Does yo-yo dieting drive compulsive eating?
There may be a connection.
According to Boston University School of Medicine (BUSM) researchers the chronic cyclic pattern of overeating followed by undereating, reduces the brain’s ability to feel reward and may drive compulsive eating.
This finding suggests that future research into treatment of compulsive eating behavior should focus on rebalancing the mesolimbic dopamine system—the part of the brain responsible for feeling reward or pleasure.
An estimated 15 million people compulsively eat in the U.S. It is a common feature of obesity and eating disorders, most notably, binge eating disorder. People often overeat because it is pleasurable in the short term, but then attempt to compensate by dieting, reducing calorie intake and limiting themselves to “safe”, less palatable food. However, diets often fail, causing frequent “relapse” to overeating of foods high in fat and sugar (palatable foods).
“We are just now beginning to understand the addictive-like properties of food and how repeated overconsumption of high sugar—similar to taking drugs—may affect our brains and cause compulsive behaviors,” said corresponding author Pietro Cottone, Ph.D., associate professor of pharmacology & experimental therapeutics at BUSM and co-director of the Laboratory of Addictive Disorders.
In order to better understand compulsive and uncontrollable eating, Cottone and his team performed a series of experiments on two experimental models: one group received a high sugar chocolate-flavored diet for two days each week and a standard control diet the remaining days of the week (cycled group), while the other group, received the control diet all of the time (control group).
The group that cycled between the palatable food and the less palatable, spontaneously developed compulsive, binge eating on the sweet food and refused to eat regular food. Both groups were then injected with a psychostimulant amphetamine, a drug that releases dopamine and produces reward, and their behavior in a battery of behavioral tests was then observed.
While the control group predictably became very hyperactive after receiving amphetamine, the cycled group did not. Furthermore, in a test of the conditioning properties of amphetamine, the control group was attracted to environments where they previously received amphetamine, whereas the cycled group were not. Finally, when measuring the effects of amphetamine while directly stimulating the brain reward circuit, the control group was responsive to amphetamine, while the cycled group was not.
After investigating the biochemical and molecular properties of the mesolimbic dopamine system of both groups, the researchers determined that the cycled group had less dopamine overall, released less dopamine in response to amphetamine and had dysfunctional dopamine transporters (protein that carries dopamine back into brain cells) due to deficits in their mesolimbic dopamine system.
“We found that the cycled group display similar behavioral and neurobiological changes observed in drug addiction: specifically, a “crash” in the brain reward system,” explained Cottone. “This study adds to our understanding of the neurobiology of compulsive eating behavior. Compulsive eating may derive from the reduced ability to feel reward. These findings also provide support to the theory that compulsive eating has similarities to drug addiction.”
“Our data suggest that a chronic cyclic pattern of overeating will reduce the brain’s ability to feel reward—feeling satiated. This results in a vicious circle, where diminished reward sensitivity may in turn be driving further compulsive eating,” said lead author Catherine (Cassie) Moore, Ph.D., former graduate student in the Laboratory of Addictive Disorders at BUSM.
The researchers hope these findings spark new avenues of research into compulsive eating that will lead to more effective treatments for obesity and eating disorders.
Eating disorders (ED), defined as disturbances in eating habits characterised by insufficient or excessive food intake causing energy imbalance, are associated with high comorbidity and have serious health consequences.
Therefore, although the prevalence for ED has remained stable, the high mortality rate, the association with other psychiatric disorders, and an increased level of awareness of eating disorders between the general population and clinicians have encouraged researchers to investigate the genetic, neurochemical, and physiological substrates implicated in ED [1,2].
In a highly obesogenic environment, much attention has been given to ED characterised by compulsivity and overeating, including binge eating disorder (BED), certain forms of obesity, and the newly proposed construct of “food eating addiction” .
Throughout this review we will briefly cover current knowledge of the neurobiology of feeding behaviour, focusing on non-homeostatic circuits, and we will look over the controversy about the misconception of “food addiction”.
Finally, we will explore new evidences learned from animal models in order to get a better understanding of BED, recently integrated as a novel diagnosis into the Diagnostic and Statistical Manual of Mental Disorders (DSM-5).
Understanding the Neurobiology of Eating Behaviour. Eating beyond Metabolic Needs
Food intake is an essential behaviour for survival and highly regulated by homeostatic, hedonic and learned cues. Consequently, eating behaviour depends on a simultaneous functioning of homeostatic pathway together with a more flexible non-homeostatic one, whose functions can vary between individuals according to previous experiences and/or epigenetic variations [4,5,6,7,8].
New insights argue that the impact of the modern food environment is mainly on cortico-limbic brain systems dealing with reward, emotion and cognition. Signals from the cognitive and rewarding brain may override classic homeostatic regulation leading to development of obesity or eating disorders.
These stimuli follow pathways that include, but are not limited to, corticolimbic regions within the amygdala, hippocampus, and thalamus; mesostriatal dopamine (DA)-gated circuits within the nucleus accumbens (NAc) and the ventral tegmental area (VTA); and prefrontal cortex (PFC) regions predominantly within the orbitofrontal projections [9,10,11,12].
In this context, Berridge and collaborators described three aspects of reward: liking, wanting, and learning, that despite being tightly linked they can be dissociable in terms of their neural substrates yet.
Animal models have mainly associated opioid, cannabinoid, orexin and γ-aminobutyric acid (GABA) systems as mediators in the “liking” experience, via coordinated activity in a network of hedonic hotspots in the nucleus accumbens, ventral pallidum and brainstem. Moreover, these neurotransmitters can be also implicated in other processes of the reward regulation, as opioids enhancing the “wanting” .
On the other hand, the mesolimbic dopamine system is crucial in the “wanting” and “learning” components [13,16]. It should be pointed out that “liking” and “wanting” systems are essentially pure “go” systems.
That means, once they are activated cannot be diminished by satiety influences, they never generate a strong “stop” signal to halt intake, they only tone down the intensity of the “go” .
Interestingly, the incentive sensitization theory of addiction proposed by Robinson and Berridge, is based on a pathological incentive motivation (wanting) for drugs even after the discontinuation of drug use, that can be manifest in behaviour via either implicit (as unconscious wanting) or explicit (as conscious craving) processes, depending on circumstances.
Likewise, excessive “wanting” and “liking” for food, notably hyper-palatable food, may play a role in overeating. Moreover, as in in drug dependence, the attractive and rewarding properties of hyper-palatable foods do not remain confined to the reward itself. Reward-related cues, in this case food cues, can be attributed with excessive incentive salience and become signals that draw attention and trigger overconsumption .
In other words, the wanting of addiction is connected less with pleasure (liking foods), and more with the negative reinforcement created by their withdrawal. Consequently, wanting and craving high sugar and high sugar/high fat foods are more about trying to prevent the return of negative feelings [22,23].
The Role of Opioid System, More Than “Liking” Regulation
‘Liking’ and ‘disliking’ of a food was determined by carefully observing the orofacial expressions of rats drinking caloric test solutions.
The original idea to observe rats’ facial expressions to measure how much pleasure (or aversion) they are getting from a given food was inspired by earlier human studies and later adapted for rodents [24,25,26].
As we already have commented above, endogenous opioid system is crucial in “liking” aspect of the reward process.
Despite of opioids are involved in a broadly distributed neural network, affecting both homeostatic and hedonic mechanisms; the dominant view is that opioids, especially the mu-opioid system, regulate the “hedonics of feeding” by their modulation of the palatability of food regardless of the caloric value presented. Opioids stimulated ingestion of “attractive” diets in sated rats, enhanced the “dessert effect” when a palatable food was offered at the end of a regular meal .
All these conclusions are largely based on evidence obtained from animal models. Some studies tried to elucidate if opioids stimulate intake of specific macronutrients or of preferred foods.
Several works showed that when rats could select the macro composition of the diets, after an injection of morphine, a µ-opioid receptor agonist, animals had higher preference for fat intake, instead of carbohydrate intake , in contrast, the opioid antagonist, naloxone preferentially decreased fat intake .
According to these data, opioids regulate the intake of specific macronutrients. However, when the baseline dietary preferences of the rats were considered, after morphine injections, it was observed that opioidergic modulation of feeding may be driven more by individual preference than by macronutrient, since morphine primarily stimulated carbohydrate intake in the carbohydrate-preferers, and stimulated fat intake in the fat-preferers .
Glass et al. reported a complementary result with naloxone injections: the intake of the preferred diet was reduced by naloxone at lower doses than those required to reduce intake of the less-preferred diet . Lately, it was observed that the opioid effects of preferred versus non-preferred food were dependent on the site of injection. Therefore, while the naltrexone (NTX, preferential μ-opioid receptor antagonist) injections in the central nucleus of the amygdala caused a decrease in intake of the preferred food, injections in the paraventricular nucleus of hypothalamus caused a decrease in the intake of both foods .
The role of µ-opioid stimulation is not only region-dependent. While µ-opioid stimulation via microinjection of DAMGO (µ-agonist) within the rostrodorsal quadrant of NAc medial shell can double the hedonic impact of sweet tastes, stimulation in other sub regions of medial shell does not increase “liking” reactions to sweet food [35,36]. Finally, it should be noted that hedonic enhancement is also receptor (mu, delta, kappa) dependent ; although μ- and κ-opioid receptor stimulation increased the ‘liking’, only the μ-opioid receptor stimulation increased the incentive motivation for food . Other researchers also provided some evidence of the role of opioid system either macronutrient or preference-specific effects on food intake.
These studies showed that nucleus accumbens opioidergic effects were influenced by the relative preference for specific foods, foods high in fat or sugar. When both high-fat and high-sugar foods were available simultaneously, opioid stimulation increased intake more for the high-fat food [39,40,41].
However, Mena and collaborators showed that intra-PFC μ-receptor stimulation augments the reward valuation of carbohydrate-enriched foods, along with several behavioural changes such as a high-arousal and stress-like state. The authors suggest that this carbohydrate hyperphagia could represent an attempt to suppress a stress-like aversive state, since PFC is significantly activated by stress . One mechanism that could explained the opioid-mediated overconsumption of palatable foods is through delaying the satiety systems; either the melanocortin or oxytocin systems . These preferences and craving for appetizing foods were also observed in several human studies [43,44,45,46].
Beyond “liking” process, opioid stimulation directly causes increased cue-triggered ‘wanting’ as well as dopamine stimulation. Therefore, opioid mechanisms can also regulate incentive motivational, to wit propensity to seek palatable foods [37,38,47].
In fact, for example, the injection of non-selective opioid receptor antagonist, nalmefene, blocked the anticipatory negative contrast in the binge-eating procedure as well as highly palatable food binge eating .
External food-related cues, learned cues, can also precipitate the desire for food, increasing the food craving independently of homeostatic needs. In this context, using Pavlovian-instrumental transfer (PIT) paradigm, that can model mechanisms responsible for producing “cue-triggered wanting” or craving , it was shown that opioid stimulation caused “wanting” as well as dopamine .
Additionally, NTX, and GSK1521498 (μ-opioid receptor antagonist) were tested on food seeking behaviour using chocolate-flavoured pellet reinforcement. Both compounds reduced food intake, but only GSK1521498 reduced the seeking responses for chocolate before ingestion, suggesting that μ-opioid system has a crucial role on incentive motivational mechanisms controlling food seeking .
Noteworthy, the same compound also reduced motivational responding in binge-eating obese people in a placebo-controlled trial, although subjective liking increased following drug treatment . It has been hypothesized that μ-opioid receptors localized on the GABAergic interneurons in the VTA may act to decrease dopamine release in the NAc to reduce food seeking and incentive motivation for food [38,52].
Additional studies have also confirmed the role of opioid systems in other brain structures in the motivational mechanisms underlying eating behaviour.
Therefore, when central amygdala (CeA) was stimulated by DAMGO infusions (µ-opioid agonist) caused elevated incentive motivation in subjects naturally attracted both by a predictive cue (sign-trackers) and by a reward contiguous goal cue (goal-trackers); also, an increase in “wanting” behaviour under PIT model [53,54].
On the other hand, likewise, μ-opioid receptors within the medial prefrontal cortex (mPFC) mediate an important function in overeating. In fact, naltrexone microinfused into the mPFC selectively reduced the consumption and the motivation to obtain highly palatable food, but not standard chow , while intra-PFC DAMGO engendered “high-drive-like” effects .
In this context, Baldo group has suggested that neuroadaptations of the opioid system in mPFC could explain partly the development of binge-like eating [42,57,58]. Such as it has been recently demonstrated, mPFC exerts top-down control over midbrain dopaminergic interactions with the striatum and an increase in the mPFC can suppress natural reward-related behaviour . Finally, it has been also reported that δ-opioid receptors in the NAc-Shell are involved in the effects of predictive learning on choice between actions. In fact, under a PIT paradigm, the treatment with δ-opioid receptor antagonist naltrindole, blocked this behaviour [60,61].
Dopamine System, the “Want” Pathway
The association between dopamine and food intake appears to date from ancient times and it has been postulated to be linked to and play a key role in human evolution. Specifically, it was proposed that that increased levels of dopamine were part of a general physiological adaptation of our ancestors due to an increased consumption of meat around two million years ago and later enhanced by further changes in macronutrient intake about 80,000 years ago .
At present there are clear evidences that dopamine pathways and dopamine receptors are involved in energy homeostasis. In fact, there are clear evidences linking all the five subtypes dopamine-receptors to energy balance and metabolic homeostasis. In this regard the FDA have approved a D2-agonist, bromocriptine, as adjunctive treatment for type 2 diabetes .
Moreover, most of all the FDA-approved antiobesity drugs, including liraglutide, appears to act largely through dopaminergic pathways . On the contrary chronic consumption of dopamine antagonists as in patients with schizophrenia leads to enhanced eating and weight gain.
In general, the relevance of dopamine in the integrated control of the homeostatic pathways involved in food intake is beyond any doubt. Detailed studies on dopamine involvement in leptin- and ghrelin-elicited changes in food intake are well established .
This dopamine effects appear to be mediated by both D1R and D2R, being the latest one the most relevant. Moreover, heteromers of GHSR1:DRD2 have been implicated in obsessive eating associated to Prader-Willi Syndrome .
While “liking” is closer to sensory processes, “wanting” is closer to decision making and motor action, by reflecting the cue-driven inclination to choose one behaviour over another to optimize reward. Dopaminergic projections from the VTA to the NAc and prefrontal cortex are the most important component of the implicit or unconscious “wanting” system .
Part of dopamine hypothesis of reward is based on the initial work conducted by Wise and collaborators, where animals subjected to DA antagonist pimozide (specially D2 receptor) showed a decrease in self-stimulation in ways that implied a devaluation of reward, a decrease in the pleasure of the reinforcer [72,73].
Through the years many other groups have arrived at same conclusions, DA is required for normal motivation and reward, and has a crucial role in feeding behaviour; in fact, animals lacking dopamine throughout the brain and body do not eat [74,75], although as it has been confirmed its role is brain-region dependent.
The NAc is a brain region in the ventral striatum that appears to play a crucial role in behaviours related to natural reinforcers and incentive as well as initiating key intracellular plasticity mechanisms required for learning about food resources [76,77].
Moreover, dopamine dynamics differ substantially between the NAc core and NAc shell in relation to distinct aspects of appetitive and aversive motivational states . Pharmacological blockade of D1 and D2 dopamine receptors in the NAc affects motor conduct and has small effects on feeding patterns, but does not reduce the amount of food consumed.
These effects can be interpreted as reflecting a more selective role for dopamine transmission in the anticipatory/approach phase versus the consummatory phase of feeding . According to this idea, Salamone and colleagues carried out several interesting studies examining in deep the behavioural effects of moderate NAc dopamine depletions. They found that dopamine depletion reduced the motor effort to obtain food reward, but approach or intake did not decrease when food was clearly available [80,81,82], and still animals can have hedonic responses for food in the absence of dopamine .
Therefore, dopamine system would be responsive to reward predictors and seemingly unresponsive to the reward “itself” [73,84]. Nevertheless, restoring dopamine signalling selectively to the dorsal striatum, composed of the caudate and putamen, is sufficient to allow feeding, locomotion, and reward-based learning . Besides, an increase of D2 receptors in the striatum are correlated with an optimal goal-directed behaviours and motivation [86,87].
In this context, the downregulation of striatal dopamine D2 function has been proposed to explain the reward deficiency or reward hyposensitivity theory .
According to this, a reduced D2R expression in the striatum, observed both in human and animal models, is a neuroadaptive response in order to compensate the overconsumption of palatable foods [89,90,91,92]. Furthermore, this reduced sensitivity potentially could predict a cause of excessive eating and/or obesity [89,90,92,93].
Consistent with the reward deficiency theory, some studies reported obese versus lean adults show lower striatal DA D2-like receptor availability [94,95], moreover obese adults have less capacity of nigrostriatal neurons to synthesize DA  and less striatal responsivity to tastes of high-fat/sugar beverages . Also, patients with BED tend to have reduced level of DA in the brain .
Contrarily, other groups have shown higher striatal DA in obese individuals [77,99]. Despite some discrepancies regarding to striatal dopaminergic levels, the A1 allele of the D2/ANKK1 Taq1 polymorphism has been correlated with reduced D2R availability in the striatum, obesity and compulsive behaviour [77,100]. Likewise, D2 receptors were reported to provide a target for ameliorating binge eating behaviour in a rat model after NAc deep brain stimulation [101,102]. Therefore, while DA receptor 2 antagonism in NAc increases binge-like feeding , activation of serotonin 2C receptors in DA neurons inhibits this binge behaviour in mice .
Changes in learning pathways might change rewarded responses. Learning processes are highly influenced by emotional and motivational components and required for reward prediction, for making anticipatory responses, for guidance by cues, and for goal-directed action.
These cues from the environment, such as the sight and smell of food, or even advertisements for food, are learned and become associated with future reward [11,14]. A range of animal studies has demonstrated that food-associated cues can promote eating in the absence of metabolic requirements [106,107]. Thus, food-predictive cues can stimulate eating in adults and children, even when they are full [6,108]. The basis of cue-potentiated feeding (CPF) behaviour is Pavlovian conditioning.
A feature of CPF is that it tends to be specific for the cued food and does not increase intake generally, which has similarities with the “cravings” experienced by binge eaters . Recent studies have shown that dopamine has a selective role in stimulus–reward learning that is specifically associated with the attribution of incentive salience to reward cues. This fact could explain as individuals who attribute reward cues with incentive salience find it more difficult to resist such cues, a feature associated with reduced impulse control .
The mPFC receives information about cues in the environment via the sensory cortices, but also about the internal motivation to eat via dopamine neurones of the VTA [56,58]. A group of mPFC neurons, which contain the dopamine D1R receptors, are activated during hunger-induced food intake, and their stimulation and inhibition, increases and reduces feeding respectively.
The main target of the D1R-containing subset of mPFC neurones is the medial BLA . On the other hand, it has been hypothesised that the transition from voluntary drug use to more habitual and compulsive drug use represents a transition at the neural level from PFC to striatum; and a progression in the striatum from ventral to more dorsal domains, involving its dopaminergic innervation . Moreover, reduced dopaminergic modulation has been suggested to impair inhibitory control over food intake and to increase risk of overeating in humans .
Are There Other Neurotransmitters or Hormones That Can Modify “Liking” and/or “Wanting” Behaviours?
In addition to the endogenous dopamine and opioid systems, various hormonal and neuropeptide systems influence performance in one or more of the food motivation behavioural paradigms described earlier. Signals such as leptin, insulin, ghrelin, glucagon-like peptide-1 (GLP-1) and melanin concentrating-hormone (MCH), orexins, oxytocin, serotonin between others, are involved in hunger and satiety signalling as well as reward-related neurocircuitry.
Can We Talk about “Food Addiction”?
The concept of food addiction has been suggested for the first time by Randolph in 1956 , but only recently it has received due attention, mainly because of its correlation to the increasing rate of obesity. “Addiction is defined as a chronic, relapsing brain disease that is characterised by compulsive drug seeking and use, regardless of unhealthy consequences” .
This chronic relapsing disorder is comprised of three steps: preoccupation/anticipation (craving), binge/intoxication, and withdrawal/negative effect. These three stages interact with each other, becoming more intense, and eventually leading to the pathological state known as addiction. Not all drugs produce the same pattern of addiction, but lately the progression of this behaviour triggers alterations in normal brain function and consequently induces neuroplasticity in all the structures implicated [22,23,126,127].
Remarkably, addiction induces neuronal changes in prefrontal cortical and basal ganglia activities, leading to reductions in control and decision-making skills, and causes a chronic perturbation in brain reward homeostasis mainly in the mesolimbic dopamine system. Moreover, the opioid, GABAergic and glutamatergic neurocircuitries play a key role in the development of addiction [92,128,129].
In the context of environments saturated with food, where clearly, obesity has become a worldwide problem in a short time, and the binge eating disease is the eating disorder with more incidence [130,131] several questions have been put forward. Can overeating become a pathologic attachment to food?
If so, can clinicians and researchers assert that food addiction is a new category of psychiatric disorder or brain disease? Compulsive sexual behaviour, pathologic gambling, and hedonic overeating are important problems, but are they addictions?
Although in some cases excessive consumption of food can fit with all DSM-required criteria according to some, food addiction was not included in newest edition of the of the DSM manual . Although it can share some symptoms with BED and obesity, and includes behavioural patterns similarly to substance use disorders . One fundamental distinction between currently accepted addictive substances and food is the fact that food is necessary for survival.
Notably, there is a still an ongoing debate between the scientific community about whether food addiction is a misnomer and the phenomenon could be more accurately categorised by an alternative designation [92,129,133,134,135,136,137,138,139,140,141].
Therefore, while some researchers see overeating as substance use disorders, where people are addicted to sugar, salt, additives and high fat content [140,142,143]; other suggest increased food intake related to obesity or eating disorders should be considered as a behavioural addiction [136,144,145]. Different views on this topic are almost unavoidable if we accept that it is quite unlikely that any animal model of food addiction can mimicked to a large extent food/eating addiction in humans.
However, some researchers have tried to study the food addiction in animals by following the three-criteria model proposed by Deroche-Gamonet et al. . Tolerance, reduction in the effect of a drug resulting from of repeated exposure to the substance, was observed after extended access to a palatable diet  and also some data suggest that there is a cross-tolerance between sweet solutions and opioids . In this context, Woods hypothesizes humans must learn to tolerate the intake of food in order to minimize its impact on the body, as they learn responses to help them tolerate the administration of dangerous drugs .
Regarding to the withdrawal component of addiction, negative effects after the abrupt discontinuation or decrease in intake of drugs, conflicting results have been reported depending on the kind of food. This is discussed below in Section 4.1.3. Many other aspects related to “food addition” concept have been studied in animal models too.
By using a so-called time-out model Ghizta and colleagues have attempted to elucidate the difficulty to limit intake or food seeking; after prolonged training, animals exposed to palatable diet increase their food seeking responses . Moreover, some studies have demonstrated that animals continue to seek the food despite adverse consequences [89,151].
For the first time, the DSM-5 grouped a disorder not involving substance use (gambling disorder) together with substance use disorders in a new category entitled: “Substance Related and Addictive Disorders” .
In this context, our group agreed to other researchers, and based on the current studies, consider that there is no enough evidence to conclude that a specific food, food ingredient or food additive can be addictive. Although is too early to draw definitive conclusions regarding the “food addiction” concept and further work is necessary, we consider that we should talk about “eating addiction” or more precise “addictive eating behaviour” [136,152].
In agreement with other authors , we believe that some animal models from drug addiction research, as the three-criteria model , should be applied in eating disorder field. These models may provide us new neuronal mechanisms in order to elucidate differences between “food addiction” and other compulsive eating behaviours.
More information: Catherine F. Moore et al, Reward sensitivity deficits in a rat model of compulsive eating behavior, Neuropsychopharmacology (2019). DOI: 10.1038/s41386-019-0550-1
Journal information: Neuropsychopharmacology
Provided by Boston University School of Medicine