Endogenous opioids and µ-opioid receptors (MORs) have long been implicated in the mechanism of appetite control and, in particular, hedonic processes associated with food evaluation, consumption and orosensory reward processes. In animal models of binge eating, selective MOR antagonists suppress food consumption. In humans, non-selective opioid receptor antagonists reduce hedonic taste preferences and food intake, particularly for palatable foods, and cause short-term weight loss. These effects have been linked to direct stimulation of MORs and modulation of dopamine release within the reward circuitry including the nucleus accumbens. These findings suggest that reduction of MOR-mediated hedonic and motivation processes driving consumption of highly palatable foods may be a promising therapeutic approach and provide a strong rationale for developing safer and more selective MOR antagonists or inverse agonists for disorders of ‘appetitive motivation’ including obesity and binge-eating disorder.
Excessive binge-eating is a fundamental behavioural abnormality in disorders of ‘appetitive motivation’ including obesity and binge-eating disorder (BED). The opioid system and, in particular, µ-opioid receptors (MORs) have been linked to hedonic aspects of eating and orosensory reward processes. This paper reviews findings linking MORs with binge-eating behaviour. Findings with MOR agonists and antagonists in animal models of food consumption and binge-eating behaviour, and clinical findings of energy intake, binge eating and weight loss in obesity, bulimia, and BED are discussed. These findings are integrated with the literature on MOR regulation of hedonic and motivational processes within critical reward pathways including the nucleus accumbens.
Opioid receptors and endogenous opioids
The first report of a mammalian endogenous opioid receptor that bound to opiates was made more than 35 years ago (Pert & Snyder, 1973). Subsequently, three subtypes of opioid receptors [mu (µ), delta (δ) and kappa (κ)] have been pharmacologically characterized and genes encoding these receptors have been cloned (Kieffer, 1995; Waldhoer et al.2004). In addition three major endogenous opioid peptide gene-related families have been identified [i.e. Pre-proopiomelanocortin (β-endorphins), Pre-proenkephalin (enkephalins), Pre-prodynorphin (dynorphins)], with varying affinities for µ, δ and κ receptors. Together, these endogenous peptides and their receptors encompass the endogenous opioid system (EOS). MORs are G protein-coupled receptors (GPCRs) that are predominantly coupled to heterotrimeric Gi/Go proteins. Activation of these receptors triggers multiple cellular effector systems including inhibition of adenylyl cyclases and activation of mitogen-activated protein kinase (MAPK) cascade (Waldhoer et al.2004). MORs are largely distributed within brain regions mediating food intake and reward including nucleus accumbens and amygdala (Mansour et al.1995).
Binge-eating behaviour and disorders of binge eating
Binge eating is a maladaptive feeding behaviour associated with a number of eating disorders such as obesity, bulimia and BED. It usually consists of eating highly palatable, highly caloric foods that are rich in sweets, fats or both and may have little nutritional value. According to DSM-IV criteria, binge eating is defined as eating a large amount of food (compared to normal eating in most individuals) in a limited period of time (i.e. <2 h) (APA, 1994).
Recent studies suggest that as much as 6.6% of the normal population engages in binge-eating behaviour (Grucza et al.2007; Hudson et al.2007). Binge-eating behaviour is also a key component of obesity (Stunkard, 1959), which affects up to 33% of the US adult population (Ogden et al.2007) and bulimia, which affects 0.50–1.5% of the general population (Hudson et al.2007). A review of studies in obese patients suggests that binge-eating frequency is between 23% and 46% of those seeking treatment for weight reduction (De Zwaan & Mitchell, 1992a, b). BED is a newly identified clinical syndrome that is characterized by recurrent episodes of binge eating accompanied by feelings of loss of control and marked distress in the absence of regular compensatory behaviours such as self-induced vomiting. Based on the DSM-IV-TR research criteria, epidemiological studies have shown BED to be the most common of the eating disorders with life-time prevalence of 2–3% in the general population (3% in women and 2% in men) (Hudson et al.2007). Further, obesity is seen in 65% of patients with BED with increased progression over time and continued binge eating (Striegel-Moore et al.2001).
Treatments for binge-eating behaviour
A number of pharmacological treatments with differing mechanisms of action have been examined for binge eating in a number of disorders, particularly BED (Reas & Grilo, 2008). These include: the selective serotonin reuptake inhibitors (SSRIs), fluvoxamine (Hudson et al.1998; Pearlstein et al.2003), sertraline (McElroy et al.2000), fluoxetine (Arnold et al.2002; Devlin et al.2005; Grilo et al.2005), citalopram (McElroy et al.2003a) and escitalopam (Guerdjikova et al.2008); the noradrenaline reuptake inhibitor (NRI), atomoxetine (McElroy et al.2007a); the anti-obesity agents, d-fenfluramine (a serotonin releaser and reuptake inhibitor) (Stunkard et al.1996), sibutramine (a serotonin and noradrenaline reuptake inhibitor) (Appolinario et al.2003; Milano et al.2005; Mitchell et al.2003; Wilfley et al.2007) and orlistat (Golay et al.2005); and the anticonvulsants, topiramate (McElroy et al.2003b, 2007b) and sonisamide (McElroy et al.2006). All of these drugs have been shown to exert positive effects (of moderate effect size) including reductions in binge-eating frequency and short-term reductions in weight (∼3 kg) (Reas & Grilo, 2008), with larger effects on weight observed only with sibutramine, orlistat, topiramate and sonisamide. However, most of these drugs have been associated with increased adverse effects and relatively high discontinuation rates. Furthermore, while modest effects on binge eating were observed, effects on weight loss have been poor and the long-term effects are yet to be determined.
Animal models of binge eating and effects of µ-opioid antagonists
Endogenous opioids have long been implicated in the mechanism of appetite control. The earliest evidence implicating endogenous opioids in the neurobiology of ingestive behaviour was a report by Holtzman (1979), which showed that the non-selective opioid receptor antagonist, naloxone, caused significant reduction in short-term food intake. More recent studies have examined effects of opioid antagonists on binge-eating behaviour.
Various binge-eating models employing dietary restraint have been used to probe binge behaviour in animals (Howard & Porzelius, 1999). The models posit that intermittent disruption of food intake, whether by restricting daily intake of caloric ration (Hagan et al.2003), the duration of daily food access (Giraudo et al.1993; Inoue et al.2004), combining food restriction and environmental stress (Hagan et al.2002), or intermittent sugar and chow (Avena, 2007), is critical for bingeing. Restriction models of binge eating are designed to simulate dieting and cessation or ‘breaking’ of a diet with palatable food that is typical of binge eating. In one such model, rats are given 66% of the mean daily chow intake for a number of days (i.e. 5 d) and then ad-libitum chow and highly palatable food for 2 d then only chow for 4 more days in their home cages (Hagan et al.2003). Food restriction and stress models of binge eating are designed to induce synergistic interactions between dieting and stress. In this model rats are given 66% of mean daily chow for 4 d and then re-fed for 6 consecutive days on ad-libitum chow. On the final day of refeeding, animals are exposed to stress (i.e. intermittent shock) (Hagan et al.2003). Chronic feeding restriction models of binge eating are designed to examine the adaptive behavioural changes. In these models, rats undergo time-restricted scheduled feeding, consisting of chronic food restriction (i.e. 2-h access to palatable chow per day for 2 wk) with subsequent free refeeding for 2 wk (Inoue et al.2004). The intermittent sugar and chow model of binge eating is also a chronic feeding restriction model. In this model, rats are maintained for 1 month on a diet of 12-h daily access to an aqueous 10% sucrose solution and laboratory chow, followed by 12-h deprivation (Avena, 2007).
Using these models, opioid receptor antagonists (in particular MOR antagonists) have been shown to suppress food bingeing (Barbano & Cador, 2006; Bodnar et al.1995; Cooper, 1980; Davis et al.1983; Giraudo et al.1993; Glass et al.2001; Hadjimarkou et al.2004; Hagan et al.1997; Kelley et al.1996; Levine & Billington, 1997; Marks-Kaufman & Kanarek, 1981; Marks-Kaufman et al.1985; Morley et al.1980; Pecina et al.2000). An alternative method that emphasizes the qualitative aspects of dietary restraint is restricting binge eaters from ‘forbidden’ palatable foods (Corwin, 2006; Fletcher et al.2007; Gonzalez & Vitousek, 2004; Kales, 1990; Knight & Boland, 1989; Mitchell & Brunstrom, 2005; Stirling & Yeomans, 2004). This approach leads to ‘relapse’ intake limited to very brief binge episodes. In addition to binge hyperphagia of forbidden palatable foods, hypophagia for otherwise acceptable alternative foods develops (known as anticipatory negative contrast or devaluation in animals and analogous to human finickiness) (Pliner et al.1990). This has been thought to be related to associative learning and behavioural adaptations to sensory-hedonic experience with food across time. Using this model, Cottone et al. (2008) recently found that the µ/κ-opioid receptor antagonist nalmefene not only attenuated binge behaviour (for the preferred diet of chocolate-flavoured high sucrose), but also increased food intake of the less preferred diet (i.e. chow). It was hypothesized that these effects may be mediated by inhibiting MORs in the ventral tegmental area (VTA), leading to disinhibition of GABAergic interneurons and subsequently leading to decreased dopamine release in the shell of the nucleus accumbens (MacDonald et al.2003, 2004; Taber et al.1998) (Fig. 1 and Discussion in section on Opioid neural circuitry).
MORs and hedonic and motivational aspects of eating
It is widely agreed that elements of the EOS within a circuitry involving the lateral and ventral regions of the striatum (ventrolateral striatum, lateral/rostrodorsal medial shell and core), hypothalamic areas, VTA, the substantia nigra and the nucleus of the solitary tract plays a particular role in hedonic processes associated with food evaluation, consumption and orosensory reward processes (Baldo & Kelley, 2007; Berridge, 1996; Herz, 1998; Kelley et al.2002; Levine & Billington, 1997; Reid, 1985, Zhang et al.1998; Zhang & Kelley, 2000). This is supported by a number of studies that have shown that opioid peptides regulate the intake of highly palatable foods, and opioid receptor antagonists like naloxone reduce food intake of palatable diets and are more potent at reducing intake of palatable food and drink compared to non-palatable food and water, respectively (Apfelbaum & Mandenoff, 1981; Barbano & Cador, 2006; Cooper, 1983; Cooper & Turkish, 1989; Cooper et al.1985; Giraudo et al.1993; Glass et al.1996, 2001; Hayward et al.2006; Kelley et al.2002; Levine et al.1982, 2003; Lynch & Libby, 1983; Ward et al.2006; Zhang & Kelley, 1997). The essential role of MORs on hedonic eating is further supported by the evidence that decreased consumption of palatable sucrose in rabbits following MOR antagonism is associated with specific loss of coupling of MORs to their G proteins in the nucleus accumbens (Ward et al.2006) and efficacy of opioid antagonists at reducing food intake is substantially reduced in MOR knockout (KO) animals (Zhang et al.2006).
In direct support of the studies with MOR antagonists, endogenous opioids and opioid agonists administered within the nucleus accumbens have been linked to the hedonic or pleasurable aspects of palatable food consumption (Baldo & Kelley, 2007). Studies have shown that direct stimulation of MORs with MOR agonists such as morphine or DAMGO ([d-Ala2, N-Me-Phe4,Gly5-ol5]-enkephalin) within the nucleus accumbens of rats preferentially increases intake of energy rich foods such as fat and sucrose, as well as tasty non-caloric foods such as saccharin and salt (Bakshi & Kelley, 1993; Bodnar et al.2005; Evans & Vaccarino, 1990; Kelley et al.2002; Majeed et al.1986a; Mucha & Iversen, 1986; Will et al.2003; Zhang et al.1998; Zhang & Kelley, 1997, 2000). In addition, findings indicate that opioids administered in the nucleus accumbens (rostrodorsal medial shell), increased or amplified positive affective reactions (i.e. liking reactions) to sucrose taste (Pecina & Berridge, 2005), suggesting that the EOS encodes the hedonic properties of food intake.
At least at the level of the nucleus accumbens, there is evidence for dissociation between opioid regulation of hedonic aspects of palatable food ingestion and dopamine-regulated goal-directed motivational behaviour (Baldo & Kelley, 2007). However, it is possible that hedonic and motivational aspects of food intake are interconnected physiological processes with changes in hedonic aspects leading to modulation of motivational incentive value of rewarding stimuli. For example, Papaleo et al. (2007) showed a decrease in motivational aspects of food intake in MOR-deficient mice (i.e. MOR KO mice), in a nose-poke operant paradigm involving fixed-ratio and progressive-ratio reinforcement, where less operant behaviour to obtain food pellets was observed in the KO mice. MOR KO mice also show diminished food anticipatory activity which is thought to be associated with dopamine neurotransmission and motivational processes (Kas et al.2004). While intra-accumbens infusion of a µ/δ-receptor agonist failed to enhance instrumental responding for conditioned food reinforcement (i.e. lever pressing for a stimulus previously associated with food) (Cunningham & Kelley, 1992), and does not support learning of an instrumental response in rats fed ad libitum (Hanlon et al.2004), administration of the more selective MOR agonist DAMGO, has been shown to increase break point (i.e. the reduction in lever-press responding to sucrose in a progressive-ratio schedule of an operant task used to assay incentive motivational properties of food) of sated rats working for palatable food. Furthermore, mice lacking β-endorphin, enkephalin or both have been shown to have decreased levels of motivation to obtain food when sated but not when food deprived (Hayward et al.2002). Finally, a number of operant studies have reported that opioid receptor antagonism with naloxone decreases intake of preferred (sucrose based) and non-preferred diets in rats and monkeys (Cleary et al.1996; Rudski et al.1994, 1997; Williams et al.1998; Zhang et al.2003) and decreases running performance for palatable food in sated rats (Barbano & Cador, 2006). These findings suggest that MORs may indeed play a role in the translation of the hedonic properties of palatable food into increased motivation to eat when in contact with palatable food; however, administered in isolation, they are incapable of triggering a motivational response to support new learning.
Opioid neural circuitry controlling hedonic and motivational eating
Neurochemical studies in reward-related neural circuits provide support for the possibility that hedonic and motivational aspects of food intake are physiologically interconnected processes. A dense distribution of MORs is found in hedonic ‘hot-spot’ areas of the nucleus accumbens, such as the rostral and dorsal medial shell (Tempel & Zukin, 1987). The nucleus accumbens, located within the ventral striatum, is ideally positioned to integrate the affective assessment of food with executive and cognitive processes (Cardinal et al.2002; Kelley, 1999; Kelley et al.2005). It receives taste and visceral information from the brainstem (i.e. nucleus of the solitary tract), signals conveying internal homeostasis from the arcuate nucleus (ARC) and lateral hypothalamus (LH), and convergent input from cortico-limbic areas associated with higher-order cognitive and emotional processing, including the amygdala, prefrontal cortex, gustatory cortex, and hypothalamus. The nucleus accumbens also sends efferent projections to the extrapyramidal motor control regions, with output from the accumbens core reaching basal ganglia motor control circuits while outputs from the accumbens shell reach the ventral pallidum and LH, the outputs of which reach structures that modulate brainstem control of eating (Kelley et al.2005). Areas like the amygdala are thought to amplify the hedonic value of palatable food, linking motivational value of the stimulus with goal-directed behaviour (Cardinal et al.2002; Kelley et al.2005; Will et al.2004). Similarly, the LH has been shown to be necessary for expression of opioid-induced intake of palatable food (Will et al.2003). This anatomical network is suggested to underlie the translation of motivational signals into behavioural output (Mogenson et al.1980). In support of the receptor distribution within this neural network, consumption of a high-fat diet or excessive sugar intake has been shown to be associated with increases in MORs in the nucleus accumbens, cingulate cortex, hippocampus and the hypothalamus (Barnes et al.2003; Colantuoni et al.2001; Smith et al.2002).
There is also substantial evidence for interactions between the opioid system and the dopaminergic system, which is critical for goal-directed motivational behaviour. MORs are localized on GABAergic interneurons in the VTA that normally inhibit the dopamine system. MOR stimulation inhibits this input leading to disinhibition of the dopaminergic system, causing greater dopamine release in the nucleus accumbens and other dopaminergic target areas (Kalivas, 1993). Inhibiting opioid receptors in the VTA causes disinhibition of GABAergic interneurons and subsequently leads to decreased dopamine release (Fig. 1). Similarly, in the nucleus accumbens, MORs are found on GABAergic interneurons that receive input from the mesolimbic dopamine system. Furthermore, MOR KO mice demonstrate decreased firing frequency (including reduced bursting activity) of midbrain dopamine neurons (Mathon et al.2005a) and these dopamine neurons also receive increased GABAergic input (Mathon et al.2005b). It is possible that reduced burst activity in dopamine neurons in MOR KO mice may in part be due to increased GABAergic inhibition of these neurons. These electrophysiological observations are further supported by findings of decreased dopamine reuptake in the nucleus accumbens in MOR KO mice (Chefer et al.2003).
The acquisition of hedonic feeding is also thought to involve the activation of the mesolimbic dopamine system. For example, behavioural studies in animals show that during certain feeding conditions, including binge eating and limited access to palatable food, there is an increase in dopamine release in the nucleus accumbens (shell) (Bassareo & Di Chiara, 1997, 1999; Gambarana et al.2003; Rada et al.2005; Sahr et al.2008; Taber et al.1998). Interestingly, the effects on consumption of palatable food (Sahr et al.2008), as well as food-induced dopamine release in the accumbens were also shown to be attenuated by the opioid receptor antagonists, naltrexone (Taber et al.1998) and LY255582 (Sahr et al.2008). Together, these findings provide further evidence for an interaction between the opioid and dopaminergic systems, with the possibility that enhancement of the hedonic (‘liking’) properties of palatable food (µ-opioid related) might be translated, potentially via associative learning mechanisms, into increased motivation (‘wanting’) to eat (dopamine related).
In animals food restriction followed by access to food has been shown to increase (i.e. sensitize) MOR binding in limbic regions, including the cingulate cortex, hippocampus and the nucleus accumbens shell (Colantuoni et al.2001). Food deprivation (for 48 h but not shorter periods) in rats has also been shown to increase mRNA expression of MORs in the ventral medial hypothalamus and ARC (Barnes et al.2008). These finding suggests that repeated activation of MORs (and associated signalling) by food restriction and binge eating may help sustain behaviour, including desire to acquire a high-fat diet (Barnes et al.2008). It is possible that sensitization or up-regulation of MOR and interactions with dopamine in areas such as the nucleus accumbens, which are thought to play a role in incentive motivation and future goal-directed behaviour, may mediate such behaviours. This is supported by the finding of increased dopamine D1 receptors in the accumbens core and shell in the former study which also saw an increase in MORs (Colantuoni et al.2001).
In humans, MORs have been quantified using [11C]carfentanil PET imaging. Contrary to the animal studies described above, a reduction in MOR binding in the insula cortex in patients with bulimia has been reported and this correlated inversely with fasting behaviour (Bencherif et al.2005). The reduction in MORs may be related to a state-related down-regulation of the receptors subsequent to fasting or may even reflect a state of craving as shown in one study linking craving for alcohol to reduced MOR binding in the right dorsolateral prefrontal cortex, right anterior frontal cortex and right parietal cortex (Bencherif et al.2004). However, other studies found elevated MOR binding in several brain regions that correlated with either alcohol or cocaine craving (Gorelick et al.2005; Heinz et al.2005; Zubieta et al.1996). For example, elevated MOR availability has been reported in the ventral striatum following 1–3 wk alcohol abstinence, and this also correlated with the severity of craving (Heinz et al.2005). Similarly, 1–4 days of cocaine abstinence has been shown to increase MOR binding in a number of regions including the frontal cortex and cingulate cortex and these changes correlated positively with severity of craving (and persisted after 4 wk of abstinence) (Gorelick et al.2005; Zubieta et al.1996). These studies support observations in animals linking compensatory up-regulation of MORs with excessive alcohol intake (Cowen & Lawrence, 1999; Pecina & Berridge, 2000).
Endogenous opioids and hedonic eating
Although there is overwhelming evidence for µ-opioid regulation of palatable food intake, relatively little is known about the relationship between endogenous opioids and food intake or food deprivation, mainly due to technical difficulties in conducting microdialysis studies of peptides in areas such as the nucleus accumbens. An alternative approach has been to measure gene expression of opioids including enkephalins (ENK) and dynorphins (DYN). Using this approach, Chang and colleagues showed that an acute 4-h exposure to high fat compared to low fat increased ENK and DYN mRNA in the paraventricular nucleus (PVN) (Chang et al.2004), although the increase in ENK was not noted following chronic ingestion of high-fat diet (Welch et al.1996). In the ARC no changes in ERK was noted following acute or chronic diet consumption (Chang et al.2004; Welch et al.1996), while in the nucleus accumbens, a suppression of ENK mRNA was shown following chronic high-fat food intake (3 h/d for 2 wk), with no changes following acute intake (Kelley et al.2003). In a study examining variable periods of food intake, high-fat diet was found to increase PVN ENK and DYN mRNA following 15 min, 60 min, 1 d and 1 wk (Chang et al.2007). The DYN findings support other studies showing similar increases in PVN DYN mRNA following chronic intake (1–7 wk) (Levin et al.2002; Welch et al.1996). Compared to PVN, ARC was shown to be less responsive to high-fat intake with no changes in ENK and DYN mRNA in the shorter periods (15 min, 60 min) and small increases following 1 d and 1 wk of food intake (Chang et al.2007). The findings on hypothalamic peptides following high-fat food intake is supported by findings showing opposite changes following food deprivation for ⩾48 h, where decreases in mRNA for β-endorphin as well as proDYN, proENK and proopiomelanocortin (POMC) in the hypothalamus including the ARC have been reported following 7–14 d food deprivation (Bi et al.2003; Kim et al.1996; Knuth & Friesen, 1983; Welch et al.1996). This is also consistent with the recent evidence that mRNA expression of MORs is increased in the ARC and ventromedial hypothalamus following 48 h food restriction (Barnes et al.2008). However, the studies also suggest that the duration of food restriction plays an important role in levels of peptides as 24-h food restriction has been shown to increase β-endorphin in the hypothalamus (Majeed et al.1986b) and no changes in MOR mRNA expression has been observed (Barnes et al.2008).
The findings reviewed above suggest that the responses of specific hypothalamic and striatal regions to high-fat food intake vary considerably, and these changes in turn are modulated by duration of food intake. The precise nature of the changes in peptide release in specific hypothalamic and striatal areas, their timing and how these relate to changes in MORs and changes in food intake still remain to be established. One possibility is that an increase in energy state signalling from the ARC and LH may lead to increased opioid peptide gene expression (ENK and DYN) in the PVN and this may serve to stimulate feeding. Within the nucleus accumbens, the up-regulation of MORs following chronic food ingestion may contribute to the compulsive intake of highly palatable foods (potentially via interactions with the dopamine system) and increased signalling via MORs might subsequently lead to compensatory down-regulation of ENK transcription within the nucleus accumbens (Kelley et al.2003). This down-regulation within the nucleus accumbens (and subsequent disinhibition of LH via enkephalin-releasing axon collaterals arsing from striatal medium spiny neurons) has been suggested to be a possible mechanism leading to a modulation of hedonic aspects of food and resultant impact on food intake (Kelley et al.2003, 2005), although this hypothesis requires further testing and confirmation.
Hedonic binge eating in humans and effects of µ-opioid antagonists
A number of human studies have provided some support for the animal studies linking the EOS and food intake as well as the hedonic aspects of food reinforcement. The opioid receptor antagonists, naloxone, naltrexone and nalmefene (administered acutely) have been shown to be effective in decreasing short-term food intake (decreases between 11% and 29%) (Yeomans & Gray, 2002). Similarly, studies investigating sensory properties of food have shown that acute administration of opioid receptor antagonists cause a small, but significant reduction in affective or subjective pleasantness of palatable foods (including sucrose solution, sugar and fat mixtures, sweetened milk and actual food) in healthy subjects (Arbisi et al.1999; Fantino et al.1986; Yeomans et al.1990; Yeomans & Gray, 1996, 2002; Yeomans & Wright, 1991) and binge eaters (Drewnowski et al.1992, 1995). These studies support the animal studies indicating that the EOS may modulate physiological processes underlying the expression of hedonic responses to food. However, a number of other studies have shown no effect of the opioid receptor antagonist naltrexone on the palatability of sweet nutrients/food (Bertino et al.1991; Hetherington et al.1991). Furthermore, minimal effects have been reported on gustatory or olfactory perception, taste detection or recognition thresholds or rated blandness, sweetness and saltiness ratings of food stimuli, following opioid receptor antagonists (Drewnowski et al.1995; Yeomans & Gray, 2002). Overall the findings suggest that opioid receptor antagonists have a small effect on hedonic taste preference and short-term food intake, but have little effect on gustatory or olfactory perception or recognition of taste per se.
Opioid receptor antagonists have been shown to be effective in reducing the frequency and severity of binge eating. Early open-labelled studies in bulimic patients showed reduced binge size and frequency following naltrexone administration (of up to 6 wk administration) (Jonas & Gold, 1986, 1988a, b). A subsequent double-blind placebo-controlled trial in bulimic patients with naltrexone (100 mg) showed improvements in most patients on binge-related indices, including number of binges and purges and a ratio of binge to normal eating (Marrazzi et al.1995). Further, naltrexone (100 mg, 8 wk) was also found to reduce binge duration in bulimic patients and obese binge eaters (Alger et al.1991). However, other studies have reported no effects of naltrexone (up to 8 wk treatment) on long-term food intake in obese (Mitchell et al.1987) and bulimic (Mitchell et al.1989) patients, although the latter study examined effects in patients of normal weight and the findings were on energy intake and not binge behaviour.
From binge eating to long-term weight loss: effects of MOR antagonists
Evidence linking MORs to control of eating in obesity and the reduction of body weight with opioid antagonists has largely come from findings in animals. Animals susceptible to diet-induced obesity have increased expression of MORs in the hypothalamus (Barnes et al.2006), while MOR KO mice have been shown to be resistant to diet-induced obesity (Tabarin et al.2005). These findings are supported by studies showing that naltrexone reduces long-term body-weight gain in rats (Glass et al.2002; Marks-Kaufman et al.1984; Shaw, 1993). However, other studies have shown that short-acting opioid antagonists including naloxone and naltrexone only transiently decrease body weight (Cole et al.1995), have no effect (Pfeiffer et al.1984) or even increased food intake and body weight in animals (i.e. nalmefene in mice) (Chen et al.2004). More recently, in a conditioned place preference learning model of ‘food craving’, rats genetically prone or resistant to diet-induced obesity were also resistant to the effects of naltrexone on food intake and body weight (Jarosz et al.2007).
In humans, opioid antagonists have demonstrated efficacy in suppressing short-term appetite following acute treatment as discussed previously (Yeomans & Gray, 2002). A number of clinical trials have also been conducted to examine the effects of chronic oral administration of naltrexone and naloxone on body weight, but the findings have been mostly negative. The drug of choice for most of these studies has been naltrexone, with treatment durations between 4 wk and 52 wk (Atkinson et al.1985; Fruzzetti et al.2002, Guido et al.1998; Maggio et al.1985; Malcolm et al.1985; Mitchell et al.1987; Novi et al.1990; Raingeard et al.2004). Most of these studies have been negative and the ones that have reported weight loss have reported modest effects, mainly in females (Atkinson et al.1985; Fruzzetti et al.2002; Novi et al.1990; Raingeard et al.2004). It is also important to highlight a number of methodological factors in the above studies that are critical for the interpretation of the findings. First, two of the latter studies that found positive effects were not placebo-controlled (Fruzzetti et al.2002; Raingeard et al.2004) and were conducted in women with polycystic ovary syndrome and type 1 diabetes, respectively. Second, all of the studies had relatively small subjects numbers (between 8 and 20 subjects per group) and were of short treatment durations (mostly 8–16 wk). Finally, none of these studies examined the effects of naltrexone in a stratified group of obese patients such as obese binge eaters. Given that obesity is a complex disorder with metabolic and central mechanisms contributing to its aetiology, a centrally acting compound such as a MOR antagonist is unlikely to be successful in long-term weight reduction in all patients with obesity. An alternative strategy would be targeting behaviours that are associated with weight gain (such as binge eating) and examining the efficacy in a stratified group of obese binge eaters. To our knowledge, these types of studies in stratified groups of obese patients have not been conducted with MOR antagonists following short- or long-term administration. Interestingly, a recent study has reported greater frequency of the ‘gain of function’ G-allele of the A118G single nucleotide polymorphism (SNP) of the MOR (OPRM1) in obese patients with BED. This group of patients with BED also reported greater scores on a self-report measure of hedonic eating (Davis et al. 2009). These findings support our hypothesis of a more targeted therapeutic approach in obese patients that are prone to binge eating, a behaviour that is likely to be driven by hyperactivity to the hedonic properties of palatable foods through a central µ-opioid mediated mechanism involving the reward circuitry.
Genetics and therapeutic response to MOR antagonists
There is some evidence in the literature suggesting that individual differences both in bingeing behaviour and response to opioid antagonism may in part be influenced by genetic factors. The frequently occurring functional A118G SNP of OPRM1 (which results in an amino-acid change from asparagine to aspartic acid at position 40 of the N terminus) (overall allelic frequency 10.5%) has been associated with 3-fold higher affinity for β-endorphin and concurrent second-messenger activation (Bond et al.1998). An association between the 118G allele and heroin (Bart et al.2004), alcohol (Town et al.1999) and opioid addiction (Kapur et al.2007) and therapeutic response to naloxone and naltrexone (Ray et al.2007; Wand et al.2002) have been reported (Kreek & LaForge, 2007). The relationships between the 118G allele, food intake and responses to µ-opioid antagonists are yet to be determined.
Future strategies for drug development
The studies conducted to date have largely used non-selective opioid receptor antagonists. For example, naltrexone and naloxone display nm affinity for κ- and δ-opioid receptors and such a mechanism may offset therapeutic effects mediated via MOR antagonism. Hence there is a need to develop more selective MOR antagonists to maximize MOR-mediated therapeutic effects. An alternative strategy is the development of MOR-selective inverse agonists. Antagonists such as naltrexone are neutral antagonists when measured in vitro using assays measuring constitutive activity (i.e. 35S-labelled GTPγS assay). An inverse agonist would be expected to have negative intrinsic activity by reducing constitutive activity. In comparison to neutral antagonists, inverse agonists may show greater efficacy particularly at lower receptor occupancies. The advantage of therapeutic efficacy at lower receptor occupancies is the potential for a better safety and tolerability profile. Finally, there is increasing evidence for interactions between opioid receptors, including µ- and δ-opioid receptors and formation of heterodimers (Gomes et al.2000; Jordan & Devi, 1999). Such interactions have been shown to exert synergistic effects leading to increased intrinsic efficacy. For example, it has been shown that a δ-receptor antagonist enhances the potency and efficacy of MOR-mediated signalling (Gomes et al.2000, 2004). These findings suggest that future drug screenings could be designed to identify ligands that target a specific set of heterodimers (within the opioid receptor family or between opioid receptors and other related GPCRs), which is likely to increase specificity as well as facilitate more direct identification of drugs that potentiate a common pharmacological mechanism (Gupta et al.2006).
Statement of Interest
Pradeep J. Nathan and Edward T. Bullmore have fractional appointments with GlaxoSmithKline. Pradeep Nathan is also a member of the editorial board of the International Journal of Neuropsychopharmacology, but was not involved in the review of this paper.