefly highlight the most compelling evidence for this hypothesis here. 596
Histamine hypothesis Histamine (HA) is a monoamine that binds four HA receptors (H1R–H4R), three of which (H1R–H3R) are expressed in the brain. Histaminergic neurons are abundant in the posterior hypothalamus and project throughout the brain, leading to speculation that HA may mediate homeostatic and hedonic aspects of feeding. Intracerebroventricular (ICV) HA infusion in rodents reduces food intake, probably mediated by H1R, as specific H1R agonists recapitulate the anorectic effect of HA, and H1R knockout mice consume more food after stimulation with the orexigenic neuropeptide NPY [38]. Consistent with these observations, APDs with the greatest antagonist H1R affinity, clozapine and olanzapine, preferentially stimulate hypothalamic AMPactivated protein kinase (AMPK) activity in proportion to their H1R antagonism [39]. AMPK activity is negatively regulated by the anorexigenic hormone leptin and leptin receptor signaling [40]; APD antagonism of this effect could contribute to APD-induced weight gain. In support of this theory, clozapine does not stimulate hypothalamic AMPK phosphorylation and activity in H1R-knockout mice [39]; effects on food intake or weight gain in these mice were not reported – this was likely to be related to the difficulty in modeling APD action in rodent models, as discussed above. Serotonin hypothesis Serotonergic (5-HT) neurotransmission is a common target of psychotropic drugs, particularly for antidepressants and atypical APDs. Notably, 5-HT agents have a long history as anorexiants, depending on their receptor specificity and affinity. The serotonin secretagogue and re-uptake inhibitor fenfluramine was widely prescribed for obesity treatment before being withdrawn in 1997 as a result of the risk of pulmonary hypertension. Lorcaserin, a selective 5HT2c receptor agonist, was recently approved for short-term treatment of obesity [41]. Conversely, mice with a hypomorphic 5HT2c receptor mutation are hyperphagic and insulin-resistant [42]. Second-generation APDs have variable activity at presynaptic 5-HT receptors in nigrostriatal neurons. 5HT2a and 5HT2c receptor single-nucleotide polymorphisms (SNPs) have also been associated with obesity, glucose intolerance, and susceptibility to APDinduced weight gain [43]. These observations led to the suggestion that APD blockade of hypothalamic 5HT2a and 5HT2c receptors causes drug-induced weight gain, but this hypothesis has not been rigorously tested. Obesity-independent effects of APDs on glucose metabolism Temporal relationships of hyperglycemia and new-onset diabetes to initiation of APD treatment and their reversibility on drug withdrawal suggest a causal relationship. Most new-onset diabetes occurs within the first 6 months after drug initiation, but increased risk of hyperglycemia persists with extended treatment. Hyperglycemia may be quite severe – presentation in diabetic ketoacidosis, although rare, has been observed with most APDs and may represent the first clinical manifestation of APD-induced T2D or even incipient type 1 diabetes (T1D). The most parsimonious explanation for this increased diabetes risk would be the APD-associated weight gain;
Review
Trends in Endocrinology and Metabolism November 2014, Vol. 25, No. 11
Table 1. Weight gain results from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) [3] Weight change from baseline (kg) Weight gain >7% from baseline Hemoglobin A1c % change from baseline
Olanzapine 4.3 (0.4) 92/307 (30%) 0.41 (0.09)
Quetiapine 0.5 (0.4) 49/305 (16%) 0.05 (0.05)
however, the mechanisms driving the obesogenic and diabetogenic effects of APD appear to be partially independent. Clozapine has been shown to cause increased insulin secretion independent of body mass [44]. Even relatively weight-neutral APDs show higher than expected rates of hyperglycemia (Table 1), although the risk increased with the more obesogenic drugs [3]. This association does not preclude the possibility that increased adiposity with chronic treatment exacerbates early APD-induced insulin resistance and/or beta cell dysfunction, leading to greater hyperglycemia and worsening diabetes. Nevertheless, there is clear evidence that early metabolic dysfunction precedes significant weight gain. In a well-conducted study, 12 healthy volunteers (18–30 years) were administered olanzapine 10 mg/day or placebo for 3 days, evaluated for multiple metabolic parameters, then crossed-over to the other treatment after a sufficient drug washout period [45]. No significant differences were detected between placebo and olanzapine treatments in terms of body weight, but olanzapine treatment caused a marked deterioration (42% increase of glucose area under the curve) of oral glucose tolerance (OGTT). Interestingly, neither fasting insulin levels nor insulin concentration during the OGTT was increased in olanzapine-treated subjects, suggesting that a compensatory insulin response to hyperglycemia was absent in this short-term exposure experiment [45]. Plasma leptin levels were significantly increased with olanzapine treatment [45], which is suggestive of an increase in body fat and/or increased leptin production as a result of the anabolic effects on adipocytes. Many of these findings have been corroborated in a recently published inpatient study with healthy volunteers, where slightly longer olanzapine treatment (9 days) led to weight-independent impaired glucose homeostasis [46]. In this study, olanzapine raised fasting insulin levels, suggesting increased insulin resistance, but both olanzapine and aripiprazole (which is considered weight-neutral) increased post-prandial insulin levels and reduced glucose disposal in hyperinsulinemic–euglycemic clamp studies independently of body weight effects when compared to placebo-treated volunteers [46]. These complementary studies confirm the weight-independent adverse effects on glucose homeostasis seen in animal models. However, the molecular mechanisms of APD-associated hyperglycemia remain as unclear as those underlying APD-associated weight gain. Further, it remains uncertain whether schizophrenia per se predisposes patients to respond differently to the metabolic effects of these agents. Molecular mechanisms of antipsychotic-induced hyperglycemia The hypothesis that APDs provoke weight gain-independent disturbances in glucose/insulin homeostasis has been tested in animals. In a seminal study, healthy rats treated
Risperidone 0.4 (0.4) 42/300 (14%) 0.08 (0.04)
Perphenazine 0.9 (0.5) 29/243 (12%) 0.10 (0.06)
Ziprasidone 0.7 (0.5) 12/161 (7%) 0.10 (0.14)
P
