INTRODUCTION

Obesity has reached pandemic proportions and it is rapidly surpassing smoking as a number one killer in the industrialized world (Sturm, 2002; Skidmore and Yarnell, 2004). Its annual cost to the American society is staggering, and is estimated to be around $117 billion owing to related illnesses and loss of productivity (National Institute of Diabetes and Digestive and Kidney Diseases, 2005).

In schizophrenia, obesity is twice as prevalent as in the general public, afflicting over half this patient population (Allison et al, 1999a; Dixon et al, 2000; American Diabetes Association, 2004; Marder et al, 2004; Wirshing, 2004). Besides negative psychosocial impacts (distorted self-esteem and societal stigmatization) and medications noncompliance, schizophrenics appear to be particularly susceptible to the detrimental medical sequelae of obesity such as the ‘Metabolic Syndrome’, which is a cluster of cardiovascular risk factors, including abdominal adiposity, insulin resistance, impaired, glucose tolerance, dyslipidemia, and hypertension (McKee et al, 1986; Mukherjee et al, 1996; Brown et al, 2000; Haupt and Newcomer, 2002; Ryan and Thakore, 2002; Ryan et al, 2003; Holt et al, 2004; Kohen, 2004; Marder et al, 2004; Lieberman et al, 2005).

Excessive body weight gain (BWG) could be attributed to a constellation of endocrine, molecular, genetic, demographic, and lifestyle-related factors. Provided that a common trait may be unhealthy eating, this review considers the dietary aspects of obesity that are pertinent to schizophrenia. First, we provide clinical context on the association between schizophrenia and consumption of the foodstuffs implicated in the development of BWG. Next are discussed the putative neuroanatomy and neurochemistry of mechanisms involved in the regulation of food intake and reward, as they relate to ‘reward deficiency’ state (Comings and Blum, 2000) and unhealthy eating patterns (Peet, 2004a) associated with schizophrenia.

Subsequent sections compare the data on the effects of food, addictive substances, and second-generation antipsychotic agents (SGAs) on homeostasis- and reward-related function, and this comparison is taken as a basis for the analysis of the role of the latter in worsening of BWG. The last subsection covers specific background for a testable hypothesis on alleviation of these emerging abnormalities via augmentation of the SGAs with opioid receptor antagonists such as naltrexone. As this review draws heavily upon preclinical research, it is important to highlight many uncertainties inherent in extrapolating from animal data cited here to schizophrenia (or any other neuropsychiatric syndrome) such as absence of well-defined pathognomonic neurobiological abnormalities and sole reliance on behavioral symptoms and subjective self-reports for the illness diagnosis (Lieberman, 2000). More objective phenotypic (eg, neuroimaging) or genotypic (eg, gene markers or single-nucleotide polymorphisms analyses) tests would provide further evidence on the pathophysiological mechanisms underlying metabolic disturbances in patients with schizophrenia.

SEARCH TERMS AND METHODOLOGY

English language literature search of reward function in schizophrenia, mechanisms of food hedonics and reward and metabolic effects of the SGAs was undertaken using PubMed. Keywords used included schizophrenia plus obesity, sugar, opioid, fat, food, palatable, fast food, eating, appetite, pain, hunger and reward. Each specific SGA was combined with opioid, reward, appetite, hunger, pain and obesity terms. Data on reward, homeostatic mechanisms and SGAs’ metabolic effects were also drawn from recent seminal reviews of these topics (Grace 2000; Koob and Le Moal, 2001; Saper et al, 2002; Berridge and Robinson, 2003; Wise, 2004; Kelley, 2004a, 2004b; Kalra and Kalra, 2004; Nestler, 2004; Newcomer, 2005; Volkow and Wise, 2005; Kalivas and Volkow, 2005; Hyman, 2005; Kalivas et al, 2005; Schwartz and Porte, 2005). Additional strategies included manual searches for relevant articles from the selected papers’ reference lists as well as utilization of PubMed's related articles function.

WEIGHT STATUS OF DRUG NAIVE/FREE SCHIZOPHRENIC PATIENTS

Obesity and associated pathologic conditions (eg, insulin resistance, impaired glucose tolerance, and dyslipidemia) tend to be more prevalent (Finney, 1989; Ryan et al, 2003, 2004), albeit not universally present (Allison et al, 1999a; Zhang et al, 2004) in patients with schizophrenia. The basic mechanisms behind these metabolic irregularities have not been fully elucidated, but it appears that antipsychotic medications could play an important role (Newcomer, 2005). Moreover, high rates of obesity and type II diabetes mellitus, observed in drug-naive/free patients (Mukherjee et al, 1996; Allison et al, 1999a; Thakore et al, 2002; Ryan et al, 2003, 2004) before (Haupt and Newcomer, 2002; Kohen, 2004) and after the advent of antipsychotics and in nonschizophrenic blood relatives (Dynes, 1969; Mukherjee et al, 1989; Cheta et al, 1990; Martins et al, 2001; Lamberti et al, 2004), were potentially attributed to genetic factors (Stone et al, 2004), illness neurobiology (Thakore, 2005) and to unhealthy lifestyle (Brown et al, 1999). The interpretability of the preneuroleptic era data (reviewed in Kohen, 2004) is, however, limited by flaws in epidemiological methodology including lack of evaluation of and adjustments for adiposity, lifestyle, and anthropometric measures together with inconsistent diagnostic criteria for schizophrenia and glucose/insulin abnormalities (Haupt and Newcomer, 2002; Newcomer, 2005).

Seeking to tackle serious methodological limitations of the early research, Thakore et al (2002) applied a well-validated technique of computed tomography (Mayo-Smith et al, 1989) to directly measure abdominal adiposity in a sample of drug-naive/free schizophrenic patients and age- and sex-matched healthy controls. They found that schizophrenic patients had higher body mass index (BMI) and over three times more abdominal fat (Thakore et al, 2002), accumulation of which poses even greater cardiovascular/diabetes risks than overall obesity (Wang et al, 2005). A similar magnitude of increases in abdominal adiposity was observed in a tandem study where drug naive, first episode schizophrenic subjects (N=19) were matched to healthy controls in terms of BMI, age, gender, exercise, smoking, and alcohol use (Ryan et al, 2004).

Although generally consistent with yet another observation by the same group (Ryan et al, 2003), these results have not been replicated in an ethnically distinct sample of drug-naive Chinese patients where increased obesity and adiposity was not detected using low field strength (0.2 Tesla) magnetic resonance imaging (Zhang et al, 2004). This apparent discrepancy in the studies’ outcomes underscores the importance of socioeconomic, cultural, and ethnic variables for the development of excessive weight and abdominal fat deposits. Replication of both groups (Thakore et al, 2002; Zhang et al, 2004) findings may be, however, warranted for obtaining more conclusive evidence because of several confounds including hypercortisolemia and older age (that could reflect a longer duration of untreated psychosis) in the participants of the Thakore et al studies (Thakore et al, 2002; Ryan et al, 2003, 2004) as well as use of a nonstandardized imaging protocol and of unmatched schizophrenics/controls samples for the determination of abdominal adiposity by Zhang et al (2004). Additionally, correction for geographic region-specific normative values will be required in future studies as average BMIs tend to be lower in Chinese vs Western population (Cheng, 2005).

DIETARY HABITS OF SCHIZOPHRENIC PATIENTS

Obesity and adiposity represent a complex pathophysiological condition that could result from increased caloric intake (Strassnig et al, 2003a, 2003b) and decreased energy expenditure owing to reduced physical activity (Strassnig et al, 2003a) or a combination of both (Peet, 2004a). Dietary intake, a relatively easily modifiable (Brar et al, 2005) behavioral phenomenon, could provide a rich framework for preventive/therapeutic clinical interventions.

A rather scarce (relatively to the scope of the problem) literature on schizophrenic eating habits describes exaggerated preference for unhealthy predominately fast food-type (Peet, 2004a, 2004c; Strassnig et al, 2005) palatable nutrition, poor in fiber (Brown et al, 1999; McCreadie et al, 1998), in fruits and in vegetables (McCreadie et al, 1998, 2005; McCreadie, 2003), but excessive in saturated fat (Brown et al, 1999; Strassnig et al, 2003a, 2005; Ryan et al, 2003, 2004), in carbohydrates (Strassnig et al, 2003b), and in high glycemic index (Foster-Powell et al, 2002) foodstuffs such as cereal (McCreadie, 2003). These patterns may be further worsened by treatment with SGAs such as clozapine, olanzapine, and to lesser degree quetiapine and risperidone, but not ziprasidone or aripiprazole (Allison et al, 1999b; American Diabetes Association, 2004; Kane et al, 2004; Newcomer 2005).

The pathophysiological significance of these findings remains unclear. Environmental factors play an obvious role as patients are frequently uninformed about dietary issues (Meyer, 2002) and are unable to access/afford healthy food choices (Holt et al, 2004). They also tend to exercise less (Brown et al, 1999; Strassnig et al, 2005), and therefore even slight excess in caloric intake could result in the accumulation of fat.

In this article, we propose a complementary view that in addition to lifestyle-related factors and similar to its comorbidity with substance use disorders (SUDs; Chambers et al, 2001), schizophrenia and unhealthy food preferences may be potentially explained by recursive partly shared neural systems. Direct evidence for such a neuropathological link is missing. There is, however, consistent (McIntosh and Lawrie, 2004) correlational evidence at a population level revealing a link (not causal-effect relations) between high dietary levels of refined sugar (Peet, 2004b) and of saturated fat (Christensen and Christensen, 1988; Gupta, 1992; Peet, 2004b) with the illness poor prognosis (Christensen and Christensen, 1988; Gupta, 1992), including increased length of hospitalizations and severe social deficit symptoms (Peet, 2004b).

These correlational findings were obtained from two types of international databases. The first contains dietary information, that is, national dietary report provided by the Food and Agriculture Organization of the United Nations (2002). The second source involves three large-scale multicenter surveys of various schizophrenia outcome measures, namely the International Pilot Study of Schizophrenia (World Health Organization, 1979; Gupta, 1992; Peet, 2004b), the Determinants of Outcomes of Severe Mental Disorders (Jablensky et al, 1992; Gupta, 1992; Peet, 2004b), and the World Health Organization International Follow-up Study (World Health Organization, 1979; Christensen and Christensen, 1988). Although the reasons behind transcultural differences in schizophrenia outcomes detected in these and other studies (Kulhara, 1994; Sartorius et al, 1996) have received less attention than the outcomes themselves, it may be relevant that, taken as a whole, schizophrenia prognosis tends to be substantially worse in the developed Western nations, in comparison to the developing countries, that have not yet adopted the Westernized diet (Peet, 2004a).

To gain insight into this entity, research must isolate the interaction of schizophrenia with voracious appetite for food containing high quantities of sugar and fat. One such possible overlap area is the neural pathways mediating reward (subjective experience of pleasure) and reinforcement (increased rate of recurrent actions), a relevant etiologic factor in both schizophrenia (Green et al, 1999; Chambers et al, 2001; Brady and Sinha, 2005) and eating behavior (Wang et al, 2001, 2002, 2003, 2004). The following sections discuss energy balance systems and how they intersect with the brain reward and reinforcement mechanisms.

NEUROBIOLOGY OF SYSTEMS MEDIATING CALORIC BALANCE

Eating is imperative for continued existence of all living organisms (Maslow, 1943), and even brief (days) starvation leads to detrimental physical and psychological changes (Phillips, 1994; Fessler, 2003). Hence, eating behavior is insured via powerful neural systems responsible for the maintenance of caloric balance and for the rewarding nature of food intake. These systems are schematically overviewed in Figures 1 and 2, respectively.

Figure 1
figure 1

Schematic overview of the metabolic, endocrine, and neural pathways involved in the homeostatic control of eating behavior (ie, maintenance of energy balance). Peripheral adiposity- and nutrient-related signals converge on the level of the medial hypothalamus (other CNS sites, for example, nucleus of the solitary tract and the area postrema are also involved but are not represented here). Fasting or decrease in body energy stores and in circulating nutrients, signaled by gastric hormone, ghrelin, and/or by hypothalamic intracellular agent, AMP-activated protein kinase, promote anabolic processes (ie, appetite and conservation of energy) and restrain catabolic processes via hypothalamic orexigenic pathways. Food intake, conversely, increases body fat content and plasma levels of the nutrients’ metabolites (eg, glucose and free fatty acids; FFA) and triggers (among other metabolic events) release of catabolic hormones, insulin (from pancreas), and leptin (from adipose tissues). These hormones, respectively, suppress and activate hypothalamic orexigenic and anorexigenic networks, thus inhibiting further eating and enhancing energy expenditure. The neural orexigenic networks, originating in the arcuate nucleus (ARC) of the hypothalamus and projecting to the paraventricular nucleus (PVN), coexpress neuropeptide, gamma-aminobutyric acid, and agouti-related protein. The anorexigenic networks also project from ARC to PVN and are comprised of neurons coreleasing pro-opiomelanocortin, alpha-melanocyte-stimulating hormone, brain-derived neurotrophic factor, and cocaine- and amphetamine-regulated transcript (Gerozissis, 2004; Kalra and Kalra, 2004; Saper et al, 2002; Schwartz and Porte, 2005). Orexin and melanin-concentrating hormone (MCH) from neurons in the lateral hypothalamus (LH) stimulate the orexigenic networks; ventromedial nucleus (VMN) exerts an opposite effect potentially via histaminergic mechanisms (Mercer et al, 1994; Sakata et al, 1997). Histamine also has an important role in catabolic processes by mediating the effects of leptin (Masaki et al, 2004), which promotes its release (Itateyama et al, 2003). Other important anorexigenic factors include serotonin (5HT) via 5HT1b, 5HT2A and 5HT2C receptors (Makarenko et al, 2002; Muller et al, 2004) and corticotropin-releasing hormone (CRH; Kalra et al, 1999), whereas orexigenic processes are favored by endogenous opioid peptides (Horvath et al, 1992; Kalra et al, 1999; Morley, 1987) and glucocorticoids (Kalra et al, 1999). Lastly, dopamine in the dorsal striatum (basal ganglia) is involved in nonhedonic homeostatic drive and goal-directed motor behavior towards procurement of food (Cannon et al, 2004; Volkow et al, 2002c). For the clarity of presentation, the scheme was rendered out-of-scale and simplified to reduce the numbers of the displayed links and structures to those of direct relevance to the main themes of this review.

Full size image
Figure 2
figure 2

Schematic overview of the ‘motivational system’ for ingestive behavior (Swanson, 2000; Kelley, 2004a) formed from hierarchically organized clusters of hypothalamic and brainstem nuclei (behavioral control column) along with the associated corticolimbic regions. Hypothalamus plays the critical role in the maintenance of body energy balance (see Figure 1), whereas brainstem mediates orosensory sensations and motor function involved in the act of eating, which is ‘powered’ through motivational states and emotions generated by the corticolimbic reward circuitry. The major interface between the homeostatic, motivational/emotional, and motor components is located within the lateral hypothalamus (LH; DiLeone et al, 2003; Saper et al, 2002). This area is involved in reinforcement (Berridge and Valenstein, 1991; Wise, 1996) and is interconnected with the major hypothalamic nuclei (ie, ARC and PVN) implicated in the metabolic sensing (Kalra and Kalra, 2004) and with the brainstem nuclei, mediating eating-related motor function and autonomic activity. Pleasure and positive affective states produced by consumption of palatable food, that is, food ‘liking’ is conveyed to the frontotemporal cortical structures (Berridge, 2003; Kelley, 2004b) through μ-opioid neurotransmission within the network of subcortical and brain stem nuclei (Berridge, 2003; Berridge and Robinson, 2003; Berthoud, 2004b; Kelley, 2004b; Pecina et al, 2003; Saper et al, 2002; Tanda and Di Chiara, 1998; Will et al, 2004), including the nucleus accumbens (NAc), vental tegmental area (VTA), ventral pallidum (VP), nucleus of the solitary tract (NTS), parabrachial nucleus (PBN), and the amygdala (Amy). Mesolimbic dopaminergic pathways projecting from the VTA to the NAc, along with the LH, Amy and prefrontal cortex (PFC; Goldstein and Volkow, 2002; Volkow et al, 2004a) are responsible for food ‘wanting’ (Kelley, 2004b; Kelley and Berridge, 2002; Pecina et al, 2003). The liking and wanting together constitute full reward system (Berridge, 2004). Although these are dissociable phenomena, they are closely linked as liking food usually evokes wanting and eventually eating it (Berthoud, 2004a). Food reward is modulated by a variety of peripheral signals such as nutrition and adiposity, smell, taste, orosensory, and visual stimuli along with environmental factors (eg, food composition and availability, media effects, and social situations). More specifically, metabolic signals of nutrient metabolites and of adiposity, that is, insulin and leptin decrease rewarding value of food (and other rewards) by binding insulin and leptin receptors, in the NAc, LH, and VTA (Figlewicz, 2003a, 2003b, 2004). For the clarity of presentation the scheme was rendered out-of-scale and simplified to reduce the numbers of the displayed neuroanatomical/neurochemical links and structures to those of direct relevance to the main themes of this review.

Full size image

As depicted in Figure 1, peripheral adiposity- and nutrient-related signals converge on the level of the medial hypothalamus (Schwartz et al, 2000). Fasting or decrease in body energy stores and in circulating nutrients, signaled by gastric hormone, ghrelin, and/or by hypothalamic intracellular agent, AMP-activated protein kinase, promote anabolic processes (ie, hunger and conservation of energy) and restrain catabolic processes (ie, satiety signals) via hypothalamic orexigenic pathways (Schwartz and Porte, 2005). Food intake, conversely, increases body fat content and plasma levels of the nutrients’ metabolites (eg, glucose and free fatty acids) and triggers (among other metabolic events) release of catabolic hormones, insulin (from pancreas) and leptin (from adipose tissues). These hormones, respectively, suppress and activate hypothalamic orexigenic and anorexigenic networks, thus inhibiting further eating and enhancing energy expenditure (Morton et al, 2005; Schwartz and Porte, 2005). In sum, brain metabolic control apparatus is tightly regulated via antagonistic ying- and yang-type processes (Wynne et al, 2005).

The neural orexigenic networks, originating in the arcuate nucleus (ARC) of the hypothalamus and projecting to the paraventricular nucleus (PVN), coexpress neuropeptide Y, gamma-aminobutyric acid, and agouti-related protein (AgRP). The anorexigenic networks also project from ARC to PVN and are comprised of neurons coreleasing proopiomelanocortin, alpha-melanocyte-stimulating hormone, brain-derived neurotrophic factor (BDNF) and cocaine- and amphetamine-regulated transcript (Saper et al, 2002; Gerozissis, 2004; Kalra and Kalra, 2004; Schwartz and Porte, 2005). Orexin and melanin concentrating hormone (MCH) from neurons in the lateral hypothalamus (LH) stimulate the orexigenic networks; ventromedial nucleus exerts an opposite effect via histaminergic mechanisms (Mercer et al, 1994; Sakata et al, 1997). Histamine also has an important role in catabolic processes by mediating the effects of leptin (Masaki et al, 2004) that promotes histamine release (Itateyama et al, 2003). Other important anorexigenic factors include serotonin (via 5HT1b, 5HT2A, and 5HT2C receptors) and corticotropin-releasing hormone (CRH; Kalra et al, 1999; Makarenko et al, 2002; Muller et al, 2004), whereas orexigenic processes are favored by endogenous opiate peptides (Morley, 1987; Horvath et al, 1992; Kalra et al, 1999) and glucocorticoids (Kalra et al, 1999). Lastly, basal ganglia dopamine is involved in nonhedonic metabolic drive and goal-directed motor behavior towards procurement of food (Volkow et al, 2002c; Cannon et al, 2004).

As a whole, this well-orchestrated machinery is almost infallible in preventing caloric deficit states as it sets of hunger (and supresses satiety), accompanied by negative subjective experiences ranging from slight discomfort to restlessness and food obsession. Eating can therefore be construed as a negative reinforcer as it brings an end to the discomfort induced by hunger (Saper et al, 2002). Food intake is, however, not merely an automatic action aimed at regaining comfort and tranquility. It is highly rewarding and positively reinforcing process. Indeed, palatability, derived from high levels of sugar and fat, is the most important (aside from the energy need) factor determining the amount of consumed food (Berthoud, 2004a). This is perhaps to encourage ingestion of scarce (at the earlier stages of evolution) high-energy content foods (Ulijaszek, 2002; Kelley, 2004b), and to secure steady supply of sugar, which is the primary energy source throughout the central nervous system that is neither stored nor produced there.

NEUROBIOLOGY OF SYSTEMS MEDIATING FOOD REWARD

Neuroanatomy, neurophysiology, and neurochemistry of reward have been at the outset investigated using drug addiction paradigms. Over the last decade, these findings were expended and generalized to food reward (Figure 2), given the well-described substantial overlap between both phenomena (Mercer and Holder, 1997b; Kelley and Berridge, 2002; Berridge and Robinson, 2003; Wang et al, 2004; Volkow and Wise, 2005).

Incentive motivation theory (Berridge and Robinson, 2003), largely based on rodent research, appears to be applicable for both food and drug reward even though it has not yet received substantial clinical support. According to this theory, brain reward function is not a homogeneous entity. Rather, it can be dissected, using neurochemical, neuroanatomical, and functional criteria, into two core processes, namely ‘liking’ and ‘wanting’ (Berridge and Robinson, 2003). Drug-induced changes in the mesolimbic dopaminergic circuitry underlying the wanting but not liking purposes is purportedly responsible for transforming regular wanting responses into heightened incentive salience assigned to drugs or drug-related cues. This incentive sensitization process is construed to be an animal homolog of human craving (Berridge and Robinson, 2003).

A closely related concept, derived from primate work, is the aberrant learning theory suggesting that learning of new rewards is encoded via interactions between tonic (baseline) and phasic spikes in dopaminegic neurons; phasic firing predicts new rewards (Schultz, 2001). Hence, neural adaptations to excessive dopaminergic trafficking in response to drugs (Breiter et al, 1997) or during psychotic episodes (Grace, 1991; Lieberman et al, 1997; Laruelle, 2000; West et al, 2003) may lead to low signal-to-noise detection capability for natural rewards along with ‘profound overlearning of the motivational significance of cues that [either] predict delivery of drugs’ (Hyman, 2005) or are associated with psychotic contents resulting in ‘aberrant salience’ (Kapur, 2003) for drug- or psychosis-related stimuli (detailed in the following sections).

Mesolimbic dopaminergic pathways projecting from the vental tegmental area (VTA) to the nucleus accumbens (NAc), along with the LH, amygdala, and prefrontal cortex (PFC; Goldstein and Volkow, 2002; Volkow et al, 2004a), are responsible for the incentive motivational aspects of reward function. These are collectively termed ‘wanting’ processes (Kelley and Berridge, 2002; Pecina et al, 2003; Kelley, 2004b) and include conditioned learning of stimulus–reward association (McClure et al, 2003; Tobler et al, 2003; Berridge, 2004; Wise, 2004), reward prediction, and attribution of incentive salience to rewarding stimuli (Horvitz, 2000; McClure et al, 2003) by ‘gating’ of stimulus representations in the PFC (Montague et al, 2004). The latter region also integrates peripheral and environmental stimuli and exercises cognitive control over drives and emotions gated from the limbic regions (Goldstein and Volkow, 2002).

Consumption of palatable food produces pleasure and positive affective states. Food ‘liking’ is conveyed to the frontotemporal cortical structures (Berridge, 2003; Kelley, 2004b) through μ-opioid neurotransmission within the scattered network of subcortical and brainstem nuclei (Tanda and Di Chiara, 1998; Saper et al, 2002; Berridge, 2003; Berridge and Robinson, 2003; Pecina et al, 2003; Berthoud, 2004b; Kelley, 2004b; Will et al, 2004), including the NAc, VTA, ventral pallidum, nucleus of the solitary tract, parabrachial nucleus, and the amygdala. Opioids enhance food pleasantness (Doyle et al, 1993; Pecina and Berridge, 1995; Kelley et al, 2002; Will et al, 2003), whereas consumption of high-energy sweet and fat food increases endogenous opiates release (ie, feedforward interaction; Tanda and Di Chiara, 1998; Colantuoni et al, 2002; Grigson, 2002) so that analgesia (Mercer and Holder, 1997a; Segato et al, 1997; Lewkowski et al, 2003), opioid dependence (Schoenbaum et al, 1989, 1990; Colantuoni et al, 2001, 2002), and compensatory decrease of opiate gene expression in the reward structures (Kelley et al, 2003, 2005) may ensue.

The liking and wanting processes, contributing to the fully functional reward system (Berridge, 2004), are dissociable but closely related phenomena. This is obvious on the intuitive level as food desire is usually triggered by palatable, but not bland types of cuisine (Yeomans et al, 1997, 2004; Berthoud, 2004a). Accordingly, opioid-induced feeding depends on the NAc's dopamine receptors signaling (Schmauss and Emrich, 1985; Nylander and Terenius, 1987; Sivam, 1989; You et al, 1994; MacDonald et al, 2004), and in the reversed order, opioids cause dopamine releases in the NAc (Devine et al, 1993a, 1993b; Tanda and Di Chiara, 1998; Yoshida et al, 1999; Schad et al, 2002).

CENTRAL MECHANISMS FOR ENERGY BALANCE AND FOR FOOD REWARD ACT TOGETHER TO ENSURE ADEQUATE NUTRITION

As presented in Figure 3, appetite or incentive motivational drive to seek and consume food is determined by interrelated psychobiological factors including food's rewarding properties, individual's homeostatic needs, and cognitive ability to favor alternative (to eating) behaviors (Saper et al, 2002; Hinton et al, 2004; Kelley, 2004b). Regular (ie, predominately bland) food does not fully engage reward centers and its consumption is effectively terminated with sensation of satiety (Erlanson-Albertsson, 2005) when fulfillment of homeostatic needs is relayed to the brainstem nucleus of the solitary tract (Appleyard et al, 2005; Broberger, 2005) by cranial afferents from the alimentary tract organs (Harding and Leek, 1973).

Figure 3
figure 3

Schematic overview of the interface between biopsychological factors governing eating behaviors. Appetite, that is, incentive motivational drive to seek and consume food, is derived from interrelated psychobiological factors including food's rewarding properties, individual's homeostatic needs, and cognitive ability to favor alternative (to eating) behaviors (Saper et al, 2002; Hinton et al, 2004; Kelley, 2004b). Satiety is associated with decreased reward function secondary to insulin and leptin actions (Figlewicz, 2003a, 2003b, 2004; Berthoud, 2004b) during caloric overflow signaled by these hormones, whereas caloric restriction, reflected in subjective sensation of hunger, increases food reward (Carr, 2002; Fulton et al, 2004) by augmenting dopaminergic (Carr et al, 2003) and opioidergic (Carr and Wolinsky, 1993) neurotransmission. On the other hand, reward centers activation by high sugar and fat content palatable foods promotes desire ‘to come back for more’ (Kelley et al, 2002) by upregulating of hunger signaling (orexin, agouti-related protein), melanin concentrating hormone) in hypothalamic orexigenic networks and by blunting brain responses to the peripheral satiety hormones, insulin and leptin (Erlanson-Albertsson, 2005; Isganaitis and Lustig, 2005). Prefrontal cortex (PFC) modulates these effects via reciprocal innervation with the hypothalamic–limbic areas (Berthoud, 2004a) by forming subjective hedonic perceptions and integrating them with metabolic signals and with intrapsychic and environmental contexts (Kringelbach, 2005). For the clarity of presentation, the scheme was rendered out-of-scale and simplified to reduce the numbers of the displayed links and structures to those of direct relevance to the main themes of this review.

Full size image

Satiety is associated with decreased reward function secondary to insulin and leptin actions in the NAc, LH, VTA, taste receptor cells, and cortical areas (Figlewicz, 2003a, 2003b, 2004; Fulton et al, 2004; Berthoud, 2004b) during caloric overflow signaled by these hormones, whereas caloric restriction, reflected in subjective sensation of hunger, increases food reward (Carr, 2002; Fulton et al, 2004) by augmenting dopaminergic (Carr et al, 2003) and opioidergic (Carr and Wolinsky, 1993) neurotransmissions.

On the other hand, reward centers activation by high sugar and fat content palatable foods promotes desire ‘to come back for more’ (Kelley et al, 2002) by upregulating of hunger signaling (orexin, AgRP, and MCH) in hypothalamic orexigenic networks and by blunting brain responses to the peripheral satiety hormones, insulin and leptin (Erlanson-Albertsson, 2005; Isganaitis and Lustig, 2005). Prefrontal cortex modulates these effects via reciprocal innervation with the hypothalamic–limbic areas (Berthoud, 2004a) by forming subjective hedonic perceptions and integrating them with metabolic signals and with intrapsychic and environmental contexts (Kringelbach, 2005).

REPETITIVE PALATABLE FOOD CONSUMPTION MAY DYSREGULATE HOMEOSTATIC AND REWARD PROCESSES

To maintain stability of the body adipose content, energy expenditure should tightly match caloric intake. This is accomplished through negative feedback regulation exerted by the endocrine markers of body fat mass, insulin and leptin; both are also affected by recent food intake and its macronutrient composition (Havel, 2001; Schwartz et al, 2003). In the basal state (Figure 1), insulin and leptin inhibit anabolic pathways (appetite and energy conservation) while maintaining tonic activation of the catabolic pathways (satiety and energy outflow).

Such regulatory system may be adequate for regular diets but is ‘inherently biased toward weight gain’ (Havel, 2001; Schwartz et al, 2003; Schwartz and Niswender, 2004) in ‘obesigenic’ environments abound in ‘supersized’ meals, more than twice caloric density (energy per weight) and content of healthy diets (Prentice and Jebb, 2003). High caloric density food contributes to abnormally elevated total caloric intake as weight and volume of meals ingested by humans tend to remain relatively steady (Rolls et al, 1998). Such rich in caloric content diets robustly activate reward centers and fail to produce proportional (to the amount of ingested calories) suppression of hunger signals, as CNS insulin and leptin stop boosting and restraining, respectively, already activated (in the basal state) catabolic pathways and already inhibited (in the basal state) anabolic pathways, that is, ceiling effect (Schwartz et al, 2003). Thus, metabolic restraint on reward function via insulin/leptin mechanism becomes inefficient at about 40% fat concentration, typical of the Western diet (Figlewicz, 2003a, 2003b, 2004), allowing reward mechanisms to over-ride metabolic requirements (perhaps owing to evolutionary pressure giving preferentiality to food intake over fasting) and underscoring the role of cortical cognitive control mechanisms in the determination of the amount of eaten food.

Pleasurable hedonic impact and inefficient suppression of hunger by palatable foods predictably brings increases in the amount of consumed food and weight gain (in the absence of compensatory caloric loss) with ensuing hyperinsulinemia and insulin resistance (Kahn and Flier, 2000; Homko et al, 2003), which drive weight further upward (Sigal et al, 1997; Odeleye et al, 1997). Moreover, weight gain and insulin resistance from habitual consumption of palatable foods decrease insulin and leptin blood–brain barrier (BBB) penetrability (Caro et al, 1996; Kaiyala et al, 2000; Banks, 2003; Banks and Farrell, 2003; Woods et al, 2003) and their CNS effects (Couce et al, 2001; Banks and Farrell, 2003; Lindqvist et al, 2005; Porte et al, 2005), albeit some homeostatic hypothalamic areas lack BBB (Peruzzo et al, 2000; Ganong, 2000). The resultant brain insulin and leptin ‘resistance’ renders normal satiety signals even more ineffective (Erlanson-Albertsson, 2005; Isganaitis and Lustig, 2005) leading to further impairments in physiologic mechanisms regulating food intake (eg, overeating) and shifting the set point for energy homeostasis towards the development of overweight and obesity (Levine et al, 2003; Erlanson-Albertsson, 2005; Isganaitis and Lustig, 2005). It is quite difficult to reverse the developing vicious cycle (schematically illustrated in Figure 4) as attempted weight loss by caloric restriction results in insulin and leptin declines triggering robust activation of anabolic pathways and suppression of catabolic pathways; the magnitude of these responses exceeds by far that during caloric overflow (Niswender et al, 2004).

Figure 4
figure 4

Diagram describing repetitive and compulsive nature of palatable food consumption. The figure depicts transition from occasional palatable food intake to dysregulation of homeostatic and reward processes leading to impairments in normal signals of hunger and satiety along with adaptations and deviations of physiological set points (cf, Koob and Le Moal, 2001).

Full size image

STRESS EXPOSURE ALSO AFFECTS EATING HABITS

Stress is another factor controlling eating behavior and its effects on reward processes are quite similar to addicitive drugs (see Table 1 for comparison of these effects). Acute stress activates dopamine transmission in the reward circuitry through enhancement of extracellular dopamine release and/or potentiation of dopamine receptors’ affinity and their activity (Thierry et al, 1976; Imperato et al, 1989). Chronic stress exerts an opposite action by decreasing dopaminergic (Puglisi-Allegra et al, 1991; Imperato et al, 1993) neurotransmission and is accompanied by decreased motivation towards normally pleasurable stimuli (Willner et al, 1987; Papp et al, 1991). Furthermore, stress-related negative affective states drive comfort or palatable food consumption as it can alleviate these states (ie, ‘self-medication’) via temporary normalization of reward function (Wang et al, 2001), via inhibition of hypothalamic CRF secretion involved in stress-induced anxiety and depression (Dallman et al, 2003) or via priming effects (Wise, 2004) on the mesocorticolimbic dopaminergic system (Thierry et al, 1976; Imperato et al, 1989).

Table 1 Effects of Stress and Addictive Drugs on Reward Circuitry and Psychological Correlates of these Effects
Full size table

Finally, stress-induced hypercortisolemia contributes to the development of obesity as a result of visceral fat accumulation (Ryan and Thakore, 2002) consequentially to more abundant intra-abdominal adipose tissue's (vs subcutaneous fat mass) concentration of glucocorticoid receptors that activate cortisol-sensitive lipoprotein lipase (Rebuffe-Scrive et al, 1985; Pedersen et al, 1994; Fernandez-Real et al, 1997), an anabolic enzyme promoting buildup of fat (Bjorntorp, 1996). Cotisol's role in stress-induced obesity in schizophrenic patients has been undermined, though, by (1) the lack of consistent hypercortisolemia in nonacutely ill schizophrenia samples (Gispen-de Wied, 2000, 2) the lack of visceral fat accumulation and obesity in drug-naive schizophrenia samples from non-Westernized populations (Zhang et al, 2004), and (3) weight gain commonly occurring in the setting of cortisol decreases during SGAs treatment (Newcomer et al, 2002; Mann et al, 2006).

KEY FACTORS INVOLVED IN THE NORMAL REWARD FUNCTION ARE IMPAIRED IN SCHIZOPHRENIA

Given the metabolic abnormalities arising in healthy people exposed to fast food diets (see above), a compelling a fortiori argument could be that propensity for weight gain in ‘obesigenic’ environments is worsened by neuropathology afflicting the same factors involved in energy homeostasis and reward regulation (Thakore, 2005). Schizophrenia could be considered such a disorder (Figure 5a). One obvious factor is higher rates of peripheral glucose intolerance and insulin resistance observed in some schizophrenic patients (Mukherjee et al, 1996; Ryan et al, 2003) that could lead to CNS insulin and leptin resistance with the ensuing consequences as discussed above.

Figure 5
figure 5

(a) Schematic overview of the psychological signals underlying manifestation of appetite and of the potential sites where schizophrenia neuropathology predisposes for overeating and obesity. Eating patterns are largely determined by homeostatic system responsible for total body energetic balance and by reward system ensuring repetitive activities essential for the survival of individuals and species. To maintain stability of the body adipose content, energy expenditure should tightly match caloric intake. This is accomplished through negative feedback regulation exerted by the endocrine markers of body fat mass, insulin and leptin; both are also affected by recent food intake and its macronutrient composition (Havel, 2001; Schwartz et al, 2003). Propensity for weight gain could be worsened by schizophrenic neuropathology afflicting the same factors involved in energy homeostasis and reward regulation (Thakore, 2005). One such cause is higher rates of peripheral glucose intolerance and insulin resistance in schizophrenic populations (Mukherjee et al, 1996; Ryan et al, 2003) that could lead to CNS insulin and leptin resistance. Mesolimbic hyperdopaminergic state presumably renders motivational/incentive reward system insensitive to low salience/palatability food. Also, palatable food may produce dysregulated dopamine release (Lieberman et al, 1997; Laruelle, 2000) in the ventral striatum, creating a rigid motivational state fixated on procurement of this type of food. These, together with poor cognitive control from the hypofunctional PFC and enhanced hedonic impact of food, owing to exaggerated opioidergic drive, may underlie increased intake of palatable food in patients with schizophrenia. (b) Treatment with SGAs may improve dopamine-mediated reward aspects, but at the cost of increased appetite and insulin resistance along with worsened or at least not improved opiodergic capacity; these effects can further increase intake of palatable food.

Full size image

There are several lines of evidence that point to deficits of the brain circuits mediating reward and reinforcement in schizophrenic patients (Voruganti et al, 2001; Juckel et al, 2003; Voruganti and Awad, 2004). From a neuroanatomical perspective, reward circuitry structures including NAc, amygdala, and PFC have been reported to be abnormal in schizophrenia in neuropathological (Beckmann and Lauer, 1997; Bogerts et al, 1985; Lauer and Beckmann, 1997; Lauer et al, 2001; Selemon et al, 2002), in structural (Breier et al, 1992; Sigmundsson et al, 2001; Kawasaki et al, 2004), and in functional (see Table 2 for summary) neuroimaging studies.

Table 2 Neuroimaging Studies Pertinent to Reward Function in Patients with Schizophrenia
Full size table

The association between reward deficits and schizophrenia is also supported by the clinical phenomena typical of this illness: anhedonia, affective flattening, social isolation, avolition, apathy, high comorbidity of SUDs (Regier et al, 1990; American Psychiatric Association, 2000), and insensitivity to pain (discussed below). These characteristics support inclusion of schizophrenia within the spectrum of the ‘reward deficiency syndrome’ (Green et al, 1999), which is comprised of a broad range of personality traits and mental disorders (including, but not limited to schizoid/avoidant character styles, glucose binging and posttraumatic stress disorder) allegedly characterized by hypofunctionality of the reward circuitry, clinically manifested as diminution of drive/motivation and of capacity to enjoy or to experience pleasure (Comings and Blum, 2000).

SCHIZOPHRENIA AND UNHEALTHY FOOD PREFERENCES MAY BE EXPLAINED BY SHARED NEURAL SYSTEMS

Presently, no clinical studies directly link schizophrenia neuropathology and excessive consumption of palatable food. A cumulative body of evidence suggests, however, that (1) ‘predilection for addictive behavior’ is ingrained in schizophrenia neuropathology owing to ‘neural and motivational changes [that are] similar to long-term substance abuse but without the necessity of prior drug exposure’ (reviewed in Chambers et al, 2001) and (2) there is a ‘considerable overlap between brain circuitry that evolved in the service of body-weight regulation and brain circuitry that is usurped by exogenous drugs of abuse’ (reviewed in Volkow and Wise, 2005). The adaptation of these two ideas (discussed in the following sections) lends strength to the hypothesis that schizophrenia and unhealthy food preferences may be potentially explained by recursive partly shared neural systems.

NEURAL AND MOTIVATIONAL CHANGES IN SCHIZOPHRENIA AND IN SUDS

The first of the above-mentioned ideas (Table 3) is based on the neurodevelopmental model of schizophrenia (Weinberger, 1987) and is supported by neuroimaging research (Laruelle, 2000), and by ketamine and amphetamine challenge studies (Malhotra et al, 1997; Krystal et al, 2005). In addition, extensive preclinical work (Grace, 1991, 2000; Pierce and Kalivas, 1997) provides supportive evidence for distinct roles for tonic vs phasic dopaminergic firing and glutamatergic signaling in clinical symptomatology, motivational states, and in the outcomes of schizophrenia and SUDs (Lieberman et al, 1997; Kalivas and Volkow, 2005).

Table 3 Neural and Respective Motivational Changes in Schizophrenia and in SUDs
Full size table

‘Hypofrontality’ or decrease in cerebral metabolism and blood flow have been observed in both schizophrenia (Weinberger and Berman, 1996) and in SUDs (Volkow et al, 1992; Kalivas et al, 2005; Garavan et al, 2000) at baseline or when challenged by cognitive tasks or when exposed to natural reinforcers. The hypofrontality phenomenon has been potentially related to the disease process- (Lieberman et al, 1997; West et al, 2003; Laruelle, 2000) or drug- (Kalivas and Volkow, 2005) related diminished dopaminergic tone with corresponding decrease in the tonic glutamatergic activity.

On the background of this diminished activity, hallucinations (McGuire et al, 1993; Silbersweig et al, 1995; Shergill et al, 2000) or other positive symptoms (Sabri et al, 1997) in schizophrenic patients or exposure to drugs (Volkow et al, 2005) or drug-related cues (Grant et al, 1996; Wang et al, 1999; Childress et al, 1999; Garavan et al, 2000) in drug addicts produce strong activations in the PFC. Prefrontal activations caused by drugs or drug cues in addicts profoundly increase PFC's glutamatergic output (Kalivas and Volkow, 2005; Kalivas et al, 2005) to already hypofunctional NAc (Volkow et al, 1997, 2005), thus further decreasing its dopaminergic activity. This triggers the drive to seek and consume drugs (ie, craving) with resultant drug-induced phasic dopaminergic bombardment (Grace, 2000).

Similarly, in schizophrenics, homeostatic adjustments (Laruelle et al, 2003) to diminished resting state tonic dopaminergic neurotransmission in the PFC and in ventral striatum result in robust augmentations of phasic dopamine responses to drugs and to behaviorally relevant stimuli. The limbic hyperdopaminergic state is further amplified through the long-term plasticity (Pierce and Kalivas, 1997) process of endogenous sensitization (Lieberman et al, 1997; Laruelle et al, 1999; Laruelle, 2000; Chambers et al, 2001; West et al, 2003). This is an autonomous, self-sustaining feedforward loop whereby a trivial event (eg, psychosocial stress) can mount escalating dopamine releases in the striatum, further worsening symptoms of psychosis (Laruelle, 2000).

OVERLAP BETWEEN FOOD AND DRUG REWARD SYSTEMS

Table 4 is a summary of basic neuroscience and limited clinical findings that support the assumed overlap between food and drug reward systems (Berridge, 1996; Volkow and Wise, 2005). Consumption of palatable food (Hajnal et al, 2004) and drugs (Hernandez and Hoebel, 1988) activates dopamine transmission in the mesolimbic dopaminergic system, whereas repeated exposure to both stimuli decreases dopaminergic firing (Bassareo and Di Chiara, 1999; Nestler, 2004) in the same neural structures (ie, habituation and tolerance). Repeated intermittent exposure to food (Rada et al, 2005) or to drugs (Steketee, 2003) conversely results in sensitization, that is, increased dopaminergic release in the NAc. The commonality of the circuitries is further supported by crosssensitization between palatable food with drugs and alcohol (Carr et al, 2003; Avena and Hoebel, 2003a, 2003b; Avena et al, 2004).

Table 4 Neurotransmitter and Neuropeptide Systems Involved in Palatable Food- and Drug-Motivated Behaviors
Full size table

Palatable foods do not only engage dopaminergic reward system but also change it (Bello et al, 2002; Fetissov et al, 2002; Hajnal and Norgren, 2002; Bello et al, 2003). That is to say, qualitatively similar to the effects of psychosis or euphorogenic drugs, repeated artificial enhancement of dopaminergic neurotransmission in the reward system leads to a dysfunctional (Wang et al, 2001; Barrot et al, 2002; Bello et al, 2002; Hajnal and Norgren, 2002; Spangler et al, 2004) neuroadaptational state that renders it even less responsive to other natural reinforcers (Wang et al, 2001; Volkow et al, 2004b). Therefore, a common result in vulnerable individuals could be in a form of contribution to the ‘spiraling distress cycle’ (Goldstein and McEwen, 2002; Koob and Le Moal, 2001) depicted in Figure 4, whereby excessive palatable food consumption provoked by its enhanced salience and conditioned learning produces additional deterioration in emotional and behavioral problems, leading to further consumption that may eventually produce a transition from excessive eating to food addiction (Tuomisto et al, 1999).

Both palatable food (Tanda and Di Chiara, 1998; Colantuoni et al, 2002) and drugs (Tanda and Di Chiara, 1998; Roth-Deri et al, 2003) also release endogenous opiates, that is, endorphins and enkephalins. Their chronic consumption leads to neuroadaptations in opioid receptors accompanied by signs and symptoms of opioid dependence (Zubieta et al, 1996; Colantuoni et al, 2002; Bailey et al, 2005; Heinz et al, 2005). Only limited human data (Ibanez-Rojo et al, 1993; Bencherif et al, 2005) are available, though, to extend preclinical palatable food opioid findings to humans.

Neuroimaging studies in obese people reported increased PFC activity when food-related stimuli were presented (Del Parigi et al, 2002) in the face of decreased performance on a cognitive task recruiting PFC (Davis et al, 2004). As discussed above, hypo- (Volkow et al, 1992; Kalivas et al, 2005; Garavan et al, 2000) and hyperfrontality have been reported in drug addicts at baseline and during exposure to drugs (Volkow et al, 2005) or drug-related cues (Grant et al, 1996; Wang et al, 1999; Childress et al, 1999; Garavan et al, 2000), respectively. While the nature of these effects in obesity has not been elucidated, in addicts they have been attributed to drug-induced disintegration between dopaminergic and glutamatergic firing (Grace, 2000; Kelley and Schiltz, 2004; Kalivas and Volkow, 2005; Goto and Grace, 2005).

Although BDNF has been widely recognized as a neurotrophic growth factor (Huang and Reichardt, 2001) and orexin as a regulator of sleep/wake cycle (Takahashi, 1999), the opposing actions (anorexigenic and orexigenic, respectively) of these two neuropeptides in food and drug reward have been also identified. In mesolimbic dopaminergic pathways, BDNF regulates appetitive behavior (Eisch et al, 2003; Itoh et al, 2004) along with behavioral sensitization to drugs (Guillin et al, 2001) and potentially to palatable food. Through its hypothalamic effects, BDNF controls energy balance by enhancing catabolic processes (Xu et al, 2003). Similarly important are the actions of orexin neurons, situated in the LH and projecting to the NAc and VTA (Fadel and Deutch, 2002), to control food intake (Edwards et al, 1999; Harris et al, 2005) as well as drug reward and craving (Harris et al, 2005). In fact, a finding of a correlation between SGAs weight liability and the degree of LH orexin neurons stimulation (Fadel et al, 2002) is provocative and may offer important insights into homeostatic dysregulation associated with antipsychotic agents.

ABNORMAL DOPAMINERGIC FUNCTION IMPAIRS WANTING PROCESSES IN SCHIZOPHRENIA

As discussed above, dopamine theory of psychosis postulates that hyperactivity of mesolimbic dopaminergic pathways mediates the key characteristic symptoms of schizophrenia, delusions and hallucination (Laruelle et al, 1996; Breier et al, 1997; Abi-Dargham et al, 1998; Moore et al, 1999; Laruelle et al, 2003). To put it in more metaphoric terms: mesolimbic dopamine is conceptualized to be ‘the wind’ (Laruelle and Abi-Dargham, 1999), whereas delusions and hallucinations could be the smoke ‘of the psychotic fire’.

Given the central role played by mesolimbic dopamine in the wanting processes, psychotic symptoms in schizophrenic patients could originate from sporadic or stress-induced dopamine spikes causing attribution of salience to random events and/or to trivial subjective experiences (Taylor and Liberzon, 1999; Floresco et al, 2003; Kapur, 2003; Taylor et al, 2005), that is, aberrant conditioned learning (Schultz, 2001). Deficits of the neural mechanism involved in attribution of emotional salience in schizophrenics (Kohler et al, 2003; Moberg et al, 2003; Turetsky et al, 2003) are supported by functional brain imaging (summarized in Table 2) and electrophysiological (Pfefferbaum et al, 1989) studies, employing various natural probes ranging from odors to facial images and aversive vs neutral scenes (Taylor et al, 2002, 2005; Gur et al, 2002; Paradiso et al, 2003; Takahashi et al, 2004; Williams et al, 2004).

Similarly to SUDs, which is another condition associated with erratic supraphysiological dopamine surges in the limbic system (Volkow et al, 2004a), aberrant salience in schizophrenia creates rigid motivational states fixated on irrelevant and idiosyncratic stimuli with diminution of mesolimbic neurons’ ‘signal detection capability’ (Green et al, 2002) for normal salience (Taylor and Liberzon, 1999; Taylor et al, 2005) and with the loss of normal modulation of reinforcers’ values by nonpsychotic contexts (eg, caloric need state; Breier, 1989). Thus, palatable food (eg, ice cream) may be salient enough to compete with the psychosis-related reinforcers as a result of its effects on reward circuitry (Hajnal et al, 2004) and owing to aggressive and ubiquitous advertisement (Wansink, 2004), whereas less salient and palatable nutrition (eg, lettuce and broccoli), lacking these characteristics, may fail to elicit incentive motivation towards its consumption even within the context of caloric deficits (Breier, 1989). Preference for highly palatable food may be further enhanced, despite considerable efforts exerted by public health authorities to explain the risks of unhealthy diets, owing to a hypofunctional PFC involved in the inhibitory control (Volkow and Fowler, 2000; Goldstein and Volkow, 2002; Volkow et al, 2004a) over eating behavior.

ABNORMAL OPIOIDERGIC FUNCTION MAY IMPAIR LIKING PROCESSES IN SCHIZOPHRENIA

Robust elevations of endogenous opiates concentrations in the cerebral spinal fluid and in plasma (Terenius et al, 1976; Lindstrom et al, 1978, 1986; Rimon et al, 1980; Brambilla et al, 1984; Marchesi et al, 1995) is a relatively consistent clinical finding in schizophrenia. Opiates levels tend to parallel the severity of psychosis (Terenius et al, 1976; Lindstrom et al, 1978; Rimon et al, 1980; Marchesi et al, 1995), suggesting involvement of this neurochemical effect in schizophrenia neuropathology (Volavka et al, 1979; Schmauss and Emrich, 1985; Nemeroff and Bissette, 1988).

One possible involvement could be through opiate's interference with the neurotrophines (eg, BDNF) supporting neuronal survival (Weickert et al, 2003; Angelucci et al, 2005) with consequential damage of mesolimbic dopaminergic neurons (Sklair-Tavron et al, 1996). Furthermore, similarly to methadone-maintained patients (Willenbring et al, 1989; Zador et al, 1996), exaggerated opioidergic activity could enhance hedonic preference (ie, liking) for sweet and fatty foods (Doyle et al, 1993; Pecina and Berridge, 1995; Kelley et al, 2002; Will et al, 2003). The consumption of these foods further reduces BDNF efficiency in preventing neuronal death (Molteni et al, 2002) and in regulating reward function (Horger et al, 1999; Kernie et al, 2000; Nakagawa et al, 2003), glucose metabolism (Tonra et al, 1999; Nakagawa et al, 2000; Ono et al, 2000), appetitive behaviors (Eisch et al, 2003; Itoh et al, 2004), and other important homeostatic processes (Xu et al, 2003).

Another manifestation of alterations in endogenous opioids in schizophrenics may relate to pain insensitivity. This and other aberrations in protective mechanisms afforded by the pain system, noted in a number of reports (Davis et al, 1979; Evans, 1980; Davis and Buchsbaum, 1981; Fishbain, 1982; Bickerstaff et al, 1988; Rosenthal et al, 1990; Dworkin, 1994; Kudoh et al, 2000; Torrey, 2002), some of which date back to the days of Haslam (Haslam, 1798; 1809 cited in Torrey, 2002), Kraepelin (Kraepelin, 1919 cited in Hooley and Delgado, 2001), and Bleuler (Bleuler, 1924 cited in Hooley and Delgado, 2001), could be yet another aspect of excessive/altered endogenous opioid function in schizophrenia (Davis et al, 1979; Davis and Buchsbaum, 1981; Davis, 1983; Wiegant et al, 1992). Such a notion is supported clinically by reversal of pain insensitivity by opioid antagonism (Davis et al, 1979) and by molecular abnormalities in specific opioid genes, for example, prodynorphin (Ventriglia et al, 2002) or proenkephalin (Mikesell et al, 1996) in schizophrenic patients. Importantly, pain problems in schizophrenics are apparent in the context of real-life situations, for example, tissue damage following surgical treatment (Kudoh et al, 2000; Murthy et al, 2004), and their consequences range from finger burns caused by cigarette smoking (Jenkins et al, 1996) to such grave medical outcomes as silent myocardial infarction (Marchand, 1955) or delays in management of abdominal emergencies (Katz et al, 1990; Bickerstaff et al, 1988) that could be followed by perforated bowel (Rosenthal et al, 1990) and ruptured appendix (Geschwind, 1977).

ADVANTAGEOUS THERAPEUTIC PROPERTIES OF SGAS HAVE BEEN QUESTIONED

The first-generation antipsychotics (FGAs) improve the perceptual disturbances and other positive symptoms and can quieten down the ‘the wind of psychotic fire’ (Laruelle and Abi-Dargham, 1999) by aborting stress-induced dopamine releases (Laruelle, 2000). Their therapeutic properties are, however, tarnished by serious side effects, including those resulting from dopamine blockade in the cortical and subcortical areas with secondary worsening of reward and motivational deficits (Knable et al, 1997; Voruganti et al, 2001; Voruganti and Awad, 2004). In contrast, SGAs produce net increases in dopamine activity in the cortical dopaminergic pathways and substantially lesser blockade (in comparison to FGAs) of the limbic dopaminergic areas (Stahl, 2002).

Thus, all in all, SGAs share advantageous therapeutic properties for the treatment of motivational, emotional, and cognitive deficits (Stahl, 2002; Stahl and Grady, 2004) as well as for the reduction of dopaminergic sensitivity to stress (Dazzi et al, 2004). These qualities partially explain the emergence of SGAs during the past decade as the first line of choice in the pharmacotherapy of schizophrenia and related disorders (Stahl and Grady, 2004).

However, SGAs beneficial features have been put into question by recent studies. Numerous epidemiological surveys have found BWG to be a prevalent outcome in patients treated with clozapine, olanzapine, and to lesser degree quetiapine and risperidone, but not ziprasidone or aripiprazole (Allison et al, 1999b; American Diabetes Association, 2004; Kane et al, 2004; Newcomer, 2005). For clozapine, one study suggested that its advantageous therapeutic properties may be offset by serious morbidity and mortality associated with the BWG (Fontaine et al, 2001). The longest and the most comprehensive of the prospective trials randomized 1493 schizophrenics to treatment with olanzapine, risperidone, quetiapine, ziprasidone, or perphenazine, and found no differences in discontinuation rates (about 74% at 18 month) between the SGAs and the FGA, perphenazine (Lieberman et al, 2005). Although the medications were discontinued because of different reasons, olanzapine was mostly stopped because of the BWG or related metabolic side effects.

PUTATIVE MECHANISMS OF SGAS-INDUCED WEIGHT GAIN

The mechanisms of SGAs-induced weight gain are likely to be multifactorial and to involve both peripheral and central factors (summarized in Table 5). Among peripheral factors, glucoregulatory abnormalities probably play an important role (Newcomer, 2005). In vitro (Melkersson et al, 2001; Melkersson, 2004; Johnson et al, 2005, but Melkersson and Jansson, 2005) and preclinical studies (Ader et al, 2005) demonstrated that SGAs alter insulin secretion. However, in clinical samples, SGAs appear to spare pancreatic cells function (Ebenbichler et al, 2003; Laimer et al, 2005), but to change insulin signaling (Engl et al, 2005) and produce glucose intolerance and insulin resistance (Ebenbichler et al, 2003; Newcomer et al, 2002; Graham et al, 2005; Henderson et al, 2005). Although majority of glucose metabolism irregularities associated with SGAs therapy are secondary to weight gain (Newcomer, 2005), that causality could run in the opposite direction (Figure 4). That is to say, in addition to decreasing the ability of homeostatic and reward regions sensing of adiposity- or food-related signals (Figure 5b; Figlewicz, 2003a, 2003b, 2004; Berthoud, 2004b), insulin resistance increases body fat mass as it tends not to afflict adipose tissue that retains insulin sensitivity even in the face of other tissues’ resistance (Mizuno et al, 2004; Virtanen et al, 2005; Isganaitis and Lustig, 2005).

Table 5 Putative Mechanisms of SGA-Induced Weight Gain
Full size table

In contrast, SGAs’ leptin effects (Monteleone et al, 2002; Melkersson and Dahl, 2003; Ebenbichler et al, 2005; Murashita et al, 2005; Sporn et al, 2005; Theisen et al, 2005) seem to be entirely attributable to increased adiposity (Haupt et al, 2005; Sporn et al, 2005). In a cross-sectional study of plasma leptin and adiposity in schizophrenic patients treated with olanzapine (N=27), risperidone (N=24), or a typical antipsychotic (N=21), and in 124 healthy controls, BMI correlated with leptin levels in the entire sample. No SGAs effects on leptin were observed, despite adequate power, when SGAs-treated patients were individually matched to the controls by their BMI values (Haupt et al, 2005).

Increases (Murashita et al, 2005; Palik et al, 2005), decreases (Togo et al, 2004), and no change (Sporn et al, 2005; Theisen et al, 2005; Himmerich et al, 2005) in plasma ghrelin levels were reported in patients treated with SGAs; a consensus regarding clinical significance of these changes has not yet transpired. Likewise, potential contributions to weight gain of norepinephrine (Elman et al, 2002, 2004) and prolactin (Melkersson, 2005; Mann et al, 2006) increases associated with SGA treatment are still unclear. The findings of SGA effects on the brain BDNF (Bai et al, 2003; Luo et al, 2004; Angelucci et al, 2005) and orexin (Fadel et al, 2002) in preclinical literature may be also of potential importance as both neurochemicals are involved in reward and appetitive functions; however, their precise role and how they are influenced by SGAs in humans (Dalal et al, 2003; Weickert et al, 2005) are intriguing topics for future research. In addition, further studies are needed to examine a potential involvement of the endogenous cannabinoid system that has been implicated in schizophrenia (Weiser and Noy, 2005), in obesity (Engeli et al, 2005), and in the mechanisms of action of SGAs, for example, clozapine (Sundram et al, 2005).

EFFECTS OF SGAS ON FOOD INTAKE

Observational preclinical (Thornton-Jones et al, 2002; Ota et al, 2002; Hartfield et al, 2003a; Arjona et al, 2004; Cope et al, 2005) and clinical data suggest that one of the ways SGAs can lead to changes in the patterns of food intake is by increasing appetite (Figure 5b; Bromel et al, 1998; Briffa and Meehan, 1998; Eder et al, 2001; Basson et al, 2001; Kinon et al, 2005) and subsequent consumption of the previously preferred diet (Gothelf et al, 2002; Kane et al, 2004).

Additionally, SGAs can contribute to decreased caloric expenditure and subsequent BWG via diminished physical activity (Arjona et al, 2004) owing to their pronounced sedative effects (Lublin et al, 2005). However, the relative contribution of SGAs effects on energy intake and/or expenditure remains largely unquantified by experimental studies. To begin exploring this issue, Gothelf et al (2002) prospectively measured BMIs, caloric intake, and energy expenditure in 10 olanzapine-treated subjects and compared them to a matched group of patients treated with haloperidol. At the 4-week follow-up time point, the patients had increases in both caloric intake and in the BMI. Their resting and daily energy expenditure remained unchanged. The results of this pilot study add further support to the hypothesis that caloric intake plays an important role in the SGAs-induced weight gain.

An important candidate mechanism for SGAs-induced increases in appetite is blockade of receptors systems such as 5-HT2A, 5-HT2C, serotonin; D1, D2, D3, and D4 dopamine; H1 histamine, α1 and α2 adrenergic and acetylcholine (Van Tol et al, 1991; Zeng et al, 1997; Seeman et al, 1997; Wirshing et al, 1999; Meltzer, 1999; Richelson, 1999; Richelson and Souder, 2000; Casey and Zorn, 2001; Muller et al, 2004; Kroeze et al, 2003; Templeman et al, 2005; Johnson et al, 2005; Reynolds et al, 2005), involved in appetite regulation in hypothalamus and elsewhere in the brain. Greater part of the receptor data are based on preclinical models (eg, knockout mice). However, some but not all (Tsai et al, 2002; Basile et al, 2002) clinical studies implicate polymorphism of the 5-HT2C receptor gene in BWG in response to SGAs (Reynolds et al, 2002, 2005; Ellingrod et al, 2005). In addition, one study (Wirshing et al, 1999) found a logarithmic relationship between H1 receptor affinity in patients treated with clozapine (N=20), olanzapine (N=13), risperidone (N=38), haloperidol (N=43), and sertindole (N=8). This study is echoed by another article (Kroeze et al, 2003), demonstrating that H1 receptor affinity predicted weight gain for both FGA- and SGA-treated patients enrolled in meta-analysis of 418 antipsychotic drug-induced estimates of weight change (Allison et al, 1999b).

Nevertheless, rank ordering of specific SGAs’ propensities to increase body weight against all their binding affinities at the human brain receptors argues against antagonism of a single receptor as a sufficient cause for SGA-induced BWG (Richelson, 1999; Richelson and Souder, 2000). For instance, an SGA that does not appear to produce BWG, ziprasidone, also binds with substantial affinity to the 5-HT2C and H1 receptors (Richelson, 1999; Richelson and Souder, 2000; Pouzet et al, 2003; Kirk et al, 2004). Hence, it is possible that simultaneous blockade of several (rather than single) receptor, for example, H1, 5-HT2, D2, and acetylcholine receptors may have a synergistic effect on increased appetite and food intake (Masand, 1999; Wetterling, 2001; Hartfield et al, 2003b).

OPIOID ANTAGONISTS MAY IMPROVE HEDONIC DEFICITS IN PATIENTS WITH SCHIZOPHRENIA

Dopamine remains the most extensively investigated neurotransmitter system in the mechanism of the SGAs action (Kapur and Mamo, 2003). However, among the complex concoction of their pharmacological properties, some SGAs (clozapine, olanzapine, and risperidone) may also be associated with enhancements in the opioidergic activity, which could further deteriorate the pre-existing hedonic alterations in patients with schizophrenia.

Several lines of clinical and preclinical evidence support this sort of effect. These include clinical presentation of olanzapine overdose, which is indistinguishable from opioid intoxication (O’Malley et al, 1999; Kochhar et al, 2002; Palenzona et al, 2004) and analgesic/antinociceptive properties of clozapine, olanzapine, and rispiridone observed in both human (Kiser et al, 2001; Silberstein et al, 2002; Fe-Bornstein et al, 2002; Khojainova et al, 2002; Gorski and Willis, 2003) and rodent (Schreiber et al, 1997, 1999; Weizman et al, 2003) models and ascribed to the opioid mechanisms (Schreiber et al, 1997, 1999; Weizman et al, 2003). Taken together, these results suggest that SGAs are likely to have positive impact on the incentive/motivational reward aspects (including drive to procure food), but at the cost of worsened, or at least not improved opioid-mediated hedonic capacity.

If excess of central opioid activity, which is consequential to the schizophrenia neuropathology with or without SGAs pharmacotherapy, creates metabolic problems for the patients, it is reasonable to expect amelioration of the symptoms through blockade of opioid receptors. Opioid antagonism enhances sensitivity of opioid receptors (Zukin et al, 1982; Lesscher et al, 2003) and diminishes tolerance and physical dependence on opioids (Powell et al, 2002). In rodent studies, opioid antagonists block hedonic responses to sweet foods and sugar-induced analgesia (Blass et al, 1987; Blass and Fitzgerald, 1988) without affecting total energy intake (Berridge, 1996; Rodefer et al, 1999).

Moreover, all but one human study (Hetherington et al, 1991) employing opioid receptor antagonists, including naloxone (Thompson et al, 1982; Trenchard and Silverstone, 1983; Cohen et al, 1985; Drewnowski et al, 1992; MacIntosh et al, 2001), naltrexone (Fantino et al, 1986; Jonas and Gold, 1986; Melchior et al, 1989; Bertino et al, 1991; Chatoor et al, 1994; Yeomans and Gray, 1996, 1997), and nalmefene (Yeomans et al, 1990; Yeomans and Wright, 1991), found significant decreases in caloric intake (Yeomans and Gray, 2002).

In clinical populations, opioid antagonism successfully curbed excessive food intake in bulimic patients (Jonas and Gold, 1986; Chatoor et al, 1994; Marrazzi et al, 1995a, 1995b) as well as opioid (Shufman et al, 1994) and alcohol (O’Mara and Wesley, 1994) abuse. Above and beyond their dietary effects, opioid antagonists may also contribute to weight reduction by reasons of diminished hypeinsulinemia and improved insulin sensitivity (Cucinelli et al, 2002; Fruzzetti et al, 2002). Overall, these findings support the hypotheses that contrary to dopamine antagonism that worsen reward function (Knable et al, 1997; Voruganti et al, 2001; Voruganti and Awad, 2004), blockade of opioid receptors may actually improve it and that these agents can be successfully used in clinical settings.

On the basis of the hypothesis that opioidergic mechanisms are involved in the pathophysiology and symptoms of schizophrenia (Volavka et al, 1979; Schmauss and Emrich, 1985; Nemeroff and Bissette, 1988), opioid antagonists were evaluated for potential antipsychotic efficacy in numerous clinical trials (McNicholas and Martin, 1984; Nemeroff and Bissette, 1988; Welch and Thompson, 1994). On the whole, they were safe and well tolerated and resulted in significant improvements of various aspects of schizophrenic symtomatology such as positive and deficit symptoms, polydipsia, alcoholism, and tardive dyskinesia in some (Watson et al, 1978; Lehmann et al, 1979; Berger et al, 1981; Nishikawa et al, 1994a, 1994b; Marchesi et al, 1995; Petrakis et al, 2004; Wonodi et al, 2004), but not all (Volavka et al, 1977; Pickar et al, 1982) participants.

Possible explanations for these partial treatment responses may have to do with varying dosing strategies, oral vs injectable preparations, open-label vs blind designs, different lengths of treatments, and use of the antagonists as sole agents vs augmentation strategies of various antipsychotic medications. Thus, double-blind placebo-controlled dose–response trials employing the only available (and therefore mostly clinically relevant) oral antagonist form, naltrexone, as an add-on treatment to a standard SGA (eg, olanzapine) may be an important future direction. The potential clinical utility of naltrexone for reward deficiency spectrum problems in schizophrenia is supported by its ability to curb alcohol consumption when administered in a double-blind placebo-controlled manner to comorbid schizophrenic/alcohol-dependent patients on both SGAs and FGAs (Petrakis et al, 2004). In short, these preliminary data suggest that blockade of opioid receptors may have beneficial effects in patients with schizophrenia, especially in the domain of hedonic alterations, but without impairing motivational aspects of reward or aggravating positive symptoms. Definite treatment recommendations are, however, pending rigorously designed clinical trials.

CONCLUSIONS

Overeating, driven by ubiquitously available fast food, appears to be a major cause of the obesity epidemic in the industrialized world. The coalescence of preclinical, neuroimaging, and clinical data suggests that schizophrenic patients may have a predilection for excessive consumption of fast food-type nutrition owing to functionally impaired neural substrate comprising homeostatic and reward mechanisms. Treatment with SGAs does not fully address this problem and may even worsen it.

If confirmed in clinical studies, our insights could have important implications for the primary and secondary prevention of obesity in schizophrenia. If neurobiologic vulnerability factors for obesity could be identified, they might be used to screen patients at risk for the development of such condition. Patients found to possess high vulnerability for the development of obesity owing to schizophrenia-related alterations in glucoregulatory or reward function might be counseled to avoid excessive sweet and fat consumption (primary prevention), or targeted for early intervention even in the presence of mild weight problems (secondary prevention).

The proposed model of excessive BWG in schizophrenia could also have treatment implications, as it implies use of opioid receptors antagonists for amelioration of weight problems in schizophrenics. Clinical experience suggests that opioid antagonists utilization is lower than what could be expected from positive outcomes of clinical trials (Yeomans and Gray, 2002). We therefore believe that this Perspective Article will equip practitioners with the essential knowledge base for understanding the rationale for naltrexone therapy and will sharpen their sensitivity to the unmet medical needs of their schizophrenic patients. Lastly, this model may also provide important leads for recognition and treatment of hazardous eating habits in patients with other disorders, including obesity (Wang et al, 2001), SUDs (Kampov-Polevoy et al, 2001), and major depression (Papakostas et al, 2005).